AMD Radeon HD 7990 Review: 7990 Gets Official
by Ryan Smith on April 24, 2013 12:01 AM EST- Posted in
- GPUs
- AMD
- Radeon
- Radeon HD 7000
- Tahiti
Compute
As always we'll start with our DirectCompute game example, Civilization V, which uses DirectCompute to decompress textures on the fly. Civ V includes a sub-benchmark that exclusively tests the speed of their texture decompression algorithm by repeatedly decompressing the textures required for one of the game’s leader scenes. While DirectCompute is used in many games, this is one of the only games with a benchmark that can isolate the use of DirectCompute and its resulting performance.
Civ V's texture compression routines are technically mutli-GPU capable, but multi-GPU scaling has never been particularly impressive here. So this test mostly reinforces what we already know about the Tahiti GPU being very capable in most DirectCompute workloads.
Our next benchmark is LuxMark2.0, the official benchmark of SmallLuxGPU 2.0. SmallLuxGPU is an OpenCL accelerated ray tracer that is part of the larger LuxRender suite. Ray tracing has become a stronghold for GPUs in recent years as ray tracing maps well to GPU pipelines, allowing artists to render scenes much more quickly than with CPUs alone.
The 7990 isn’t billed as a compute product, but that doesn’t mean it’s at all weak at compute. On the contrary, as LuxMark doesn’t hit the ROPs hard the 7990 has no trouble staying under its 375W target, allowing it to sustain 1000MHz indefinitely. As a result the 7990 takes AMD’s compute advantage and runs with it. The 7990 is a bit more 2x the cost of a 680, but it’s 8.5x faster. Even against GTX Titan the difference is just short of 4x; NVIDIA simply doesn’t do well in our OpenCL tests.
Our 3rd benchmark set comes from CLBenchmark 1.1. CLBenchmark contains a number of subtests; we’re focusing on the most practical of them, the computer vision test and the fluid simulation test. The former being a useful proxy for computer imaging tasks where systems are required to parse images and identify features (e.g. humans), while fluid simulations are common in professional graphics work and games alike.
These two CLBenchmark sub-tests aren’t multi-GPU capable, so the performance of the 7990 is essentially dictated by its first GPU. All that means however is that the 7990 is once again at the top of the charts, well ahead of NVIIDA’s other cards and beating Titan by 50%-100%. At the same time this is a good reminder that not every compute task scales well across multiple GPUs, which is why single-GPU products still have a strong place in the GPU compute world.
Moving on, our 4th compute benchmark is FAHBench, the official Folding @ Home benchmark. Folding @ Home is the popular Stanford-backed research and distributed computing initiative that has work distributed to millions of volunteer computers over the internet, each of which is responsible for a tiny slice of a protein folding simulation. FAHBench can test both single precision and double precision floating point performance, with single precision being the most useful metric for most consumer cards due to their low double precision performance. Each precision has two modes, explicit and implicit, the difference being whether water atoms are included in the simulation, which adds quite a bit of work and overhead. This is another OpenCL test, as Folding @ Home is moving exclusively OpenCL this year with FAHCore 17.
Only FAHBench’s explicit mode is multi-GPU capable, and even then the scaling to multiple GPUs isn’t great. Still, it’s enough to help the 7990 take the top spot on this benchmark, even with NVIDIA’s latest drivers slightly closing the gap. What’s particularly interesting here though is that the 7990 is faster than the 7970GE CF, and that’s not a fluke in our results. The 7990 should by all means be at least a bit slower, and more if throttling kicks in. It looks like we’re seeing one of those rare cases where the GPUs are benefitting from the presence of the PLX bridge, as going through the relatively close-by bridge is faster than in a two-card setup where the GPUs would have to communicate through the CPU/Northbridge. However this is the only time we see such an advantage; in most other compute scenarios – let alone gaming – the PLX bridge won’t have any kind of notable impact.
