Gigabyte GeForce RTX 2080 Ti Gaming OC Review
UK price (as reviewed): £1,109.99 (inc. VAT)
US price (as reviewed): Currently unavailable
As manufacturers come around to the idea that RGB is pointless or at least very questionable without a neutral backdrop to act as a “blank canvas”, components are starting to become pretty uniform aesthetically speaking. The first wave of Threadripper boards were quite an obvious indication of this, and now the same is coming true for graphics cards. Take the RTX 2080 Ti Gaming OC from Gigabyte here, for instance, and compare it to the RTX 2080 Duke OC from MSI or the RTX 2080 Ti Amp Edition from Zotac. Still, skin-deep similarities do not mean much, so it’s worth seeing what features partners are bringing to the table for the RTX cards, especially as the new generation of Founders Edition cards are themselves very respectable.
This is currently Gigabyte’s most premium GPU, as it is the higher specced model of the two RTX 2080 Ti cards in its lineup. Like many RTX 2080 Ti cards, stock for this SKU is very thin on the ground, with the vast majority of stores only accepting pre-orders at the time of writing, and prices could easily vary at the same store by £100 or more within a day or two. As such, we’ve fallen back on the MSRP, which is £1,099 i.e. the same as the Founders Edition (also out of stock). This puts it £100 above the MSRP for reference boards, but most MSRPs are pretty meaningless for this GPU right now. Regardless, for £1,100 it’s really a shame not to see an all-metal shroud, with Gigabyte instead relying as usual on plastic. Now, it is solid enough, but it feels like a pretty standard graphics card next to the premium craftsmanship exhibited by the FE cards. The metal backplate, meanwhile, is certainly welcome, but it’s plain and unexciting and you don’t get a matching cover for the NVLink SLI connector either.
The cooler extends the card’s length beyond the PCB edge a little to a total of 287mm, which actually isn’t that long by today’s standards, though this overhang is not covered by the backplate. It’s a 2.5-slot design (triple-slot, effectively speaking), but the height of the card is standard. No anti-sag mechanism is supplied, though it doesn’t need it as much as some even bulkier cards do.
As you would hope, the RTX 2080 Ti Gaming OC comes with a factory overclock. Specifically, the card ships with a boost clock of 1,650MHz, a mere 15MHz more than the FE but 105MHz more than any cards that happen to have reference clocks. With software, you can activate the card’s OC Mode, which raises the boost clock to 1,665MHz, but this is not how the card runs out of the box. As usual, the 11GB of GDDR6 memory is left at the default speed of 14Gbps in both modes.
The display connectors here are in keeping with the reference design, and more than suitable for a premium GPU in 2018.
Likewise, the dual eight-pin PCIe plugs are what you get on reference and Founders Edition designs. Combined with the PCIe slot, you get 375W of peak available power, so there’s no need for anything more. The plugs are top-mounted and easy to access, and there are white LEDs that will blink in case of abnormal power supply; when everything is okay they stay off. A cable that converts two six-pin PCIe connectors to one eight-pin PCIe plug is supplied, and the cabling is all black for a consistent finish.
Also along the top edge is the singular area that gets treated to RGB lighting, the Gigabyte logo. This is supported by the RGB Fusion ecosystem, although when we tried it on our system we were only able to pick a static colour, and the “off” functionality refused to work. It’s advertised as supporting extra colour modes, so we suspect this is a software bug.
The Windforce 3X cooler uses three 11-blade 82mm fans, with the central one spinning in the opposite direction to the other two, which Gigabyte claims reduces turbulence, the idea being that where one fan edge meets another, airflow will be in the same direction rather than opposing. The fan blades have a triangular edge to “split” airflow, while grooves on the outer surface are designed to guide the air down in an even spread. The fans also support switching off under a certain temperature, so you can expect silence when the GPU is idle or under low load only.
The cooler itself relies on six 6mm copper heat pipes. Sadly, these do not have a nickel coating, but that’s more of an aesthetic bonus. They all make direct contact with the GPU and twist and turn in order to feed the three separate aluminium fin stacks. Additionally, contact plates that are bonded directly to the fins ensure that all VRM components and memory packages are directly cooled via thermal pads, which are also used to help dissipate some heat to the backplate.
The PCB is pretty much a clone of reference models that are used on the Founders Edition cards, although Gigabyte claims to have certified the chokes and capacitors on the 13+3 phase power system as ‘Ultra Durable’.
While the standard warranty with this card is three years and thus in line with the FE variant, Gigabyte allows customers to extend it to four years free of charge if they simply register the card within 30 days of purchase.
- Graphics processor Nvidia GeForce RTX 2080 Ti, 1,350MHz (1,650MHz boost) (1,365MHz/1,665MHz in OC Mode)
- Pipeline 4,352 stream processors, 544 Tensor Cores, 68 RT Cores, 272 texture units, 88 ROPs
- Memory 11GB GDDR6, 14Gbps effective
- Bandwidth 616GB/sec, 352-bit interface
- Compatibility DirectX 12, Vulcan, OpenGL 4.5
- Outputs 3 x DisplayPort 1.4a, 1 x HDMI 2.0b, 1 x USB-C VirtualLink
- Power connections 2 x eight-pin PCIe, top-mounted
- Size 287mm long, 115mm tall, 50.2mm tall (~2.5-slot)
- Warranty Three years (user-extendable to four years if registered within 30 days of purchase)
1 — Gigabyte GeForce RTX 2080 Ti Gaming OC Review2 — Test Setup3 — 3DMark and VRMark4 — Battlefield 15 — Deus Ex: Mankind Divided6 — Middle-Earth: Shadow of War7 — Total War: Warhammer II8 — Wolfenstein II: The New Colossus9 — World of Tanks Encore10 — Power and Thermals11 — Overclocking12 — Performance Analysis and Conclusion
Gigabyte GeForce RTX 2080 Gaming OC 8G Review: Turing Goes Semi-Passive
Tom’s Hardware Verdict
Gigabyte’s GeForce RTX 2080 Gaming OC 8G offers great performance at 2560×1440 on high-refresh displays, the luxury of a semi-passive fan mode, and a quiet cooling solution. Now that it costs less than Nvidia’s sturdily-built Founders Edition model, we’re willing to consider it a viable alternative for enthusiasts looking to save some money.
Runs cooler than Nvidia’s Founders Edition design
Higher GPU Boost rating than Nvidia’s Founders Edition design
Semi-passive fan mode facilitates silence and low power consumption at idle
Configurable lighting via bundled RGB Fusion app
Lowest price among third-party GeForce RTX 2080s
Today’s best Gigabyte GeForce RTX 2080 Gaming deals
164 Amazon customer reviews
No price information
No price information
For more information visit their website
Gigabyte GeForce RTX 2080 Gaming OC 8G Review
Update, 11/23/18: Due to depleted inventory of GeForce GTX 1080 Ti cards and falling prices on third-party Turing-based models, we are revisiting our impressions of Gigabyte’s GeForce RTX 2080 Gaming OC 8G and updating value comparisons throughout the review.
Just weeks after the launch of Nvidia’s GeForce RTX 2080 and 2080 Ti cards, previous-gen 1080 Ti boards based on the Pascal architecture have almost completely disappeared.
Perhaps that’s alright, though. The least-expensive GeForce RTX 2080 now sells for $750 (£585 in the UK), which is still higher than the price Nvidia told us to expect when its Turing-based line-up debuted in the U.S., but substantially less expensive in the UK. And the Gigabyte GeForce RTX 2080 Gaming OC 8G we’re reviewing today even includes a copy of Battlefield V. Clearly, Nvidia’s board partners are trying harder to drum up interest in GeForce RTX 2070 and 2080 (the 2080 Ti remains woefully overpriced).
When we originally published our review of the GeForce RTX 2080 Gaming OC 8G, it was priced around $830 (£750 in the UK). In the weeks that followed, however, it fell to $750 (£585 in the UK). Those prices are far more attractive compared to Nvidia’s own Founders Edition design. After all, Gigabyte offers a highly-capable Windforce 3X thermal solution, a semi-passive fan mode for absolute silence at idle, configurable lighting, and a four-year warranty.
Meet The GeForce RTX 2080 Gaming OC 8G
Despite the GeForce RTX 2080 Gaming OC 8G’s commanding size, it’s not that heavy of a graphics card. Nvidia’s Founders Edition weighs 2.8 pounds (1.3 kg). The Zotac Gaming GeForce RTX 2080 AMP dips in at 2.5 pounds (1.2 kg). Meanwhile, our scale claims the Gigabyte card weighs just 2.12 pounds (0.98 kg). Less heft usually means a lighter heat sink cooling the GPU. But good fans can help counter a lack of mass.
The GeForce RTX 2080 Gaming OC 8G measures 11.3 x 4.4 x 2 inches (28.7 x 11.3 x 5 cm), meaning it occupies three expansion slots worth of space on your motherboard (along with a bit of room above/behind the card due to its backplate). For some enthusiasts, this isn’t an issue. The extra width is used for a taller heat sink, which improves cooling. Gamers with smaller cases or multiple GPUs will have a harder time accommodating such a configuration though, especially when it exhausts waste heat out the card’s top and down toward the motherboard through vertically-oriented fins.
Gigabyte uses plastic gratuitously across its shroud, which houses three 82mm fans. These fans blow down through an array of aluminum fins split into three sections. The section closest to the display outputs sits up over the PCB. It doesn’t make contact with any on-board components, but rather helps dissipate thermal energy from four heat pipes touching the TU104 processor. The middle section rests on top of Nvidia’s GPU. Six pipes cross through it. Below the fins and heat pipes, a plate mounts to the PCB. Thermal pads between it and the GDDR6 modules help cool Micron’s memory. The third section is the largest, extending from the power circuitry out over the PCB’s back edge by almost a centimeter. It also has four pipes passing through, along with a shaped metal plate underneath drawing heat from the VRMs through pads.
The trio of fans, the six copper heat pipes, and the direct-touch sink combine to form what Gigabyte calls its Windforce 3X cooling system. As part of this system, the outside fans spin counter-clockwise, while the middle fan rotates clockwise. Turbulence is purportedly kept to a minimum, generating less competing airflow from adjacent fans. Then, at idle, the fans stop spinning altogether through a feature that Gigabyte calls 3D Active Fan. Enthusiasts who prefer to maintain lower idle temperatures can disable 3D Active Fan through Gigabyte’s Aorus Engine software. Frankly, the fans make so little noise at idle that we’d prefer to keep them spinning (even though the semi-passive mode is one of this card’s competitive advantages).
The Nvidia and Zotac cards we already reviewed are big and heavy, justifying base plates that help keep both cards rigid. Gigabyte’s doesn’t weigh as much, so it doesn’t run quite the same risk of flexing in a mobile LAN box. However, the GeForce RTX 2080 Gaming OC 8G’s thermal solution moves a lot more relative to its PCB when you press on one side of the fan shroud or the other.
Gigabyte does add a metal plate to the back of its Gaming OC card. Seven screws keep it pinned up against the PCB, with thick chunks of thermal pad behind the GDDR6 memory and power circuitry helping circumvent hot-spots.
Up top, Gigabyte’s GeForce RTX 2080 Gaming OC 8G hosts a single NVLink connection covered by a think piece of plastic, a company logo in the middle, and GeForce RTX branding on the other end. Downloadable RGB Fusion software allows you to control the color and effect of LED back-lighting the Gigabyte logo, similar to what we saw on Zotac’s Gaming GeForce RTX 2080 AMP. Eight- and six-pin power connectors are rotated 180 degrees to avoid conflict with the form-fitted heat sink, and special white LEDs mounted to the PCB light up to tell you if something is wrong with the auxiliary power. We only saw these illuminate at boot.
Display outputs on Gigabyte’s card match the GeForce RTX 2080 Founders Edition: you get three full-sized DisplayPort 1. 4 connectors, one HDMI 2.0 port, and VirtualLink support via USB Type-C. A fairly free-flowing grille isn’t functionally significant, unfortunately, since the cooler’s fins move air perpendicular to the bracket.
|GeForce RTX 2080 Ti FE||Gigabyte GeForce RTX 2080 Gaming OC 8G||GeForce RTX 2080 FE||GeForce GTX 1080 Ti FE|
|Architecture (GPU)||Turing (TU102)||Turing (TU104)||Turing (TU104)||Pascal (GP102)|
|Peak FP32 Compute||14.2 TFLOPS||10.7 GFLOPS||10.6 TFLOPS||11.3 TFLOPS|
|Base Clock Rate||1350 MHz||1515 MHz||1515 MHz||1480 MHz|
|GPU Boost Rate||1635 MHz||1815 MHz||1800 MHz||1582 MHz|
|Memory Capacity||11GB GDDR6||8GB GDDR6||8GB GDDR6||11GB GDDR5X|
|Memory Bandwidth||616 GB/s||448 GB/s||448 GB/s||484 GB/s|
|L2 Cache||5. 5MB||4MB||4MB||2.75MB|
|Transistor Count||18.6 billion||13.6 billion||13.6 billion||12 billion|
|Die Size||754 mm²||545 mm²||545 mm²||471 mm²|
|SLI Support||Yes (x8 NVLink, x2)||Yes (x8 NVLink)||Yes (x8 NVLink)||Yes (MIO)|
What lives under the GeForce RTX 2080 Gaming OC 8G’s hood is already well-known. We dug deep into the TU104 graphics processor and its underlying architecture in Nvidia’s Turing Architecture Explored: Inside the GeForce RTX 2080. Gigabyte takes the same graphics processor with 2,944 of its CUDA cores enabled and bumps the typical GPU Boost rating up slightly to 1,815 MHz in Gaming mode and 1,830 MHz in OC mode (versus the Founders Edition card’s 1800 MHz). Eight gigabytes of GDDR6 memory move data at 14 Gb/s, matching Nvidia’s reference design. As you might expect, then, performance comparisons between the two models fall within a single-digit percentage variance.
How We Tested Gigabyte’s GeForce RTX 2080 Gaming OC 8G
While many users will attach Gigabyte’s card to a system with the latest Intel or AMD processor, our graphics station still employs an MSI Z170 Gaming M7 motherboard with an Intel Core i7-7700K CPU at 4.2 GHz. The processor is complemented by G.Skill’s F4-3000C15Q-16GRR memory kit. Crucial’s MX200 SSD remains, joined by a 1.4TB Intel DC P3700 loaded down with games.
