nVidia - Why We’re Investing Heavily in GameWorks

[humbug]

Yes, the devs can send the game for profiling, however AMD claim that this makes it impossible to optimise, without source code.

Working with developers does not equal source code access, but that is what AMD would uave you believe.

At the end of the day, any code being run through Dx in to YOUR hardware via YOUR drivers, you can profile and see what it is doing, to say it is impossible to do any driver optimisation without source code is nonsense.

Its not nonsense, without the source code you cannot optimise for your hardware, at all. they might as well not bother as you can't do anything with it.

Whats to optimise if you don't know what its doing.

 

[Rroff]

Its not nonsense, without the source code you cannot optimise for your hardware, at all. they might as well not bother as you can't do anything with it.

Whats to optimise if you don't know what its doing.

Maybe AMD needs to ask nVidia really nicely for a copy of perfkit

 

[humbug]

Maybe AMD needs to ask nVidia really nicely for a copy of perfkit

Or pretend to be an Nvidia partner.

 

[KillBoY_UK]

Its not nonsense, without the source code you cannot optimise for your hardware, at all. they might as well not bother as you can't do anything with it.

Whats to optimise if you don't know what its doing.

Source code by it self is not to useful unless is well documented/commented otherwise it can be extremely hard to read to someone who has not wrote said section of code.

I bet my life on it that both AMD and Nvidia have fancy debuggers to see what and how a application is interacting with there drivers and data gathered from that could be use to optimise there drivers to a application. Ofc Source code access (which is readable) is ideal when you plenty of man hours to optimise against it.

 

[KillBoY_UK]

It's because AMD are the good guys and NVidia are the devil incarnated on Earth to ruin PC gaming, I thought everybody knew that by now.

Did NVidia users complain about an AMD conspiracy when Tomb Raider (an AMD sponsored game) with TressFX ran poorly at launch on NVidia? nope.

Did NVidia themselves launch a massive media campaign trying to make out AMD as the bad guys and absolving themselves of all responsibility that it ran poorly? nope.

NVidia went quietly to the developers, worked with them to patch the game and then released a new set of drivers thus bringing the game up to par on NVidia hardware.

All you ever get from AMD is "it's NVidia's fault", "it's the developers fault", it's always everyone elses fault and they don't have to lift a finger because their army of fanboys always back them and criticise everybody else.

I tend to agree when looking at this forum sometimes

 

[Marine-RX179]

Just for the record, I have grabbed a couple of bench results from our threads.

Sniper Elite V3

Score 75.4 GPU 290X @1344/1740, CPU 4930k @4.79, Pulse88 Link Mantle 14.9 Drivers
Score 75.3 GPU 980 @1610/2120, CPU 5960X @4.86, Silent_Scone Link DX11 344.16 Drivers
Score 74.3 GPU 980 @1603/2115, CPU 3930k @5.0, Besty Link DX11 344.16 Drivers
Score 73.6 GPU 290X @1356/1740, CPU 4930k @4.79, Pulse88 Link DX11 14.9 Drivers
Score 73.5 GPU 290X @1255/1625, CPU 3970X @4.9, AMDMatt Link Mantle 14.9 Drivers
Score 72.5 GPU 290X @1344/1725, CPU 4930k @4.75, Pulse88 Link DX11 14.9 Drivers
Score 71.3 GPU 290X @1270/1625, CPU 4930k @4.8, Kaapstad Link Mantle 14.9 Drivers
Score 71.1 GPU 290P @1285/1625, CPU 2600k @5.0, Uncle Petey Link Mantle 14.9 Drivers
Score 70.0 GPU 290P @1225/1650, CPU 3770k @4.6, win8.1 Link Mantle 14.9.1 Drivers
Score 69.9 GPU 290X @1250/1625, CPU 3970X @4.9, AMDMatt Link DX11 14.9 Drivers
Score 69.9 GPU 290P @1240/1625, CPU 3970X @4.9, AMDMatt Link Mantle 14.9 Drivers
Score 68.6 GPU 290X @1280/1625, CPU 4930k @4.8, Kaapstad Link DX11 14.9 Drivers
Score 64.1 GPU 290P @1180/1600, CPU 4790k @4.8, MadMatty Link Mantle 14.9 Drivers
Score 63.2 GPU 290P @1200/1500, CPU FX-9590 @4.7, Sgt Bilko Link Mantle 14.9 Drivers
Score 62.5 GPU 290P @1200/1450, CPU FX-8350 @4.5, humbug Link Mantle 14.9.2 Drivers
Score 61.9 GPU 290X @1200/1500, CPU 3770k @4.5, tommybhoy Link Mantle 14.9 Drivers
Score 59.7 GPU 290P @1200/1450, CPU FX-8350 @4.5, humbug Link DX11 14.9.2 Drivers
Score 58.8 GPU 980 @1241/1753, CPU 3930k @4.4, whyscotty Link DX11 344.16 Drivers
Score 58.6 GPU 980 @1241/1753, CPU 4770k @3.5, andrewohare Link DX11 344.16 Drivers
Score 55.0 GPU 290P @1050/1400, CPU 4770k @4.5, Robzere31 Link DX11 14.9 Drivers
Score 51.4 GPU 970 @1190/1342, CPU 4790k @4.7, byron_hinson Link DX11 344.16 Drivers
Score 44.1 GPU 280X @1110/1575, CPU FX-8320 @4.5, thebennyboy Link Mantle 14.9.2 Drivers

