AMD Radeon HD 7950 3GB Review
Written by
Harry Butler
January 31, 2012 | 07:30
Companies: #amd
1 — AMD Radeon HD 7950 3GB Review2 — AMD Radeon HD 7950 3GB — The Card3 — Test Setup4 — AMD Radeon HD 7950 3GB Arma II Performance5 — AMD Radeon HD 7950 3GB Battlefield 3 Performance6 — AMD Radeon HD 7950 3GB Dirt 3 Performance7 — AMD Radeon HD 7950 3GB Skyrim Performance 8 — AMD Radeon HD 7950 3GB Power and Thermals9 — AMD Radeon HD 7950 3GB — Overclocking10 — AMD Radeon HD 7950 3GB — Performance Analysis11 — AMD Radeon HD 7950 3GB — Conclusion
Manufacturer: AMD
UK Price (as reviewed): approx. £350 (inc VAT)
US Price (as reviewed): approx. $550 (ex. tax)
Well, we suppose there was only so long AMD and Nvidia could keep making graphics cards before they had to start recycling names, although we suspect AMD’s latest addition to its line-up, the Radon HD 7950 3GB, might just have the performance advantage over Nvidia’s six year old GeForce 7950 GT 512MB.
Click to enlarge — We’ll be reviewing the one on the left, not the one on the right.
Jokes aside, the HD 7950 3GB is AMD’s second 28nm GPU to hit retailers, following the launch of the HD 7970 3GB on January 9th. We said in our original review of the HD 7970 3GB that its new Tahiti XT architecture, based around 64 separate compute units, was extremely well suited to incremental hardware scaling, and the HD 7950 3GB is the first taste of that.
Its GPU, code-named Tahiti Pro, has been understandably cut down considering the new card’s £350 price tag, but it’s hardly been a hack-and-slash job. AMD has simply disabled four of the card’s 64 compute units (AMD insists on called them GCNs, we don’t), thus reducing the GPU’s stream processor count from 2048 to 1792 and its texture unit count from 128 to 112.
Click to enlarge — The HD 7950 3GB and its Tahiti Pro architecture.
And, architecturally at least, that’s the only difference. Based on the same 28nm production process and with the same 4.3 billion transistor count, the HD 7950 3GB retains all of its bigger sibling’s 32 ROPs, all six of its 64-bit dual channel memory controllers, and an un-touched dual front-end engine. The HD 7950 3GB is, as with all AMD’s 7-series GPUs, also equipped with a PCI-E 3.0 interface. Initially available with 3GB of GDDR5, even the card’s memory is the same as its more prestigious sibling, although a lower price HD 7950 1.5GB will arrive soon to offer a cheaper alternative for those less concerned with ultra-high resolution or multi-monitor gaming.
Of course, there’s more to the HD 7950 3GB than the 12.5 per cent drop in stream processor count. AMD has also reduced the card’s stock speeds to match. While the HD 7970 3GB ships with a core frequency of 925MHz, the HD 7950 3GB’s core clock speed is only 800MHz, a 13.5 per cent reduction.
The memory frequency has also been reduced, from 1,375MHz (5.5GHz effective) to 1,250MHz (5GHz effective), a drop of nine per cent. We’re not too worried by the cuts though; we found the full-fat HD 7970 3GB to be roughly twenty per cent faster than Nvidia’s current fastest single GPU, the GTX 580 1.5GB, so a 12 per cent cut across the board should still see the HD 7950 3GB able to overhaul the GTX 580 1.5GB, and that’s before we get busy overclocking it.
1 — AMD Radeon HD 7950 3GB Review2 — AMD Radeon HD 7950 3GB — The Card3 — Test Setup4 — AMD Radeon HD 7950 3GB Arma II Performance5 — AMD Radeon HD 7950 3GB Battlefield 3 Performance6 — AMD Radeon HD 7950 3GB Dirt 3 Performance7 — AMD Radeon HD 7950 3GB Skyrim Performance 8 — AMD Radeon HD 7950 3GB Power and Thermals9 — AMD Radeon HD 7950 3GB — Overclocking10 — AMD Radeon HD 7950 3GB — Performance Analysis11 — AMD Radeon HD 7950 3GB — Conclusion
Gigabyte Radeon HD 7950 3GB specifications
Ultra Durable VGA
Features & Benefits
GPU Temperature 5%~10% down
Ultra Durable VGA board provides dramatic cooling effect on lowering both GPU and memory temperature by doubling the copper inner layer of PCB.
