ShintaiDK
Lifer
- Apr 22, 2012
- 20,378
- 145
- 106
Yep, its far from as easy as some people think. Specially when you cant limit it to selected cards
Makes all those claims of "Advanced API" support plastered all over the marketing materials and packaging of a certain IHV even more laughable. If there was the slightest chance that dx12 would do better on serial API optimised(by design) hardware, the dx 12 implementation would be much further along. Just a pity the marketing doesn't match the reality.
Some users in this forum said that Maxwell is finished and end of life ,however, that end life product is man handling AMD best GPU.
A GTX 980 TI is 20-30% then fury x.
http://www.pcgameshardware.de/Battl...chnik-Systemanforderungen-Benchmarks-1206197/
A GTX 980 TI is 20-30% then fury x.
http://www.pcgameshardware.de/Battl...chnik-Systemanforderungen-Benchmarks-1206197/
According to the link the low level aspect seems to be working well in the beta- FX8350+Nano is ~50% faster at 720p under DX12. They probably still have work to do on the graphics side of things and ofcourse drivers to match and surpass DX11 under higher resolutions. Lets not forget its just a beta.
Right, which explains why the GTX 1080 with its 2560 shaders and 256 bit bus is killing the Fury X with its 4096 shaders and 4096 bit bus in Ashes of the Singularity, the greatest and most polished example of DX12 optimization so far.
The reason why NVidia has the edge in DX11, is because they've tuned their driver to create worker threads using the CPU to parallelize rendering. Basically, NVidia's DX11 driver has been doing what DX12 was designed to do for years, which is to lower CPU overhead and parallelize rendering. Ironically, this also explains why NVidia doesn't get as large a gain when going to DX12 as AMD.
That's not to say that NVidia's DX11 driver can equal what an expertly optimized DX12 implementation can do however. But it gets close enough to matter..
Last edited:
ShintaiDK
Lifer
- Apr 22, 2012
- 20,378
- 145
- 106
Right, which explains why the GTX 1080 with its 2560 shaders and 256 bit bus is killing the Fury X with its 4096 shaders and 4096 bit bus in Ashes of the Singularity, the greatest and most polished example of DX12 optimization so far.
Not sure what the meaning with your bus width and SP count is.
Could you elaborate?
My point is that FuryX is a much wider GPU than the GTX 1080 and should theoretically be capable of greater exploitation of "parallel APIs" like DX12, but it still loses against the latter which is a narrower but deeper design.
Basically, I'm making fun of the whole NVidia" serial API optimized design" nonsense.
crisium
Platinum Member
- Aug 19, 2001
- 2,643
- 615
- 136
1440 maxed out (default resolutions scale) runs fine. I also increased the FOV from 70 HOR to 90 HOR. 60s and 70s framerate with high action scenes sometimes running in the 50s. With Freesync no problems here. Fine with 4GB of VRAM.
Not sure if I want FXAA or TAA for this one. I didn't get to play much last night so maybe I'll compare them today (well I played over an hour actually but had too much fun to bother comparing). But neither of them really seemed that great here at first glance.
And as far as gameplay goes, I'm pleased with what they did with the bolt action rifles. Body shot 1 hit kills would be too strong but the higher ROF helps make bolt action more dominant which they were at the time, especially since much of the automatic weapons were prototypes or limited 1918 German only use. Bolt action rifles and riding around as an Ottoman soldier on horseback with a saber are a fun concept.
Not sure if I want FXAA or TAA for this one. I didn't get to play much last night so maybe I'll compare them today (well I played over an hour actually but had too much fun to bother comparing). But neither of them really seemed that great here at first glance.
And as far as gameplay goes, I'm pleased with what they did with the bolt action rifles. Body shot 1 hit kills would be too strong but the higher ROF helps make bolt action more dominant which they were at the time, especially since much of the automatic weapons were prototypes or limited 1918 German only use. Bolt action rifles and riding around as an Ottoman soldier on horseback with a saber are a fun concept.
Last edited:
DamZe
Member
- May 18, 2016
- 188
- 84
- 101
TAA without a sharpening filter is just garbage. SweetFX via ReShade should do wonders here.
custom 980TI is faster than custom 1070 so no surprise here.
EDIT:Custom 1070 is only 10% faster than furyx in 1440p.Stock 1070 will be probably slower than furyx..
I've done limited comparisons myself now, and at least on my Kepler Titan + i7 4770k, DX11 and DX12 performance seem to be close to eachother, can't say one is constantly providing the better performance. Settings were high at 1440. EDIT: Seems like DX11 is a bit better
Also gave the older drivers a go in both DX11 and DX12 before updating to the game ready drivers, possible the newest drivers increased performance by a few FPS.
Also gave the older drivers a go in both DX11 and DX12 before updating to the game ready drivers, possible the newest drivers increased performance by a few FPS.
Last edited:
Arachnotronic
Lifer
- Mar 10, 2006
- 11,715
- 2,012
- 126
Arachnotronic
Lifer
- Mar 10, 2006
- 11,715
- 2,012
- 126
980 Ti is a wider machine, more SPs, more texture units, more ROPS, greater memory bandwidth, etc.
Think of GTX 1070 as a cheaper-to-build 980 Ti with more VRAM and much lower power consumption.
Wow - this is beta. How difficult is that to understand? If you really believe the DX12 path will be slower than DX11 for the life of this game, I've got a bridge to sell you.
Source desperately needed. This sounds 100% fake. No way this is true. There is zero chance any DX11 implementation can reach the same low levels of overhead as DX12 (which is what DX12 is "designed to do," among other things) by the very nature of it.
I dont know where the tribes got off the rails and partisanized low overhead APIs but the fact of the matter is DX11 is a high level of abstraction with high levels of overhead and DX12 is low abstraction, low overhead. It's a very simple and very well understood trade off that's been happening in software development for 30 years. C++ vs Assembly is the same thing. Python vs C++. Writing your own webserver in C is probably faster than using whatever one comes out of the box in your PHP framework of choice but it takes a lot more time and skill
computerbase benchmarks...Surprise ref 1070 is 10% faster than ref 980TI.Too bad no1 buy reference models.Aftermarket 980TI is faster than aftermarket 1070.I dont know why they even test ref models.
They should only test aftermarket cards.Those cards actually people buy
https://www.computerbase.de/2016-08/battlefield-1-beta-benchmark/
They should only test aftermarket cards.Those cards actually people buy
https://www.computerbase.de/2016-08/battlefield-1-beta-benchmark/
tweakboy
Diamond Member
It's not fake and more importantly, you aren't even getting what he said. No where did he say DX11 is a low level API. Relax, read it again.
http://www.overclock.net/t/1573982/amd-gpu-drivers-the-real-truth
Last edited:
I always prefer tests with reference models. Then you know what the normal model gets and know that any higher clocked aftermarket GPU will be better.
Azix
Golden Member
- Apr 18, 2014
- 1,438
- 67
- 91
What are you talking about really? narrower? wider? deeper? huh? The 1080 clocks high enough to be faster, end of story.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter