Oxide now says that NVIDIA does support Async Compute

PRIME1 said:
Luckily there are factory overclocked cards.

You must be new to this whole pc hardware thing. Please do a little research before posting.

As far as the Furry X goes:

http://techreport.com/review/28513/amd-radeon-r9-fury-x-graphics-card-reviewed/14
Click to expand...

When did I say that the 980 GTX isn't smoother then a fury X? IF you actually read what I said, I SAID that no one should recommend a fury/fury X to anyone right now period.

And yes all cards can get Factory overclocks, but then you pay a $10-50 premium.

We are talking about Stock vs Stock normal retail cards. You must be knew to reading posts properly.

P.S. LOVE how to linked an HEAVILY Biased Nvidia website who didn't get a Nano sample because of that fact LOL
 
AMD fans had to grasp to that last little bit of hope that they might make a resurgance during the DX12 era.

Can't wait for Pascal.
 
DASHlT said:
Actually kepler performs worse now then it did before. And the 290x is now faster/equal to the 780ti.

So much for 780ti future proofing lol

http://www.hardwarecanucks.com/foru...iews/70125-gtx-780-ti-vs-r9-290x-rematch.html (Didn't you post this think Prime1 showing the 290x actually being faster/equal to a 780ti?)
Click to expand...
From the article you linked...
mHwQlen.jpg

At launch
980 5% faster than 780ti (happened 2nd)
780ti 9% faster than 290X (happened 1st)
Currently
980 7% faster than 780ti (happened 3rd)
780ti 6% faster than 290X (happened 3rd)
So if a 780ti which is still faster than a 290X is not future proofing then a 290X is not future proofing.
 
That's just another one of the anti-propagandist movement's mythical fairy tales.
At this point nobody cares about the truth. Just whatever they can hype and spin to meet their agenda.

I've stopped fighting it. They'll just call you a liar. Someone else told them it was true, so it must be true... Echo chamber.
 
Ocre said:
an Alpha stage benchmark is sent to the media first, before anyone could even download it.

This just so happens to be the same benchmark AMD was using to showcase mantle on a 290x, 6 months ago, https://www.youtube.com/watch?v=t9UACXikdR0

Might there be a reason this benchmark does so well the 290x?

the developer has an agreement with AMD, just download the benchmark and you will see the AMD logo. But you will see the people making the loudest noise ignoring this entirely.

AMD and Oxide have an agreement, this alpha stage benchmark is rushed to the media and you would have to be dense to not see why. Clearly this is an AMD strategy, dont let anyone tell you otherwise
Click to expand...

If Oxide was sabotaging the benchmark nVidia would run the ass in DX11 mode too. It's nVidia's fault their DX12 pathway is limited. Their drivers can expose Async compute but their hardware doesn't run it. The way Oxide does Async compute is by the book. I'll be curious to see what they have to do to make it run.
 
Last edited:
Igroock said:
I just dont see how a Async chips in AMD cards can give so much performance. A little add-on cannot overtake the primary function of the main GPU chip. If that was the case then why would AMD release Fiji in the fist place when their previous gen cards can get 30% more performance with DX12?

If they can squeeze 30% more performance out of 290/x cards then Fiji is basically pointless. They would have waited until HBM2 was mature and release the next iteration of the Fiji Architecture which surpasses HBM1 performance by a ton in terms of capacity and bandwidth.

What I think is going on is Oxides game is DEEPLY coded in Mantle code, purely optimized for AMD related cards to make a statement perhaps.
Click to expand...


Again, DX11 runs great on nVidia hardware. It's not the game code.
 
Relayer said:
If Oxide was sabotaging the benchmark nVidia would run the ass in DX11 mode too. It's nVidia's fault their DX12 pathway is limited. Their drivers can expose AS but their hardware doesn't run it. The way Oxide does AS is by the book. I'll be curious to see what they have to do to make it run.
Click to expand...

So AMD needs to fix their drivers for released games and NVIDIA needs to fix their drivers for alpha level unreleased games. I will stick to the cards that work well in released games.
 
 
Last edited:
Relayer said:
If Oxide was sabotaging the benchmark nVidia would run the ass in DX11 mode too. It's nVidia's fault their DX12 pathway is limited. Their drivers can expose AS but their hardware doesn't run it. The way Oxide does AS is by the book. I'll be curious to see what they have to do to make it run.
Click to expand...
:confused:
 
Igroock said:
I just dont see how a Async chips in AMD cards can give so much performance. A little add-on cannot overtake the primary function of the main GPU chip. If that was the case then why would AMD release Fiji in the fist place when their previous gen cards can get 30% more performance with DX12?