Wrapping things up, our final compute benchmark is an in-house project developed by our very own Dr. Ian Cutress. SystemCompute is our first C++ AMP benchmark, utilizing Microsoft’s simple C++ extensions to allow the easy use of GPU computing in C++ programs. SystemCompute in turn is a collection of benchmarks for several different fundamental compute algorithms, as described in this previous article, with the final score represented in points. DirectCompute is the compute backend for C++ AMP on Windows, so this forms our other DirectCompute test.
SystemCompute isn’t multi-GPU capable, so once again we’re leaning on the 7990’s first GPU. To that end we find the 7990 in second place, but we also see the 7790 clearly trailing the 7970GE by more than we’ve seen in our other compute benchmarks. SystemCompute does do a lot of I/O, so if FAHBench is the ideal case for showcasing the benefits of the PLX bridge in GPU to GPU I/O, then SystemCompute is good case for showcasing the drawbacks of the PLX bridge, mainly the higher I/O latency. It’s not enough to cripple the 7990 – it’s faster than the GTX Titan even here – but it does cost the 7990 some performance.
Facebook
Twitter
RSS
91 Comments
View All Comments
TheJian - Saturday, April 27, 2013 - link
I said a week or two ago you'd find an excuse to leave out fcat...LOL. You did. Even come up with a problem nobody else has, and show a large boost to AMD that places like tomshardware don't show while showing NV cards slower than their benchmarks in the exact same games and resolutions.Just like the percent gains in the driver post a few days ago (as I posted all of your %'s were lower than NV's own page which I linked to), you have NV scoring lower than other reviews.
http://www.anandtech.com/show/6913/nvidia-r319-ser...
Read the last comment people...They seem to lie a lot here and it's not surprising AMD chose to talk ONLY to this site recently and avoided access even to these guys to the driver team working on the "ISSUES" AMD has.
Heck toms last page title is this:
"Can The World’s Best Bundle Save Radeon HD 7990?"
ROFL...That pretty much sums up what they found...AMD sucks currently when FCAT is thrown in the picture.
But even leaving that out they put 5 gamers in a pepsi/coke style blind play test. ALL 5 CHOSE NV 690 gtx as the better card. They even note both Titan and 690GTX gave a better gaming experience.
It's funny you didn't mention what tom's found with noise:
"It’s only unfortunate that power-related vibrations generate more volume than the coolers themselves."
hmmm....
I guess you guys will keep claiming something about fcat problems until AMD solves their issues just as I've said before. This site is really losing credibility.
http://www.tomshardware.com/reviews/radeon-hd-7990...
http://www.pcper.com/reviews/Graphics-Cards/AMD-Ra...
"Perhaps they are using multiple displays for Eyefinity/Surround setups - in that case, the HD 7990 has different but equally traumatic problems. In our testing, nearly every other frame generated by all of our games tested are dropped and never shown to the gamer, resulting in frame rates at about half of what they should be and half of what is being reported by some other testing methods. That's just not acceptable. We are working on another story that directly compares the issues of Eyefinity and Surround that we hope to have up by next week along with some videos to demonstrate those complicated issues. "
Other people draw the exact OPPOSITE conclusions from this site's conclusions...LOL. “traumatic problems” in multi-monitor setups is a pretty harsh thing to say correct?
But then you come up with this zinger:
"As we’ve seen in the past AMD does a bit better than NVIDIA at multi-monitor resolutions, so while the 7990 trails the GTX 690 at 2560, it immediately jumps ahead at 5760."
From your shogun page...How much does AMD pay you guys to say complete opposites of other review sites using FCAT results? Are they completely funding your site at this point or what?
"It’s not a problem at this moment, but 2GB at multi-monitor resolutions in particular is going to be a wall very soon, and 2560 may not be too far behind. 2GB per GPU was reasonable in 2012, but now it’s a year later and things change. So for as little as we can predict something to be future-proof, AMD certainly seems to have an edge."
