AMD Radeon HD 7990 Review: Dual GPU Comeback

Crytek: AMD Graphics enable amazing visual experiences on Crysis 3 tribute by Cevat Yerli and Co.

youtube.com/watch?v=cUe7PNe5cI8
 
So you link me to a video that AMD uploaded? It's just one massive advert anyway.
In your eyes AMD is better than Nvidia, that's fine, I disagree though, as per the 200 odd comments above me.

And as per JC173's 200th post, we'll leave this horse to rest :)
 
Wanted to add to this, as seen in the GTX 780 review, the frame latency issue seems to already have made huge progress in being resolved and moved the 7990 to the top of the charts in almost every test.

Go back and read my quote that give it some time after release and some of the issues will be resolved with that. I think I have proven my point.
 
What was your point? And please start from the beginning :p
Please. No.
Wanted to add to this, as seen in the GTX 780 review, the frame latency issue seems to already have made huge progress in being resolved and moved the 7990 to the top of the charts in almost every test.
I must be reading the charts wrong then. Isn't the GTX 690 the HD 7990's principle competitor ? From my understanding the 690 is ahead in 10 of the 12 benches concerning frame latency
 
Please. No.

I must be reading the charts wrong then. Isn't the GTX 690 the HD 7990's principle competitor ? From my understanding the 690 is ahead in 10 of the 12 benches concerning frame latency

I know I am creating work for you (the world's most patient man) but I sometimes enjoy following these GPU threads ;)
 
I guess 13.5 beta 2 solved some issues. But GhostRyder, GPU reviews/benchmarks are meant to be published on the day of release.

Let us keep this at post #210! lol
 
The point was originally GIVING THE CARD a little time after release to solve the issues. Basing all results when comparing a card that has been out for months to a card fresh out on drivers is not proving more than they have had time to work out the bugs. At the launch of the 690, Nvidia also had some issues and given a little time they resolved it. Divide and a few others argument was that the card was bad because on its pre-release beta drivers when it was not available to the public yet that the card was a failure because the only test that it was losing in was frame time latency. Now with the recent updates, its beats most cards including the 690 in those tests excluding a few and if nothing else it is only like a difference of 5% on some of the frame latency when it smokes almost all the other cards by sometimes a 10% or higher margin.

I believe in giving cards more than one day after launch to fix things and update things because we cant nor can any company predict how the card is always gonna act, and sometimes they don't have time to fix it for certain games because games get released so fast. The 7990 is now on every type of gaming related test at the top of the food chain. It resoled the issue like I said after giving it some time.

So lets recap, the 7990 now has more FPS sometimes by as much as 10 or more and its frame time latency is better in many cases or withing 5% of the 690 which is its primary competitor.

JC713 Ill give you that, but at the same time, what card has not improved as drivers mature. Both companies improve each and every card constantly after release with new drivers and support.
 
The provided prototype driver was all reviewers had. Hopefully 13.6 will improve the microstuttering and such.
 
Divide and a few others argument was that the card was bad because on its pre-release beta drivers when it was not available to the public yet that the card was a failure because the only test that it was losing in was frame time latency.
Well that is patently false as far as my comments are concerned. What I actually said was:

1. Why release a card (when you're under no pressure to do so) with less than optimum driver support? The fact that a month after the launch the driver support is much better points to a launch that should have been delayed until now. CHECK
In hindsight it would have mitigated the GTX 780/770/760 Ti launches also.

2. The card is at the whim of Crossfire profiles. CHECK
ac3_2560_1600.gif


3. Limited overclocking due to the board design being hard limited to 375W regardless that the PCI-E SIG spec allows for an extra 75W (150W total) to be drawn from the PCIe x16 slot.
There are around two dozen HD 7990 reviews on the net- around half investigated overclocking and a few GTX 690 comparisons by the same sites where available just to show that not all dual-GPU cards are as limited in OC potential as the 7990:
Bit-tech: 3.3% gain from 10% core OC
Guru3D: 0 - 1.4% gain from 10% core OC ( 0 - 16.7% gain from 17.4% core OC for the GTX 690)
TPU : 4.1% gain from 11% core OC (13.7% gain from 19% core OC for the GTX 690)
HotHardware: 1.7% gain from 10% core OC ( 7.2-7.6% gain from 19% boost OC for the GTX 690)
ComputerBase: 4.2% max gain from 8% core OC ( 17.3% max gain from 14.2% core OC for GTX 690)
Tweaktown: negative scaling to 7.6% gain from 10% core OC
Hardware Canucks: 6.6 -7.8% from 10% core OC
CHECK.
BTW:
the reviewers have seen in a range of 5-15% depending on the game for the 7990 which is around the 1100 on the core overclock
15% ? Which reviews were these. Please link, I'd be keen to read a HD 7990 review I haven't already seen.

