RTX 2080 vs RX 5700 XT ⚙️
0x1011.635MH/s137W0.085MH/W
Allium9.434MH/s87W0.108MH/W
Argon2d-dyn260.09KH/s198W1.314KH/W
Argon2d-ninja—KH/s—W—KH/W
Argon2d2501046.225KH/s220W4.756KH/W
Argon2d409643.687KH/s201W0.217KH/W
Astralhash15.359MH/s145W0.106MH/W
Autolykos272.828MH/s158W0.461MH/W
BCD31.841MH/s112W0.284MH/W
BMW512—MH/s—W—MH/W
BeamHashII33. 06H/s167W0.198H/W
BeamHashIII27.85H/s215W0.13H/W
Blake (14r)4.629GH/s—W—GH/W
Blake (2b)—GH/s—W—GH/W
Blake (2s)6.448GH/s232W0.028GH/W
Blake (2s-Kadena)1.148GH/s102W0.011GH/W
C1134.105MH/s122W0.28MH/W
CNReverseWaltz1030.35H/s150W6.869H/W
Chukwa99.502KH/s219W0.454KH/W
Chukwa235.493KH/s219W0.162KH/W
ChukwaWRKZ162.
18KH/s—W—KH/W
Circcash—KH/s—W—KH/W
Cortex0.073H/s172W0H/W
CryptoNightAlloy455.92H/s—W—H/W
CryptoNightArto873.11H/s—W—H/W
CryptoNightConceal1584.6H/s112W14.148H/W
CryptoNightFast1500.83H/s138W10.876H/W
CryptoNightGPU2239.95H/s220W10.182H/W
CryptoNightHaven863.18H/s140W6.166H/W
CryptoNightHeavy966.1H/s141W6.852H/W
CryptoNightHeavyX420.
09H/s125W3.361H/W
CryptoNightLiteV71.673KH/s141W0.012KH/W
CryptoNightR837.7H/s157W5.336H/W
CryptoNightSaber854.62H/s121W7.063H/W
CryptoNightStelliteV41468.95H/s—W—H/W
CryptoNightStelliteV51611.52H/s—W—H/W
CryptoNightTalleo—H/s—W—H/W
CryptoNightTurtle6.571KH/s149W0.044KH/W
CryptoNightV70.831KH/s113W0.007KH/W
CryptoNightV8834.1H/s156W5.
347H/W
CryptoNightWOW795.88H/s—W—H/W
CryptoNightZLS1051.83H/s127W8.282H/W
Cuckaroo29S4.42H/s151W0.029H/W
Cuckaroo29b4.46H/s149W0.03H/W
CuckooBFC32.97H/s113W0.292H/W
CuckooCycle8.39H/s140W0.06H/W
Curvehash5581252H/s192W29069.021H/W
Darkcoin—H/s—W—H/W
Dedal10.795MH/s185W0.058MH/W
Eaglesong0.001GH/s197W0GH/W
Equihash0.
762KH/s—W—KH/W
Equihash(125,4)54.86H/s230W0.239H/W
Equihash(144,5)70H/s218W0.321H/W
Equihash(150,5)35.6H/s121W0.294H/W
Equihash(192,7)37.059H/s227W0.163H/W
Equihash(210,9)320H/s211W1.517H/W
Equihash(96,5)29.011KH/s156W0.186KH/W
Equihash+Scrypt32.68KH/s216W0.151KH/W
EquihashBTCZ68H/s158W0.43H/W
EquihashBTG70H/s209W0.
335H/W
EquihashSAFE69H/s157W0.439H/W
EquihashZEL72H/s—W—H/W
Etchash37.12MH/s196W0.189MH/W
Ethash37.273MH/s147W0.254MH/W
FiroPoW18.739MH/s153W0.122MH/W
GhostRider685H/s107W6.402H/W
Globalhash32.79MH/s90W0.364MH/W
HMQ17256.032MH/s171W0.035MH/W
Handshake—MH/s—W—MH/W
HeavyHash390.252MH/s240W1.
626MH/W
Hex11.983MH/s181W0.066MH/W
HoneyComb0.524MH/s52W0.01MH/W
Jeonghash8.99MH/s185W0.049MH/W
KAWPOW24.996MH/s224W0.112MH/W
KHeavyHash492.187MH/s194W2.537MH/W
KangarooTwelve—MH/s—W—MH/W
Keccak1.236GH/s213W0.006GH/W
Keccak-C1.213GH/s235W0.005GH/W
Lyra2REv273.13MH/s154W0.475MH/W
Lyra2REv367.465MH/s251W0.