Yeah, get back to me when you're not making up stuff that isn't happening. As toms 5 gamers showed, NV was better all around and all 5 chose NV without knowing which was which. That kind of data matched their FCAT results exactly, even with the beta2 latest driver and the Prototype driver that won't be out until who knows when. After Q2 isn't specific...that could mean 2014. While the prototype driver helped at toms they still had issues and the gamers still chose NV even after testing again with prototype.
“But when we combine the quantitative data enabled by video capture-based performance analysis and the subjective judgments of a panel of gaming enthusiasts who simply want to play their favorite titles on the best hardware possible, Nvidia’s thousand-dollar GeForce GTX 690 outshines the similarly-priced Radeon HD 7990. Our early look at AMD’s prototype driver suggests that more evenly pacing the rate at which frames are shown on-screen helps minimize frame time variance, which our gamers definitely noticed. But that release isn’t expected for months—the second half of 2013 is as specific as AMD gets.”
Also, you tested all the way to 5760...Memory isn't a problem and won't be for a long time (don’t forget the titan has 6GB!). They don't program games for cards that barely exist. I'm guessing this is a limited run just like asus' card only had 1000, unlike NV's first titan run which was 100,000 and is sold out still today after more runs...LOL.
Without the games as incentive toms gives this on the conclusion page:
“And so we’re faced with a card that represents a huge improvement over its predecessor, but still comes up shy of its competition, and is priced like an equal.”
The games get old or I already have them. But I have to live with the card itself (especially at these prices) for probably 4-5 years or maybe more.
“For buyers that have faith in multi-GPU scaling and AMD’s ability to deliver timely profile updates the 7990 can and will handily surpass Titan by about 20%”
Umm…Have you read Hardocp’s driver review of AMD’s situation for the last year? They were 6 months late on some game optimizations and it took them a year to get drivers even near right and Enduro still sucks according to notebookcheck’s 7970 rehash review.
http://www.hardocp.com/article/2013/03/04/2012_amd...
They paint a pretty POOR driver picture for a year for AMD. While you’re there read their review of the 7990 which also points to multimonitor problems:
“We have sent our experiences to AMD, and hopefully there will be a fix in the future and then we can bring you TressFX performance under Eyefinity and CrossFire.”
LOL. But you think it’s all good with multi-monitor’s and think the memory future proofs it eh ryan? Everyone else thinks it sucks and has issues. Comically they found a slide showing AMD claiming to be able to do 3840x2160 in tombraider. They couldn’t get it to run there at all:
“AMD is claiming playable performance in Tomb Raider with TressFX enabled at a resolution of 3840x2160. We wanted to test this ourselves, and found where Tomb Raider stores its graphics settings in the registry. We attempted to force 3840x2160, unfortunately the game would not start at this resolution.” Followed by this little gem, “Therefore, we find it highly unlikely that AMD's claim is valid.” Well ok then. They made the same claim for crisis 3…Which again Hardocp attempted to prove (right or wrong):
“Above is our run-through at "max settings" at 3840x2160. You can clearly see that performance is very poor on the AMD Radeon HD 7990. The minimum framerate is 10 FPS, and the average is 19.4 FPS, well under 30 FPS. This level of performance is not playable in Crysis 3.”…Well ok then…Another lie shot down eh? “This claim is clearly invalid.”
Yeah, we hear you loud and clear. Two AMD claims, two AMD lies. It didn’t even work with 1.3mil pixels less res of 5760x1200 for them. They got below 30fps.
Farcry note: “GeForce GTX 680 SLI has the lowest framerates, but the smoothest overall experience thanks to SLI's non-stutter gameplay compared to CrossFire in this particular title.”
Yeah, that’s kind of what tomshardware and their 5 game players said. No stutter=BETTER gameplay.
And when they found your memory claim kicking in:
“Hitman is sensitive to VRAM capacity. In this case, the AMD Radeon HD 7990 and Radeon HD 7970 GHz Edition CrossFire have the advantage thanks to the larger 3GB of VRAM per GPU. GeForce GTX 680 SLI only has 2GB per GPU.”
