Discussion in 'Videocards - AMD Radeon Drivers Section' started by oGow89, Feb 20, 2016.
yeah, im not sure where that came from.
Dat 7870=680 WTF.
just a heads up fiji and tonga are not buggy.. direct flip is. It is frame locking them ergo vsync cant be turned off.. amd needs to bring a hotfix fast
Nope, only 1 fps won from dx12 and the avg fps remained the same. Which means it doesn't.
It came from the graph.
Where do you see 7870 = 680?
The 390X only gains 3FPS average. I think it just means the game is mostly GPU bound and the lower overhead isn't benefitting much. It doesn't mean other games won't though.
R9 390 gains 7 fps on the min and 6 on avg. R9 390x wasn't benched on dx11. Don't know where you get the 3fps from.
Sure maybe for this game tahiti isn't benefiting from the new api, but that is why we are on this thread talking about this game exclusively.
Just saying I must be going blind.
I think it might be your wording.
Tahiti does not benefit from optimizations in this driver
Tahiti doesnt benefit from DX12.
the second statement is wildly incorrect.
I don't think the driver is the one to blame. I am talking about tahiti not benefiting from dx12 in this game. Nothing wrong with my wording if you are following the context and the title of this thread. And also if you looked at the numbers from the dx11 and dx12. I meant what i said.
I confused the threads, I have no idea for what I replied, but it apparently had the 7870 equal a 680. I believe it was The Division
I believe that the game saturates Tahiti as a GPU, so it doesn't really benefit. On the other hand I'm hearing that it has general optimization problems.
Here some tests of Final Build
Yes, and they still have a bug with Fiji/Tonga, which are running with a 60 FPS cap only, even on 144Hz displays. It works well with Hawaii for example, and they are in contact with AMD about this. Also Robert Hallock from AMD should probably know, because he commented on a Reddit thread about this.
Hawaii is doing awesome, which cannot be said for Fiji. Also the 10% difference between 290 and 390 is intriguing. Maybe held back by 4GB?
No that one looks weird. The r9 290 scores similar in both dx11 and dx12. I don't think it is the extra vram, but more likely some sort of BS on amd's part. The 200 extra mhz on the memory wouldn't play a part either. I feel like it has do with the driver itself being more focused on the r9 300 series of cards. Something is fishy.
Look under dx11 the r9 290 and 390 are within 1-2 fps of each other across all resolutions but 4k (here 4gb is then the reason why) which is reasonable due to the higher memory clock. But in dx12 it just screams ahead.
If you check DX11 vs DX12 all 4GB cards barely gain any FPS. Fury being exception.
And then theres Gigabyte GTX 770 Windforce OC/2G and Asus R9 270X DC2 TOP/2G where DX12 performance plummets!
Well in that case, screw it, i am gonna go for an rma due to fan ratteling and hope to get an r9 390 because the 200 series ist outsold.
Uhh, 390 vs 980 Ti @ Stock clocks :gape:
German Computerbase found a workaround for Fiji/Tonga, and they have some new Benchmarks. Poor NVidia ...
Edit: Ohh, someone already posted that? I must be blind today ... sorry.
Anyone remember TR2013 and Hitman Absolution?
Both were Amd optimized, few months later they started to lag behind nv a lot. Just saying.
NVidia fixed TR2013 performance pretty quickly..I remember that.. I had a titan back then ..
Count how many GE titles had major problems with NVIDIA and count how many gimpworks titles had\still have problems when running on AMD.. yeah...actually not only on amd but on keplers....I'm pretty sure NV is gonna kill off maxwell the same way after pascal gets released.. Huang wants ur money
And if we take into consideration
nvidias hair vs tressfx 3.0(purehair)..
closed source vs open course...oh boy...
Nvidia has always played dirty... that's why I'm still a bit concerned for AMD even with all these dx12\async compute "wonders"..
Gimpworks+ tons of money. and publishers\developers just looove money... it's hard to lose lol