AMD Radeon HD 7990 Video Card Review @ [H]

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,532
AMD Radeon HD 7990 Video Card Review - Today marks the launch of AMD's Radeon HD 7990. The Radeon HD 7990 is a dual-GPU video card that has its two GPUs down on a single PCB that uses CrossFire to operate the two Radeon HD 7970 GPUs. We will test this video card in the latest games, comparing it to GeForce GTX 680 SLI and Radeon HD 7970 GHz Edition CrossFire.
 
Sorry it is late, we had to double and triple check most of the performance numbers to make sure what we had was correct in gameplay.
 
Well.. it looks spectacular(in appearance) at least...

Performance is... bleh.

Good thorough review as always. Thanks!
 
Hi Kyle,

I was wondering if you would consider using the reg hack to see if the card would act the same at 50% PowerTune . When I ran my 7950 at 20% it was just able to keep at the boost speeds, but at higher temps it would still sometimes still throttle, might be something to try. I wrote about it here for more info: http://hardforum.com/showthread.php?t=1756303

Thank you for your time on testing this card it was an excellent read.
 
Thank you as always, jefe. And credit for just going "Lower resolutions? Riiight."

That said: it hangs for the most part with SLI 680s, but loses out to twin 7970GEs? At the price tag it's at, and even as someone who prefers single-card solutions in general, that's not encouraging right now...
 
Great review.

The anecdotal narrative about the discrepancy between playability and expected fps is exactly what I experienced by going from dual 670s to a single Titan. The Titan fails the frame rate competition but provided a far superior gaming experience in triple monitor.

I still like mgpu at 2560 and below.

I'm really glad respected sites like [H] are calling out multi GPU issues. Hopefully in a year or two we can all be rocking dual $250 cards with excellent playability.
 
pirce is to be expected for a "halo" card but performance expectations for this card make the real world benchmarks bleh
 
Why not compare this to 690 and Titan, or other $1000 alternatives?

I didn't see the Titan getting points docked for it losing out to alternatives half (or less) its price.
 
AMD CrossFire. It stuttered badly at very high settings and you knew, you could feel it that it wasn't playable as it negatively affected the overall gameplay and immersion.
AMD must fix this, otherwise 7990 is not even relevant.
 
Pointless card really, overpriced, and came out a hell of a lot later than it should have.
 
Why not compare this to 690 and Titan, or other $1000 alternatives?

I didn't see the Titan getting points docked for it losing out to alternatives half (or less) its price.
Titan didn't lose. It has lowest latency in the business at the moment, more than twice as fast delivering FULLY RENDERED frames to the monitor than SLI, 3-4x faster than Crossfire.
 
great review.

just a poor performing card though. I mean this cards comes out 16+ months after the initial 7970s and is 20% slower?

It seems to me, this "engineering feat" they accomplished by getting this card to run on two 8-pin plugs could have been achieved by just binning regular 7970 chips and lowering the power.

On second though, I bet if you binned chips the whole time, 16 months would give you enough time to collect a bunch of good performing chips....
 
This and the Titan are both useless overpriced crap.

Then again 6 games included makes it more like $800 card.

tomb raider graph fixed, thanks, -brent
 
Last edited by a moderator:
I have a question about the test setup GTX cards.
Were they they 2GB or 4GB 680GTX cards?

I kinda wonder how the GTX680 with 4GB would fair against this 7990 by itself.

Also I wonder how this 7990 would do against that one-off (only 500 made) DEVIL card that MSI released a while back.
 
Over on Tom's Hardware AMD gave them a special 'prototype' driver that AMD is planning to launch at the end of the year.

It is supposed to add more smoothness to CrossFire using similar techniques to NVIDIA's driver.

They also did a blind test with a bunch of gamers and they noticed that the prototype driver is much smoother than the current AMD drivers.
 
good review. will there be an article previewing AMD's frame pacing driver. few sites like hardwarecanucks and pcper previewed this driver even though its an early prototype. the disappointing thing is AMD's hardware is being let down by software and it seems like it will take till July for AMD to release the frame pacing driver to bring smoothness on par with SLI.
 
Nice read. Thanks for being so thorough and testing the marketing claims too. Kinda disappointed AMD couldn't retake the performance crown.

As with all AMD products, I wonder if performance will improve in a month or two when AMD releases some new drivers.

By the way, love the new batch of games for testing.
 
Ok, I'm confused.
In the Tombraider run-through it says that you used the same settings for all 3 configurations, but the graph shows that you used normal shadows for the Radeon HD 7970 GHz Edition CrossFire table.
 