“The drops in performance make gameplay not possible because it hitches and lags behind as you are playing.”
So when your situation actually happens, it isn’t playable already anyway…LOL Not the minimums on all cards were under 3fps…LOL. That’s not a type man, it’s they all scored 1, 2 & 3 fps…ROFL. But you keep trying to help AMD any way you can ryan…They need it…LOL.
Their conclusion page:
“Hopefully after reading this evaluation you've come to the same question we have, "Where's the missing performance?"….So the card sucks then there too? They even manually turned up powertune 20% and got the same results. And again another site saying it: “Quite simply, AMD CrossFire is the stutterer in the room.” Yeah we know the 5 gamers said the same at toms. His 680 GTX experience however is described as this on the same page:
“I was in awe. It felt smooth, really smooth, it felt like it was running at a faster framerate than it really was.”…Yeah kind of why the 5 gamers at toms picked it too. He calls the 7990 a “BAD VALUE”.
Again, 80 pages of complaints at notebookcheck’s forums don’t lie either. People need to take all the AMD driver complaints into major consideration. Who here thinks AMD’s drivers are great and on time? We’ll be waiting months for them to fix a problem they’ve had for a year+ and enduro has had issues for as long also (notebookcheck’s first review was awful and the second fixed some but not everything, they still say it has a ways to go compared to NV). Just google 7970m & notebookcheck you’ll get there people. It’s laughable hardocp crashed trying to prove AMD’s driver claims.
PCPER’s review conclusion page says the same about drivers:
“With its performance completely dependent on CrossFire technology, the HD 7990 as a $1000 graphics card has a very hard time justifying its price. With our early testing of the Catalyst prototype driver showing positive results though, there is yet hope for CrossFire to be fixed in this generation, at least for single monitor users! But until that driver is perfected, is bug free and is presented to buyers as a made-for-primetime solution, I just cannot recommend an investment this large on the Radeon HD 7990.”
LOL…HOPE…At least for SINGLE monitor users. Ryan Shrout says TITAN for multi-monitor and 690GTX for single monitor. It seems everyone says quite the opposite of you Ryan Smith. I could keep going but hopefully people get the point. NOBODY sees it like anandtech.
Check out Techpowerup’s review also people…Shows a TON of games these guys leave out. Skyrim, Assassins Creed 3, F1 2012, Borderlands 2, StarCraft2 Heart of the Swarm, Diablo3, COD Black ops2, and World of Warcraft (awful scores for radeons on WOW). The 690GTX basically wins everything at nearly ever res. Guess that’s why NONE of these are tested at anandtech even though they are all HUGE sellers compared to crap like Warhead, Dirt Showdown. Nobody PLAYS Warhead based on server checks, and which nobody bought in the case of showdown. Makes you wonder why these two games get tested here don’t it people? Why leave out multi-million units popular sellers like these that they leave out? It’s because you’d get a very clear picture of why NV owns 65% of the gaming market. Overall in their 20 or so games Titan was 4% faster and GTX 690 was 14% faster.
The negatives they state?:
• High price
• Annoying coil noise
• Requires CrossFire game support to reach proper performance
• Many games do not scale properly in CrossFire
• Very high multi-monitor and Blu-ray power consumption
• Slim performance gains from overclocking
• Long card
Again with the COIL NOISE and of course they note all the games with crossfire issues in their review.:
“Out of eighteen titles, five did not scale, or worse, showed negative scaling. These are not small titles, but big AAA games: Assassin's Creed 3, Batman: Arkham City, F1 2012, StarCraft II, Skyrim, and World of Warcraft. What really surprises me is that this long list is the same as the one we had with our reviews of HD 7990 "New Zealand" implementations by board partners, like the ASUS ROG ARES II and PowerColor Devil 13. So either AMD does not care or can't fix CrossFire support with these games millions of people play.”
Jeez…5 out of 18 games running like CRAP and they’re all AAA titles. Is this why they are all OUT of Anandtech’s gaming suite ryan? Again I ask, how much is AMD paying you guys or do you just really love them that much?
