apparently even 980Ti can't handle async-compute without a crippling context switch

[Carfax83]

the context switches aren't baked into the hardware.

Not sure what you mean by that.. Context switches are always done in hardware in GPUs. AMD likely has faster context switching than NVidia when it comes to AC though..

 

[RussianSensation]

Too much rumors and people talking about things they have no idea how they actually work. I'll rather wait for a reputable source to research into this.. like Anandtech, although they are quite slow these days.

I'll admit that watching various forum flame wars is kinda entertaining tho.

Guru3D article here has a lot of the details summarized in 1 post:

http://www.guru3d.com/news-story/nv...12-benchmark-to-disable-certain-settings.html

If what's outlined there is true, it's a big deal even if DX12 games wont' come out for 2-3 years because it would mean NV will have falsely misrepresented full DX12 support on its products. In the grand scheme, a lot of PC gamers will upgrade to next gen GPUs for true DX12 games but a lot of NV loyalists on here proclaimed how NV's DX12.1 feature set is an advantage and NV's own marketing I am sure resulted in some gamers buying GTX900 cards over GCN 1.0-1.2 cards for "better/full DX12 support."

Some GTX670/680/770/780/HD5000/7000 gamers will surely upgrade to 16nm HBM2 GPUs but what about all those gamers buying GTX950/960/970/980/980Ti now? I doubt they would have been pleased if this is true:

"In short, here's the thing, everybody expected NVIDIA Maxwell architecture to have full DX12 support, as it now turns out, that is not the case. AMD offers support on their Fury and Hawaii/Grenada/Tonga (GCN 1.2) architecture for DX12 asynchronous compute shaders. The rather startling news is that Nvidia's Maxwell architecture, and yeah that would be the entire 900 range does not support it."~ Guru3D

With GTX970 false specifications advertising and now if THIS is also true, I'd say it's pretty shady use of marketing business practices. We still need to see 2-3 more DX12 games to validate Oxide's point of view. It's going to be a while before we can evaluate DX12 performance as most AAA DX12 games are some time away. Also, surely GW DX12 titles will be trying to cover up NV's Maxwell's DX12 weaknesses, if they exist, because GW is a NV paid for co-marketing program.

Seems like a great selling point for NVIDIA's next GPU relative to prior generation ones

GTX970 3.5GB by design. Kepler compute and driver crippled by design. NV charging premiums for extra VRAM to get gamers to buy 2GB GPUs like 670/680/770/950/960, fully knowing those are obsolete. I guess a CEO would call it smart business strategies. Apple is upgrading its rear camera to 12MP from 8MP and moving to 4K but keeping the base 6S/6S+ models at 16GB so that it either forces consumers to pay $100 extra for the useful 64GB version or gives people incentive for iPhone 7 upgrade. NV is learning a lot from Apple

I don't think this will matter for GTX980Ti owners as those cards have a solid 18-25% advantage in max overclocked states vs. Fury X, but for someone who can't afford to buy new high-end cards every 2 years, it's going to be hard to recommend GTX950/960/970 over R9 380/280X/290/390.

 

[thilanliyan]

Of course the GTX 980ti being $650 is a scam in the first place, flagship graphic cards have always been around $500, we've never really had overpriced turds until the first OVERPRICED TURD the GTX TITAN. That was the first overpriced turd at $1000.[/COLOR][/COLOR]

What? Weren't cards like the X1900xtx and 6800ultra more like $600/650?

 

[railven]

Saw this post on reddit. Summarizes my feelings. I don't regret buying my 970s. The performance for what I paid is worth it. I don't expect a first gen DX12 part to be flawless in DX12. That wasn't the case for DX11 or DX10 either, it took newer cards to really push the performance up.

Pretty much how I feel, too. By the time this really matters to me as a consumer, there will be new cards to pick from. And like the last two times instead of just buying AMD's top single GPU card, I'll compare the two camps.

Funny thing is, the type of consumer I am is what is giving Nvidia revenues in a diminishing market. AMD users will probably opt to hold on to their cards a tad longer and see what the new wave are cards bring about.

For AMD's sake, I hope Pascal doesn't decimate them at their own game.

 

[cmdrdredd]

Pretty much how I feel, too. By the time this really matters to me as a consumer, there will be new cards to pick from. And like the last two times instead of just buying AMD's top single GPU card, I'll compare the two camps.

Funny thing is, the type of consumer I am is what is giving Nvidia revenues in a diminishing market. AMD users will probably opt to hold on to their cards a tad longer and see what the new wave are cards bring about.

For AMD's sake, I hope Pascal doesn't decimate them at their own game.

Yeah competition is good, but looking at these first DX12 benchmarks the performance is close now. Even without the Async-Compute benefit(maybe part of the reason Ashes sees huge AMD boosts and Nvidia doesn't get much). If Pascal boosts performance by any significant amount it can hurt AMD who just released cards recently. Then prices can go crazy as a result. Just not a good situation.