As far as competition goes, we can assume that GeForce RTX 2080 and all of the partner boards based on the same design are bested by GeForce RTX 2080 Ti and Titan V, both of which we have in our test pool. We also compare GeForce GTX 1080 Ti, Titan X, GeForce GTX 1080, GeForce GTX 1070 Ti, and GeForce GTX 1070 from Nvidia. AMD is represented by the Radeon RX Vega 64 and 56.
Our benchmark selection now includes Ashes of the Singularity: Escalation, Battlefield 1, Civilization VI, Destiny 2, Doom, Far Cry 5,Forza Motorsport 7, Grand Theft Auto V, Metro: Last Light Redux, Rise of the Tomb Raider, Tom Clancy’s The Division, Tom Clancy’s Ghost Recon Wildlands, The Witcher 3 and World of Warcraft: Battle for Azeroth.
The testing methodology we’re using comes from PresentMon: Performance In DirectX, OpenGL, And Vulkan. In short, these games are evaluated using a combination of OCAT and our own in-house GUI for PresentMon, with logging via AIDA64.
All of the numbers you see in today’s piece are fresh, using updated drivers. For Nvidia, we’re using build 411.51 for GeForce RTX 2080 Ti and 2080. Zotac’s Gaming GeForce RTX 2080 AMP is tested on 411.70, while Gigabyte’s GeForce RTX 2080 Gaming OC 8G employs 416.34. The other cards were tested with build 398.82. Titan V’s results were spot-checked with 411.51 to ensure performance didn’t change. AMD’s cards utilize Crimson Adrenalin Edition 18.8.1, which was the latest at test time.
Interestingly, there is a bug in The Witcher 3 that was introduced a couple of builds ago. It causes flickering through our benchmark scene, where the background appears to go white and come back. This issue doesn’t seem to affect performance, but it’s certainly distracting. Nvidia released a hotfix driver on October 28 to address it.
MORE: Best Graphics Cards
MORE: Desktop GPU Performance Hierarchy Table
MORE: All Graphics Content
Gigabyte GeForce RTX 2080 Gaming OC 8G Review
Next Page Performance Results: 2560×1440
Chris Angelini is an Editor Emeritus at Tom’s Hardware US. He edits hardware reviews and covers high-profile CPU and GPU launches.
Gigabyte GeForce RTX 2080 Ti WindForce OC review
Gigabyte GeForce RTX 2080 Ti WindForce OC
Gigabyte GeForce RTX 2080 Ti WindForce OC
Why is Gigabyte GeForce RTX 2080 Ti WindForce OC better than the average?
- GPU clock speed?
- Floating-point performance?
14.1 TFLOPSvs9.47 TFLOPS
- Pixel rate?
142.6 GPixel/svs85.22 GPixel/s
- GPU memory speed?
- Effective memory speed?
- Texture rate?
440.6 GTexels/svs223.62 GTexels/s
- Maximum memory bandwidth?
Which are the most popular comparisons?
Gigabyte GeForce RTX 2080 Ti WindForce OC
MSI GeForce RTX 2080 Ti Gaming X Trio
Gigabyte GeForce RTX 2080 Ti WindForce OC
Gigabyte GeForce RTX 2080 Ti Gaming OC
Gigabyte GeForce RTX 2080 Ti WindForce OC
Asus TUF GeForce RTX 3090 Ti Gaming OC
Gigabyte GeForce RTX 2080 Ti WindForce OC
Nvidia GeForce RTX 2060
Gigabyte GeForce RTX 2080 Ti WindForce OC
Nvidia GeForce RTX 3070
Gigabyte GeForce RTX 2080 Ti WindForce OC
Gigabyte GeForce RTX 3060 Ti Eagle
Gigabyte GeForce RTX 2080 Ti WindForce OC
Asus ROG Strix GeForce RTX 3070 Gaming
Gigabyte GeForce RTX 2080 Ti WindForce OC
EVGA GeForce RTX 3070 XC3 Ultra Gaming
Gigabyte GeForce RTX 2080 Ti WindForce OC
EVGA GeForce RTX 2080 Ti Gaming
Gigabyte GeForce RTX 2080 Ti WindForce OC
Gigabyte Aorus GeForce RTX 3090 Ti Xtreme Waterforce
1. GPU clock speed
The graphics processing unit (GPU) has a higher clock speed.
When the GPU is running below its limitations, it can boost to a higher clock speed in order to give increased performance.
The number of pixels that can be rendered to the screen every second.
Floating-point performance is a measurement of the raw processing power of the GPU.
The number of textured pixels that can be rendered to the screen every second.
6.GPU memory speed
The memory clock speed is one aspect that determines the memory bandwidth.
Shading units (or stream processors) are small processors within the graphics card that are responsible for processing different aspects of the image.
8.texture mapping units (TMUs)
TMUs take textures and map them to the geometry of a 3D scene. More TMUs will typically mean that texture information is processed faster.
9.render output units (ROPs)
The ROPs are responsible for some of the final steps of the rendering process, writing the final pixel data to memory and carrying out other tasks such as anti-aliasing to improve the look of graphics.
1.effective memory speed
The effective memory clock speed is calculated from the size and data rate of the memory. Higher clock speeds can give increased performance in games and other apps.
2.maximum memory bandwidth
This is the maximum rate that data can be read from or stored into memory.
VRAM (video RAM) is the dedicated memory of a graphics card. More VRAM generally allows you to run games at higher settings, especially for things like texture resolution.
4.memory bus width
A wider bus width means that it can carry more data per cycle. It is an important factor of memory performance, and therefore the general performance of the graphics card.
5.version of GDDR memory
Newer versions of GDDR memory offer improvements such as higher transfer rates that give increased performance.
6.Supports ECC memory
✖Gigabyte GeForce RTX 2080 Ti WindForce OC
Error-correcting code memory can detect and correct data corruption. It is used when is it essential to avoid corruption, such as scientific computing or when running a server.
DirectX is used in games, with newer versions supporting better graphics.
2. OpenGL version
OpenGL is used in games, with newer versions supporting better graphics.
Some apps use OpenCL to apply the power of the graphics processing unit (GPU) for non-graphical computing. Newer versions introduce more functionality and better performance.
4.Supports multi-display technology
✔Gigabyte GeForce RTX 2080 Ti WindForce OC
The graphics card supports multi-display technology. This allows you to configure multiple monitors in order to create a more immersive gaming experience, such as having a wider field of view.
5.load GPU temperature
Unknown. Help us by suggesting a value.
A lower load temperature means that the card produces less heat and its cooling system performs better.
6.supports ray tracing
✖Gigabyte GeForce RTX 2080 Ti WindForce OC
Ray tracing is an advanced light rendering technique that provides more realistic lighting, shadows, and reflections in games.
✔Gigabyte GeForce RTX 2080 Ti WindForce OC
Allows you to view in 3D (if you have a 3D display and glasses).
✔Gigabyte GeForce RTX 2080 Ti WindForce OC
DLSS (Deep Learning Super Sampling) is an upscaling technology powered by AI. It allows the graphics card to render games at a lower resolution and upscale them to a higher resolution with near-native visual quality and increased performance. DLSS is only available on select games.
9.PassMark (G3D) result
Unknown. Help us by suggesting a value.
This benchmark measures the graphics performance of a video card. Source: PassMark.
1.has an HDMI output
✔Gigabyte GeForce RTX 2080 Ti WindForce OC
Devices with a HDMI or mini HDMI port can transfer high definition video and audio to a display.
2. HDMI ports
More HDMI ports mean that you can simultaneously connect numerous devices, such as video game consoles and set-top boxes.
Newer versions of HDMI support higher bandwidth, which allows for higher resolutions and frame rates.
Allows you to connect to a display using DisplayPort.
Allows you to connect to a display using DVI.
6.mini DisplayPort outputs
Allows you to connect to a display using mini-DisplayPort.
1.Has USB Type-C
✔Gigabyte GeForce RTX 2080 Ti WindForce OC
The USB Type-C features reversible plug orientation and cable direction.
With more USB ports, you are able to connect more devices.
Which are the best graphics cards?
Gigabyte RTX 2080 and RTX 2080 Ti Review
NVIDIA just launched their GeForce RTX 20 series and have introduced their RTX 2080 and RTX 2080 Ti. It comes at a specific time when the silicon-fabrication technology is not progressing similarly to how it was about four years ago, breaking the architecture roadmaps of some of the giant names like NVIDIA, which is urging them to create innovative and fresh architectures on current nodes.
The brute transistor count boosts are not considered as available options today, so NVIDIA needed to have a really notable feature to be able to sell fresh new GPUs. This feature we speak of is the RTX Technology, which is so huge for the brand that it was able to alter the classification of the client segment graphics cards. This is also with the launching of GeForce RTX 20 Series.
While the lineup of Gigabyte’s RTX 20 series graphics card is coming in fast, the series from AORUS has seen some short delay; yet today, we’re going to check out two extreme products namely the RTX 2080 and the RTX 2080 Ti. On offer, these feature a factory overclock to the memory and GPU and triple fan coolers.
With the inclusion of the Ray Tracing and DLSS, the contemporary Turing architecture, plus a power-filled boot, the RTX 2080 and RTX 2080 Ti are surely exciting. Today, we’re going to see what these cards have to offer us, and if these are products that you’d need for your PC build.
Gigabyte RTX 2080 and RTX 2080 Ti Review – What’s in the Box?
Gigabyte RTX 2080
The RTX 2080 comes in a glossy black packaging with a mechanical eye of a robot printed in front. At the corner, you will find a summary of the product’s key features. One side of the box has a few details about its extended warranty when you register to AORUS Care online.
The back of the box displays more details and photos about the highlights of the RTX 2080 such as its triple fan setup with the middle fan rotating in the opposite direction. According to Gigabyte, this helps unify the direction of the wind between the fans since it reduces turbulence. This results in wider coverage when cooling. Other highlights are the Super Overclocking via one click and its compatibility with AORUS engine. The side of the GPU shroud also has the RGB for the logo o GIGABYTE.
When you open the box, you will see the RTX 2080 itself together with a utility / driver and a quick-start guide. The larger black leaflet tells about the additional one year to the product’s 3-year warranty. We think that this is a nice touch that the company made.
Gigabyte RTX 2080 Ti
Let’s begin with the package of the RTX 2080 Ti. It comes in a rather large cardboard box; and if you think about it, it’s actually bigger than the rest of the cards under the RTX 20 series.
The front portion of the box features a huge GeForce RTX logo, together with the AORUS logo right at the top left area of the package. You’ll also read the AORUS X-treme series logo right at the center.
The company put a lot of emphasis on their packaging, especially on the RTX side where the first feature listed by AIBs was the Ray Tracing. The next would be the GDD-R6, the Direct-X 12, then the Ansel support. NVIDIA took the support of Ray Tracing and bet the future of their GPUs with them. Of course, they highlighted other features on the package like the RGB fusion support, 8 GB GDD-R6, Windforce cooling, plus the 4-year warranty.
The back of the package features your typical information and highlights; these are the primary features as well as the card specifications. The main features of the top-notch custom cards from AORUS would be its blazing and excellent performance which is accomplished by having a fully custom design, the lighting systems featuring the Stack x3 fans, the Windforce cooling, and the latest angular fin heatsink. The latter offers enhanced cooling performance unlike the classic flat-surfaced fin heatsinks that most people know of.
They also focused on the site, GeForce.com, on the AIB cards where you can easily download all the newest drivers, as well as GeForce Experience applications. These are vital for every gamer to be able to access all the feature sets of their brand-new cards.
When you check out the sides of the package, you’ll be again, greeted by the large branding from GeForce RTX. There’s also the mention of their 8 GB GDD-R6 memory which is available on these cards.
The increased bandwidth dispatched through the latest GDD-R6 interface would greatly boost the efficiency in gaming titles at much higher resolutions over the GDD-R5X and GDD-R5 based graphics cards.
Now that we’re done with the outer portion of the package, it’s time for us to check the good stuff and see what’s inside the box of the RTX 2080 Ti. Within the box is another box that comes with a dotted texture plus a reflective AORUS logo at its center. AORUS is obviously aiming for the more premium and sophisticated look with this specific product, and we’re not complaining. It really does look nice and appealing.
Here, the accessory package and graphics card are both held in place by a sturdy foam container. For the graphics card, it comes with a couple of manuals and accessories that might not really be useful if you’re a hardcore enthusiast; but of course, it is something that will prove to be useful for the regular gaming audiences. Following is a list of accessories included in the box: the AORUS VGA Holder, AORUS Metal Sticker, a quick guide, an I / O guide, the 4-year warranty registration, and a driver CD.
Gigabyte RTX 2080 and RTX 2080 Ti Review
The appearance of the RTX 2080 is obviously inspired by the G1 Gaming GTX 1080 which was released previously. It is made of plastic with a matte black finish while some of its sections are grey. What makes this different from other Gigabyte cards is that it doesn’t have orange accents. The grey and black theme is great since no one will have any problems when creating a build with matching colors.
For the triple fan setup, it is part of the cooling solution called Windforce 3X. For its measurements, each of the fans have a diameter of 82 millimeters. As mentioned earlier, the fan at the center rotates at the opposing direction of the two other fans.
Now when it comes to removing the heatsink from the PCB, all you have to do is take the seven screws from the back and you’re good to go. Upon seeing the PCB, you will see that the design of NVIDIA has been used. It retains the similar 8 + 2 power phase layout of the Founders card. The eight GDDR6 chips are produced by Micron.
Now for the cooler, it is completely different from that of the Founders card. The cooler of the RTX 2080 makes use of six heatpipes made of copper. Each of these have a diameter of 6 millimeters and these have contact with the core of the GPU. For the large fin array that is made of aluminum, this makes up the bulk of the cooler. A cold plate with thermal pads is also present for the MOSFETs and the VRAM chips.
When looking at the front portion of the card, you will find a little Gigabyte logo that RGB LEDs illuminate. You can configure this using the RGB Fusion software. At the back is a full-length backplate made of metal that looks and feels like anodized aluminum. It also has Gigabyte’s logo printed in white. The logo is not an RGB zone and this portion is kept simple.
The RTX 2080 from Gigabyte is fitted with the Turing TU-104-based GPU; it is also a product series that is created and released to take the place of the GeForce GTX 1080. Its graphics card is fully customized and fitted with Gigabyte’s newest model – the Windforce 3x cooler.
Once again, the cooler is direct heatpipe touch-based – this means that the heatpipes are really fixed on top of the GPU. The Windforce 2x is installed with a top side RGB LED logo that can be animated and color-controlled.
RTX 2080 Ti
The thickness of this card is quite evenly-split between the fan shroud and its heatsink. This device mostly has matte black plastic which is accented with nice silver swoops together with gleaming black edges.
Beneath its adorning cover, a total of three 100-millimeter fans in a stacked setting overlaps each other across the length of the cooler. By spinning and coinciding in the opposite course, Gigabyte notes that the Windforce Stack 3x system yields costless airflow from each of the fans instead of having ineffective turbulence.
Interestingly, based on the company’s own experiments and diagram, the right and middle fans blow down on the motherboard, while the middle and left fans push the air out of the card’s top.
Its heatsink is divided into two similar sized portions which are shaped uniquely to welcome the surface mounted elements just underneath them. Right at its core, the fins are cut to have the same height to maximize the surface area; however, about ⅔ of the outer portions of both sinks utilize an angular-fin style with every other fin present, sitting lower. Gigabyte claims that this aids in channeling air via fins while simultaneously decreasing noise.
The heatsink that is nearest the display outputs is set at the top of the TU-102 processor of NVIDIA. A total of 7 heatpipes make direct contact with the GPU where two of these bends back into the heatsink’s bottom and top edges to guarantee evenly dissipated thermal energy.
Also, there are 5 pipes that go through the other sink, running throughout its far edge. A spacer located beneath the GPU sink squeezes the thermal pads between itself and the GDD-R6 memory modules that surround the TU-102. Additionally, both sides o link to the voltage-regulation circuitry on the board to mitigate hot spots.
Instead of establishing its thermal solution to a metallic frame that is almost similar to the RTX 2080 Ti Extreme 11G’s PCB, the backplate from Gigabyte directly bolts to the heatsink in 10 areas. A metallic frame located at the top of the sink enhances the rigidity by maintaining the thermal solution and preventing it from flexing. Despite this, it is a bit weighty to hang from a PCIe slot; so what Gigabyte did was to bundle a stand for supporting the card. It pushes up from the floor of your chassis or your bottom mounted PSU.
Above the Fan Stop sign, you will see a pair of 8-pin power connectors rotating at 180-degrees to bypass hanging up on the heatsinks beneath them.
The LEDs at the top of each connector aids in diagnosing power supply concerns. When they’re switched off, it means that all’s good; if you end up forgetting to run a cable from the PSU, these will light up. It also works when the power is intermittent; these blink to point out that there’s something wrong. At the other end of the top portion, you’ll see an NV-Link interface which is safeguarded by an orange piece made of plastic. As for the built-in cover on NVIDIA’s Founders Edition card, it sure looks better and classier.
Since the heatsink’s fins have the ability to direct air up and down, you won’t really need to have ventilation on the expansion bracket. Instead, Gigabyte fills the two-slot cover with a total of seven display outputs: 3 HDMI ports, 3 Display Port connectors, plus 1 USB Type C interface. Of course, the TU-102 only simultaneously drives four displays; so, a total of 7 connectors should be used.
At any time, either the two DP interfaces or the two HDMI ports are switched off. We really liked that Gigabyte gave gamers all that connectivity which saves everyone from having to utilize those annoying adapters. Additionally, there is a metal backplate that provides 2 heatsink pieces something solid and sturdy to screw into. This also hosts the logo of AORUS which is controllable and backlit via the RGB Fusion 2.0 software from Gigabyte. The plate is also solid, concealing the PCB from one end to another.
The Gigabyte RTX 2080 Ti is the latest and leading consumer graphics card coming from NVIDIA. This series has been integrated with the TU-102 GPU which has active shader processors which is vital when you compare this with the GTX 1080 Ti.
The device will get 11 GB GDD-R6 graphics memory; and together with this, you’ll also get the 352-bit -wide memory bus. The GPU comes with a total of six GPCs (Graphics Processing Clusters), 72 SMs (Streaming Multiprocessors), and 36 TPCs (Texture Processing Clusters). Each of the SMs has a total of 64 CUDA cores, 4 texture units, 8 Tensor Cores, plus 96 KB of L1 / shared memory. This can be set up for different capacities based on the graphics or the computer workloads.
Ray tracing acceleration is done via the new RT Core processing, the TU-102 has a total of 72 of these with 576 tensor cores plus 96 ROP units. When it comes to the clock frequencies, we’re looking at a base frequency of around 1,350 MHz that has a turbo allowance that’s going to 1,635 MHz.
The Windforce 3x – Cooling System
This cooling system features the 3x 82-millimeter distinct blade fans, 6 combined copper heat pipes, 3D active fan functions, heatpipe direct touch, and alternate spinning fans. Together, these deliver a highly effective and efficient heat dissipating capacity for better performance at decreased temperatures.
Extreme Durability + Overclocking
The graphics card utilizes the 8 + 22 power phase style to let the MOSFET operate at the lowest temperature as possible, load balancing and over temperature protection style for each of the MOSFET available, and the Ultra Durable approved capacitors & chokes to give you a longer system life and excellent performance.
Alternate Spinning Patent
The patent of Gigabyte which is “Alternate Spinning” is the one and only solution that can readily deal with the rough airflow of three fans.
Usually, the biggest issue with three fans is the turbulence. Since these rotate simultaneously in the same direction, the airflow direction becomes opposite with between the fans. This will cause turbulent and rather rough airflow, thus, decreasing the heat dissipation efficiency.
Gigabyte made the fan turn the opposite direction so the airflow direction between the other two fans are the same. This decreases the turbulence while also enhancing the airflow pressure.
NVIDIA NV-Link Support
Support for the NVIDIA NV-Link TM technology combines two similar GPU’s with a GeForce RTX NV-Link Bridge to provide the best experience in VR or 4K gaming. You can also readily modify the light effects on the bridge via the AORUS engine.
Distinct Blade Fan + the 3D Active Fans
Airflow is divided by the triangular fan’s edge while being seamlessly guided through the 3D stripe curve found on the fan’s surface. This, in turn, effectively improves the airflow. The 3D Active Fan gives you semi passive cooling so the fans will stay off when the GPU is on low power game or low load. This will allow you to fully enjoy your gameplay in total silence even when your system is running idle or light.
The Direct Touch Heat Pipes
The structure of the copper heat pipe increases the direct contact area together with the GPU, thus, improving the transfer of heat. Also, the heat pipe conceals the VRAM via large metallic plate contact to guarantee adequate cooling.
Composite Heat Pipes
The available composite heat pipes associate phase transition and thermal conductivity to thoroughly control the heat transfers occurring between the two solid interfaces that boost cooling capacity.
Using the Product
Are you wondering if the RTX 2080 can play games using the highest possible settings at a 1080p resolution? Well yes, it sure does, and it will be able to handle pretty much any game you can throw at it.
Another great thing about the card is that it is extremely quiet. Not only does this benefit from the “3D Active Fan”. This means that the spinning of the fans stop at low-load instances. However, even if we ran our stress tests on loop, the card was still hardly audible. Honestly, the fact that we could barely hear the fans spinning already tells you a whole lot of things you’d need to know about how silent this specific card is.
And while performance is a given to the RTX 2080 graphics card from Gigabyte, the company’s continuous effort on ensuring that things stay cool and quiet is extremely commendable. These are indeed little details that greatly affect your overall gaming experience, and we really loved that the company focused on this specific feature. Further, by having an efficient and effective cooling system, the card was able to readily overclock appropriately, and even on higher speeds. Plus, it allowed all games to perform much better than how it usually would.
RTX 2080 Ti
As for the RTX 2080 Ti, it provides you with an overall excellent performance. We’re seeing a couple of caps when on the decreased resolution of gaming.
Depending on the workload and title, you will see about 25% to even 40% boosts in performance for this specific resolution. The RTX 2080 Ti really begins to flex after a resolution of 2560 x 1440, and anything that is below will be normalized because of processor and driver bottlenecks.
In all honesty, we can’t really think of a game that won’t operate well combined with the most excellent image quality settings. The more challenging you make it on the card, the better it will perform if you compare this to the last generation products. When it comes to gaming, you should do away with at least a monitor that is 24 inches at 2,560 / 1600, or higher to make it a better fit. All in all, the 11 GB graphics memory is perfect, thus, making the products extremely future proof.
A benefit of the 2080 Ti is that it features a zero RPM fan mode. During low load situations, we were really amazed that it was able to stay completely silent. Even when on full load, it was still performing quietly, hitting only 43 dBA in total.
Though this may not be the most appealing cooler around, the company clearly made the perfect choice for their brand-new GPU. So, without questions, this surely is a fast card. It also set most of everyone’s second highest scores to date in the synthetic benchmarks.
When we speak of the 4K gaming, it is absolutely one of the best around, thus considered as a beast by many. Even if it’s considered cheaper, it was able to run and perform quietly and it wasn’t also wasteful when it comes to the power. To top it off, it was able to overclock really well and effectively.
So after testing both products for some time, we can definitely say that we highly suggest both the RTX 2080 and RTX 2080 Ti to you.
AORUS Engine Utility
The company bunched the RTX 2080 and their RTX 2080 Ti with its own service for overclocking, lighting, and fan control. Its newest version of the software, which is called the AORUS Engine Utility, has a cutting edge and instinctive interface that lets you alter the clock speed, fan performance, voltage, as well as the power target based on your needs.
From the home screen, the software lets you easily toggle between the default Gaming Mode (1815 MHz), OC Mode (1830 MHz), the Silent Mode (1710 MHz), and the User Mode which can be customized based on your preferences. The buttons located on the bottom-right area corresponds to the RGB Fusion or lighting control, toggle, and hardware monitoring to switch the semi-passive fan mode either off or on.
If you click on the “Professional Mode” text located at the top right portion, the AORUS Engine will switch to another interface that is complete with the memory clock rate slider, GPU boost control, manual fan speed, GPU voltage adjustment, and connected power target / temperature target sliders.
For the single-click super overclocking feature, all you need to do is click on the AORUS engine to easily modify the card to meet a variety of requirements even without knowledge on overclocking.
The RTX 2080 and RTX 2080 Ti delivers superb gaming performance. We would recommend the RTX 2080 over the RTX 2080 Ti if you are a gamer who uses 1440p as native resolution especially if you want to limit your expenses.
On the other hand, if you are an ultra HD gamer, go for the RTX 2080 Ti. It works excellently and we think that it’s an enthusiast class graphics card. As we mentioned earlier, it’s really quiet and delivers a nice cooling performance.
Even under load, its cooling system is able to maintain proper temperature levels throughout every area. The overall results of its performance is great. Aside from ultra HD gaming, we recommend the RTX 2080 Ti if you’re willing to pay for its higher price. Either one you pick you will sure as heck be onto a winner.
Where to Buy
You can pick up the Gigabyte RTX 2080 for around $1200 AUD whilse the RTX 2080 Ti can be had for around $1999 AUD. For more head on over to the RTX 2080 and RTX 2080 Ti product pages.
AORUS RTX 2080 Ti Gaming Box Review — Cooler than Cool
Table of Contents
Gigabyte has upped the ante with their release of the AORUS RTX 2080 Ti Gaming Box, the very first liquid-cooled external graphics card solution. It showcases Gigabyte’s commitment to offering the latest and greatest in eGFX technology. At $1,500 the RTX 2080 Ti Gaming Box is more expensive than many Thunderbolt 3 ultrabooks. This raises the question: is the performance gain worth the price of admission?
Aorus RTX 2080 Ti Gaming Box + Logi G900 + 8bitdo N30pro
|PSU max power||450W|
|GPU max power||300W|
|Power delivery (PD)||100W|
|TB3 USB-C ports||1|
|Size (in/mm, LxWxH)||11. 8 x 5.5 x 6.80
300 x 140 x 173
|Max GPU len (in/cm)||12.60/32.0|
|Updated firmware||44.44 ✔|
|TB3 cable length (cm)||50|
AORUS’ new Gaming Box is no longer the diminutive-sized ITX box that made its first generation a popular eGPU choice. The larger size is a necessary evolution to incorporate an AIO closed-loop liquid cooling system and more importantly to reduce overall noise emission. Surprisingly this second generation Gaming Box remains compact for the power and components it contains. Its footprint is barely larger than a 240mm radiator. For reference, here are some photos to compare the RTX 2080 Ti Gaming Box to the RX Vega 64 LC and the RX 580 Gaming Box.
AORUS RTX 2080 Ti Gaming Box vs RX Vega 64 LC
RTX 2080 Ti Gaming Box vs RX 580 Gaming Box
A unique feature of all Gaming Boxes are their travel cases. It’s no exception with the latest AORUS RTX 2080 Ti version. The new bag retains the same color scheme of a bright orange liner and black exterior. There’s plenty of padding to keep the enclosure protected, and it makes for a truly portable solution. A small compartment contains the power cable and a Thunderbolt 3 cable. It would be a nice touch if Gigabyte had included a longer 40Gbps 100W PD Thunderbolt 3 cable rather than the .5m.
Aorus Rtx 2080 Ti Gaming Box Travel Bag
Top Panel & 240mm Radiator
Unlike the original Gigabyte/AORUS Gaming Box that was rather flimsy with thin sheet metal construction, the RTX 2080 Ti Gaming Box is much more solid with aluminum panels reenforcing an inner metal frame. Foam padding and dust filters provide premium touches. Finished in graphite, it matches many ultrabooks that are constructed out of aluminum. Another refinement is the placement of RGB. It is now in a more appropriate location underneath the front fascia. The lighting zone doesn’t strike you straight in the eyeballs like the bare LED diodes in the 1st generation Gaming Box.
Better build quality means it’s harder to service, especially for compact units such as this Gaming Box. Gigabyte used many more screws in this new enclosure. I counted over 50 before I lost track. At the core of the second generation Gaming Box is an RTX 2080 Ti PCB and its all-in-one liquid cooler. A 240mm radiator is mounted to an L-shaped metal frame. The two 120mm cooling fans have cables that are just long enough to reach the Y-adapter that eventually connects to the graphics board. If you ever need to take this eGPU apart, make sure this fan connector is securely in place when you put it back together (ask me how I know).
AORUS RTX 2080 Ti Gaming Box Right Side
AORUS RTX 2080 Ti Gaming Box Left Side
Due to the horizontal placement of the graphics card, the Thunderbolt 3 mainboard sits vertically. The two boards when connected perpendicular to each other create enough space for an fATX power supply. The PSU is a 12V single-rail unit, producing a max 450W output. One 24-pin and one 4-pin EPS power connector go to the mainboard while two 6+2-pin PCIe power connectors go to the graphics card. This power supply is longer than the unit in the first generation Gaming Box. Another addition with the new AORUS RTX 2080 Ti Gaming Box is a power switch/reset button. We had observed power-related issues with the 1st gen that required a reset, so having a dedicated button is appreciated.
450W fATX PSU Label
PSU Cables & Connectors
One major complaint with the first generation Gaming Box is the three tiny 40mm fans, two of which are in the front and run constantly. Gigabyte issued a firmware update that keeps them off during light use. In this 2nd generation box, Gigabyte switched to using one 70mm enclosure fan to replace those two 40mm units. The remaining 40mm fan is inside the power supply. Overall noise level from the enclosure fans has reduced greatly as a result of these changes.
Enclosure 70x10mm Fan
Rear Ports & PSU 40mm Fan
The RTX 2080 Ti itself is a delicate and expensive PCB, suspended in midair by the x16 physical PCIe slot and five tiny screws attached to the monitor output ports. Given how short the coolant hoses are, it’s unlikely this graphics card and its AIO components can be repurposed inside a standard PC case. ITX cases with a PCIe extension cable may be able to use it.
As expected in a premium eGFX, the AORUS RTX 2080 Ti Gaming Box has a dual Thunderbolt 3 controller setup to provide stability to expansion ports. The Thunderbolt 3 mainboard has a primary JHL6540 TB3 controller hosting the eGPU and a secondary JHL6340 TB3 controller hosting the expansion ports. Other crucial components as found on a typical Thunderbolt 3 enclosure are Texas Instrument TPS65983 USB-C controller and Winbond 25XX series EEPROM firmware chipset.
AORUS RTX 2080 Ti Gaming Box Board Components
Full Teardown Component Layout
Due to a dual Thunderbolt 3 controller setup, the RTX 2080 Ti Gaming Box has two sets of Intel Alpine Ridge and their accompanied firmware EEPROM chips. Both the primary TB3 controller, JHL6540 and the secondary TB3 controller, JHL6340 are running firmware version 44.44. Upstream Power Delivery remains at the maximum 100W to the host laptop. This allows the AORUS RTX 2080 Ti Gaming Box to charge all Thunderbolt 3 laptops, including the latest 2019 16-in MacBook Pro (96W).
AORUS RTX 2080 Ti Gaming Box macOS Thunderbolt Tree
100W Power Delivery macOS Power System Info
Firmware flashing on Thunderbolt 3 enclosures is possible through Intel FW Update Tool in Windows. However this process is not straightforward and was almost impossible with a Mac. We had been trying to to come up with a solution for a while. With the release of the Blackmagic eGPU firmware update, @mac_editor was able to reverse-engineer and develop a very effective solution to flash Thunderbolt 3 enclosure firmware in macOS. Our community is also in the process of collecting firmware that is not available from the Thunderbolt vendor websites. It is our wish that Apple and Intel would release an official tool to bake our own firmware to make the most of these eGPU enclosures.
Ch441a Eeprom Programmer
TBT-Flash – Thunderbolt Firmware macOS Tool
Ch441a EEPROM Programmer USB EPP 24cxx25xx
Ethernet port works plug-and-play in Windows at 1Gbps. The Network Interface Controller attaches to a x1 PCIe connection which is hosted by the secondary Thunderbolt 3 controller, JHL6340. This same TB3 controller also hosts 3x USB-A expansion ports via another x1 PCIe connection. As we learned with the first generation Gaming Box, using a single TB3 controller for both eGPU and expansion ports can cause lagging issues for low-latency peripherals such as mouse and keyboard. In the two months I tested this RTX 2080 Ti Gaming Box, I didn’t experience any lag.
Regarding noise and thermal emission, this 2nd generation Gaming Box is cooler and quieter than the original Gaming Box. Placement of the 70mm fan draws air primarily into the power supply. The eGPU never reached higher than mid-60s degree Celsius. Overall noise level is typical for an AIO 240mm liquid cooler. When the Thunderbolt connection becomes active and powers on the Gaming Box, you hear a rush of liquid flowing through the hoses and radiator. This settles down quickly and the next noise you hear is from the 40mm fan inside the PSU. This fan noise is a weak point in an otherwise great cooling system. There are two software utilities to monitor and control different features of the RTX 2080 Ti Gaming Box. AORUS Engine provides monitoring and overclocking of the graphics card. RBG Fusion is the other software with the sole purpose of changing RGB settings.
I was excited to test the AORUS RTX 2080 Ti Gaming Box with newly available Ice Lake processor ultrabooks. The 10th generation Intel U-CPU has an on-die Thunderbolt 3 controller that eliminates the shared connection through PCH in previous iterations. The host laptop of choice with i7-1065G7 CPU was a Razer Blade Stealth. The Mercury White model I got is the base configuration with no discrete graphics, which allows the CPU to run at 25W TDP. This makes it an ideal eGPU-host.
Another in-demand host laptop with the RTX 2080 Ti eGPU is the 2019 16-in MacBook Pro. I had tried both the i9 and i7 models but found the 6-core i7 more tolerable in terms of thermal management and overall noise during heavy use. Due to no Nvidia support since macOS Mojave 10.14, we can only use GeForce graphics cards in Windows through Boot Camp mode. If you were wondering if eGPU support for Boot Camp has gotten better, I hate to say that it’s muddy at best.
2019 Razer Blade Stealth + RTX 2080 Ti Gaming Box
Rtx 2080 Ti Gaming Box + AORUS Fi27q-P Monitor
Through Nvidia Optimus, gaming with the internal display is possible and is an overall excellent experience. Given how powerful this Gaming Box is, it really needs an external monitor to achieve full potential. I have a fairly capable Samsung CHG90 – 144Hz FreeSync 2 32:9 FHD monitor that is G-Sync compatible and works well with the RTX 2080 Ti eGPU. However to do it justice, I got an Acer Predator XB273K – 144Hz G-Sync certified 16:9 4K monitor. The Gigabyte team also shipped me an AORUS FI27Q-P – 165Hz G-Sync compatible 16:9 QHD monitor for additional testing. My plan was to observe any noticeable differences between G-Sync certified and G-Sync compatible monitors with the AORUS RTX 2080 Ti Gaming Box.
The two main selling features of Nvidia Turing graphics cards are NVIDIA DLSS (deep learning supersampling) and RTX (real-time ray tracing). At launch there were not many titles optimized for these GPUs. The library of ray tracing games are getting better. The ones I could test with the AORUS RTX 2080 Ti Gaming Box were Battlefield V, Control, and Wolfenstein: Youngblood. As seen in the one-minute gameplay video above, the AORUS RTX 2080 Ti Gaming Box enables eGPU gaming at 4K 60FPS RTXOn with an ultrabook. Testing with all three monitors mentioned above, I could not tell the difference in terms of smoothness between G-Sync certified and G-Sync compatible.
Battlefield V Settings
Wolfenstein: Youngblood Settings
One thing we’ve learned over the years is that eGPU can be a great solution for compute tasks. Unlike the performance loss through gaming, the loss through compute is minimal. We’ve seen strong interest in using eGPU with machine learning platforms such as TensorFlow. With 4,352 CUDA cores, the RTX 2080 Ti is the perfect match for this application. Distributed computing is another great use for external GPU. Our community recently lends a hand in the fight against COVID-19 through [email protected] and BOINC. The SARS-COV2 projects through FAH use graphics cards specifically. A 3x eGPU setup I recent built can post 2,293,210 points per day.
Triple Radeon eGPUs FAH Advanced Control
RX 580 + RX 5600 XT + RX 5700 XT eGPUs [email protected] Radeon Performance Monitoring
Using eGPU with Linux is also of great interest. My distro of choice is Pop!_OS by System76. I installed the Nvidia ISO that includes the graphics drivers for this RTX 2080 Ti Gaming Box. Pop!_OS is based on Ubuntu which has very good Thunderbolt hot-plug support. Gaming in Linux has gotten a lot better too. Thanks to Proton and Steam Play, we can now play many Windows titles in Linux with ease. I tried Witcher III and GTAV on this eGPU setup. The installation process and gaming experience was no different compared to Windows.
Linux Egpu Switcher
Gigabyte’s AORUS RTX 2080 Ti Gaming Box is an affordable solution relative to piecing together an RTX 2080 Ti and a premium external GPU enclosure yourself. By no means is it the best value for eGPU users. Yet by all means, it is the highest-performing eGPU available today and the only liquid-cooled option. The ultrabook and monitor you pair will need to be equally as capable in order to make the most use of this eGFX. Enabling eGPU gaming at 144Hz in FHD, 100Hz in QHD, and 60Hz in UHD, this Gaming Box keeps up with desktop alternatives while retaining the versatility of the ultrabook. Ultimately to choose the AORUS RTX 2080 Ti Gaming Box is to answer Outkast’s call for rebellious originality, “What’s cooler than being cool?”
Share this Post
This article is also published as a forum topic here »
Gigabyte AORUS GeForce RTX 2080 Ti XTREME 11G Review
Christopher Coke Updated: Posted:
Category: Hardware Reviews 0
If you’re ready to step into the world of high-end 4K gaming, the RTX 2080 Ti is the only way to go. If you’re ready to make the jump, your next choice is which 2080 Ti you should choose. Today, we’re looking at a showpiece GPU if ever we’ve seen one with the Gigabyte AORUS GeForce RTX 2080 TI XTREME 11G. It features innovative triple-fan cooling, a speedy factory overclock, and one of the most impressive lighting solutions we’ve ever seen, but is it worth $1299? Join us as we find out.
- Current Price: $1299
- Graphics Processing: GeForce RTX 2080 Ti
- Core Clock: 1770 MHz (Reference Card: 1545 MHz)
- RTX-OPS: 81
- CUDA Cores: 4352
- Memory Clock: 14140 MHz
- Memory Size: 11 GB
- Memory Type: GDDR6
- Memory Bus: 352 bit
- Memory Bandwidth (GB/sec): 616 GB/s
- Card Bus: PCI-E 3.0 x 16
- Digital max resolution: [email protected]
- Multi-view: 4
- Card size: L=290 W=134.31 H=59.9 mm
- PCB Form: ATX
- DirectX: 12
- OpenGL: 4.6
- Recommended PSU: 750W
- Power Connectors: 8 Pin x 2
- Output: DisplayPort 1.4 x 3, HDMI 2.0b x 3, USB Type-CTM(support VirtualLinkTM) x 1
- SLI support: 2-way NVIDIA NVLINKTM
- 1. AORUS VGA holder
- 2. AORUS metal sticker
- 3. Quick guide
- 4. 4-year warranty registration
- 5. I/O guide
- 6. Driver CD
Buying a graphics card today is more complicated than its ever been. You have a lot to consider. What resolution are you playing at, what kinds of games do you plan to play, do you want ray tracing, warranties, future-proofing, RGB, cooling, G-Sync, HDR… the list goes on. Thankfully, things become a whole lot simpler once you decide you’re ready to make the jump to 4K. At the moment, only the RTX 2080 and RTX 2080 Ti are true “4K cards” and, if we’re looking at new AAA games, the 2080 Ti becomes the only real option. The question then becomes which RTX 2080 Ti you should buy.
That’s where the AORUS XTREME comes in with a suite of features to make the case that it’s the 2080 Ti for you. Right off the bat, this model comes with a hefty factory overclock to 1770 MHz, up from the reference speed of 1545 MHz, and is one of the absolute fastest factory overclocks I’ve seen or could find. Out of the box performance is definitely enhanced on this one, assuming it’s not being thermal throttled.
To that end, Gigabyte has packed the RTX 2080 Ti XTREME 11G with a massive heatsink an innovative triple-fan cooling solution. The copper heat pipes included in the heatsink make direct contact with the GPU and MOSFETs to effectively draw heat up into the fin array. They’re also hollowed out to utilize phase transition thermodynamics for more efficient heat transfer. Up top, Gigabyte uses three 100mm fans but what makes this solution interesting is that the center fan actually spins counter to the other two to reduce air turbulence for improved cooling potential. In practice, this actually works quite well as you’ll see in our performance results.
The card also features a great backplate to support its additional weight. At 11.4 x 5.4 x 2.4 inches (LxWxD) and 2.5 slots. It also makes direct contact with the board for modest heat dissipation abilities but even more importantly adds a lot of rigidity to the card to keep it from flexing. Gigabyte actually bundles a metal support arm with the card for additional support to ward off GPU sag. The backplate also has a bright AORUS logo, too, which we’ll get to soon.
The card packs a punch. Under the hood, the AORUS is features 4352 CUDA cores and 11GB of GDDR6 clocked to 14GB/s. On the 352-bit bus, the total memory bandwidth is 616 GB/s. Thanks to the pre-set overclock and 544 tensor cores, the RTX-Ops receive a nice bump to 81 from the 78 found on the Founders Edition. Likewise, you also get an extra 1.2 TFLOPS of performance for a total of 15.4.
What all of that really boils down to is that the AORUS GeForce RTX 2080 Ti XTREME 11G is perhaps the most capable air-cooled card on the market.
The other big element with this GPU is its lighting. Yes, yes, lighting doesn’t affect the performance of the card, I know. But still, what we have here is one of the best light shows ever seen on a GPU. The AORUS 2080 Ti XTREME it kitted out with LEDs. Each fan has a set long the fins, creating an eye-catching ring when they’re in motion (they turn off when stopped). In the center is the AORUS falcon, whose color changes depending on your preset. Pictures really don’t do it justice.
The card is clearly made to look great in a vertical mount as pictured above but also looks great traditionally mounted. A logo on the side flows in RGB rather than the single-color options used on their cheaper cards. The gloss looks great too. The falcon logo on the back is also customizable and is quite bright. When mounted vertically, it washes the area behind it in colored light.
The fan lighting is also very customizable with a lot of great-looking presets. My favorite was definitely the tri-color preset which blends three colors in a flowing gradient. This is definitely the best RGB I’ve ever seen on a GPU and definitely the best I’ve seen from Gigabyte. My only qualms are that it makes you want to disable intelligent fan stop to keep the lights on. Likewise, the falcon logo in the front is based on how fast the fans are spinning, so running them below 40% results in some noticeable flickering. Thankfully, in any situation where you’d want the fans to run for gaming, they’ll be running fast enough to keep the lighting consistent.