I was really hoping to see this card in the $799-$849 price range. At $999, it seems way over priced. Even with all the games included. If you could actually get retail price out of the games it would be a little different, but most games sold I've seen go for half or less than half MSRP. They should offer the card without the bundle and just drop the price a couple hundred bucks.

I'd like to see the same thing with all of the AMD cards. Drop the never settle bundle and pass the savings on. $400 for a 7970 minus -$120 (est. retail for 2 games) put it at $280. I'd jump all over a 7970 for that price.
(note: i understand AMD doesn't pay MSRP for the games included, but one can dream of such savings!)
 
Last edited:
I started the run-through, and as I was playing through the entire level of "Safeties Off" I was borderline calling it playable at that setting. I didn't know the FPS, as when I'm recording with FRAPS the FPS is not displayed. (We do this purposefully.) Therefore, all I could feel was the game itself and I wasn't burdened with thinking in FPS at all. It was a fluid and immersive gaming experience. The game to me felt like it was playing at 40-50 FPS. Finishing up the run-through I stopped FRAPS and it read an average of 33 FPS.

I was in awe. It felt smooth, really smooth, it felt like it was running at a faster framerate than it really was. If not for some minor lag throughout the run-through, I would have called that very high setting playable based on the experience alone, without looking at the framerate. It was a shock to see the framerate so low, I really thought it was higher as I was playing. This kind of smoothness we often experience with NVIDIA SLI.

My experiences exactly. In Crysis 3, I'm playing and everything is smooth, and my frames are between 25-30fps @ 5900x1080. Quite impressive.
Nvidia controls the stutter issue through hardware, which is a more superior approach than AMD software/driver upcoming fix.

This and the Titan are both useless overpriced crap.

Then again 6 games included makes it more like $800 card.

Your member name is appropriate for that comment.
 
And it didn't get head to head with the GTX690 because...
 
My experiences exactly. In Crysis 3, I'm playing and everything is smooth, and my frames are between 25-30fps @ 5900x1080. Quite impressive.
Nvidia controls the stutter issue through hardware, which is a more superior approach than AMD software/driver upcoming fix.

I think its part of software load balancing. I read somewhere its not hardware related.
 
The power consumption numbers here make me think some other sites use completely different measuring tools, or they tried double or triple 7990 and reported it as a single.
(I know H's got the accurate numbers, mind you)
 
Regardless of how you "feel", they do not benchmark the same, nor do they cost the same. By that notion, why have graphs or benchmark figures at all if you're going to revolve reviews around guesstimates?
 
Regardless of how you "feel", they do not benchmark the same, nor do they cost the same. By that notion, why have graphs or benchmark figures at all if you're going to revolve reviews around guesstimates?

Because readers demand those. I would suggest to you that frame rates and such focal attention on those are fundamentally bad for the industry at this point in time. We use the data to help analyze what is going on and draw a better informed opinion about gaming experience and immersion that is demanded by our readers. I suggest the gaming experience delivered by the hardware is of far greater importance than anything else. If you are a "framerate" kind of guy, certainly there are hundreds of websites that will deliver exactly what you want if you are not finding it here.
 
Does the 7990 work in Triple-Crossfire with a 7970?

The 6990 worked in Trifire with a 6970, the 5970 worked in Trifire with a 5870, etc so it would make sense that it would, but i'm still curious.
 
I like the way the reviews are done, gives me a good idea of how the game will play on my system. Well done guys!
 
I purchased 2x 7970s right when they came out. Spent over $1k and got crap, unsupported xfire drivers and ridiculous stuttering. Sold 1 and my experience with 1 card was great. No stuttering, great performance. Sold that one eventually and got a 690. No problems at all.

Pure and simple....1 card = AMD, 2+ cards = Nvidia.
 
I was disappointed to read that the crossfire issues have still not been worked out to a satisfactory degree. It looks like that at this point with the last release of hardware before the next gen gear arrives in the fall that we will never see a reasonable fix on this generation of hardware from AMD.

The decrease in performance over 2 ghz edition cards is really astounding, thanks for verifying the data to confirm your numbers, I know that running the same benchmarks for so many years is only exciting for so long.

I think it would be interesting if a 3rd party wanted to put 2 7950 gpus on a card and get it out the door for under 600$. I think that there would be a larger market for that type of price performance than a 1000$ crown of rust. It's really a shame that the gameplay experience is such a hindrance in crossfire still.

the bundle is nice but with sales on several of those games in the under 15$ range already, it's hardly worth an offset of 300$ that would be needed to balance out the performance ratio reduction.
 
A question for either Brent or Kyle: Was there significant coil whine while using the card? Tom's (which may as well start reviews with Sponsored by Nvidia) was really moaning about severe noise issues separate from the fans.