 

[happy medium]

it's going to be hard to recommend GTX950/960/970 over R9 380/280X/290/390.

Like someone had said earlier in this thread, I doubt these mid range cards will play real direct X 12 games anyway at playable framerates, especially at everyones magic 60fps number.

Next year AMD and Nvidia will have the next gen that will be fully compatable with direct X 12 and have the power to run those games.

This sounds more like the AMD bicycle trying to stop the Nvidia train from taking more than the already 80% of the GPU marketshare that Nvidia has.
It also provides a good platform to point out the same stuff that the AMD shills and fanboys have been crying about for months.

1. gtx970 has only 3.5gb of memory
2. Nvidia is killing gaming with gameworks, ect ect ect.
3. Nvidia stopped supporting Kepler so don't buy Maxwell because even a 290x will be faster.
4. AMD's cards will play direct x 12 games better.
Yada Yada Yada..........
5.the only reason Nvidia has more market share is their marketing.

Take a good look at the post, these shills are as plain as day. Not just in this thread in every thread.

 

[railven]

Yeah competition is good, but looking at these first DX12 benchmarks the performance is close now. Even without the Async-Compute benefit(maybe part of the reason Ashes sees huge AMD boosts and Nvidia doesn't get much). If Pascal boosts performance by any significant amount it can hurt AMD who just released cards recently. Then prices can go crazy as a result. Just not a good situation.

Price wise, I've already accepted the climbing ceiling. Once cards get out of my comfort zone, guess I'll buy more console games :/

Start of a new gen - DX12 era. See how it goes. If an AMD sponsored game is indicative of anything, is that AMD sponsored tech works on AMD hardware. So I guess, when Gameworks games work on Nvidia cards better, we're getting closer to the console battle lines.

If I have to own an AMD and Nvidia card, I got no problem with that

 

[cmdrdredd]

Price wise, I've already accepted the climbing ceiling. Once cards get out of my comfort zone, guess I'll buy more console games :/

Start of a new gen - DX12 era. See how it goes. If an AMD sponsored game is indicative of anything, is that AMD sponsored tech works on AMD hardware. So I guess, when Gameworks games work on Nvidia cards better, we're getting closer to the console battle lines.

If I have to own an AMD and Nvidia card, I got no problem with that

Lol I don't wanna own one of each heh.

 

[railven]

Lol I don't wanna own one of each heh.

Understandable, but I got two PC gamers in my house hold. She's an NV fangirl and I prefer Radeons.

We can just go back to the status quo for our household .

 

[monstercameron]

Price wise, I've already accepted the climbing ceiling. Once cards get out of my comfort zone, guess I'll buy more console games :/

Start of a new gen - DX12 era. See how it goes. If an AMD sponsored game is indicative of anything, is that AMD sponsored tech works on AMD hardware. So I guess, when Gameworks games work on Nvidia cards better, we're getting closer to the console battle lines.

If I have to own an AMD and Nvidia card, I got no problem with that

why are you promoting this fallacy?

 

[railven]

why are you promoting this fallacy?

How is that a fallacy? A game that is sponsored by AMD is working great on their hardware. Where is the fallacy?

 

[monstercameron]

How is that a fallacy? A game that is sponsored by AMD is working great on their hardware. Where is the fallacy?

sigh, so the misinformation continues.

 

[railven]

sigh. and so the misinformation continues.

Haha, I pretty much said the same thing for Nvidia right after. Oh well.

EDIT: Since you edited your post:
What misinformation? Correct me if I said anything wrong instead of going emo on me. Woof.

 

[cmdrdredd]

why are you promoting this fallacy?

http://www.ashesofthesingularity.com/

Look at the bottom of the page and notice the AMD logo. Whether AMD is putting any resources into the game or not, they are getting advertising. It was also showcased at AMD's Booth at GDC.

 

[railven]

http://www.ashesofthesingularity.com/

Look at the bottom of the page and notice the AMD logo.

Dat misinformation!!!

Railven said:
AMD sponsored game works great on AMD hardware, Nvidia sponsored game works great on Nvidia hardware. Battle lines are drawn!

He said:
Sigh, so much misinformation!

Sometimes I don't get people around here. I pretty much said water is wet and he acts like I drowned his puppy in the process.

 

[cmdrdredd]

Dat misinformation!!!

Railven said:
AMD sponsored game works great on AMD hardware, Nvidia sponsored game works great on Nvidia hardware. Battle lines are drawn!

He said:
Sigh, so much misinformation!

Sometimes I don't get people around here. I pretty much said water is wet and he acts like I drowned his puppy in the process.

It might be semantics. Maybe AMD isn't sponsoring it or putting any money into it, but Stardock did work with AMD pretty closely on other projects. Maybe that means something, maybe it means nothing.

 

[monstercameron]

Dat misinformation!!!

Railven said:
AMD sponsored game works great on AMD hardware, Nvidia sponsored game works great on Nvidia hardware. Battle lines are drawn!

