The NVIDIA GeForce GTX 980 Review: Maxwell Mark 2
by Ryan Smith on September 18, 2014 10:30 PM ESTSynthetics
As always we’ll also take a quick look at synthetic performance. These tests mainly serve as a canary for finding important architectural changes, and with the exception of pixel throughput we are not expecting any major changes for GTX 980 and GM204.
GM204 is designed to have an ever-so-slightly higher triangle throughput rate than GK110 – 16 tris/clock versus 15 tris/clock, and sure enough the GTX 980 comes out on top in TessMark, slightly edging out the GTX 780 Ti. The difference is only very slight here, and though GM204 should be a bit more powerful than GK110 in practice it’s a dead heat.
Moving on, we have our 3DMark Vantage texture and pixel fillrate tests, which present our cards with massive amounts of texturing and color blending work. These aren’t results we suggest comparing across different vendors, but they’re good for tracking improvements and changes within a single product family.
Beginning with Maxwell NVIDIA reduced their texture-to-compute ratio from 12:1 to 16:1. As a result of this change Maxwell GPUs have fewer texture units than comparable Kepler GPUs. Compounding this effect is the fact that Maxwell CUDA cores are more efficient than Kepler CUDA cores, leading to NVIDIA placing fewer cores overall and further reducing the texture fill rate.
As a result the GTX 980 is not texture fillrate competitive with any of the GK110 cards. It is competitive with the GK104 cards, but only because these cards had the same number of texture units at 128. NVIDIA has told us that they believe this new ratio is a better fit for modern workloads, and judging from the performance we’re seeing elsewhere it would appear that NVIDIA is right.
On the other hand, thanks to NVIDIA’s newer 3rd generation delta color compression technology, our 3DMark pixel fillrate performance is through the roof. GTX 980 comes very close to doubling the throughput of our GK110 cards and more than doubles the throughput of the GK104 cards, reflecting the fact that it has 64 ROPs and more importantly has the available bandwidth to put them to good use.
This benchmark in a nutshell is why NVIDIA can deliver chart-topping performance despite having only 2/3rds the memory bandwidth of GTX 780 Ti. By improving their color compression to this point, NVIDIA can significantly reduce their memory bandwidth requirements Maxwell 2, allowing them to do more with less. In real games the result won’t be anywhere near this remarkable since this is a pure pixel fillrate test, but it goes to show that NVIDIA has been able to expand their effective memory bandwidth in concert with their ROP and shader performance improvements.
Facebook
Twitter
RSS
274 Comments
View All Comments
nathanddrews - Friday, September 19, 2014 - link
http://www.pcper.com/files/review/2014-09-18/power...kron123456789 - Friday, September 19, 2014 - link
Different tests, different results. That's nothing new.kron123456789 - Friday, September 19, 2014 - link
But, i still think that Nvidia isn't understated TDP of the 980 and 970.Friendly0Fire - Friday, September 19, 2014 - link
Misleading. If a card pumps out more frames (which the 980 most certainly does), it's going to drive up requirements for every other part of the system, AND it's going to obviously draw its maximum possible power. If you were to lock the framerate to a fixed value that all GPUs could reach the power savings would be more evident.Also, TDP is the heat generation, as has been said earlier here, which is correlated but not equal to power draw. Heat is waste energy, so the less heat you put out the more energy you actually use to work. All this means is that (surprise surprise) the Maxwell 2 cards are a lot more efficient than AMD's GCN.
shtldr - Wednesday, September 24, 2014 - link
"TDP is the heat generation, as has been said earlier here, which is correlated but not equal to power draw."The GPU is a system which consumes energy. Since the GPU does not use that energy to create mass (materialization) or chemical bonds (battery), where the energy goes is easily observed from the outside.
1) waste heat
2) moving air mass through the heatsink (fan)
3) signalling over connects (PCIe and monitor cable)
4) EM waves
5) degradation/burning out of card's components (GPU silicon damage, fan bearing wear etc.)
And that's it. The 1) is very dominant compared to the rest. There's no "hidden" work being done by the card. It would be against the law of conservation of energy (which is still valid, as far as I know).
Frenetic Pony - Friday, September 19, 2014 - link
That's a misunderstanding of what TDP has to do with desktop cards. Now for mobile stuff, that's great. But the bottlenecks for "Maxwell 2" isn't in TDP, it's in clockspeeds. Meaning the efficiency argument is useless if the end user doesn't care.Now, for certain fields the end user cares very much. Miners have apparently all moved onto ASIC stuff, but for other compute workloads any end user is going to choose NVIDIA currently, just to save on their electricity bill. For the consumer end user, TDP doesn't matter nearly as much unless you're really "Green" conscious or something. In that case AMD's 1 year old 290x competes on price for performance, and whatever AMD's update is it will do better.
It's hardly a death knell of AMD, not the best thing considering they were just outclassed for corporate type compute work. But for your typical consumer end user they aren't going to see any difference unless they're a fanboy one way or another, and why bother going after a strongly biased market like that?
pendantry - Friday, September 19, 2014 - link
While it's a fair argument that unless you're environmentally inclined the energy savings from lower TDP don't matter, I'd say a lot more people do care about reduced noise and heat. People generally might not care about saving $30 a year on their electricity bill, but why would you choose a hotter noisier component when there's no price or performance benefit to that choice.AMD GPUs now mirror the CPU situation where you can get close to performance parity if you're willing to accept a fairly large (~100W) power increase. Without heavy price incentives it's hard to convince the consumer to tolerate what is jokingly termed the "space heater" or "wind turbine" inconvenience that the AMD product presents.
Laststop311 - Friday, September 19, 2014 - link
actually the gpu's from amd do not mirror the cpu situation at all. amd' fx 9xxx with the huge tdp and all gets so outperformed by even the i7-4790k on almost everything and the 8 core i7-5960x obliterates it in everything, the performance of it's cpu's are NOT close to intels performance even with 100 extra watts. At least with the GPU's the performance is close to nvidias even if the power usage is not.TLDR amd's gpu situation does not mirror is cpu situation. cpu situation is far worse.
Laststop311 - Friday, September 19, 2014 - link
I as a consumer greatly care about the efficinecy and tdp and heat and noise not just the performance. I do not like hearing my PC. I switched to all noctua fans, all ssd storage, and platinum rated psu that only turns on its fan over 500 watts load. The only noise coming from my PC is my radeon 5870 card basically. So the fact this GPU is super quiet means no matter what amd does performance wise if it cant keep up noise wise they lose a sale with me as i'm sure many others.And im not a fanboy of either company i chose the 5870 over the gtx 480 when nvidia botched that card and made it a loud hot behemoth. And i'll just as quickly ditch amd for nvidia for the same reason.
Kvaern - Friday, September 19, 2014 - link
"For the consumer end user, TDP doesn't matter nearly as much unless you're really "Green""Or live in a country where taxes make up 75% of your power bill \