The NVIDIA GeForce GTX 980 Ti Review
by Ryan Smith on May 31, 2015 6:00 PM ESTSynthetics
As always we’ll also take a quick look at synthetic performance. Being a virtual copy of the GTX Titan X, GTX 980 Ti should perform very similarly here, just as we've seen in our gaming tests.
Compared to GTX Titan X, GTX 980 Ti does technically lose 2 Polymorph Engines as a result of losing 2 SMMs. However as with our games, this doesn’t really hinder GTX 980 Ti, leading it being within a few percent of GTX Titan X on tessellation performance.
As for texel and pixel fillrates, the results are both as-expected and a bit surprising. On the expected side, we see the GTX 980 Ti trail GTX Titan X by a bit, again taking a hit from the SMM loss. On the other hand we’re seeing a larger than expected drop in the pixel fill rates. GTX 980 Ti loses some rasterization throughput from the SMM loss, but a 15% drop in this test is much larger than 2 SMMs. Just to be sure we checked to make sure the ROP/MC configuration of GTX 980 Ti was unchanged at 96 ROPs, so while we can explain 10% or so (GTX 980 Ti doesn't have its clockspeed advantage in such a short test), we're at a loss to fully explain the last 5%. The short run time of the test also makes it more varaible than other tests, so that may be the last 5%.
Though in either case, despite what 3DMark is telling us, we aren’t seeing any signs of GTX 980 Ti struggling at 4K versus GTX Titan X. So if there is a meaningful difference in pixel fillrates, it’s not impacting game performance.
Facebook
Twitter
RSS
290 Comments
View All Comments
przemo_li - Monday, June 1, 2015 - link
Fiji is NOT gpu name.Its gpu segment name.
Just like VI, SI, and some more.
Its chip name if anything.
If we for a moment switched to CPU-speak, You just claimed that Puma is single CPU from AMD ;)
Refuge - Monday, June 1, 2015 - link
It is the name of a GPU Architecture. Which could be a one run chip, or it could have multiple versions based on binning.dragonsqrrl - Monday, June 1, 2015 - link
It's not a GPU architecture. GCN is a GPU architecture, Maxwell is a GPU architecture. Fiji is a GPU, GM204 is a GPU. This isn't exactly a new paradigm we're dealing with here. Oh dear, I think I might be telling someone something (aka trolling***). I've done it again.dragonsqrrl - Wednesday, June 3, 2015 - link
First, that's a terrible analogy. Puma is not a CPU, it's a CPU architecture, successor to Jaguar. Fiji is a GPU, I never once assumed or suggested that there will be a single Fiji SKU (that was all you), right now it's likely there will be 2 for the consumer market.I'm honestly not sure what you meant by "segment", perhaps you could clarify? Are you talking about AMD's XT/PRO convention? They're still the same GPU, pro is typically just a harvested XT.
Refuge - Monday, June 1, 2015 - link
They released more than four if you include mobile GPU's I believe it goes up 2 more to 6.dragonsqrrl - Monday, June 1, 2015 - link
What do you mean mobile GPU's? Are you talking about the 900m series? There are no mobile specific GPU's in that lineup. It's all binned GM204 and GM107 SKU's.eanazag - Sunday, May 31, 2015 - link
Interesting observation. I see the same behavior, but the situations are different. Both major graphics vendors are stuck on 28 nm. The 285 is a new product. AMD's graphic situation is not even close to the same as CPU. They are not even competitive in most of the markets for CPU. AMD will likely release a very competitive GPU, which is why NV is releasing the Ti now.dragonsqrrl - Sunday, May 31, 2015 - link
Yes, the 285 is a new product, and while it is an improvement and a step in the right direction, Tonga just isn't enough to address the issue of AMD's profit margins this coming generation, or make them anymore competitive in mobile (M295X). It would be as though Nvidia were selling the 980 at the $200 price point. Not exactly, but from a memory interface, die size, PCB complexity, power consumption perspective, that's basically what AMD is doing right now, with no solution forthcoming. But I guess it's better than selling Tahiti for $200."AMD's graphic situation is not even close to the same as CPU. They are not even competitive in most of the markets for CPU. AMD will likely release a very competitive GPU, which is why NV is releasing the Ti now."
Some might argue they aren't competitive in the dGPU market with Nvidia market share approaching 80%... some might say that's like Intel levels of dominance...
And I didn't say it's the same as their CPU situation, I said it's becoming more similar. While AMD will likely be competitive in raw performance, as I've tried to explain in my past 2 comments, that's kind of besides the point.
chizow - Sunday, May 31, 2015 - link
Yes, both are stuck on 28nm, but only Nvidia came out with a comprehensive ASIC line-up knowing we'd be stuck here for another 2 years (going back to 2H 2014). It is obvious now that AMD's cost-cutting in staff and R&D is starting to manifest itself as they simply can't keep up while losing ground on 2 fronts (CPU and GPU).The culmination of this will be AMD going to market with a full series of rebrands of mostly old parts going back to 2011 with a single new ASIC at the top of their stack, while Nvidia has fully laid out its arsenal with GM107 (1 SKU), GM206 (1 SKU), GM204 (2 SKU), and now GM200 (2 SKU).
Kevin G - Monday, June 1, 2015 - link
Is it unprecedented? I recall the Geforce 9000, GTS 100 and most of the GTX 200 series being various rebrands from 'generation' to 'generation'. In fact, the 8800GTS 512 MB, 9800GTX, GTS 150 and GTS 250 were all the same chip design (the GTS 250 had a die shrink but was functionally unchanged).nVidia has gotten better since then, well with the exception of the GF108 that plagued the low end for far too long.