Separate question: Did you have access to the "prototype" drivers that some other 7990 reviewers were using and if so, was there any difference?
 
This review sets up a no win situation for the Reference 7990. The metrics measured were price to performance ratio. We all know dual GPU cards and high end cards in general always fall off the price performance curve. The silly thing is, it trade blows with the GTX690 in almost every game, performance is very comparable...it is also a direct competitor for the $1000 graphics card market segment. Why was the GTX690 excluded from the review? What about the Titan?

Also, HardOCP gave the ASUS 2x7970 $1500 version a glowing review...but price/performance was not a judging factor. Sure that card is about 10-15 percent faster than this card, but also it was 50% more expensive. So...why was that card not graded on price/performance. It was apparently terrible on that metric?

Back
to the GTX690 review, this unit was reviewed as a luxury item combining two GPUs into a single card, and for that it was measured as GOLD based on its ability to combine two cards and do better than the last gen GTX590. Sure lets set that premise. Why was the 7990 judged on different standards? Better yet why IS ths 7990 judged on price/performance? We are all aware that if we're paying 1000 dollars, we are not worried about P/P vs looking at other $1000 single card solutions.

Both cards sell at 1000. Both cards are withing 5-10% on any given game or graphics setting (according to combined benchmarks for every other review website). Nvidia gets gold, AMD gets a crappy review.

I don't get it guys, please explain this to me.
 
Last edited:
great review.
why is AMD touting performance in games that is not true? not that i can afford to buy that card, but if i had the funds, why would i?
i have not read other reviews on the 7990, i wonder if anandtech had the same results?
maybe they need some bake time for the drivers.
 
http://www.anandtech.com/show/6915/amd-radeon-hd-7990-review-7990-gets-official

from the review
Ultimately we’re not looking to be dour on the 7990, it just hasn’t particularly excited us. It’s a solid piece of engineering on AMD’s part to get a pair of Tahiti GPUs into a dual-GPU card like this, and the noise improvements over the 6990 are simply incredible. The 7990 is the natural progression of AMD’s dual-GPU cards, and with AMD’s improvements it’s a fine addition to the Radeon family.

As always however, our usual reservations about multi-GPU cards apply. We continue to favor strong single-GPU cards for their consistency, and right now at the $1000 price point that’s Titan. 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%.; just be aware of the pros and cons that come with such a card.

Bringing things to a close, in a few days we’ll be following up today’s launch article with an in-depth look at the 7990’s performance under the microscope of FCAT. AMD has laid out a plan to improve their frame pacing situation, and with AMD’s alpha drivers in hand we can take our first look at AMD’s changes and what kind of an impact those will bring. So please stay tuned for that.


i would say that the general conciseness is the 7990 is not the card we would have expected.... yet
 
This review sets up a no win situation for the Reference 7990. The metrics measured were price to performance ratio. We all know dual GPU cards and high end cards in general always fall off the price performance curve. The silly thing is, it trade blows with the GTX690 in almost every game, performance is very comparable...it is also a direct competitor for the $1000 graphics card market segment. Why was the GTX690 excluded from the review? What about the Titan?

Also, HardOCP gave the ASUS 2x7970 $1500 version a glowing review...but price/performance was not a judging factor. Sure that card is about 10-15 percent faster than this card, but also it was 50% more expensive. So...why was that card not graded on price/performance. It was apparently terrible on that metric?

Back
to the GTX690 review, this unit was reviewed as a luxury item combining two GPUs into a single card, and for that it was measured as GOLD based on its ability to combine two cards and do better than the last gen GTX590. Sure lets set that premise. Why was the 7990 judged on different standards? Better yet why IS ths 7990 judged on price/performance? We are all aware that if we're paying 1000 dollars, we are not worried about P/P vs looking at other $1000 single card solutions.

Both cards sell at 1000. Both cards are withing 5-10% on any given game or graphics setting (according to combined benchmarks for every other review website). Nvidia gets gold, AMD gets a crappy review.

I don't get it guys, please explain this to me.
When the 690 was released at $999, GTX680s were still $500, so it was much more competitive with the single GPU versions, whereas the 7990 is $200 more than comparable 7970GE boards and performs noticeably worse.

The Titan is a much different product. It's by far the single fastest GPU you can buy.

I'm not going to dig through past conclusions to try and explain the H justification for their conclusions, because prices and performance change over time with price cuts and drive updates. This isn't a limited edition card, it's a mass produced card and it just doesn't have very exciting performance. The game bundle is pretty much the only reason I can see someone buying this.
 
When the 690 was released at $999, GTX680s were still $500, so it was much more competitive with the single GPU versions, whereas the 7990 is $200 more than comparable 7970GE boards and performs noticeably worse.