4. High price in comparison with it's immediate competition:
HD 7990......$1050 - 1100
2 x HD 7970 GE....$840
CHECK
Now with the recent updates, its beats most cards including the 690 in those tests excluding a few and if nothing else it is only like a difference of 5% on some of the frame latency
I see. So the HD 7990 being behind by 5% in frame latency tests is no biggie. You do realize that the average frame rate difference between the 7990 and GTX 690 where the 7990 "wins" is less than 5% in half the benchmarks and the margin between the two cards is 5.6% (at 1920) and 9% (at 2560) - or 1% and 3.3% without the GCN optimized Sleeping Dogs ? Even the worst case scenario of 9% is less than the 10% difference in price...or does performance-per-$ not count in this instance ? or maybe non-gaming features do count in this instance ?
So lets recap, the 7990 now has more FPS sometimes by as much as 10 or more
Or is lower by 6.5 and 10.2% in Crysis 3, less than 5% in six further benchmarks, and exceeds 10 in a further three.
and its frame time latency is better in many cases or withing 5% of the 690 which is its primary competitor.
While costing 10% more. Yes, that's about it.

Thanks for playing.

Steve
You see what you made me do!
Sadist !
 
Oh look you picked one of the few "Nvidia Biased" games to base this of shock and awe.

Your trying to provide the few over many, since you like to post games showing the fps ill return the favor
BF3_01.png


Dirt3_01.png

FC3_01.png

MP3_01.png

SD_01.png


Try those of for size.
 
Oh look you picked one of the few "Nvidia Biased" games to base this of shock and awe.
You mean using Assassins Creed 3 as an example of less than stellar Crossfire support in games? Yes, I suppose I was rather remiss since it is an isolated case....well, isolated except for F1 2012, Star Craft 2, Planetside 2, Project Cars, WoW, not to mention odd quirks at different resolutions and image quality settings for other games...
Crossfire_Scaling_oddities.jpg


Your trying to provide the few over many, since you like to post games showing the fps ill return the favor
[snip]
Try those of for size.
Seems to be exactly what I was talking about. So, shall we "try them for size"
Battlefield 3...HD 7990 @ 152.0 fps....GTX 690 @ 146.3 fps....difference of 3.9%
DiRT 3...........HD 7990 @ 162.0 fps....GTX 690 @ 160.3 fps....difference of 1.06%
Far Cry 3.......HD 7990 @ 90.3 fps......GTX 690 @ 85.4 fps......difference of 5.7%
Max Payne3..HD 7990 @ 113.2 fps....GTX 690 @ 112.3 fps....difference of 0.8%
Sl. Dogs........HD 7990 @ 95.5 fps......GTX 690 @ 74.5 fps......difference of 28.2%

With the exception of Sleeping Dogs (and Crysis 3 which some odd reason you've omitted- couldn't be because the GTX 690 heads the chart could it?) the numbers are pretty damn close.

BTW: Maybe you could consider just linking to the pages, or maybe scaling the images down, or maybe, presenting an argument that makes sense.
 
Yeah just like you omitted all these because they were better, picking biased games does not help your case. I picked most of the ones posted with the exception of two which were crysis and Sleeping Dogs. Both of which were better with the 690, that's still 2 of the 7 tested here which for some reason I see that 5 > 2 by your logic.

Since I know exactly where your going next since your so easily predictable now, the Benchmarks on the frame latency will show still that the 690 is better in a majority at the moment in that respect (Though once again, AMD is obviously working on it and improving it which we have already seen). But the range is once again like seen here, so were apparently at an impass in that reguards. Oh right, but wasn't your argument that the 690 is cheaper than the 7990 true true, I mean its not like theres a 300+ dollar game bundle with one of those.

Oh wait there is.
 
As for the games you posted, ok those games do show better performance on nvidia on crossfire support (A few like the 7970ghz edition better when compared to a single 680). As for the games selected, I know planet side 2 has a bad crossfire profile though with Radeon Pro I was able to boost that number on my 6990's a bit. I dont know what Project cars is though I would guess a racing game along with F1. As for Starcraft 2, yea they do need to fix that. Last WoW, umm that I did not know had that poor support on AMD cards, huh interesting to note, I had not known that WoW did not have a profile or had that much issues on that game, thats good to note and is a wonder.

When you add information without being an ***, ill be happy to take your information in like you did up there. Next time I will be happy to scale them down.

Yes I omitted the ones that beat it, I did because I was showing ones where it won, the other it did like I posted stating that it won in 5 of 7. I also stand by that many companies on both sides of the market flat out endorve nvidia or AMD so many times you really cant do comparisons between the 2 because the company is obviously going to optimize it for that companies cards.

Example: Sleeping Dogs is AMD Biased
Assasins Creed is Nvidia Biased
 
Interesting piece over at Tom's Hardware.
Seems as though the 7990 is exhibiting some interesting thermal characteristics. As a side note, the premium OEM's don't seem to be getting much customer interest from the card- which might be just as well given the findings of the article.
 
I'm from Russia and practically all our so to say hardware laboratories recommend 7990 when you have a choose of 7990, 690 or Titan. Well I think I'll agree with that. To my mind the Titan is no more than a piece of **** really. You pay the same amount of money for a single gpu as for a double gpu. You should really check out the palit gtx 780 super jetstream reviews. It beats the Titan almost everywhere because of its overclocking capabilities. Personally I'm choosing between a 7990 which you can find for 850 $ or so in my country and a super jetstream 780 which is cheaper but not by much. The only thing that concerns me is that double gpus have problems with some games and work slower in them than they should... Maybe someone can give some advice?
 
Back