Sapphire 7970GE Toxic Review

[Bobisuruncle54]

Yeah, but AMD snuck in their own GPU boost which I believe gets it to 1050mhz?

The 7970GE won more than it lost (factor in the driver update) but it is so marginal all nVidia has to do is give a little more stock Boost or release a performance driver.

It really is semantics, but if you were going to draw a line in the sand about excluding products then you'll have to look at what is currently drawn in the sand.

The "stock" GE model boosts to 1050MHz and is a little faster overall than the standard 680's the reviewers had. This one is faster than the fastest 680 that boosted to over 1200MHz in KitGuru's review. Again, by a bit.

Obviously, there's no current limiting or shenanigans sensing .exe's and throttling for the benchmarks going on with this card. It can use, and handle, massive amts. of power!

Thanks for the info fellas, pretty interesting to see how things have shaped up over the last few months. Makes me wonder what AMD and Nvidia's next move will be on 28nm. We've already seen a potential monster from Nvidia from the compute side of things but the fact that both sides are practically even this generation changes the game somewhat. It'll be interesting to see if either company manages to make their architectures more efficient - I reckon AMD has more room to grow in this regard.

 

[RussianSensation]

Thanks for the info fellas, pretty interesting to see how things have shaped up over the last few months.

Ya, the interesting part is with latest drivers and 1050mhz clocks on the 7970 GE, it is slightly ahead of GTX680 on average but GTX680 performs very well in Frostbite 2.0 games (BF3, and upcoming Medal of Honor game).

As you add more AA and increase resolution, HD7970 pulls away even more (but I admit at 2560x1600, a lot of enthusiasts are probably using 2 GPUs).

1920x1080 8AA HD7970 @ 1.05ghz is 15% faster than GTX680
2560x1600 8AA HD7970 @ 1.05ghz is 20% faster than GTX680
http://www.computerbase.de/artikel/grafikkarten/2012/test-amd-radeon-hd-7970-ghz-edition/4/

And HD7970 with 1.1ghz clocks is available for sale for ~$450 with free games too. So it's cheaper and faster than a 680. But because early drivers had many issues with 7900 series and performance was poor, the reputation that 670/680 cards are faster is set in stone. Not many people revisit reviews 5-6 months later.

Makes me wonder what AMD and Nvidia's next move will be on 28nm. We've already seen a potential monster from Nvidia from the compute side of things but the fact that both sides are practically even this generation changes the game somewhat. It'll be interesting to see if either company manages to make their architectures more efficient - I reckon AMD has more room to grow in this regard.

For gaming, NV is half a step ahead I would say. They have an excellent track record of making huge 450-530mm^2 die chips. GK104 is just 294mm^2. They can easily increase performance by 50% for games without too much sweat by just adding more SPs, ROPs, TMUs and widening the memory bus width to help with AA. Their power consumption is also better this round and Kepler architecture seems to handle deferred MSAA better.

AMD has a much more difficult task with HD8900 series. Their Tahiti XT chip is 365mm^2, and their strategy involves heterogeneous computing (so including FP64 and dynamic scheduling into gaming cards). They have already implemented a wide memory bandwidth bus (little room to grow there without GDDR6 on the market).

Die size wise, AMD has very little room to grow if they stay with Tahiti XT chip and expand it. If they ditch FP64, rebalance the ROPs to 48 and maybe add some enhancements to cache, rasterization/geometry engines and start off with Pitcairn, they can probably add a lot more gaming performance. But if AMD keeps FP64 and full compute intact, I have no idea how they'll add 40% more performance over 7970 GE on 28nm without sending power consumption past GTX480 levels. My gut feeling is in 2013 GTX780 will go back to GTX580/6970 (or GTX480/5870) days with NV having 10-15% more performance. NV might not want to go to large die if they don't need to since I bet they are making more $ on Kepler than on Fermi. We'll have to see though.

 

[3DVagabond]

One more review. This time from Hexus.

This review is a bit better than the others, IMO. Still not a big benchmark suite, but at least the games are relevant without any real outliers (Hawk2 or Dirt Showdown types) to skew the results too terribly bad.

Power numbers = Fugly
Noise = Reasonable considering the performance.
Price = They're guessing, but if they are right someone call the cops 'cause we've been robbed. (Kidding. I know it's a limited edition card.)

Found another. OverclockersClub Maybe I missed it in the other reviews but they say it's a "Flex" model. I had thought from the output configuration that it could be.

 

[Silverforce11]

@Russian, see NV is doing the correct approach, neuter compute on their "gaming" GPU. But this approach may change, when more compute features are implemented in a wide variety of software which people commonly use. I see a bright future with GPUs accelerating pretty much every non-gaming program, with openCL being the key API devs support.

I'm willing to have a less efficient (perf/w) product if its very useful outside of gaming as well.

 

[DigitalWolf]

I know this is a bit off from the "discussion" going on but...


This is from Sapphires little PR statement:


"The SAPPHIRE HD 7970 6GB TOXIC Edition is a limited edition product that will be available to order from SAPPHIRE etailers and retailers from 17th July 2012."


Is there any translations available as to what date you will actually be able to order one of these? Because as far as I know July 17th 2012 was yesterday... and I couldn't find any e/retailer taking orders on the card.

 

[p_monks33]

Yep. I want one. Where the hell are they?

 

[RussianSensation]

@Russian, see NV is doing the correct approach, neuter compute on their "gaming" GPU. But this approach may change, when more compute features are implemented in a wide variety of software which people commonly use. I see a bright future with GPUs accelerating pretty much every non-gaming program, with openCL being the key API devs support.

I'm willing to have a less efficient (perf/w) product if its very useful outside of gaming as well.

I guess it depends. If more programs start using compute on the desktop, then AMD's strategy will pay off.

Look at WinZip. Also, Photoshop is starting to use specific features of the 7900 series.

NV has the resources and 95% market share in the professional space to be able to split their line into a gaming and professional GPUs. I didn't think they'd actually do it which is why I was surprised when they neutered Kepler's compute. I don't think AMD has that luxury at the moment. Also, remember that their strategy involves APUs. So they probably want the full compute capability in their APUs as well on the struggling CPU side.

AMD's engineers have it much tougher. They have to make a card that's power efficient, fast for games and retain strong compute. NV struggled with that with GTX470/480 series. I think AMD did a better job with waiting to 28nm for 7900 series but obviously it's not as lean as the 28nm Kepler gaming chips. To me the extra 50-60W of power consumption is nothing if it means having 1 Tflops of double precision performance in a $450 card. Nvidia's K20 will have about that and probably cost 7-10x that. D:

-----------

Hexus just posted a follow-up review for the Toxic using 3 screens. The GPU runs out of power well before it needs more than 3GB of VRAM:

For the most part, you need 2 GPUs for triple-monitor gaming.


"...the TOXIC remains the best of our high-end trio, though even it doesn't have the pixel chutzpah to run the latest titles at silky-smooth framerates. Examination of the per-second and per-frame times indicates that the TOXIC's performance lead over a Radeon HD 7970 GHz Edition is more down to elevated clocks than jumbo framebuffer considerations, and we feel that gaming situations would need to be artificially manufactured to show the real-world worth of an extra 3GB of card memory; heck, the GeForce GTX 680 does well enough with 'only' 2GB."