• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Oxide Developer: “NVIDIA Was Putting Pressure On Us To Disable Certain Settings In The Benchmark”

Rroff said:
People need to buy like me :P the 780 GHZ will last me til DX11 is done and mean time I've saved a load of money for when I need a card for DX12. :smug:
Click to expand...


i get bored with a GPU after a year max and get an itch i need to scratch, actually i'm on my forth GPU in 2 years.

I'm not as bad as Kaap or Boom tho.....
 
humbug said:
This is interesting.

Direct Compute and Shader Combined rendering latency

GTX 980TI, a clear delay here



Tahiti (GCN 1.0) 7970 / R9 280, nothing, nada, nichts..... absolutely simultaneous and parallel. That looks like AMD HSA for GPU's at work here, awesome.

Click to expand...


Those are interesting graphs, being that compute plus rendering is higher than just compute on the NVidia graph, we have to assume that the higher the score on the vertical scale the worse it is, which doesn't say a lot for the AMD graph with a score nearly double the NVidia one.
Not saying that AMD's GCN doesn't use Async compute much better NVidia Maxwell, just saying that those graphs don't really make AMD look too good. Great that there is no added latency when you do rendering and compute on AMD, but the latency is twice as bad to begin with, according to those graphs.
 
bru said:
Those are interesting graphs, being that compute plus rendering is higher than just compute on the NVidia graph, we have to assume that the higher the score on the vertical scale the worse it is, which doesn't say a lot for the AMD graph with a score nearly double the NVidia one.
Not saying that AMD's GCN doesn't use Async compute much better NVidia Maxwell, just saying that those graphs don't really make AMD look too good. Great that there is no added latency when you do rendering and compute on AMD, but the latency is twice as bad to begin with, according to those graphs.
Click to expand...

Keep in mind it's a 7970 v a gtx980ti. The gtx980ti compute + graphics line does eventually get upto 60 the same as the 7970.
 
keep in mind also that all gxn products perform about the same on that test. From laptops to Fury X. Test is still way too light for GCN to get any slowdowns.
 
Interesting that Nvidia has ASync Compute capabilities that shows up in the drivers but didn't work when Oxide tried to enable it. I am very interested to see what Nvidia has to say about that and if it is indeed not a possibility to have in use. If Async Compute is a massive plus for AMD then that is great as well as it gives AMD users a head start. I am sure Pascal will have it sorted if indeed Maxwell isn't capable and this could well play into the hands of Nvidia.
 
oldestgregg said:
"the only 2 cards which might benefit from DX12 are the Titan X (confirmed 2 DMA engines) and the 980ti (suspected), Maxwell Gen 1 will no see any performance boost"

Can anyone explain the bit about the DMA engines to me please? And does this affect asyncCompute?
Click to expand...

I don't know about that, but Maxwell Gen 2 cards are:

950GTX, 960 GTX, 970 GTX, 980GTX, 980Ti and Titan X.
 
Gregster said:
Interesting that Nvidia has ASync Compute capabilities that shows up in the drivers but didn't work when Oxide tried to enable it. I am very interested to see what Nvidia has to say about that and if it is indeed not a possibility to have in use. If Async Compute is a massive plus for AMD then that is great as well as it gives AMD users a head start. I am sure Pascal will have it sorted if indeed Maxwell isn't capable and this could well play into the hands of Nvidia.
Click to expand...

It did work (as it does work on the test above), but not efficiently so it was disabled. Also worth mentioning that the stuff they do with asynchronous shaders is rather light (their words) compared with what others already do on consoles.
 
Last edited:
bru said:
Those are interesting graphs, being that compute plus rendering is higher than just compute on the NVidia graph, we have to assume that the higher the score on the vertical scale the worse it is, which doesn't say a lot for the AMD graph with a score nearly double the NVidia one.
Not saying that AMD's GCN doesn't use Async compute much better NVidia Maxwell, just saying that those graphs don't really make AMD look too good. Great that there is no added latency when you do rendering and compute on AMD, but the latency is twice as bad to begin with, according to those graphs.
Click to expand...