If they can squeeze 30% more performance out of 290/x cards then Fiji is basically pointless. They would have waited until HBM2 was mature and release the next iteration of the Fiji Architecture which surpasses HBM1 performance by a ton in terms of capacity and bandwidth.

What I think is going on is Oxides game is DEEPLY coded in Mantle code, purely optimized for AMD related cards to make a statement perhaps.
Click to expand...


30% performance increase is just the tip of the iceberg as async shaders are used more we will see more than that.

As gpu's become better at general processing, they will act more like CPU's in the sense they will have similar functionalities that can be tapped into.

Think about this, on a CPU can do pretty much anything you want to programming wise, but the draw back is speed. Lets take graphics for example, it can do it, but just slow.

GPU's get around that because they are specialized in speed and this is done by parallelizing everything, but to get that speed certain design choices have to be made, so general processing is slower, features are cut out, now because transistors budgets can be increased with node drops they can increase things that were "cut out" to make them fast and fast enough to run in real time and this is why they are more complex than CPU's.
 
Last edited:
Semantics said:
From the article you linked...
mHwQlen.jpg

At launch
980 5% faster than 780ti (happened 2nd)
780ti 9% faster than 290X (happened 1st)
Currently
980 7% faster than 780ti (happened 3rd)
780ti 6% faster than 290X (happened 3rd)
So if a 780ti which is still faster than a 290X is not future proofing then a 290X is not future proofing.
Click to expand...

There is no such thing as future proofing, hasn't that been clear for the past 20 years :eek:
 
razor1 said:
There is no such thing as future proofing, hasn't that been clear for the past 20 years :eek:
Click to expand...

more like

There is no such thing as future proofing, hasn't that been clear for the past 30+ years :eek:

Yes I am getting old :eek:
 
And we heard this argument with Xbox 360 win for AMD before when people stated it those fruits will be seen in PC gaming as well, don't see fruits yet, don't even see seeds lol.

Its a lot more than just having console wins to sell discrete cards in the pc market. AMD strategy while admirable and well thought out (in my mind looks like something straight out Sun Tzu's art of war attacking from all sides), they aren't in a position to fully execute it yet, and that is because they don't have the money nor marketshare to fully do it. I don't think they expected to be in the situation they are in now in the market sense and financials. Even if they were still at lets say 40% of the graphics market they still wouldn't have the capital to do it, because the CPU division will still eat up all their profits.
 
razor1 said:
And we heard this argument with Xbox 360 win for AMD before when people stated it those fruits will be seen in PC gaming as well, don't see fruits yet, don't even see seeds lol.

Its a lot more than just having console wins to sell discrete cards in the pc market. AMD strategy while admirable and well thought out (in my mind looks like something straight out Sun Tzu's art of war attacking from all sides), they aren't in a position to fully execute it yet, and that is because they don't have the money nor marketshare to fully do it. I don't think they expected to be in the situation they are in now in the market sense and financials. Even if they were still at lets say 40% of the graphics market they still wouldn't have the capital to do it, because the CPU division will still eat up all their profits.
Click to expand...
Actually AMD is bleeding money from it's GPU division currently it's CPU division actually makes some money probably because they aren't spending shit on r&d.
 
razor1 said:
game code and shader code are two different things I agree its not game code :)
Click to expand...

Well, they are running nVidia supplied shader code. Async Compute still isn't working.

BTW That's what I meant in my previous post when I called it AS. Why I called it AS, I have no idea. Sorry. ;)
 
Semantics said:
Actually AMD is bleeding money from it's GPU division currently it's CPU division actually makes some money probably because they aren't spending shit on r&d.
Click to expand...

AMD stated that most of their R&D is currently spent on the CPU division.
 
Relayer said:
Well, they are running nVidia supplied shader code. Async Compute still isn't working.

BTW That's what I meant in my previous post when I called it AS. Why I called it AS, I have no idea. Sorry. ;)
Click to expand...


Sorry but you really don't know what you are talking about ;). Press go and try again.

Wouldn't it be nice if you could actually back your shit up with some kind of factual proof or theoretical possibilities based on factual proof? Otherwise its just nonsense posts. That way when you stated something people won't be able to argue with you, and have these threads get crapped on..
 
Semantics said:
Actually AMD is bleeding money from it's GPU division currently it's CPU division actually makes some money probably because they aren't spending shit on r&d.
Click to expand...



Both divisions are losing money right now its not limited to just GPU's.

http://csimarket.com/stocks/segments.php?code=AMD

look at the break down of where money is coming in from, enterprise embedded and semicustom is higher and both those tend to be low margin
 
DASHlT said:
Thats great....but when a normal average Joe goes to buy a video card, they don't care what overclocks a video card can get.