On paper, the AORUS RTX 2080 Ti XTREME 11G has the specs to be a showstopper. Let’s take a look and see how it pans out in real life.
Test System: Gigabyte X570 AORUS Master Motherboard, AMD Ryzen R9 3900X CPU, NZXT Kraken X72 AIO Cooler, G.Skill TridentZ Royal DDR4-3600MHz 16GB DRAM Kit, Gigabyte AORUS NVMe Gen4 SSD 2TB, Corsair HX-1050 1050 Watt Power Supply.
For all of our benchmark testing, we strive to provide real-world results akin to what you would see in your own PC. To that end, we stress test our cards through a series of modern games and compare them against one another. We also record peak temperatures within these games on stock fan settings. Be aware that the thermal results we report can usually be improved upon by creating a custom fan curve in programs like MSI Afterburner.
In today’s review, we’ve chosen to focus on 1440p and 4K gameplay. The reason for this is that the 2080 Ti is clearly not intended for 1080p and we wouldn’t recommend anyone purchase this caliber of card for FHD gameplay. Any card at this price is going to be for enthusiasts that are pushing high resolutions, frame rates or both. To that end, we don’t think it’s impossible someone might choose this card for 1440p in an effort to push a 144Hz monitor to its limit for the foreseeable future.
When it comes to settings, we turn everything up with the exception of MSAA and FXAA and instead opt for the lowest level of TAA possible. In cases where that’s not possible, we enable the existing anti-aliasing option at its lowest setting. Ray tracing is also disabled for this testing as so few games currently support it.
Let’s see how it did.
Running the game through our suite of test titles shows the value of that factory overclock in real terms. Without doing any overclocking of your own, you’re getting top of the line performance that routinely outpaces the 2080 Ti Founders Edition (which had its own factory overclock) by 5-7 frames.
At 1440p, the card will still struggle to push these games at 144 FPS without turning down some settings but it was consistently able to top the other cards we compared against. Rightfully so at this price point and the internal specs. Still, it was able to achieve triple-digit frame rates consistently with the exception of Metro: Exodus (and reducing a couple of key settings would accomplish this with minimal graphical degradation).
At 4K, the results are similar with the AORUS RTX 2080 Ti XTREME 11G leading the pack. Here, the card card was able to hold 4K60 with a bit of headroom to spare. Not much, however, and it’s important to note that this is still not the case for the latest titles out there. Soon, we’ll be refreshing our suite of test games to include these, but The Division 2, for example, pushed the card into the mid-50s for FPS fairly regularly.
What’s not reflected in those average FPS numbers is how consistent those frame rates were after multiple hours of gaming once warm air permeated the case. As you can tell in the chart above, thermal throttling was never an issue. At peak the card topped out at 76C and never progressed higher. The ambient temperature for this testing was 23C. This definitely reinforces that the triple-fan, alternate spinning solution Gigabyte used here is effective in combination with their heat pipes and that massive sink.
This is especially impressive because the card was mounted vertically with choked off airflow thanks to the side glass panel. Now, usually I would mount the card horizontally but since the XTREME was so clearly intended for a vertical mount, I changed my approach. With a traditional mount, temperatures dropped another 2-3C on average.
I am no master overclocker but a card of this caliber screams to be pushed. I started off simply by using Gigabyte’s AORUS Engine Software. Turing cards are able to utilize Nvidia’s auto-overclocking scanner to find the maximum safe overclock. After allowing AORUS Engine to scan, I came out with 87 MHz boost to the core clock. This is modest and the resulting impact on frame rates was minimal.
Instead, I retreated back to MSI Afterburner, my go-to overclocking program, and was able to push the card to a +115 MHz boost after raising the power limit. As is often the case with modern GPU overclocking, the gains were small but noticeable on paper with an average gain of 1-3 FPS.
Is it worth it? If you’re the kind of gamer looking to push his hardware to the limit (and, come on, if you buy a $1299 graphics card, that probably describes you), then yes. For the average gamer, taking the time to play and eek out those last few MHz isn’t going to net you a FPS boost you can feel, but with AORUS Engine (and other tools) Auto Scan functionality, overclocking your card is easier than ever. With one button press and ten minutes, the card will push itself to a modest to decent overclock. With that kind of system, why wouldn’t you take advantage of it?
At $1299, the Gigabyte AORUS GeForce RTX 2080 Ti XTREME isn’t going to be for everybody but, quite frankly, no 2080 Ti is for everybody. If you’re looking to make the jump to 4K, 60 FPS gameplay while also running cool and looking great, this is simply the best way to do it. The Gigabyte AORUS GeForce RTX 2080 Ti XTREME is an absolute, unequivocal BEAST of a card.
The product described in this article was provided by the manufacturer for evaluation purposes.
- Outstanding performance — top of its class
- One of, if not the best, factory overclock
- Great thermal performance, even with low airflow
- Runs fairly quiet at 80% and lower fan speeds
- Looks outstanding — hands-down, the best looking 2080 Ti
- 2.5 slots — it’s a big boy (but comes with its own support)
- Logo can flicker at low fan speeds
Review of the video card GIGABYTE GeForce RTX 3060 GAMING OC 12G
In the context of the current total shortage of video cards, the announcement of any new product is an extraordinary event that attracts the attention of both gamers who do not leave hopes of getting a suitable solution for a comfortable pastime, and virtual currency miners who are waiting for replenishment for their mining farms. Today we have a review of a novelty from the line of graphic solutions with Ampere architecture — GeForce RTX 3060 12 GB . A very promising mid-range model with plenty of memory and an attractive recommended price tag of $329. We evaluate the capabilities of a video card with a GA106 chip using the original GIGABYTE GeForce RTX 3060 GAMING OC 12G model.
- 1 NVIDIA GeForce RTX 3060
- 2 Gigabyte GeForce RTX 3060 Gaming OC 12G
- 3 Design and componing
- 4 in the work
- 5 Configuration0014
- 6 performance 4K
- 7 performance 2560 × 1440
- 8 performance 1920 × 1080
- 9 working applications on GPU
- 10 Acceleration
- 11 Energy
nvidia b 3060 uses the GA106 processor with Ampere graphics architecture. Like other GPUs in the series, the chip is manufactured by Samsung using 8nm technology. The crystal has an area of 276 mm² and includes 13.
3 billion transistors. Note that the silicon plate here is noticeably more compact than that of the older GA104 chip, which is used for the GeForce RTX 3060 Ti/3070 (392 mm², 17.4 billion). The smaller dimensions of the crystal indirectly allow us to talk about reducing the cost of manufacturing video cards.
It is curious that the desktop version of the GeForce RTX 3060 uses a GPU configuration with a slightly smaller number of functional blocks than the mobile modification of the video card. The latter has noticeably reduced operating frequencies, but the number of active computers is higher. Suppose that in the case of the desktop GeForce RTX 3060, a certain reserve is left for the possible appearance of accelerated SUPER modifications in the future.
|GeForce RTX 3060 Ti||GeForce RTX 3060||GeForce RTX 2060|
|Manufacturing process||8 nm||8 nm||12 nm|
|Crystal area||392 mm²||276 mm²||445 mm²|
|Number of transistors, billion||17. 4 billion||13.25 billion||10.8 billion|
|GPU clock speed||1410/1665 MHz||1320/1777 MHz||1365/1680 MHz|
|RT cores||38 (2nd generation)||28 (2nd generation)||30 (1st generation)|
|Tensor cores||152 (3rd generation)||112 (3rd generation)||240 (2nd generation)|
|Memory (type, capacity)||GDDR6 8 GB||GDDR6 12 GB||GDDR6 6 GB|
|Memory bus||256 bit||192 bits||192 bits|
|Effective memory clock speed||14000 MHz||15000 MHz||14000 MHz|
|Memory bandwidth||448 GB/s||360 GB/s||336 GB/s|
|Power consumption, W||200 W||170 W||160 W|
Returning to the presented version of the GeForce RTX 3060 for PC, we note that the GA106-300 includes 3584 computers, 112 texture units and 48 ROPs. In addition, the GPU also contains 112 Gen 3 Tensor Cores and 28 Gen 2 RT Cores for hardware-accelerated ray tracing. The base frequency of the GPU is defined at 1320 MHz, and the average acceleration value is 1777 MHz. In this configuration, single precision (FP32) peak performance is 12.74 TFLOPS. For comparison, the same value for the GeForce GTX 1060 6 GB is 4.38 TFLOPS.
The GeForce RTX 3060 uses a 192-bit memory bus, while the graphics cards of the line are equipped with 12 GB of local RAM. In this case, the amount of memory is even greater than that of the GeForce RTX 3080, which has 10 GB on board. The adapters are equipped with GDDR6 chips with an effective operating frequency of 15,000 MHz. Considering the bit depth of the data transmission line, the peak memory bandwidth is 360 GB / s. A nominally good figure for Full HD and 1440p graphics modes, and the amount of memory is completely in excess for games at this resolution. There are prerequisites for the fact that over time versions with 6 GB of memory on board may appear, but there is no official information on this matter — in the regular version of the GeForce RTX 3060, 12 GB of memory are equipped.
The thermal package of the novelty is slightly higher than that of its predecessor — the declared Graphics Card Power parameter of the GeForce RTX 3060 12 GB video card is 170 W, instead of 160 W for the GeForce RTX 2060 6 GB. The developer recommends using power supplies with a capacity of 550 W or more for a system with a GeForce RTX 3060.
NVIDIA does not offer a reference version of the GeForce RTX 3060, relying entirely on its partners in this matter. Video card manufacturers have presented a wide range of original models based on GA106, which traditionally differ in PCB designs, component equipment and cooling system configurations. We will use one of these models for a more substantive acquaintance with the capabilities of the GeForce RTX 3060 12 GB.
Despite the fact that the reference version of the GeForce RTX 3060 is not presented, NVIDIA, of course, has indicated the recommended cost for models in this series. For the US market, this is $329 (excluding taxes), while for Ukraine it is UAH 12,990. How useful this information will be in the current conditions is a rhetorical question, but the formalities are observed.
GIGABYTE GeForce RTX 3060 GAMING OC 12G
GIGABYTE is launching five models of the GeForce RTX 3060 series, and we are talking about four different lines at once. These are a pair of starter two-fan EAGLEs, a «creator» version of VISION, similar in layout to it, but more familiar in design, the classic GAMING with a three-fan cooler, as well as the top model AORUS ELITE.
We got a GIGABYTE GeForce RTX 3060 GAMING OC 12G for testing. As you might guess by the name, the model belongs to the GAMING series and is offered with factory overclocking.
In this case, the frequency formula for the GPU is 1320/1837 MHz, instead of the recommended 1320/1777 MHz for basic versions.
The video card comes in a medium-sized cardboard box with a recognizable design. Only a short paper manual is included.
Design and layout
The external design of the model can be called typical of the latest models of the GAMING series. For a protective plastic casing, a combination of an asymmetric panel with a black base and a pair of graphite inserts is used.
The GIGABYTE GeForce RTX 3060 GAMING OC 12G uses one of the three-fan WINDFORCE 3X cooling system. The design uses three typesetting cassettes of aluminum plates, which are interconnected by three heat pipes with a diameter of 6 mm.
The overall design of the heat sink assumes direct contact between the heat pipes and the GPU die. In this way, heat transfer from the GPU to remote areas of the radiator block is accelerated as much as possible, however, to exclude overheating zones at the junctions of the tubes, good processing and fitting of the elements is necessary.
Power elements of the power stabilization subsystem are located along the interface panel. Additional cooling is provided for the assemblies — VRM elements through thermal pads are in contact with one of the radiator cassettes.
Additional heat dissipation is also provided for memory chips, which are covered by an aluminum plate connected to the base of the radiator unit.
Despite the significant overall dimensions of the device in length (282 mm), the adapter has a fairly compact printed circuit board (173 mm), as well as a two-slot overall design.
Three 80 mm axial-type fans with a specific shape of the impeller and slight fins on the blades are used to blow the radiator block. According to the developers, such design features allow you to increase the air flow. Multidirectional rotation of the central and side fans should help to cope with the occurrence of turbulence zones.
Of the features of the fans used, it can be noted that these are models with plain bearings for which graphene grease is used. The special composition increases the reliability of the bearings and extends the life to the level of models with double ball bearings, but at a lower level of emitted noise.
One 8-pin connector is provided on the top edge of the graphics card for auxiliary power. Given the GeForce RTX 3060’s claimed power consumption of 170W, that should be plenty. Such a connector allows you to supply up to 150 W of energy, another 75 W can be obtained from the PCI Express x16 slot. Despite the factory overclocking of the GPU, the recommendations of GIGABYTE engineers regarding the power of the power supply used are unchanged — 550W.
The manufacturer has equipped the video card with an additional power line connection indication system. Next to the power connector is a bright white LED that lights up if you try to start the system without first connecting the power cable to the video card.
In the GAMING series of video cards, developers are very reserved in their approach to additional RGB lighting. In this case, only the logo on the top edge of the adapter is illuminated, as well as the light guide strip located next to it on a beveled bevel.
On the reverse side, the video card is covered by an aluminum plate — a decorative element and a protective plate, which at the same time increases the strength of the adapter design. In this case, it is not used for additional heat removal, but it does not reduce the efficiency of the cooling system. On the side farthest from the interface panel, you can clearly see a hole that allows one of the fans to blow through the radiator cassette.
A technique called Screen Cooling has recently been used quite often for three-fan video cards with a shortened PCB base. If the dimensions of the PCB really allow through ventilation of the heatsink, it is worth using
We can also note the developers’ attempt to accentuate the flow generated by the fan — the protective casing in this area fits snugly enough to the plate, forming the required air duct.
The interface panel of the video card has four video outputs — a pair of full-length DisplayPort 1.4a and HDMI 2.1. A significant part of the mounting plate is occupied by holes for additional ventilation.
The GIGABYTE GeForce RTX 3060 GAMING OC 12G cooling system supports hybrid mode, allowing the fans to stop at minimal load.
In idle mode on an open stand (+22C), the GPU temperature was kept at 35-36C.
Under load, the GPU temperature rose to ~63C. Curiously, the 2.37 version of the GPU-Z utility utility can now read the values at the hot spot (Hot Spot), and not just the average value. In this case, the temperature delta under load is about 12C.