269MH/W
Lyra2vc0ban72.276MH/s151W0.479MH/W
Lyra2z2.855MH/s108W0.026MH/W
MTP—MH/s—W—MH/W
MTP-TCR16.095MH/s240W0.067MH/W
NIST50.05GH/s216W0GH/W
NeoScrypt1.501MH/s210W0.007MH/W
Octopus47.59MH/s179W0.266MH/W
PHI161221.08MH/s187W0.113MH/W
PHI212.071MH/s—W—MH/W
Padihash0.262MH/s51W0.005MH/W
Pawelhash6.
541MH/s172W0.038MH/W
Phi5—MH/s—W—MH/W
ProgPow19.119MH/s156W0.123MH/W
ProgPowSERO16.347MH/s236W0.069MH/W
ProgPowVeil24.333MH/s135W0.18MH/W
ProgPowZ21.596MH/s157W0.138MH/W
Radiant427.086MH/s158W2.703MH/W
RandomKEVA1127.96H/s160W7.05H/W
RandomSFX1127.36H/s144W7.829H/W
RandomX1028.9H/s154W6.681H/W
SHA-256csm268.
966MH/s62W4.338MH/W
ScryptSIPC1.073KH/s—W—KH/W
Skein1.066GH/s—W—GH/W
Skein2525.855MH/s101W5.206MH/W
Skunkhash28.618MH/s156W0.183MH/W
SonoA1.757MH/s185W0.009MH/W
Tellor—MH/s—W—MH/W
Tensority0.01KH/s226W0KH/W
TimeTravel1045.612MH/s184W0.248MH/W
Tribus105.004MH/s178W0.59MH/W
Ubqhash36.
958MH/s166W0.223MH/W
Verthash—MH/s—W—MH/W
VerusHash—MH/s—W—MH/W
X11k2.257MH/s151W0.015MH/W
X1311.907MH/s—W—MH/W
X1511.208MH/s191W0.059MH/W
X16R14.449MH/s178W0.081MH/W
X16RT11.328MH/s190W0.06MH/W
X16RTVEIL10.552MH/s146W0.072MH/W
X16Rv232.366MH/s158W0.205MH/W
X16S25.274MH/s122W0.
207MH/W
X1711.303MH/s188W0.06MH/W
X17R10.816MH/s188W0.058MH/W
X180.262MH/s62W0.004MH/W
X21S8.065MH/s121W0.067MH/W
X22i0.524MH/s51W0.01MH/W
X25X5.976MH/s214W0.028MH/W
X3316.49MH/s172W0.096MH/W
Xevan3.791MH/s200W0.019MH/W
Zhash72H/s240W0.3H/W
cuckARoo298.93H/s195W0.046H/W
cuckARood292.
06H/s174W0.012H/W
cuckARoom293.94H/s159W0.025H/W
cuckARooz295.57H/s195W0.029H/W
cuckAToo310.57H/s157W0.004H/W
cuckAToo320.5H/s210W0.002H/W
vProgPow10.202MH/s122W0.084MH/W
AMD Radeon RX 5700 XT – Review and Benchmarks
The day we’ve all been waiting for has finally arrived. We can fully reveal the performance of AMD’s next-gen 7nm Navi GPU(s), and you can find out if it’s worth switching camps from Nvidia, or upgrading your older GPU. As a quick catch-up, this is the first GPU for AMD that uses an all-new architecture called RDNA, and it’s also the first midrange consumer GPU built on a 7nm process, so it’s quite a big deal.
A second-generation (or third?) variant of this chip will be powering both next-gen consoles, so all eyes are on AMD here.Though AMD announced that its new GPU was indeed faster than the RTX 2070 when it was unveiled at Computex, perennial Clever Girl Nvidia just released the RTX 2070 Super less than a week ago in order to ruin AMD’s launch plans. The fallout from that was a last minute price drop for the Radeon cards, though the Radeon GM tweeted it was all a big setup on their part, as they inflated the prices initially and Nvidia took the bait. We’ll never know if that really was the plan all along, but either way the GPU wars are officially back on in a very big way.
AMD Radeon RX 5700 XT — Design and Features
Everything about this GPU is new, from the blower cooler design to the silicon that rests beneath it, so let’s tackle it piece by piece. I also covered most of this from the big reveal at E3, but I’ll do it again for those who weren’t following along. The biggest news with this GPU is that it’s the first consumer GPU built on a 7nm process.