And besides. Who would recommend a fury at $570 when you can get a 980ti on sale for $629-$639? I mean it's pointless.

But saying (an overclocked card can be faster then a stock card) Is just stupid to even say.
Click to expand...

Overclocked performance is very relevant on forums such as this, OEMs make the decision for most consumers. If you are buying standalone cards the overclocking headroom is very important, the Fury X has practically no overclocking head room and the 290x/390x has maybe 10%. The average joe does not buy standalone video cards.
 
razor1 said:
There is no such thing as future proofing, hasn't that been clear for the past 20 years :eek:
Click to expand...

Really? My R9 290 is still working great and it is 2 years into the future from when I bought it. This is not the 1990's anymore. Also, the only card even worth upgrading to that would show a difference is the R9 Fury X or 980Ti and I will not spend more than $500 on a card. (Reference R9 290 so I cannot crossfire with it. Works great on its own though.)
 
ManofGod said:
Really? My R9 290 is still working great and it is 2 years into the future from when I bought it. This is not the 1990's anymore. Also, the only card even worth upgrading to that would show a difference is the R9 Fury X or 980Ti and I will not spend more than $500 on a card. (Reference R9 290 so I cannot crossfire with it. Works great on its own though.)
Click to expand...


yeah but graphics card gen lengths have increased, while before the 8800 and x2900 series it was 1 year per gen, now its 1.5 years per gen, if their is any delay its going up to 2 years. Also we aren't seeing the mid gen updates all the time or all segments either.
 
razor1 said:
yeah but graphics card gen lengths have increased, while before the 8800 and x2900 series it was 1 year per gen, now its 1.5 years per gen, if their is any delay its going up to 2 years. Also we aren't seeing the mid gen updates all the time or all segments either.
Click to expand...

Yeah, my guess is they pretty much are topping out on 28nm for the most part. In fact, I remember how the R9 290 at $400 was a great deal when it was first released and it is turned out to be better than that. Perhaps the game devs need to make better use of what is available first.
 
razor1 said:
Sorry but you really don't know what you are talking about ;). Press go and try again.

Wouldn't it be nice if you could actually back your shit up with some kind of factual proof or theoretical possibilities based on factual proof? Otherwise its just nonsense posts. That way when you stated something people won't be able to argue with you, and have these threads get crapped on..
Click to expand...

Baker posted that they were running nVidia shader code and it didn't use Async compute. It's been posted many times. I'm not going to bother and find it again. Especially in a few posts you'll just go back to making believe he never said it.
 
ManofGod said:
Yeah, my guess is they pretty much are topping out on 28nm for the most part. In fact, I remember how the R9 290 at $400 was a great deal when it was first released and it is turned out to be better than that. Perhaps the game devs need to make better use of what is available first.
Click to expand...


Same could be said for the 780 ti, still does well for Dx11 games, but won't do too well comparative to the r290 in the first batch of Dx12 games though.

mid 2016 when games that use more compute shaders, I don't think the r290 will be able to keep it going. With that extra performance that we get from Dx12 it is going to be used up for other things.
 
Relayer said:
Baker posted that they were running nVidia shader code and it didn't use Async compute. It's been posted many times. I'm not going to bother and find it again. Especially in a few posts you'll just go back to making believe he never said it.
Click to expand...

Yes he did, but why is this nV code going slower on Dx12 than on Dx 11 paths at times? well there is the question that you should be asking. Dx12 path should still get the cpu overhead advantage, why isn't that showing up? Another question you should be asking. Did nV put it in or did nV give it to Oxide to put it in? Another question that should be asked. When was this code put in? Was it put in as an preemptive measure for temporary reasons?, due to the time constraints on the release of the application?
 
razor1 said:
Yes he did, but why is the this nV code going slower on Dx12 than on Dx 11 paths? well there is the question that you should be asking. Dx12 path should still get the cpu overhead advantage, why isn't that showing up? Another question you should be asking.
Click to expand...

Ask nVidia. They wrote it.

Wouldn't it be nice if you could actually back your shit up with some kind of factual proof or theoretical possibilities based on factual proof?
Click to expand...
So you knew what my proof was before you wrote this.
 
Relayer said:
Ask nVidia. They wrote it.


So you knew what my proof was before you wrote this.
Click to expand...


Is it fact or just being reworded to show something you want to show? Its great you have a point of view, that is not really a very good point of view, because all those questions aren't answered by that one statement.

Writing the code isn't the same as implementation. Hint there might have been changes done to the code for implementation. I'm sure you are only conformable with plug and play, its a bit more then that.
 
You must log in or register to reply here.
Back
Top