GIGABYTE Ultra Durable VGA can lower GPU temperature by 5% to 10%.
Overclocking Capability 10%~30% Up
Ultra Durable VGA board reduces voltage ripples in normal and transient state, thus effectively lowers noises and ensures higher overclocking capability. GIGABYTE Ultra Durable VGA graphic accelerators improve overclocking capability by 10% to 30%.
Power Switching Loss 10%~30% Down
Ultra Durable VGA board allows more bandwidth for electron passage and reduces circuit impedance. The less circuit impedance, the more stable flow of current and can effectively improve power efficiency. GIGABYTE Ultra Durable VGA can lower power switching loss by 10% to 30%.
«Triangle Cool» Technology
GIGABYTE introduces the latest exclusive «Triangle Cool» Technology to reach a better cooling performance. The latest patent pending technology combines fin with clip module in a special triangle shape.
With the original anti-turbulence structure plus the new triangle cooling design, it enhances the efficiency of heat dissipation dramatically by minimizing the flow of turbulence between fans. Therefore, the «Triangle Cool» Technology provides a more efficient air flow for the cooling system.
AMD Eyefinity
Supporting multiple independent display outputs simultaneously, AMD Eyefinity technology delivers innovative graphics display capabilities enabling massive desktop workspaces and ultra-immersive visual environments for the ultimate panoramic computing experience.
AMD EyeSpeed
AMD EyeSpeed technology helps you:
— Enjoy beautifully rich and clear video playback when streaming from the web;
— Take in your favorite movies in stunning, stutter-free HD quality;
— Run multiple applications smoothly at maximum speed;
— Enjoy lightning fast game play and realistic physics effects.
CrossFireX
Upgrade to even greater 3D performance quickly and easily thanks to plug-and-play ATI CrossFireX multi-GPU technology
Avivo HD
Avivocould provide vibrant high fidelity, flawlessly smooth next-generation video playback.
Avivo technology easily allows your PCs to become digital home media center by its compelling video pipeline design that comprises the latest high definition video hardware acceleration, including H.264 and WMV9.
Microsoft Window 7
Microsoft Windows 7 is the next generation operating system that will mark a dramatic improvement in the way the OS takes advantage of the graphics processing unit (GPU) to provide a more compelling user experience. By taking advangate of the GPU for both graphics and computing, Windows 7 will not only make todays’s PCs more visual and more interactive but also ensure that they have the speed and responsiveness customers want.
RoHS Compliant
As a citizen of the global village, GIGABYTE exert ourselves to be a pioneer in environment care. Give the whole of Earth a promise that our products do not contain any of the restricted substances in concentrations and applications banned by the RoHS Directive, and are capable of being worked on at the higher temperatures required for lead free solder.
One Earth and GIGABYTE Cares!
|
3DNews AMD Radeon HD 7950 video cards — testing, unlocking… The most interesting news
⇣ Contents ⇡#Introduction
The Radeon HD 7970 is a great achievement for AMD’s graphics division, a breakthrough. For the first time since the HD 2900, Radeon graphics cards received a new processor architecture — Graphics Core Next (GCN). Compared to the VLIW4 architecture used in the Radeon HD 69 cards00, it provides a more efficient use of resources when executing shader code and is incomparably better suited for performing non-graphical calculations by the GPU. We described in detail the principles of GCN operation in the review of Radeon HD 7970. In GCN, not only shader SIMD units are arranged in a completely different way, but also the geometry and tessellation processing mechanism has been completely redesigned, which is clearly seen in benchmarks and games that use this function. ⇡#XFX R7950 Double Dissipation Specifications, Package ContentsThe Radeon HD 7970 is not only a devilishly fast and technologically advanced card, but also the most expensive single-processor Radeon in the brand’s history: a suggested price of $549. Not surprisingly, for many who want to upgrade, the younger model with a slightly trimmed GPU (Tahiti Pro) is more relevant — the Radeon HD 7950. Let’s see how much the card’s specifications have suffered. Four Compute Units out of 32 are disabled in the Tahiti Pro core, which gives us 1792 ALUs (stream processors, in AMD terminology) versus 2048 in the Tahiti XT processor. The number of texture units has been reduced from 128 to 112. The memory bus width and memory size have remained untouched, although an alternative version of the HD 7950 with 1.5 GB of memory is rumored to appear in the future. There were conflicting reports about clock speeds prior to the official release of the HD 7950. Even now, AMD’s spec slide lists the GPU frequency as «800 and higher». It seems that card manufacturers will determine this parameter within wide limits. The memory, judging by the same slide, should operate at an effective frequency of 5000 MHz. TDP due to all cuts compared to HD 7970 dropped by 50 watts. We were sent for testing Radeon HD 7950 by XFX: model R7950 Double Dissipation. Her GPU and memory frequencies are increased to 900 and 