There was a 7nm GPU before, the Radeon VII, but with 16GB of HBM memory it was kind of a hybrid gaming/workstation GPU, whereas the new Radeons are pure gaming hardware, with none of that expensive high-bandwidth memory but GDDR6 instead. Nvidia’s GPUs are currently built on a 12nm process, so by going with a smaller node AMD can put more transistors in the same space with better efficiency. Before we get too far along, let’s have a look at the spec sheet:
As you can see, AMD was specifically targeting the formerly $500 Nvidia RTX 2070 with this GPU, but it now has to contend with the RTX 2070 Super. Thankfully, AMD lowered the price of the 5700 XT to just $399 before launch, so it’s got a $100 price tag advantage over Nvidia’s latest GPU.What’s most striking about the RX 5700 is it uses a die that’s just 251mm-squared, and is going up against the TU104 chip from Nvidia (in the RTX 2070 Super), which is 545mm squared. The TU106 chip used in the RTX 2060 is 445mm squared, so AMD could still theoretically make a much, much larger chip down the road.
The Radeon RX 5700 XT is the «flagship» GPU of this series, though I wouldn’t be surprised to see bigger chips coming out with the RX 5800 and RX 5900 moniker in the future, but we’ll have to wait and see on that one. For now, the RX 5700 XT is a mid-high-end chip designed for 1440p gaming, and at $400 it’s the same price as the recently launched RTX 2060 Super.
Despite its midrange-ish status, and the fact that the actual silicon inside is quite small in comparison to Nvidia’s chips, the 5700 XT is a full-sized GPU, measuring almost 11 inches long like other high-end GPUs. For reference, the RTX 2060 is just nine inches long. The GPU uses a massive blower-style cooler that exhausts heat outside of the chassis. Though most people don’t like this kind of design as it just looks «cooler» to have multiple fans on a card, it is an effective way to make sure the hot air from the GPU doesn’t waft up into your CPU cooling apparatus. It’s usually not the most effective cooling design, however, but we’ll have boards from AIBs as well so don’t fret.
AMD has loaded this GPU up with 2,506 streaming processors, which is the same amount of CUDA cores found in the RTX 2070 Super, but 15 percent more cores than on the RTX 2060 Super. The memory subsystems on all these GPUs are all pretty much exactly the same; 8GB of GDDR6 riding on a 256-bit bus, giving them 448GB/s of memory bandwidth.
On the clock speed front the XT card has a base clock of 1,605Mhz and a «game» clock of 1,755MHz, which AMD says is the average clock speed it will operate at while gaming. It also lists a boost clock of 1,905Mhz.
As the first GPU with RDNA architecture to roll off the assembly line, AMD points out that this is an entirely new animal compared to its previous GCN architecture. The company claims its improved performance-per-clock by 1.25X, and performance-per-watt by 1.5X. In the past, GCN cards weren’t terrible, but they always consumed more power and ran much hotter than Nvidia cards, so it’ll be interesting to see how that pans out this time around.
This is also the world’s first PCI Express 4.0 GPU, which offers double the bandwidth of PCI Express 3.0, though to take advantage of that you’ll need a motherboard that supports it, which are coming soon with AMD’s new Ryzen CPUs. I do not believe the extra bandwidth afforded in PCIe 4.0 will help with gaming, but AMD does say in its press materials it will benefit things like media encoding and playback for 8K files, so we’ll have to wait and see.
One other notable addition to this GPU is it supports DisplayPort 1.4 with Display Stream Compression, which allows for combinations of refresh rate and display output that’s not possible with other GPUs. For example, you could run an 8K panel at 60Hz with HDR, or run a 4K panel at 144Hz with HDR enabled and no color loss. With today’s GPU and displays, you’d need to run 8-bit color or lower the refresh rate in these scenarios as there’s not enough bandwidth available.
AMD Radeon RX 5700 XT — New Technologies
Along with the RX 5700 series AMD is launching several new technologies; some for gamers, and some for game designers.
First up is Radeon Image Sharpening (RIS), which is based on a contrast-adaptive sharpening algorithm that doesn’t require developer involvement. It’s just a setting you can flip on or off in the Radeon software, and it should help make overall image quality better but it’ll probably be like Nvidia DLSS where you’d have to see it side-by-side to really notice it, in my opinion. RIS only works on games that use DX9, DX12, or Vulkan APIs, and HDR is also not supported. It’s recommend to be used at lower resolutions and in combination with AMD’s GPU Scaling feature, so you’d run a game at 2560 x 1440 on a 4K display, then the software would upscale it to the native resolution (4K) but you’d get 1440p performance, theoretically without any image quality penalty.
Next up we have Radeon Anti-Lag, which is designed to reduce input lag. AMD claims that by smoothing out the pacing by which the CPU is feeding the GPU frames to work on, it can theoretically reduce input lag by up to 16ms, which is roughly equivalent to one frame in a 60fps scenario.