The fan speed at the peak increased to 1850 rpm, while subjectively the level of emitted noise was lower than one would expect at the sight of such values. Plain bearings play their role here. Perhaps the presence of a second BIOS chip with firmware that assumes a lower rotation speed would not hurt here, but this option is not provided in the model under consideration. However, if necessary, you can adjust the CO operation algorithm yourself in the proprietary AORUS Engine application.
In games, the GPU was automatically accelerated to almost 2000 MHz. The GPU-Z utility fixes the power consumption of the video card at 176 watts.
Test bench configuration
|Processor||Intel Core i7-9700K (8/8; 3.6/4.9 GHz)||Intel, www.intel.com|
|Motherboard||ASUS ROG Maximus X HERO (Intel Z370)||ASUS, www.asus.ua|
|RAM||G.Skill Trident Z Royal Gold 2×8 GB (F4-3600C16D-16GTRG)||G.Skill, www.gskill.com|
|Accumulator||Kingston KC2500 1TB (SKC2500/1000G)||Kingston Technology, www.kingston.com|
|Power supply||ASUS ROG Strix 1000W, 1000W||ASUS, www.asus.ua|
|Monitor||Acer Predator XB271HK (27″, 3840×2160)||Acer, www.acer.ua|
We used the GeForce GTX 1060 6 GB to evaluate the performance of the GeForce RTX 3060 12 GB. Without exaggeration, a cult model, which is still the most popular video card among users of the Steam service. Devices based on GP106 are still used by about 10% of active gamers. Let’s see how worthy an update the GeForce RTX 3060 can be for a five-year-old hit.
Results for GeForce GTX 1660 SUPER 6 GB have also been added to the charts for comparison. Also a successful model, which was especially popular with RT naysayers, who did not particularly believe in the success of the hybrid ray-traced rendering model, well, or were not ready to pay to be among the pioneers.
|GeForce RTX 3060 Ti||GeForce RTX 3060||GeForce RTX 2080||GeForce GTX 1660 SUPER||GeForce GTX 1060|
|Manufacturing process||8 nm||8 nm||12 nm||12 nm||16 nm|
|Crystal area||392 mm²||276 mm²||545 mm²||284 mm²||200 mm²|
|Number of transistors, billion||17. 4 billion||13.25 billion||13.6 billion||6.6 billion||4.4 billion|
|GPU clock speed||1410/1665 MHz||1320/1777 MHz||1515/1710 MHz||1530/1785 MHz||1506/1709 MHz|
|RT cores||38 (2nd generation)||28 (2nd generation)||46 (1st generation)||—||—|
|Tensor cores||152 (3rd generation)||112 (3rd generation)||368 (2nd generation)||—||—|
|Memory (type, capacity)||GDDR6 8 GB||GDDR6 12 GB||GDDR6 8 GB||GDDR6 6 GB||GDDR5 6 GB|
|Memory bus||256 bit||192 bits||256 bit||192 bits||192 bits|
|Effective memory clock speed||14000 MHz||15000 MHz||14000 MHz||14000 MHz||8000 MHz|
|Memory bandwidth||448 GB/s||360 GB/s||448 GB/s||336 GB/s||192 GB/s|
|Power consumption, W||200 W||170 W||215 W||125 W||120 W|
Unfortunately, we weren’t able to include the immediate predecessors, the GeForce RTX 2060 and GeForce RTX 2060 SUPER, in our testing. Despite the information that NVIDIA plans to resume deliveries of chips for these models, video cards of these series are still in great short supply. However, but only these. Therefore, a series of models with RTX of the first generation will be supported by the GeForce RTX 2080 Founders Edition 8 GB. A slightly accelerated modification will become a reference point, starting from which it is not difficult to interpolate the results within the series.
Well, the top limiter will be the older model of the line — GeForce RTX 3060 Ti 8 GB. The novelty will certainly not be able to reach the performance of the device on the GA104, but the actual difference in performance will also be informative.
The first measurements in 3DMark determine the starting positions of all participants. It is somewhat unusual to see the GeForce GTX 1060 6 GB as a starting point, especially when such a well-deserved model is so much inferior to more advanced devices. GIGABYTE GeForce RTX 3060 GAMING OC 12G is more than twice ahead of the titled «ancestor» in the Time Spy subtests with classic rendering in DirectX 12. Well, the tenfold difference in Port Royal is just another confirmation of how important hardware support is for tracing. With the help of the API translator, rays can “move the rays” with the help of universal shader calculators, but … it’s not necessary like that.
Although the GeForce RTX 3060 is not nominally positioned by the manufacturer as a solution for gaming in 4K mode, we will move downward. High resolution allows you to identify the strengths and weaknesses of the adapters, so I would not want to lose sight of any features.
On the whole, there are no surprises here. The GeForce RTX 3060 12 GB, including in real games, has at least a twofold advantage over the GeForce GTX 1060 6 GB. At the same time, the new product is 15–22% inferior to the GeForce RTX 2080 8 GB. If we move along the Turing line, then the classic GeForce RTX 2070 8 GB demonstrates approximately similar performance, which in turn is 5–7% ahead of the GeForce RTX 2060 SUPER 8 GB and 20% faster than the starting GeForce RTX 2060 6 GB.
Models with tensor cores are expected to get a good performance boost when using DLSS (Deep Learning Super Sampling) intelligent scaling. The second generation technology is able to pull the image at a lower original rendering resolution and then restore it without losing detail.
DLSS can be especially useful in projects with ray tracing enabled, when the load on the GPU increases, and literally every additional fps counts. The younger model in the Ampere series quite skillfully uses the available range of technologies, but, again, as we see, for the 4K mode with maximum graphics quality, in principle, more productive solutions are needed, because even the GeForce RTX 3060 Ti in such conditions does not always cope.
The 1440p mode is already much more comfortable for the GeForce RTX 3060. The balance of power of the test participants is generally maintained. We note a significant difference in the performance of the GeForce RTX 3060 Ti and RTX 3060.
Even in modes with a resolution of 2560×1440, the older “titanium” version often has a noticeable advantage — 20-25%. Let’s assume that such a gap was left intentionally for the SUPER modification of the RTX 3060, which may appear in the future. There is a technological backlog for this. For example, such a model could use a full-fledged modification of the GA106 chip with all enabled computers and an increased operating frequency.
However, even in the presented version of the GeForce RTX 3060 12 GB, it shows itself perfectly at 1440p. Yes, in the heaviest projects, you still have to play around with the sliders, choosing the settings for a stable 60 fps. But too significant corrections are not needed, and often the model offers good performance even at maximum speeds.
Performance 1920 × 1080
We reduce the resolution to Full HD, and only the gluttonous Cyberpunk 2077 can force you to dig into the settings of the owner of the GeForce RTX 3060 here, and even then, this is more like a safety net.
New for very comfortable performance in this mode. In many projects, counters go off scale for 100 fps, which means that monitors with an increased screen refresh rate will no longer be idle.
Here we also want to note the capabilities of the GeForce GTX 1060 6 GB. Test results show that this adapter is running out of time. Increasingly, a graphics card cannot provide comfortable performance at high graphics quality settings even without additional tracing effects. It is curious that the twofold advantage of the GIGABYTE GeForce RTX 3060 GAMING OC 12G over the age model will continue in Full HD mode. This is a good option for an upgrade that will feel good even without additional performance measurements. Well, if you remember about the «goodies» with support for RT and DLSS, then this is just a matter already resolved. But, the price of the issue is now even more relevant.
The GeForce RTX 3060 has a huge local buffer by midrange standards. The available 12 GB allow not only to close the issue with a lack of memory in games in all modes relevant to this model, but it is also possible to use the video card more efficiently in work tasks.
We have taken a number of rapid measurements in various applications that use GPU resources for rendering, rendering and computing. It is curious that in most cases the GeForce RTX 3060 turned out to be more productive than the GeForce RTX 2080 FE, despite the fact that the latter had a stable advantage in games. Still, architectural improvements and a huge buffer in a number of tasks can play into the hands of the new Ampere, allowing it to outperform its predecessors in a higher class.
Before experimenting with overclocking GIGABYTE GeForce RTX 3060 GAMING OC 12G, we immediately note that the video card allows you to significantly increase the Power Limit parameter — by 24%. Good sign.
Indeed, the adapter made it possible to increase the base value of the GPU frequency by 170 MHz — from 1320 to 1490 MHz. At the same time, the average value of the dynamic acceleration of the chip increased proportionally — from 1837 to 2037 MHz. Given that the nominal for the GeForce RTX 3060 is 1777 MHz, we have a total frequency increase of 14.6%.
Memory chips have also been accelerated rather well. Samsung chips worked stably at effective 17,600 (2200) MHz, instead of the standard 15,000 MHz. The bandwidth of the memory subsystem has increased from 360 GB / s to 422 GB / s. A ~17% increase in memory bandwidth will not be superfluous and will certainly affect the results.
Practical measurements have shown that additional overclocking in this case is quite practical. In such an accessible way, it is quite possible to increase the number of frames / s by ~ 10%. Of course, the increase in performance is not proportional to the increase in frequency, but the resulting acceleration is no longer a “measurement error” at all.
After additional acceleration, the peak frequency of the GPU increased to 2160 MHz. At the same time, in games, values of 2100–2140 MHz were typical. GPU heating increased by 2 degrees — up to 65C on average and up to 77C at the hottest point. The rotation speed was also increased by 150 rpm. Although the power limit was initially increased by an available 24%, the Board Power Draw parameter was increased by about 12% to 197 watts.
We also measured the total power consumption of systems equipped with video cards participating in the benchmark.
The performance of the platform with the GeForce RTX 3060 12 GB is quite predictable. Obviously, the graphics card consumes slightly more than previous generation models without RTX capabilities. At the same time, the actual difference in the performance of these solutions speaks for the better energy efficiency of the novelty based on Ampere. But if we evaluate the declared TDP values with those for models based on Turung that are similar in performance, then there are no breakthroughs here either. Recall that for the GeForce RTX 2060 SUPER and GeForce RTX 2070, a thermal package of 175 watts is declared. However, the latter are equipped with 8 GB of memory, while the GeForce RTX 3060 has 12 GB.
The GeForce RTX 3060 12 GB is a great candidate for the ultimate gaming platform. The video card offers good performance in 1080p / 1440p modes, supports all the latest NVIDIA developments, but just the same, a huge 12 GB buffer will not only be a safety net for resource-intensive games, but can be useful for work tasks. If we extrapolate the results, then the younger Ampere model offers performance at the level of the GeForce RTX 2070. Here one could continue the phrase, pointing out that it costs much less than its predecessor. However, the cost of video cards is now as volatile as the price of cryptocurrencies. And if the latter fluctuates significantly in both directions, then graphics adapters are only getting more expensive. Needless to say, the success of not only this model, but any other on the market, will depend on retail prices. Just imagine how much kinder the world would be if the GeForce RTX 3060 12 GB could be bought for 13,000 UAH. Alas, no. In the best case, the indicated price tag will have to be multiplied by two, and this already fundamentally affects the interest of the gaming audience, which needs a video card for the soul, and not endless recalculations of its payback.
Good gaming performance in 1080p/1440p modes; 12 GB memory; GIGABYTE model offers good frequency potential for GPU/memory overclocking; hybrid mode of CO operation with fans stopping at rest; customizable RGB lighting; double slot design; design of a cooler with a through blowing of the radiator; two HDMI 2.1 ports
Starting retail price; no second BIOS with «silent» profile
GIGABYTE GeForce RTX 3060 GAMING OC 12G is a tuned version of the GeForce RTX 3060 with a boosted GPU, a hybrid and fairly efficient cooling system, as well as moderate intensity customizable RGB lighting. The graphics card offers the performance of the GeForce RTX 2070, doubling the performance of the very popular, but no longer always pulling GeForce GTX 1060 6 GB. 12 GB of memory allows you to not think about its lack in any mode that is lifting for the RTX 3060, and it is also a good help for tasks with non-gaming calculations. Here one could write that the success of a video card will determine its price, but now this is too variable.
GIGABYTE GeForce RTX 3070 Ti GAMING OC
Gigabyte GeForce RTX 3070 Ti Gaming OC [GV-N307TGAMING OC-8GD] is a factory overclocked custom variant of the GeForce RTX 3070 Ti, which is a step below the Gigabyte AORUS Master, and is rather looking for optimal balance between cooling, cost, overclocking and power consumption. At 1830 MHz, the card sits roughly in the middle of the frequencies available in custom designs.
Table of contents:
So, besides boxed overclocking, it is worth highlighting a decent WINDFORCE 3X cooling system, which offers an alternative rotation system for one of the three turntables (2x 90 mm; 1x 80 mm), a proprietary RGB Fusion 2. 0 application, as well as better distribution of flows thanks to slots in the backplate.
Based on current market conditions where MSRP is not available for the RTX 3070 Ti Gaming OC; the model is currently selling for around $1,200, which is about $50 more than the cheapest RTX 3070 Ti cards available right now.
|Price||Shaders||Blocks||Core frequency||Acceleration||Memory frequency||GPU||Transistors||Memory|
|RX 5700 XT||$800||2560||64||1605 MHz||1755 MHz||1750 MHz||Navi 10||10300M||8 GB GDDR6 256 bit|
|RTX 2070||$650||2304||64||1410 MHz||1620 MHz||1750 MHz||TU106||10800M||8 GB GDDR6 256 bit|
|RTX 3060||$750||3548||48||1320 MHz||1777 MHz||1875 MHz||GA106||13250M||12 GB, GDDR6, 192-bit|
|RTX 2070 Super||$700||2560||64||1605 MHz||1770 MHz||1750 MHz||TU104||13600M||8 GB GDDR6 256 bit|
|Radeon VII||$800||3840||64||1802 MHz||without||1000 MHz||Vega 20||13230M||16 GB HBM2 4096 bit|
|RX 6600 XT||$xxx||2048||64||2359 MHz||2589 MHz||2000 MHz||Navi 23||11060M||8 GB GDDR6 128-bit|
|RTX 2080||$750||2944||64||1515 MHz||1710 MHz||1750 MHz||TU104||13600M||8 GB GDDR6 256 bit|
|RTX 2080 Super||$800||3072||64||1650 MHz||1815 MHz||1940 MHz||TU104||13600M||8 GB GDDR6 256 bit|
|RTX 3060 Ti||$700||4864||80||1410 MHz||1665 MHz||1750 MHz||GA104||17400M||8 GB GDDR6 256-bit|
|RX 6700 XT||$770||2560||64||2424 MHz||2581 MHz||2000 MHz||Navi 22||17200M||12 GB GDDR6 192-bit|
|RTX 2080 Ti||$1100||4352||64||1350 MHz||1545 MHz||1750 MHz||TU104||18600M||11 GB GDDR6 352 bit|
|RTX 3070||$1000||5888||96||1500 MHz||1725 MHz||1750 MHz||GA104||17400M||8 GB GDDR6 256-bit|
|RTX 3070 Ti||$1150||6144||96||1575 MHz||1770 MHz||1188 MHz||GA104||17400M||8 GB GDDR6X 256-bit|
| Gigabyte RTX 3070 Ti
|$1200||6144||96||1575 MHz||1830 MHz||1188 MHz||GA104||17400M||8 GB GDDR6X 256-bit|
|RX 6800||$1300||3840||96||1815 MHz||2105 MHz||2000 MHz||Navi 21||26800M||16GB GDDR6 256-bit|
|RX 6800 XT||$1400||4608||128||2015 MHz||2250 MHz||2000 MHz||Navi 21||26800M||16GB GDDR6 256-bit|
|RTX 3080||$1400||8704||96||1440 MHz||1710 MHz||1188 MHz||GA102||28000M||10 GB GDDR6X 320-bit|
|RTX 3080 Ti||$1750||10240||112||1365 MHz||1665 MHz||1188 MHz||GA102||28000M||12GB GDDR6X 384-bit|
Packaging and contents
- Quick Installation Guide for
- Product Warranty Registration Card — 4 Years
The Gigabyte card uses the popular mix of black and gray colors combined with a blocky industrial design. Behind you will find a high-quality metal backplate.
- L=320 W=129 H=55 mm
- 1218 g
Installation will require 3 slots in the system.
- 2x DisplayPort (v1.4)
- 2x HDMI (2.0b)
DisplayPort 1.4a outputs support Display Stream Compression (DSC) 1.2a, allowing you to connect 4K displays at 120Hz and 8K displays at 60Hz. Ampere can drive two 8K displays at 60Hz with just one cable per display.
Ampere is the first GPU to support HDMI 2.1, increasing bandwidth to 48Gbps to support higher resolutions such as [email protected] and [email protected] with a single cable. With DSC it’s up to [email protected] and [email protected] respectively.
The new NVIDIA NVENC/NVDEC video engine is optimized to handle video tasks with minimal CPU usage. The highlight here is the added support for AV1 decoding. As with Turing, MPEG-2, VC1, VP8, VP9, H.264 and H.265 can also be decoded natively, up to [email protected] resolution.
Encoder is identical to Turing, it supports H.264, H.265 and lossless up to [email protected]
This BIOS switch allows you to switch between the default «Performance» and the «Silent» option.
Power is supplied from two 8-pin PCIe power connectors. Along with slot power, this configuration delivers up to 375W. I really like how Gigabyte placed the power connectors closer to the corner of the card, unlike almost all other solutions on the RTX 3070 Ti.
GeForce RTX 3070 Ti does not support SLI.
Disassembling the device
Gigabyte’s thermal solution uses seven heat pipes. The main heatsink not only cools the GPU, but also cools the memory chips and VRM circuits. The thermal pads on the VRM are 1.3 mm thick, while those on the memory are 2.0 mm thick.
The rear panel is made of metal and protects the card from damage during installation and operation.
These pictures are for volt modders and people who want to see finer details on a PCB.
The GPU VRM is 10-phase and controlled by the UPI uP9512R voltage controller.
GPU VRM uses Vishay SiC653 DrMOS components rated at 50A.
The memory voltage is 2-phase and is generated by the uPI uP1666Q controller.
Each memory phase uses a combination of uPI QN3106 and QN3102 MOSFETs.
GDDR6X memory chips are manufactured by Micron and have model number D8BWW, decoded as MT61K256M32JE-19G:T. They are designed to operate at a frequency of 1188 MHz (GDDR6X effective frequency is 19 Gb / s).
The NVIDIA GA104 GPU is the third chip based on the Ampere architecture, the second designed for GeForce gamers. It is manufactured by Samsung on an 8nm process, has 17.4 billion transistors and a die area of 392 mm².
|Test configuration — VGA Rev. 2021.1|
|Processor|| AMD Ryzen 7 5800X @ 4. 8GHz
(Zen 3, 32MB cache)
|Motherboard|| MSI B550-A Pro
BIOS 7C56vA5 / AGESA 188.8.131.52
|Resizable BAR:||Enabled on all AMD RDNA2 and NVIDIA Ampere cards|
|RAM|| Thermaltake TOUGHRAM, 16GB DDR4
@ 4000MHz 19-23-23-42 1T
Infinity Fabric @ 2000MHz (1:1)
|Cooling|| Corsair iCue h200i RGB Pro XT
240 mm AIO
|Accumulator||Crucial MX500 2TB SSD|
|Power:||Seasonic Prime Ultra Titanium 850W|
|Operating system|| Windows 10 Professional 64-bit
Version 20h3 (October 2020 Update)
|Drivers|| RTX 3070 Ti: 466.61 Beta
RTX 3080 Ti: 466.54 Beta
RTX 3060, 3060 Ti, 3070, 3080, 3090: 466. 27 WHQL
RX 6700 XT, 6800, 6800 XT, 6900 XT: 21.4.1 WHQL
All others AMD: 21.2.3 Beta
All others NVIDIA: 461.72 WHQL
Test results in other reviews are comparable only when using the same configuration.
- All games and cards are tested with the drivers listed above — no performance results are recycled between test systems. Only this exact system with exactly the same configuration is used.
- All games are tested using the same version.
- All games are set to their highest quality unless otherwise noted.
- AA and AF are applied through in-game settings, not through the driver’s control panel.
- Before starting measurements, the card is heated for each test to ensure that a steady state is being tested.
- This ensures that the card does not surge to unrealistically high frequencies in just a few seconds until it warms up.
Each game is tested at the following screen resolutions:
- 1920×1080: The most popular monitor resolution.
- 2560×1440: An intermediate resolution between Full HD and 4K, with reasonable performance requirements.
- 3840×2160: 4K Ultra HD resolution available on modern high-end monitors.
Test set #1
Assassin’s Creed Valhalla
Launch: 2020 — API: DirectX 12 — Engine: AnvilNext 2.0
This is 873 AD, the most violent period in Norwegian history. The leader of a Viking clan takes a group of refugee warriors across the sea to medieval England, where they camp as barely welcome migrants to the local kingdoms. You forge alliances with local lords and raid Saxon strongholds in search of resources to build and develop your settlement. The historic cities of Winchester, London and York, as well as the vast landscapes of Norway, come to life in this iconic Assassin’s Creed game.
Based on the same AnvilNext 2 engine used in other Ubisoft games, Assassin’s Creed Valhalla takes advantage of DirectX 12 and offers outstanding visuals that look amazing even without ray tracing. We tested in maximum detail using our own test scene.
Launch: 2020 — API: DirectX 12 — Engine: RED Engine
Cyberpunk 2077, the most anticipated game of 2020, maybe even this decade, is not only a game, but also a work of art. A huge open-world metropolis set in the future, it beckons you with hundreds of quests and side quests, a huge amount of content and the ability to drag you into an alternate reality. Despite launch time errors, Cyberpunk 2077 will be a technical milestone in gaming history. You play as «V», a street criminal who is trying to steal a cybernetic implant with mythical powers from a mega-corporation.
Cyberpunk 2077, powered by CD Projekt RED’s RED Engine, is one of the first games to feature true DirectX 12 Ultimate with RTX real-time ray tracing, NVIDIA DLSS support, AMD FidelityFX CAS, and more. The game can be very tiring for your car, but promises to reward you with a world where you can really get lost.
Detroit Become Human
Test Set #2
Devil May Cry 5
Divinity Original Sin II
Release: 2017 — API: DirectX 11 — Engine: Divinity Engine
Divinity Original Sin 2 by Larian Studios is a recent RPG game based on the isometric view of the game world, which was used in many classic role-playing games, most notably Baldur’s Gate. The game was received very well and is probably the best new release for the genre to date. The protagonist, a sorcerer stranded in a world where witchcraft is forbidden, leading to his capture and eventual escape, as well as the revelation of the truth about the void that threatens all life.
Although the graphics are not as demanding as 3D shutters, it still loads modern equipment with DirectX 11.
910USK Hitman III
Launch: 2021 — API: DirectX 12 — Engine: Glacier Engine
Agent 47 is back in the third and final chapter of the popular 2016 Hitman reboot franchise, brought back to life with its smooth, focused gameplay. This stealth killer simulator will take you to even more exotic locations such as the dizzying skyline of Dubai, the expanse of Dartmoor, metropolitan Berlin and the eastern Neon Noir of Chongqing. You’ll also unravel the Hitman lore tapestry that harkens back to other titles in the trilogy.
Based on the same Glacier Engine used in the other two Hitman games, the game takes advantage of DirectX 12 and can be quite taxing on your machine. When you don’t focus on your goal, you will be rewarded with many sweet moments.
Test Set #3
Red Dead Redemption 2
Sekiro: Shadows Die Twice
Shadow of the Tomb8 Raider
Star Wars Squadrons
The Witcher 3: Wild Hunt
Watch Dogs Legion
Launch: 2020 — API: DirectX 12 — Engine: Dunia
at a large pond in near-future London, where the AI CTOS manager has spread its tentacles around the city, turning it into a police state. Your faction, DedSec, is accused of a terrorist attack and a private security firm has been hired to take it down. The game involves reclaiming various areas of London from your nemesis to ultimately bring the dystopian CTOS to its knees.
Based on the most advanced version of the Ubisoft Dunia engine, which is likely also used in the upcoming Far Cry 6, the game takes advantage of DirectX 12 Ultimate and offers real-time ray tracing along with other goodies such as NVIDIA DLSS, including the debut of DLSS 8K . We used a dedicated scene and tested on the highest settings with RT off.
Average Gaming FPS
The following series of charts averages the FPS results of all 22 games suggested to show users what performance to expect from this card in terms of frames per second.
To perform this kind of test, the card is first allowed to sit idle for some time to reach thermal equilibrium. This is followed by an instantaneous 100% gaming load and key parameters are recorded, which gives a rough idea of the card’s behavior when the speed of its fans increases at the same time as the temperature rises. When the temperature reaches a stable point, that is, its increase does not occur within 2 minutes, the load stops and data is recorded during the cooling process.
Quiet BIOS test:
Noise normalized cooler test
It has always been difficult to compare video card coolers on an apples-to-apples basis. Obviously, just looking at the temperature will not work due to the fact that the heat dissipation between different models of video cards varies greatly, even from batch to batch of the same product. Another problem is that the fan control settings on different graphics cards are not identical, which also results in different noise levels. Even noise normalization leaves the fact that each card differs in power consumption, which directly affects thermal output.
To overcome these issues, Unreal Engine 4 created a custom gaming graphics load with the ability to instantly adjust the card’s workload on demand, allowing you to set a precise power consumption target. In all tests on this page, the cooler runs at a constant fan speed, which is carefully chosen to achieve a noise level of 35 dB(A).
It is now possible to test GPU coolers under certain thermal loads, which makes such testing independent of the graphics card used — the only variable is the thermal power of the cooler, which is monitored. This type of noise normalization test helps you understand how efficient your cooling solution is, how much heat it can handle, and how much more it can handle. The temperature graph in the table below is the GPU temperature recorded by the chip’s thermal sensors.
This review introduces a new methodology for testing graphics card power consumption. The data acquisition system has been updated to run at a higher sample rate, and the automation capabilities have been expanded. Now, with 40 samples per second, data is collected much faster than before, even four times faster than NVIDIA PCAT. Each data point is recorded digitally — previously only averages and peaks could be obtained.
Non-gaming power consumption is slightly higher than NVIDIA Founders Edition, likely due to added RGB lighting and changes to the VRM layout.
Gaming power consumption is about the same as the Founders Edition, but provides better performance due to factory overclocking. This is really nice — custom designs often increase the voltage too much, resulting in reduced efficiency.
However, compared to the RTX 3070 and RX 6800, the RTX 3070 Ti is significantly less power efficient.
Recommended PSU (manufacturer’s website): 750W
Performance per Watt
Voltage and frequency curve
The graph below shows all combinations of GPU frequency and GPU voltage at a resolution of 1920×1080. Constructed points allow you to arrange them to indicate more commonly used values. Light colored markers mean that the frequency/voltage combination is rarely used, and darker dots more often.
|Clock rates and voltages|
|GPU frequency||Memory frequency||GPU Voltage|
|Idle|| 210 MHz
210 to 210 MHz
| 51 MHz
51 to 51 MHz
| 0.662 V
0.662 to 0.662 V
|Multi-Monitor|| 293 MHz
210 to 1830 MHz
| 108 MHz
51 to 1188 MHz
| 0.677 V
0.662 to 0.956 V
|Video Playback|| 210 MHz
210 to 255 MHz
| 51 MHz
51 to 101 MHz
| 0.662 V
0.662 to 0.662 V
|Furmark|| 1641 MHz
1620 to 1665 MHz
| 1188 MHz
1188 to 1188 MHz
| 0. 847 V
0.837 to 0.856 V
| 1954 MHz
1950 to 1965 MHz
| 1188 MHz
1188 to 1188 MHz
| 1.063 V
1.056 to 1.081 V
| 935 MHz
915 to 975 MHz
| 625 MHz
625 to 625 MHz
0.768 to 0.781V
| 1935 MHz
1830 to 1980 MHz
|1188 MHz|| 1.043 V
0.943 to 1.081 V
|Maximum overclocking (comparison)|
|Max. GPU Clock||Max. Memory Clock||Performance|
|Gigabyte RTX 3070 Ti Gaming OC||2023 MHz||1445 MHz||212. 6 FPS|
|MSI RTX 3070 Ti Suprim X||2102 MHz||1325 MHz||214.8 FPS|
|NVIDIA RTX 3070 Ti Founders Edition||2053 MHz||1410 MHz||211.7 FPS|
|Palit RTX 3070 Ti GamingPro OC||2114 MHz||1332 MHz||214.9 FPS|
|Zotac RTX 3070 Ti AMP Extreme Holo||2087 MHz||1435 MHz||217.8 FPS|
|Zotac RTX 3070 Ti AMP Holo||2055 MHz||1415 MHz||212.2 FPS|
Manual Overclocking Efficiency
Using the obtained clock speeds, a short Unigine Heaven test was run to evaluate the gain.