The Titan is a much different product. It's by far the single fastest GPU you can buy.

I'm not going to dig through past conclusions to try and explain the H justification for their conclusions, because prices and performance change over time with price cuts and drive updates. This isn't a limited edition card, it's a mass produced card and it just doesn't have very exciting performance. The game bundle is pretty much the only reason I can see someone buying this.

I guess I can see the reasoning there...reviewing based on the same situation as the GTX580, but what I was hoping for in the review was a current market value of the card based on its current relevant competitiors. I don' t think the difference between single and multi card performance compared to the same Nvidia side at the time of the 690 review is of any value or significance to me...maybe that's why i was so dissapointed in this review.

To me, 1000 dollar single card is a market segment. A good review would review other products in the same segment. Then draw conclusions based on performance in gaming, scaling, power, and compute. Right?
 
Last edited:
This review sets up a no win situation for the Reference 7990. The metrics measured were price to performance ratio. We all know dual GPU cards and high end cards in general always fall off the price performance curve. The silly thing is, it trade blows with the GTX690 in almost every game, performance is very comparable...it is also a direct competitor for the $1000 graphics card market segment. Why was the GTX690 excluded from the review? What about the Titan?

You are correct, 7990 is in a no-win situation in this review, especially considering that 7970 CF and 680 SLI were included. This 7990 falls off both the price and the perf curve you mention. 690 and 680 are the "same" in terms of real world gaming. Also worth noting is that this product is WAY late to market. GTX 690 has been out for almost a year.

GTX 690 question was answered above. We did not have time for TITAN due to AMD getting us the card so close to the embargo date. That all said, we have more than enough reviews on the cards here for an informed reader to make an informed buying decision.

Also, HardOCP gave the ASUS 2x7970 $1500 version a glowing review...but price/performance was not a judging factor. Sure that card is about 10-15 percent faster than this card, but also it was 50% more expensive. So...why was that card not graded on price/performance. It was apparently terrible on that metric?

If you read the ASUS review you will answer your own question. But I can quote it here to better point it out.

The one drawback, we know you all are gawking at, is the price. At $1500 this is definitely not for anyone on anything resembling a budget. You will by now have realized that this price is way overpriced for a CrossFire 7970 GHz Edition setup. You can purchase two separate Radeon HD 7970 GHz Edition video cards for less than $850. That makes the ASUS ROG ARES II a $650 premium, over the cost of Radeon HD 7970 GHz Edition CrossFire configuration.

The ASUS ROG ARES II is not about price though; it is very much a "halo" product. The ROG ARES II is ASUS showing that it can lead the industry in video card design. Even on the official ASUS page, you will find it titled, "ASUS Releases Limited Edition ROG ARES II with New Hybrid Thermal Design." Its production will be very limited. There will be only 1,000 of these video cards sold, and only 100 of those in North America. In the eyes of the computer hardware enthusiast, the ROG ARES II is a triumph in design and execution. Surely value is arguable, but the ROG ARES II was neither designed nor produced with value in mind.

Back to the GTX690 review, this unit was reviewed as a luxury item combining two GPUs into a single card, and for that it was measured as GOLD based on its ability to combine two cards and do better than the last gen GTX590. Sure lets set that premise. Why was the 7990 judged on different standards? Better yet why IS ths 7990 judged on price/performance? We are all aware that if we're paying 1000 dollars, we are not worried about P/P vs looking at other $1000 single card solutions.

690 did give as good gaming perf as 680 SLI, 7990 has come up short compared to 7970 CF. Then take a look at the build quality of the 690. It is simply awesome and very much a halo product in our eyes. 7990 in a foot long $1 plastic case. I do not see the 690 and the 7990 truly being in the same class when it comes to the enthusiast hardware market. Now you may see all that as lipstick on a pig, and I could very much respect that, then in that sense 7970 CF blows the 7990 out of the water.

I do respect where you are coming from with your question, and I do see how you could consider it a valid concern, hopefully I have addressed this so you see our perspective.

Both cards sell at 1000. Both cards are withing 5-10% on any given game or graphics setting (according to combined benchmarks for every other review website). Nvidia gets gold, AMD gets a crappy review.

NVIDIA also did it a year ago....with much more style and value placement in the product segment....and without all the stutter issues. I went from high end AMD CF in my own personal system to SLI because of the stutter issues with CrossFire. HardOCP has been talking about this for YEARS now. SLI is better technology implementation right now and that has to weighed in review that take CrossFire performance into consideration.

I don't get it guys, please explain this to me.

Hopefully I have addressed the issues of our "bias," since you did title your post "Biased Review."
 
Back
Top