AMD notes that this can have an impact on frames per second, but the tradeoff in responsiveness is worth it. AMD says it’s for GPU-limited scenarios, so we’re mostly talking about 4k scenarios as opposed to 1080p where you might be CPU-limited. This is also a user setting that can be turned on or off.
AMD Radeon RX 5700 XT — Benchmarks
To test the RX 5700 XT I strapped it into the IGN test bench, which consists of an Intel Core i7-7700K CPU, an Asus Prime Z270 motherboard, 16GB of DDR4 memory from Corsair, an EVGA PSU, and a SanDisk SATA SSD. Games were run at the three most common resolutions, and comparisons to all the Nvidia cards are included here as well. All tests were run at «Ultra» settings with anti-aliasing disabled.
First off, the RX 5700 XT does very well against the $100 more expensive RTX 2070 Super, though in some games the RTX card trounces the Radeon. It’s a back and forth battle, with each card taking decisive victories in certain games.
However, both cards are really designed for 1440p gaming, so in that category, the advantage goes to Nvidia in terms of performance, the RTX 2070 Super is faster in four of the seven tests, but in the three tests where AMD is faster it’s only by a slight margin.
That said, it’s not $100 more expensive faster, as the Radeon card is over 60fps in every game except Shadow of the Tomb Raider, which oddly refuses to run on Nvidia cards nowadays (the benchmark runs then crashes to desktop before delivering a result). And still, even in that game, the Radeon card is at 58fps, so it’s a very good GPU for that midrange 1440p resolution, and is over 80fps in most tests. So even though the Nvidia card is faster, the Radeon is a better value.
Now, when compared to the RTX 2060 Super, which is $50 more expensive, it’s a similar scenario to the RTX 2070 at 1440p where in some games the Radeon is clearly faster, but in others, the RTX card takes the crown. They are quite evenly matched at 1440p, so given the price disparity, it’s advantage Radeon once again.
At 1080p things look even better for AMD, as it was able to outpace the RTX 2060 Super by wide margins in four out of six tests.
Overall, AMD doesn’t dominate across-the-board here, but it’s right in the mix and its cost advantage over the Nvidia cards does deliver more performance for the dollar compared to Nvidia. Whether it’s the RTX 2070 or 2060 Super, the AMD card is extremely competitive, and also a great GPU for 1440p gaming just in general.
Also the vanilla RX 5700 is quite close to the XT card at higher resolutions, so that might be the biggest competition to the XT card since it’s $50 less expensive.
I also tested Radeon Image Sharpening by running Heaven at 2560 x 1440 on a 4K panel, with GPU scaling enabled. To my eyes it looked indistinguishable from a native 4k image, but this is one of those things where you will have to have a side-by-side comparison of the exact image and really stare at the details to notice them. In a moving game it looks just fine to me, so it theoretically does allow for «better» performance at high resolution, simply by running a lower resolution.
Finally, I tested Anti-Lag at 4K but honestly couldn’t really tell the difference. This is one of those features that will have to be measured with a high-speed camera or something because to my fingers and eyeballs I couldn’t detect a difference with it on or off.
AMD Radeon RX 5700 XT – Thermals and Overclocking
Here’s where things take a turn for worse for the red team, because though AMD seems to have made some progress on the thermal front they are still way behind Nvidia, at least in reference card trim. The Radeon 5700 XT runs pretty hot, hitting around 82C under full load, and you can certainly hear the fan spinning as well. Every Turing card I’ve tested, and the Pascal cards too, have all been silent, and rarely ran at temps over 70C or so, even the Founder’s Edition.
The Radeon 5700 XT runs pretty hot, hitting around 82C under full load
“
I didn’t have much luck using MSI Afterburner to overclock, as the card didn’t seem to be responding to my inputs most of the time, so I used the Wattman software’s «auto» overclocking tool.
The software told me the ideal GPU clock was 2130MHz, but left to its own devices it was just able to get up to 2057MHz, which is a pretty big deal since no GCN card was ever capable of doing that (if my memory serves me correctly). The only caveat here is there’s some fan noise to accompany this extra performance, but a pair of headphones gets rid of that. In the screenshot below the Afterburner software wouldn’t let me increase the power limit beyond what it’s set at due to the auto-tuning so save your tweets.
For what it’s worth, when overclocked I was able to see a 7.5% boost in Heaven 4.0, but the fan can be heard pretty clearly. My advice: just let it overclock itself and enjoy the (near ) silence.Hopefully, this thermal issue is just with the reference design, and AIB cards will be much quieter at higher clocks. All that said, if you don’t overclock the card and just let it run under full load, the final numbers I recorded were a clock speed of 1897MHz, and a temperature of 82C.