The actual 3D performance obtained with manual overclocking was 5.9%.
All NVIDIA graphics cards have a BIOS-defined power limit that limits power consumption by adjusting Boost frequencies accordingly. There is a second limit that defines the maximum TDP setting limit for custom overclocking; that is, how far the power slider will go. In the second chart, the value (+xx%) shows the percentage increase from the default power limit for the tested card to the maximum possible manual setting — the slider adjustment range.
The Gigabyte GeForce RTX 3070 Ti Gaming OC is currently available for around $1200.
- Excellent 1440p performance
- Most titles can also be played in 4K
- Factory overclock
- No efficiency loss from overclocking out of the box
- Fan stop at idle
- Silent (Silent BIOS)
- Good temperatures (default BIOS)
- Very low memory temperatures
- Faster GDDR6X memory
- 2nd Generation Hardware Accelerated Ray Tracing
- HDMI 2.1, AV1 decoding support
- Dual BIOS
- DLSS improved
- NVIDIA Reflex Low Latency Technology
- PCI Express 4. 0
- The actual market price is much higher than the recommended retail price
- Worse price/performance than RTX 3070
- Significant increase in power consumption compared to RTX 3070 without Ti
- Default BIOS fan settings are a bit loud
Gigabyte’s GeForce RTX 3070 Ti Gaming OC is the company’s «medium» RTX 3070 Ti, completely custom built. The model comes factory overclocked to a nominal boost frequency of 1830 MHz, which is roughly in the middle of what is currently available in terms of custom designs from all manufacturers. Gigabyte Gaming OC is 2% ahead of the Founders Edition and 7% ahead of the non-Ti RTX 3070 on average across a test suite of 22 games at 1440p.
Further, relative to the Radeon RX 6800, the gap is reduced to 2%; at 4K, the Gigabyte RTX 3070 Ti can match the non-XT Radeon RX 6800. The GeForce RTX 3080 is 13% faster than Gigabyte’s card, and the recently released RTX 3080 Ti is already 21% faster. The flagship GeForce RTX 2080 Ti of the latest generation is 11% behind the RTX 3070 Ti, and the difference with the RTX 2070 Super is 30%.
With these performance numbers, the RTX 3070 Ti is the perfect choice for the huge crowd of 1440p gamers, but the card also has enough power to drive many titles at 4K 60 FPS, especially if you’re willing to lower the settings a little. The RTX 3070 Ti is also a great choice for 1080p Full HD if you use a monitor with a high refresh rate of 120 or 144Hz. For 1080p at 60Hz, this is overkill unless next-gen games overdo it with their hardware requirements, which is unlikely. The ray tracing performance of the RTX 3070 Ti is better than the Radeon RX 6800 because NVIDIA does more ray tracing in hardware and uses second generation technology. However, the differences between the titles vary. Newer consoles are built with AMD RDNA2 technology, so in the future, game developers can invest more resources into RT optimization for AMD architecture, or simply reduce RT RT effects to mitigate the performance hit that recently happened with Resident Evil 7.
In previous reviews we have seen different versions of the current Gigabyte Gaming OC cooler: on RX 6700 XT Gaming OC, RX 6900 XT Gaming OC, RTX 3060 Ti Gaming OC and RTX 3070 Gaming OC. Although the design language is the same, using different shades of gray with metallic highlights, the RTX 3070 Ti Gaming OC still uses a new style that is less tall but longer and designed to let air through the card. Nearly all RTX 3070 Ti models, including the NVIDIA Founders Edition, have power connectors in the middle of the card, making cable routing difficult. Gigabyte, on the other hand, placed the power connectors on a separate module, so they are now located in the traditional upper right corner — I really liked it.
The heatsink base uses a large copper plate, unlike the direct contact heatpipes found on some other Gaming OC coolers. This approach definitely helps keep memory temperatures down, which has proven to be the best of the RTX 3070 Ti cards we’ve tested so far. GPU temperatures are good too, hitting just 68°C under load, the second-best result of any RTX 3070 Ti card tested.
However, the noise level is not that impressive; 39dB(A) roughly corresponds to the NVIDIA Founders Edition. Not bad, but could be better. The Gigabyte Gaming OC comes with a dual BIOS that makes it easy to enable silent mode. Noise levels are now MUCH better. With 32 dB(A) in quiet mode, the card is the second quietest RTX 3070 Ti in our test group.
Although Gigabyte’s cooler isn’t the strongest thermal solution, Gigabyte has very cleverly set up a quiet BIOS. While other manufacturers only lower the fan speed, Gigabyte has also lowered the power limit. It really does cost a percent or two of performance, which you’ll never notice, but it’s the secret to achieving significant fan noise reduction without significantly impacting temperatures. Lower power consumption means less heat dissipation, giving the cooler more room to breathe. Liked the big difference between the default BIOS and the silent BIOS, so there really is a meaningful choice.
The RTX 3070 Ti Founders Edition review talked about the increased power consumption of the RTX 3070 Ti and why. The bottom line is that the RTX 3070 Ti is not as power efficient as the non-Ti RTX 3070 and cannot match the AMD Radeon. You can usually expect factory overclocked custom designs to lose some extra power efficiency due to higher clock speeds and possible voltage increases.
This does not apply to Gigabyte Gaming OC. The card offers 2% faster performance and consumes 2% more power, resulting in the same power efficiency as the NVIDIA Founders Edition. Congratulations to Gigabyte for the smart improvements to their card that don’t require any efficiency — the other RTX 3070 Tis we’ve tested don’t do as well here. 9The 1039
RTX 3070 Ti is currently selling for around $1150, while the Gigabyte Gaming OC is selling for $1200. The $50 increase is quite reasonable as you get a very decent cooler with well thought out fan and power settings. The dual-BIOS feature really makes it possible, and the power connector is better than on other cards. Factory overclocking alone can’t justify a significant price increase, 2% is $25 even at a base price of $1200. In terms of price/performance, the strong competition is the RTX 3070, which costs around $1,050 — significantly cheaper, but only slightly slower.
The RTX 3070 is also more energy efficient than the RTX 3070 Ti. The AMD Radeon RX 6800 non-XT is currently $1200 which is very similar to the RTX 3070 Ti with similar performance with much better power efficiency but slower ray tracing. If you can reckon with lower overall performance but better price/performance ratio, the «only» $800 RX 6700 XT is a great option.
Thanks for reviewing and testing the W1zzard TPU.
Review and testing GeForce RTX 2080 Ti Founders Edition
The launch of a new line of GeForce RTX graphics cards has become the main event of recent months. The senior solution in the new series is the GeForce RTX 2080 Ti accelerator, which should demonstrate the full power of the Turing architecture. In this review, we will look at the features of the GeForce RTX 2080 Ti Founders Edition, check the overclocking potential, test in a number of games and compare with the GeForce RTX 2080 and GeForce GTX 1080 Ti.
Let’s briefly recall the main features of the GeForce RTX 2080 Ti. At the heart of the video card is a 12nm TU102 GPU. This GPU consists of six GPC clusters with 72 SMs, and each SM runs 64 CUDA cores. Since the four SMs are deactivated, the GeForce RTX 2080 Ti operates with 4352 CUDA cores. Also active are 544 tensor cores and 68 RT cores for tracing. Read more about the functional purpose of these blocks in a special review of the architecture. In the block diagram, for clarity, potentially deactivated blocks are indicated.
TITAN RTX offers a complete configuration of computing units, but this is traditionally an expensive product not for the mass market.
The GeForce RTX 2080 Ti has two frequency configurations. This is a base GPU frequency of 1350 MHz with a Boost Clock of 1545 MHz and a faster combination of frequencies of 1350/1635 for the Founders Edition versions. On board the video card is 11 GB of GDDR6 memory, the memory bus is 352 bits. The memory frequency is equivalent to 14000 MHz. Declared TDP at 250W for lower clocked graphics cards and 260W for Founders Edition.
Frequency differentiation conditional. There are no restrictions on the release and faster versions, of which there are already quite a few. It’s just that initially NVIDIA provides some advantage for video cards of its own production. And this is one of the factors that justifies the higher price tag of the Founders Edition compared to the simple variants of the GeForce RTX 2080 Ti.
It’s time to look directly at the video card itself. For the first time, NVIDIA has abandoned the usual «turbine» in reference products. All older GeForce RTX Founders Editions received a new cooler with two conventional fans and a large heatsink.
Video card length 26.5 cm. It occupies two expansion slots. Against the backdrop of the fact that many manufacturers have begun releasing the GeForce RTX 2080 Ti with three-slot cooling, this option has certain advantages. Very thick video cards will leave a minimum gap for airflow in SLI, and with two-slot it is more.
The back is completely covered with a metal backplate. This plate and silver panels on the front side of the case form a single streamlined design.
The name of the graphics card is printed on the back plate.
The combination of strict design and streamlined shapes gives a special personality. Visually, it looks interesting and stylish.
GeForce RTX’s large side lettering is traditionally backlit.
There is a SLI connector on the side. The new series uses the second-generation high-speed NVLink interface to combine video cards. The connector is present only on the GeForce RTX 2080 Ti and GeForce RTX 2080. By default, it is closed with a plug.
GeForce RTX are stripped of the old standard connectors and are equipped with only new interfaces: three DisplayPort, one HDMI and one VirtualLink in USB Type-C format. The video card supports DisplayPort 1.4a with the ability to transmit a picture of 8K format at 60 Hz. VirtualLink is focused on connecting VR systems through a single port.
The overall cooling design is reminiscent of previous NVIDIA solutions. At the base of the cooler is a massive black plate that removes heat from memory chips and power elements. In the GPU area there is a window for a heatsink with an evaporation chamber. There are many different types of thermal pads on the surface, there are greenish pads, white fibrous pads and a blue thick substance.
The choice of different thermal interfaces raises questions. This is probably due to the different distance between the base surface and the elements. For example, above the chokes, the base of the cooler has a rise. Therefore, thin pads are used in the center, and a thick, thick layer of blue interface is used on the sides.
Interestingly, the chokes themselves often do without direct cooling. In this case, the manufacturer took care of both the chokes and the cooling of any chip on the board, providing it with a thermal pad for contact with the cooler.
Disassembling the cooler itself is problematic. Some of the screws are hidden under the black panel between the fans where hot melt adhesive is used. To illustrate, let’s use an image from NVIDIA. It shows that the radiator is quite large, it occupies the entire area of the video card. Since the radiator is screwed to the base, they essentially form a single structure. The use of an additional thermal interface (its traces are visible during external examination) improves contact.
The backplate also acts as an additional heatsink. There are many thermal pads on the surface for contact with the elements on the board. At the same time, the inner surface is covered with a protective film, where there are windows strictly for these gaskets.
In the area under the chip and memory chips, gaskets provide contact with the board, removing heat from the textolite.
This cooling works well on the GeForce RTX 2080, providing low temperatures and low noise. The flagship graphics card is hotter because it uses the largest and most complex modern TU102 chip.
The PCB design has noticeably become more complicated compared to the GeForce GTX 1080 Ti, which is due to the strengthening of the power supply system.
13-phase core power is officially announced. In this case, we see six lines to the left of the chip (probably three channels with phase doubling) and 10 lines to the right.
The power system uses integrated DrMOS assemblies. External power is supplied through two 8-pin connectors.
Full processor marking TU102-300A-K1-A1. It is a massive 12nm chip with an area of 754mm².
There are 11 Micron 8PA77 D9WCW memory chips soldered on the board. The total amount of memory is 11 GB.
Base core clock 1350 MHz, Boost Clock 1635 MHz. The memory operates at an effective data rate of 14,000 MHz.
Testing was carried out on an open stand at 23-24 °C indoors. Under such conditions, the GPU warmed up to 82-84 °C. Operating frequencies varied widely and depended on the load and application. In the heavy gaming benchmark of Final Fantasy XV: Windows Edition in 4K mode, the core frequencies were mostly kept in the range of 1660-1700 MHz
The maximum fan speed in this test reached 2380 rpm, which is a bit high, but without serious hum.
In Strange Brigade, GPU clocks never dropped below 1750 MHz even at 4K at a peak temperature of 82°C. Peak Boost values up to 1965 MHz.
At the moment, the EVGA Precision X1 utility, which we used for monitoring and overclocking, provides the maximum control over the parameters of RTX-series video cards. With its help, you can flexibly configure the fan control mode with separate control of each fan. NVIDIA OC Sanner helps with overclocking, which automatically checks the potential of the core for the possibility of stable operation at higher frequencies. However, once the optimal value is found, additional stability tests will not hurt. And the scanner itself on one video card can produce values with a difference of 10-20 MHz.
After a series of tests, we settled on +165 MHz for the core. This is a base value of 1515 MHz with an average Boost Clock of 1800 MHz and peak values up to 2115 MHz. The final frequency of 15540 MHz has been reached for the memory.
When the power limit was increased to the maximum, the average frequency level in Final Fantasy XV was close to 1960 MHz.
In other games, the frequencies were closer to 1980 MHz. This is illustrated in the monitoring screenshot below during Gears of War 4 4K testing.
Such overclocking is associated with significant heating. The only way to compensate is by increasing the fan speed. We fixed them at 3000 rpm, which provided sufficient stability and temperatures below 80 ° C. Due to the high noise, this mode is not suitable for everyday use. Perhaps, when using headphones, noise does not play a critical role, but for such a powerful video card, overclocking will still require a replacement cooling, and CBO will be the best option.
Let’s compare the flagship with its closest friend in the face of the GeForce RTX 2080. This will be the ASUS DUAL-RTX2080-O8G video card, which is identical in terms of frequencies to the Founders Edition. The old generation is represented by the reference version of the GeForce GTX 1080 Ti, which also belongs to the Founders Edition line, but with absolutely standard frequencies.
Characteristics of test participants
GeForce RTX 2080 Ti Founders Edition
GeForce RTX 2080 Founders Edition
GeForce GTX 1080 Ti
Number of transistors, million pieces
Technical process, nm
Core area, sq. mm
Number of CUDA stream processors
Number of tensor cores
Number of cores RT
Number of texture units ROP
Number of render units
Core clock (Base/Boost), MHz
Memory bus, bit
Effective memory frequency, MHz
Memory capacity, GB