5500 MHz, respectively. Nothing is known about the price of this and other HD 7950 cards yet. Some online stores even before the official start of sales began to accept pre-orders, at least for $481, and some offers reached the price of $685. This, of course, is a markup for novelty, in a month or two the prices will settle down at an acceptable level. The package includes a CD with software, a bunch of documentation and other pieces of paper, a CrossFireX bridge and an HDMI to DVI adapter. ⇡#DesignThe XFX R7950 DD uses a reference PCB, but the cooling system is original, with two fans. The reference Radeon HD 7950 has a cooler that looks different, and it also differs from that of the HD 7970: instead of a turbine, there is a classic fan with a flow direction perpendicular to the plane of the board. The GPU is cooled by a copper plate, while the memory chips and power components transfer heat to the massive aluminum cooler frame. Heat pipes are not used here, and it is not clear whether there is an evaporation chamber in the copper sole at all, so how much this design is better than the original HD 7950 cooling system, my grandmother also said in two. If it’s still better, then things are bad with reference cards. On the HD 7970, the cooler is much more powerful and heavy. One thing is good: at standard frequencies and with automatic fan control R79The 50 DD is pretty quiet, and even at full speed it doesn’t hit the ear. The card dispenses with the clamping cross on the GPU side. The textolite is dark brown, like on the Radeon HD 6950. The R7950 does not differ from the Radeon HD 7970 in terms of the set and location of the video outputs, XFX only inscribed its logo in the «exhaust» grille. ⇡#BoardLooking at the reference Radeon HD 7950 board, our hopes of turning the “cut” into a full-fledged HD 7970 faded. The arrangement of the elements is significantly different. Power requires two six-pin cables instead of six-pin and eight-pin. GPU voltage is regulated by the CHiL CHL8228G controller, which is used in both the HD 7970 and previous generation AMD cards. But to power the I/O buffers of the video memory, another chip is used — ANPEC APW7098, it is soldered on the back side of the board next to the CrossFireX connectors. This is a two-phase controller, and judging by the two inductors and two MOSFETs that are in the neighborhood, both phases are active. Memory chips are exactly the same as in the HD 7970 — Hynix H5GQ2h34MFR, with a nominal frequency of 1500 MHz (effective frequency — 6 GHz).
For testing, the official Radeon HD 7950 BIOS from AMD was sewn into the card, which set the GPU and video memory clock speeds to 800 and 1250 (5000) MHz, respectively. Another series of tests was carried out with the proprietary XFX BIOS and the corresponding frequencies.
The following video cards took part in testing as rivals of the novelty:
⇡#Overclocking, unlocking, temperature, power consumption AMD Overdrive limits overclocking to 1100 MHz for GPU and 1575 (6300) for memory. In the course of long experiments, we managed to achieve GPU and memory frequencies of 1150 and 1790 (7160) MHz, respectively. This means that both components have added 43-44% of the nominal frequencies that AMD has set! Moreover, the video memory again ran into the Overdrive limiter. For such an epic overclock, I had to raise the voltage on the GPU to 1.185 V and turn on the cooling system at full speed. The power consumption slider, of course, was pushed to the right by a maximum of 20%. When trying to squeeze another dozen or two megahertz out of the processor, graphic artifacts already appeared, and at higher voltages, the core overheated and benchmarks crashed. Temperature measurements under load showed that the XFX R7950 DD cooler is indeed less efficient than the reference HD 7970 cooling system. warm up less. ( Note: we were provided with two BIOSes from XFX, differing in the cooler mode, and in the tests we used the firmware with more aggressive settings. ) And the most intriguing question is unlocking disabled texture units and shader ALUs. It was possible to flash the BIOS from the reference Radeon HD 7970 into the HD 7950 without any problems, the card did not “bounce”, the frequencies increased to the appropriate level. But GPU-Z did not detect any changes in the core configuration, and the performance at the frequencies standard for the HD 7950 did not increase at all. In general, the freebie is canceled, you will have to be content with what you have. ⇡#Performance, synthetic tests 3DMark Vantage
3DMark 2011
Unigine Heaven 2
⇡#Performance, gaming tests Crysis 2
Aliens vs Predator
DiRT 3
Metro 2033
Mafia 2
Just Cause 2
Far Cry 2
⇡#Conclusions The main weapon of both cards of the Radeon HD 7900 family is an advanced process technology and high clock speeds. The stripped-down version of the flagship easily survived the loss of some of the computing units, and at close frequencies, the results of the HD 7950 and HD 7970 in most games differ slightly. Finally: if we compare the Radeon HD 7950 at the reference clock speeds (800/5000 MHz) with the GeForce GTX 580 (772/4004 MHz), it seems that the latter is something else! It even outperforms the new product in traditionally favorable conditions for the Fermi architecture, for example, with the active use of DirectX 11 tessellation.