He said:
Sigh, so much misinformation!

Sometimes I don't get people around here. I pretty much said water is wet and he acts like I drowned his puppy in the process.

read this again before you start spreading misinformation.
http://www.overclock.net/t/1569897/...ingularity-dx12-benchmarks/1200#post_24356995

Certainly I could see how one might see that we are working closer with one hardware vendor then the other, but the numbers don't really bare that out. Since we've started, I think we've had about 3 site visits from NVidia, 3 from AMD, and 2 from Intel ( and 0 from Microsoft, but they never come visit anyone ;(). Nvidia was actually a far more active collaborator over the summer then AMD was, If you judged from email traffic and code-checkins, you'd draw the conclusion we were working closer with Nvidia rather than AMD wink.gif As you've pointed out, there does exist a marketing agreement between Stardock (our publisher) for Ashes with AMD. But this is typical of almost every major PC game I've ever worked on (Civ 5 had a marketing agreement with NVidia, for example). Without getting into the specifics, I believe the primary goal of AMD is to promote D3D12 titles as they have also lined up a few other D3D12 games.

 

[tg2708]

https://www.reddit.com/r/nvidia/comments/3j5e9b/analysis_async_compute_is_it_true_nvidia_cant_do/

 

[railven]

read this again before you start spreading misinformation.
http://www.overclock.net/t/1569897/...ingularity-dx12-benchmarks/1200#post_24356995

So, I was right. Thanks

As you've pointed out, there does exist a marketing agreement between Stardock (our publisher) for Ashes with AMD.

AMD is sponsoring this game.

 

[monstercameron]

So, I was right. Thanks



AMD is sponsoring this game.

from a marketing perspective yes, not from an optimzation/performance one. You are twisting this to point to some non-existent amd advantage, when they are only co-marketing the game. Stop spreading such misinformation.

 

[cmdrdredd]

https://www.reddit.com/r/nvidia/comments/3j5e9b/analysis_async_compute_is_it_true_nvidia_cant_do/

Just read that, seems the implementation is the key? Whatever Oxide is doing doesn't work for Nvidia cards perhaps and Nvidia asked them to just turn it off because it's overloading the work limit and causing it to be slow. If it was made to stay within the limits and use methods Nvidia recommends performance remains high then.

So I guess the end result is it does work and it's fast but it has limits before it slows down. Oxide seems to have falsely claimed Nvidia cards can't do it simply because of their results that don't match the results from beyond3D.

I downloaded the test from that site and compared it to results from others. My 970 beats the Fury X in compute only but Graphics and compute starts off strong and ends up slowing down near 70 command lists to even out or be slightly behind.

 

[railven]

from a marketing perspective yes, not from an optimzation/performance one. You are twisting this to point to some non-existent amd advantage, when they are only co-marketing the game. Stop spreading such misinformation.

Okay, I guess AMD putting money in the pot - it's just coincidental they went with the API that favors AMD.

If DX12 wasn't ready for release - this game would probably have a Mantle pathway.

Unless AMD just wants to give this dev and their publisher money, because they sure have tons of it to spread. Be serious. It doesn't negatively hurt AMD if they are finally playing hardball. Hell, if anything - about damn time.

 

[Hitman928]

https://www.reddit.com/r/nvidia/comments/3j5e9b/analysis_async_compute_is_it_true_nvidia_cant_do/

Just read that, seems the implementation is the key? Whatever Oxide is doing doesn't work for Nvidia cards perhaps and Nvidia asked them to just turn it off because it's overloading the work limit and causing it to be slow. If it was made to stay within the limits and use methods Nvidia recommends performance remains high then.

So I guess the end result is it does work and it's fast but it has limits before it slows down.

The post doesn't even address the issue at hand. It completely ignores what the results show in how each handles Asynch tasks. As far as total time listed for both AMD and Nvidia, the original benchmark creator is looking into it. Someone suggested that they take GPU timestamps rather than CPU as the CPU time may give erroneous results, which may make the whole test moot and all we have again is the Oxide results.

BTW, Nvidia was contacted by wccftech for a response. Normally I don't care about the site, but I didn't see anyone else reach out for a respone.

We at WCCFTech tried to contact NVIDIA about this comment, and Senior PR Manager Brian Burke only had this to say:


"We’re glad to see DirectX 12 titles showing up. There are many titles with DirectX 12 coming before the end of the year and we are excited to see them."


Read more: http://wccftech.com/amd-nvidias-max...le-of-performing-async-compute/#ixzz3kRPysrz6

I'm not saying it's entirely settled in my mind, but from what Oxide has said and the way all the data is going coupled with Nvidia's pointed silence, it seems far more likely to be true at this point. If not, Nvidia should probably say something sooner than later.

 

[tential]

When all is said and done... this really isn't going to be a big deal. That's the best part.

 

[guskline]

I didn't put all my eggs in one basket. See my rigs below.