Now then, when you take a look at those bench results, why is one brand better than another brand? When you take Sniper Elite V3 for example, why is nobody screaming AMD are playing foul and purposefully gimping the game on nVidia to make AMD look better? because that's what it looks like to me.

Please tell me you are serious? Did you just look at the model of the cards and ignore the overclocking/clock speed entirely?

The 290P/290x in the list clocked their nuts off in order be faster than a stock clock 980, and they are (all except one at crazy clock) still slower than an overclocked 980. So what's the foul play here?

The only reason I could think of why there are so many AMD cards on the list while so few Nvidia cards could well be down to not many Nvidia user actually got that game? And that there are far more AMD users have that game due to it being part of the Never Settle free games? If there were more 980 owners actually got the game and post their overclock result for that bench, I honest wonder how many 290p/290x would still be left on that top ranking...I'd better tha green vs red ratio on that list would immediately reverse

As for your question about the AA earlier, I just want to remind you that AA is a post-process feature, meaning 0xAA is the original performance of the cards before the AA post-processing take place. Nvidia cards have FAR higher frame rate than the AMD at 0xAA. The AA wasn't "grimped" for 290x...it was actually the "saving grace" for shrinking the performance gap as the AA level go up, due to it was able to handle AA better than the Nvidia's top cards, and got less performance hit.

Don't ask me why or how AMD was able to coup better with increase in AA level than cards such as 780 and lose less frame rate as the AA level increases, as it was just what's shown in the benchmark.

You yourself said:

If nVidia were miles better in that game, it could be something that required investigating.

I then merely responded to that pointed out example of such occurance, and you simply responded with in a nutshell "no investigation needed. Gameworks is innonent and has not to do with it...we just know it, and we know for sure."

And then you just post-shifted to point finger and accuse AMD of grimping instead, when the discussion was about GameWorks...

What...?

 

[andybird123]

Its not nonsense, without the source code you cannot optimise for your hardware, at all. they might as well not bother as you can't do anything with it.

Whats to optimise if you don't know what its doing.

All you are displaying there is a complete lack of any experience writing drivers or libraries, if I write a driver for my own hardware of course I can see every call that is made to my drivers, i know exactly what the game is doing without seeing a single line of source code.

 

[Gregster]

*snip*

You do realise that I don't work for nVidia and I have no idea what you are babbling on about but I am interested in why some AMD users think that nVidia is purposefully crippling AMD hardware. Going off on one at me doesn't give the answer, so next time, instead of having a rant at someone who actually thinks GameWorks is good for the community, explain to me and everyone else what nVidia are actually doing to AMD.