⇣ Contents
If you notice an error, select it with the mouse and press CTRL+ENTER. Related materials Permanent URL: https://3dnews.ru/623769/page-1.html Tags: 1392MHz vs 810MHz 36.43 GPixel/s vs 25.9 GPixel/s 75W vs 200W 1752MHz vs 1250MHz 7008MHz vs 5000MHz  9 newer version of DirectX? 12 vs 11.1 4.6 vs 4.2 1518MHz vs 925MHz
Which comparisons are the most popular?Nvidia GeForce GTX 1050 vs Nvidia GeForce GTX 1650 Sapphire Radeon HD 7950 vs AMD Radeon RX 580 Nvidia GeForce GTX 1050 vs AMD Radeon RX Vega 8 Sapphire Radeon HD 7950 vs XFX Radeon RX 580 GTR XXX OC+ 8GB Nvidia GeForce GTX 1050 vs Nvidia GeForce RTX 3050 Laptop Sapphire Radeon HD 7950 vs Sapphire Radeon HD 7970 Dual-X Nvidia GeForce GTX 1050 VS AMD Radeon Vega 8 Sapphire Radeon HD 7950 VS NVIDIA GEFORCE GTX 760 NVIDIA GeForce GTX 1050 VS 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 9000 Sapphire Radeon HD 7950 vs AMD Radeon R9 270X Nvidia GeForce GTX 1050 vs Nvidia GeForce GTX 960 Sapphire Radeon HD 7950 vs Nvidia GeForce GTX 750 Ti Nvidia GeForce GTX 10507 VS NVIDIA GEFORCE GTX 1060 SAPPHire Radeon HD 7950 VS NVIDIA GTX 960 0 Sapphire Radeon HD 7950 vs AMD Radeon HD 7850 Nvidia GeForce GTX 1050 vs AMD Radeon RX 580 Sapphire Radeon HD 7950 vs AMD Radeon RX 580X Nvidia GeForce GTX 1050 vs Nvidia GeForce MX330 Price ComparisonUser ReviewsOverall RatingNvidia GeForce GTX 1050 2 User Reviews NVIDIA GeForce GTX 1050 Value for money 6. 2 votes No reviews yet 0007 2 votes reviews yet there is no performance 6.0 /10 2 Votes reviews yet there are no noiselessness 10.0 /10 9000 2 Votes 9000 Reliability 6.0 /10 2 votes No reviews yet 9000 Manufacturer7 clock speed1392MHz 810MHz The graphics processing unit (GPU) has a higher clock speed. turbo GPU 1518MHz 925MHz When the GPU is running below its limits, it can jump to a higher clock speed to increase performance. pixel rate 36.43 GPixel/s 25.9 GPixel/s The number of pixels that can be displayed on the screen every second. FLOPS 1.73 TFLOPS 2. FLOPS is a measure of GPU processing power. texture size 72.86 GTexels/s 90.7 GTexels/s The number of textured pixels that can be displayed on the screen every second. GPU memory speed 1752MHz 1250MHz Memory speed is one aspect that determines memory bandwidth. Shading patterns Shading units (or stream processors) are small processors in a graphics card that are responsible for processing various aspects of an image. texture units (TMUs) TMUs take texture units and map them to the geometric layout of the 3D scene. More TMUs generally means texture information is processed faster. ROPs ROPs are responsible for some of the final steps of the rendering process, such as writing the final pixel data to memory and for performing other tasks such as anti-aliasing to improve the appearance of graphics. Memoryeffective memory speed 7008MHz 5000MHz The effective memory clock is calculated from the size and data transfer rate of the memory. A higher clock speed can give better performance in games and other applications. maximum memory bandwidth 112.1GB/s 240GB/s This is the maximum rate at which data can be read from or stored in memory. VRAM (video RAM) is dedicated video memory. More VRAM usually allows you to run games at higher settings, especially for things like texture resolution. versions of GDDR memory Later versions of GDDR memory offer improvements such as higher data transfer rates, which improve performance. memory bus width 128bit 384bit Wider memory bus — this means it can carry more data per cycle. This is an important factor in memory performance, and therefore the overall performance of the graphics card. Supports memory troubleshooting code ✖Nvidia GeForce GTX 1050 ✖Sapphire Radeon HD 7950 Memory troubleshooting code can detect and fix data corruption. It is used when necessary to avoid distortion, such as in scientific computing or when starting a server. FeaturesDirectX version DirectX is used in games with a new version that supports better graphics. OpenGL version The newer version of OpenGL, the better graphics quality in games. version of OpenCL Some applications use OpenCL to use the power of the graphics processing unit (GPU) for non-graphical computing. Newer versions are more functional and better quality. Supports multi-monitor