Its already been pointed out.... but i was going to say that is a 7970 vs a 980TI.
 
Last edited:
Calin Banc said:
It did work (as it does work on the test above), but not efficiently so it was disabled. Also worth mentioning that the stuff they do with asynchronous shaders is rather light (their words) compared with what others already do on consoles.
Click to expand...

I am quite behind unusually on what DX12 can and can't bring and seeing Async bringing the Fury X on par with the 980Ti is a good thing in my books. Consumers now have a choice of The Fury X or the 980Ti (speaking high end) and what best suits them. DX12 on the level and DX11 favouring Nvidia. A healthy option indeed.
 
LoadsaMoney said:
No ones going to use the Async shaders, as the markets not big enough, its only AMD, who have only got a piddly 15% share of it or something, its not going to be worth it to them, catering for a tiny market like that.
Click to expand...

Aside from the points others made about the consoles which is a far bigger feature driver than discrete gpus in general, people need to understand that sales in a quarter has absolutely nothing to do with the current market or who devs target. Devs will target a fairly broad range of cards from the past 3-4 years in general. They don't make games based on the past quarter of sales. Nvidia hasn't sold 80% of all gpus in the past 4 years, JUST this quarter.

Devs aren't targeting people who bought a Dell box three years ago with either a AMD 7450 or a Nvidia equivalent(not even sure what that is, a 540 gt, I guessed that name, you get the idea). These aren't what a AAA title is going after in terms of support. Of all the actual discrete gpus AMD/Nvidia sell, a very small total number are used by gamers. Nvidia is currently killing it in Dell box sales but that has the statistically lowest portion of gamers. Discrete gpu sales to end users via stores like OCUK is no where near a 80/20 split and that will have a far heavier gamer demographic. You can really see this easily on any forum including this one, there isn't an 80/20 split in gaming, not even close.

Then lets talk about DX12. I'd imagine for most DX12 games the broadest level of support will be DX12_0. More AMD cards by a fairly large margin support more of the most important DX12 features. Everything actually new in the 7xxx series and forward support more DX12 features than everything Nvidia made up till Maxwell at least V1 but maybe even V2, though with one card in V1 it's not a huge difference.

In DX12 terms, AMD will have more gamers with DX12_0 support than Nvidia does. Sales of the past quarter is so utterly misleading in terms of gaming market share it's a ludicrous metric to use for this. Tack on console sales and AMD DX12 'market share' within gaming is extremely strong.
 
humbug said:
Its already been pointed out.... but i was going to say that is a 7970 vs a 980TI.
Click to expand...

Some have been saying its indicating more the case,as tasks are added in parallel performance drops.

But the thing is it also means that people who bought an aftermarket R9 290 instead of GTX780 after launch and even those with some of the older GCN based cards like the HD7970 and HD7950 are going to see some decent gains over the equivalent Nvidia Kepler cards of the day.
 
I honestly cannot see the GTX 480 / 580 getting DX12 drivers - why? they have so few features , it`ll be like the 5950 all over again - enable DX9 on that card and you have a leaf blower sounding slide show
 
drunkenmaster said:
Discrete gpu sales to end users via stores like OCUK is no where near a 80/20 split and that will have a far heavier gamer demographic. You can really see this easily on any forum including this one, there isn't an 80/20 split in gaming, not even close.
Click to expand...

erm, have you actually looked at any of the "owners" threads recently?
 
Rusty0611 said:
Yet you said nothing about DMs 'scientific' analysis which also was entirely anecdotal.

He's the one disputing the actual figures with an anecdote in the first instance so pick way at him instead.
Click to expand...

But the problem is nether of us have segment to segment sales breakdowns either,so ultimately you are only looking at an average of all segments.

Plus its a card shipment figure,not even an actual marketshare metric.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom