It's still stunning to see AMD at the top of the CPU chart. Not really gpu related but still...
Good to see that 40k is a solidly balanced title, lots of cards slotting into the "correct" order while maintaining very playable framerates. Well done Relic (although it might also be that relic hasn't really updated their engine since company of heroes)!
7700K isn't in the chart though.It's still stunning to see AMD at the top of the CPU chart. Not really gpu related but still...
They have crap cpu test anyway.4770 faster than 6700..LOL.7700K isn't in the chart though.
They have crap cpu test anyway.4770 faster than 6700..LOL.
It also shows how *NOT* selectively optimizing for certain architectures means that the 780Ti can match the 290X, which is what how things should be.Relatively well optimized game and shows a Fury X almost on par with an GTX1080 despite being 40% slower in reality
It also shows how *NOT* selectively optimizing for certain architectures means that the 780Ti can match the 290X, which is what how things should be.
Define how things "should" be. 780 Ti was a card released after the 290X, for a $150 premium. It should be faster, not a hair slower. And even when Maxwell first took over, the 780 Ti beat the 290X: https://tpucdn.com/reviews/NVIDIA/GeForce_GTX_980/images/perfrel_2560.gif
Unless that's what you meant? I just assumed you had low standards for the 780 Ti and expected it to lose.
Anyway, this title seems GCN 3 (Tonga/Fiji) favoured. Normally a 380X should not be that close to the 470, nor that far ahead of the 280X (which actually sometimes wins). And of course Fury X is doing well here. Odd to see. You normally expect GCN 2 optimization because of consoles or GCN 4 because of arch improvements and more driver focus.
The 780Ti and the 290X were the top cards in their respective generation and therefore it is expected that they should perform close to one another, with the 780Ti being slightly ahead in theory.Define how things "should" be. 780 Ti was a card released after the 290X, for a $150 premium. It should be faster, not a hair slower. And even when Maxwell first took over, the 780 Ti beat the 290X: https://tpucdn.com/reviews/NVIDIA/GeForce_GTX_980/images/perfrel_2560.gif
Unless that's what you meant? I just assumed you had low standards for the 780 Ti and expected it to lose.
Anyway, this title seems GCN 3 (Tonga/Fiji) favoured. Normally a 380X should not be that close to the 470, nor that far ahead of the 280X (which actually sometimes wins). And of course Fury X is doing well here. Odd to see. You normally expect GCN 2 optimization because of consoles or GCN 4 because of arch improvements and more driver focus.
As for the Fury's performance, looks like the game is very good at filling the massive shader array of Fiji, and avoiding the various bottlenecks found in it. In these situations, it should be sitting exactly where it is on the charts.
They wouldn't be able to have this kind of performance if they attempted to balance it. Fury X was 596mm^2, it was pretty much as large as it could be.-AMD's "Fine Wine" is so frustrating from this perspective when "one tier up" performance could have been available in July of 2015 instead of April 2017 if the GCN arch had Kepler's (and Maxwell and Pascal's) balance and scalability.
-AMD's "Fine Wine" is so frustrating from this perspective when "one tier up" performance could have been available in July of 2015 instead of April 2017 if the GCN arch had Kepler's (and Maxwell and Pascal's) balance and scalability.
Another Bench by overclock3d. Are they trustworthy enough for you ?I'd need to see these benchmarks corroborated by other websites before I believe them completely. GameGPU's benches tend to be more untrustworthy and less comprehensive than the German based websites like PCGH.de and Computerbase.de, which offer superior methodology. GameGPU is almost always the first to put up benchmarks of new games, which makes me suspicious as to whether they are actually performing the benches at all, or just making them up.