Thank you for asking this question, as it sent me off into researching how MSAA works. As it turns out, apparently MSAA literally works by performing parts of the rasterization process at a higher resolution than screen resolution, and thus appears to be rasterization rate limited:
The linear...
Isn't pixel fill rate a function of ROPs times clockspeed? I'm not sure how DSBR would affect it? Isn't the primary benefit of DSBR related to saving memory bandwidth?
True, but we do have B3D Suite results for pixel fill rate:
However, having done some research into the topic, doesn't the fact that Fiji performed relatively better at higher resolutions preclude it being ROP bound? If Vega was ROP bound, shouldn't its performance worsen relative to Pascal...
I wasn't disagreeing about their importance, I was asking whether it would be possible to evaluate if this was bottlenecking Vega's performance without front end bottlenecking issues confounding the results. Does MSAA performance not depend at all on geometry performance? If it doesn't then...
Aren't the ROPs at the back of the rendering pipeline? How can you effectively assess how much of a bottleneck having 64 ROPs is or isn't while the front end is still bottlenecked by being limited to 4 polygons per clock? Note, I'm not saying that 64 ROPs aren't a bottleneck, just that I don't...
Of course, there are a ton of other factors involved across all the games being used in benchmarking today. I'm certainly not expecting to be able to precisely correlate the geometry performance of games with their performance on Vega. Still, it strikes me as pretty suggestive that two games...
Sure, but that isn't related to the thing I'm trying to get an insight into. I'm interested in knowing the extent to which games near the bottom of the chart are very demanding of geometry performance and those near the top are not because I'm trying to investigate the extent to which Vega is...
Ignoring the continuing attempts to derail the discussion, would anyone happen to know if there is a way to tell which games that are commonly used for benchmarking are the most and least demanding of geometry performance (tessellation or otherwise)? In looking at this chart:
How many of the...
Sure, but that doesn't really address what the performance implications could be for Vega suddenly becoming much faster at polygon throughput and having tessellation performance at least on par with Pascal might be. Unigine Heaven is another benchmark that shows just how much Vega still...
Well, I can think of two games off the top of my head where Vega relatively underperforms which are known to use lots of tessellation: GTA V and Watch Dogs 2. Gameworks games in general have often been accused of deliberately using large amounts of tessellation to tank performance on AMD cards...
Well, with respect to primitive shaders that question amounts to asking what would be the performance implications for Vega of a significant increase in polygon throughput? If primitive shaders manage to achieve even half of the gains claimed in the whitepaper, you are talking about doubling...
I think this is pretty convincing evidence that Vega is severely bottlenecked at the moment, as the 8 additional CUs of Vega 64 are providing no additional performance whatsoever over Vega 56 with the same BIOS. Considering how much discussion there has been of Fiji being front end bottlenecked...
Your inability to understand that I said led them to confirm on the record, not led them to investigate, is your own issue.
It is quite obvious exactly what both you and Rifter intend to imply about the Vega whitepaper, and by extension about primitive shaders as described in the whitepaper...
I was not claiming that Ryan Smith and others looked into it as a result of seeing my posts about it it. I meant only what I said: that the discussion about it that I reignited led to finally getting some on the record confirmations of what the evidence I presented pointed to.
Excellent, so do...
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.