“What is a major issue, though, is the extremely annoying coil whine the card emits as soon as it runs a 3D application.”
He notes the 690 GTX and Titan have no such issues. Also notes it OVERPOWERS the fans just as Toms said. Forget to mention this did you ryan?
“In order to overcome the frametime issues some of our colleagues reported, AMD is working on a new driver to improve things, and has provided us with it, but it's only for Windows 8 and sacrifices some performance for more constant frame delivery.”
So win7 users won’t even be expecting this HOPE driver then? OF course it slows things down also. I hope you’ll be benchmarking it rather than just telling us how AMD smoothed out play…But I doubt it. You’ll forget to mention what they already found…IT’S SLOWER WITH THE HOPE DRIVER. Nuff said I guess...Read other reviews people and judge Ryan Smith yourself.
blackoctagon - Thursday, May 2, 2013 - link
Um...you seem to have missed the fact that Ryan didn't exactly give the 7990 a glowing review mate.TheJian - Friday, May 3, 2013 - link
“For buyers that have faith in multi-GPU scaling and AMD’s ability to deliver timely profile updates the 7990 can and will handily surpass Titan by about 20%”That doesn't sound like he'd go against it as all the others I've shown flatly did. Raise your hand if you have faith in ANY amd drivers, or them being on time. Read the stuff I linked to, every site made driver comments. Look at Ryan Shrout's comment on them ignoring millions of users and still having crap drivers. Heck he basically is recommending it for "future proof" memory crap. HOGWASH. Titan has 6GB and as hardocp showed to tap out memory you end up at 3fps or less. What good is future proof crap if you can't play when at 30fps when the future kicks in?...LOL
Also, he said FCAT article coming last week. AS I said they won't write one until AMD fixes their issue. Where is this magical article? Every other big site has dealt with FCAT but this one. Strange yes?
MartinT - Saturday, May 4, 2013 - link
Ryan, this review's been out in it's misleading form for 10 days, when do you anticipate FCAT results to be published?Finraziel - Friday, July 19, 2013 - link
Well, it's been about 3 months now, so unless I'm totally missing it... I'm guessing the answer is never? Pretty crappy, guys.StealthGhost - Wednesday, April 24, 2013 - link
It's damn late to the party to still lose to the 690 in games. I have nothing against AMD, I just wish they'd step it up, or maybe price this thing to compete.CiccioB - Wednesday, April 24, 2013 - link
If they won't sell price will drop soon enough, If they sell there's not reason to lower it.This is possible an attempt to get a bit more money on their quite underpriced flagship GPU. nvidia sells a smaller GPUs for a higher price, meaning that AMD is not getting all the potential they could with such a solution.
However I do not know how many of these cards they can sell. AMD is famous for the no-quality of their crossfire (stuttering) and for its support (there are still many games where their crossfire do not scale at all), so spending such an amount of money for something that won't deliver a good enough experience is quite useless.
It is much better a solution made up of 2 discrete cards, so you can easily get rid of them once you (soon) become disappointed with AMD crossfire support.
Finally - Wednesday, April 24, 2013 - link
Now we can compare which company comes aorund with the LESS BUGGY dual GPU $999 "hopefully it will show twice the performance - unfortunately it rarely does" mess of a product...It's the perfect card for the 1%, because they don't have to worry about price, power draw, noise or pretty much everything else (they sure as hell don't soil their hands with manual labor, but have them assembled for a few grand extra).
My suggestion is to make both of them into "limited editions" by replacing every plastic part possible with solid gold...
Torrijos - Wednesday, April 24, 2013 - link
So no FCAT yet for some test?!This is ridiculous! Why propagate possibly false information about a 1k$ card instead of waiting for the proper data?
PCPer has FCAT data and it's damning for AMD's tech and marketing.
A5 - Wednesday, April 24, 2013 - link
You could've taken like 5 minutes to read the article to see why there aren't any FCAT results and saved yourself some embarrassment. They're coming soon.