technology ✖Nvidia GeForce GTX 1050 ✔Sapphire Radeon HD 7950 The video card has the ability to connect multiple displays. GPU temperature at boot Lower temperature at boot — this means that the card generates less heat and the cooling system works better. supports ray tracing ✖Nvidia GeForce GTX 1050 ✖Sapphire Radeon HD 7950 Ray tracing is an advanced light rendering technique that provides more realistic lighting, shadows and reflections in games. Supports 3D ✔Nvidia GeForce GTX 1050 ✔Sapphire Radeon HD 7950 Allows you to view in 3D (if you have a 3D screen and glasses). supports DLSS ✖Nvidia GeForce GTX 1050 ✖Sapphire Radeon HD 7950 DLSS (Deep Learning Super Sampling) is an AI based scaling technology. This allows the graphics card to render games at lower resolutions and upscale them to higher resolutions with near-native visual quality and improved performance. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The GPU has increased the functional units and the graphics card components run at higher clock speeds compared to the Radeon HD 6970. There is support for the PCI-E 3.0 bus and many new features for 3D rendering and non-graphics computing. As a result, the Radeon HD 7970 not only easily copes with tasks that were difficult for previous generation AMD adapters, but can rightfully be considered the fastest single-processor graphics card today. And in some tests, the HD 7970 is close in performance to the “two-headed” Radeon HD 6990 and GeForce GTX 590 cards.70 is no better than the flagship model of the previous series, the HD 6970. And thanks to PowerTune technology, the GPU frequency can dynamically increase if the workload does not completely exhaust the TDP. The HD 7970 overclocks superbly. Let’s say more, in the previous review, we ran into the limits of clock frequencies that AMD Overdrive sets, and the card was completely stable without raising the voltage and increasing the speed of the cooling fan.
Looking ahead, we now have a tool to increase the Overdrive limit, and the Radeon HD 7950 we overclocked without any restrictions. Get ready for surprises. 
The box, which is very pleasing, is small, the contents are packed compactly and tightly. 
The CHiL controller still has seven phases. It turns out that in the Radeon HD 7950 is used one phase more than in the flagship card, although it is not yet clear which component got it. But who knows, maybe a new version of the HD 7970 with a nine-phase power system will soon appear, on just such a board. There is even a seat for the BIOS switch, which, as it were, hints. Is unlocking still possible? 
8 GHz (100×48) 
I. Texture Filter Quality 
Max. detail, high resolution textures, DirectX 11 
To go further, you need to hack Overdrive using third-party tools, such as the latest beta version of MSI’s Afterburner utility. It pushes the boundaries to 1620 MHz for GPU and 1790 (7160 effective) for video memory, and besides, it makes it possible to raise the processor supply voltage up to 1.3 V (by default it is 1.031 V). 
A more powerful cooler will certainly help to achieve new records, not to mention the liquid cooling system. 
The potential of Tahiti chips is such that overclocking the HD 7950 to the level of HD 7970 is no problem, and we will certainly see many models with a core frequency of 900 MHz and higher, and then there will be special overclocker versions with original PCBs and powerful cooling systems. 800 MHz is already quite poor for such a GPU. Well, if you overclock the card yourself, then without any hardware modifications and with a mediocre cooling system, you can easily achieve the level of performance that dual-processor adapters of the previous generation have in many games (Radeon HD 6990 and GeForce GTX 590). In general, from a practical point of view, the Radeon HD 7950 turned out to be even more successful than the HD 7970. more: the board should be designed for increased power consumption and the cooling system is more solid. We will check it in the next articles, stay with us. 
We look forward to seeing the first representatives of the Kepler architecture from NVIDIA. 
5 /10 
9 TFLOPS 
This allows you to set up multiple monitors at the same time to create a more immersive gaming experience, such as a wider field of view. 