18 replies to this topic

[Goose]

http://forums.guru3d...ad.php?t=388074

Let the testing begin!

[Peter2k]

Was waiting for this one since nvidia claimed they could do the same as Mantle without creating a new API

Seems like it made quite a difference for me

sporting a GTX 780 and an i5

[August55]

I feel like I'm getting the same sort of performance as before, with D3D9 performing better than D3D11 by 10-20 frames at times.

evga GTX 780 with i5-4670k.

[xWiredx]

I like numbers and graphs. Can somebody that tries these in the next few hours compare and put some before/after numbers up? Knowing what hardware and previous driver version was used would also help.

[Egomane]

From what I have read, these performance gains are pretty specific to some games and system configurations. None of those articles posted numbers, so it's pretty much useless to link them.

If someone could provide real numbers it would be nice.

[Peter2k]

As I understand it, sites are not quite yet allowed to post/talk/review yet
guess a international deadline thing

Yes the gains are very specific, the games must be CPU bound to begin with
If you're graphics is already the bottleneck than there are no gains realy, but it wouldn't be different with mantle, if you're graphics card is bad then Mantle wouldn't do a lot for you either
CPU bound games are not that common realy, and benchmarking WMO is rather difficult

Things always run extremely smooth in the testing ground, but worse actually online
Everyone has to see for himself I guess

[Summon3r]

well i just installed those drivers and put all my settings back to max on dx9 and it would seem fps is a little better, anywhere from 45 to 90+ and no stuttering, not going to attempt dx11 till pgi has optimized them a bit better.... lol? i guess i could be waiting awhile then

[xWiredx]

Summon3r, on 07 April 2014 - 12:26 PM, said:

well i just installed those drivers and put all my settings back to max on dx9 and it would seem fps is a little better, anywhere from 45 to 90+ and no stuttering, not going to attempt dx11 till pgi has optimized them a bit better.... lol? i guess i could be waiting awhile then

I'm running a similar setup (i7 2600K and Galaxy GTX 660 Ti 2GB). I'll do the most objective testing I can and post results tonight (I'm still working atm).

[Summon3r]

xWiredx, on 07 April 2014 - 12:45 PM, said:

I'm running a similar setup (i7 2600K and Galaxy GTX 660 Ti 2GB). I'll do the most objective testing I can and post results tonight (I'm still working atm).

what clock speed are u running your 2600k at?

[Jabilo]

2500K and GTX770, no improvement for me.

I still have frame drops in to the 40s, but my stuttering was fixed two patches ago so it still seems smooth.

[xWiredx]

Okay, here's the breakdown from my testing. I did 2 training ground tests where I followed the same path and attacked the same 'dummy' each time in a Sarah's Jenner. I then played 3 matches in a Stalker (completely random matches).

Relevant hardware: Intel i7 2600K (4.4GhZ, hyperthreading on), 8GB Samsung DDR3-1600 (running at DDR3-1866), Galaxy GTX 660Ti 2GB

Quote

335
---
test 1: 73, 117, 102.4
test 2: 85, 108, 97.6

test 3: 54, 106, 89.6
test 4: 44, 80, 62.6
test 5: 41, 88, 63.5

----------

337
---
test 1: 88, 119, 103.6
test 2: 87, 116, 99.9

test 3: 55, 105, 72.35
test 4: 61, 105, 85.35
test 5: 51, 92, 68.4

Noticeable right off the bat is that the minimum FPS was way better pretty much all-around. Specifically, Terra Therma (test 2) noticed a decent gain here. Crimson Strait (test 1) noticed smaller gains. In matches, things ran a little more smoothly, and there were small gains in performance.

The real take-away here is that we know this game will eat every CPU cycle you can give it, and even with a relatively high amount of CPU power there were small performance gains with the new beta 337-series drivers. One can surmise that Nvidia users with lesser CPUs will probably notice slightly more significant gains, as this driver helps CPU-bound games much more than it does GPU-bound games. I might recommend waiting for an 'official' driver release from Nvidia, but I'm pretty happy with these ones so far and anybody willing to run a beta driver (that functions really well, mind you) should give this a go.

[Summon3r]

xWiredx, on 07 April 2014 - 03:17 PM, said:

Okay, here's the breakdown from my testing. I did 2 training ground tests where I followed the same path and attacked the same 'dummy' each time in a Sarah's Jenner. I then played 3 matches in a Stalker (completely random matches).

Relevant hardware: Intel i7 2600K (4.4GhZ, hyperthreading on), 8GB Samsung DDR3-1600 (running at DDR3-1866), Galaxy GTX 660Ti 2GB



Noticeable right off the bat is that the minimum FPS was way better pretty much all-around. Specifically, Terra Therma (test 2) noticed a decent gain here. Crimson Strait (test 1) noticed smaller gains. In matches, things ran a little more smoothly, and there were small gains in performance.

The real take-away here is that we know this game will eat every CPU cycle you can give it, and even with a relatively high amount of CPU power there were small performance gains with the new beta 337-series drivers. One can surmise that Nvidia users with lesser CPUs will probably notice slightly more significant gains, as this driver helps CPU-bound games much more than it does GPU-bound games. I might recommend waiting for an 'official' driver release from Nvidia, but I'm pretty happy with these ones so far and anybody willing to run a beta driver (that functions really well, mind you) should give this a go.

nice work, were u using dx9 or 11?

also does anyone have know where a guy could find software to to monitor your avg fps throughout a game?

Edited by Summon3r, 07 April 2014 - 03:33 PM.

[xWiredx]

My testing was all done using DX11. All video settings were set to high, except texturing was set to very high. Antialiasing is currently only set to PostAA.

Also possibly relevant: I ran the MWO repair tool about a week ago. I did let it clear the shaders or whatever the heck it does by default. Other than that, I haven't altered anything else on my system at all.

[xWiredx]

Summon3r, on 07 April 2014 - 03:21 PM, said:

also does anyone have know where a guy could find software to to monitor your avg fps throughout a game?

I use FRAPS personally. I think it can also be done with MSI Afterburner.

[Smokeyjedi]

MSI afterburner for fan profiles and OC on GPU
Fraps for frame rate relay recording bechmarking(frametimes,etc)

Im about to get some here before too long.......

Edited by Smokeyjedi, 07 April 2014 - 05:07 PM.

[Monkeystador]

Just install msi afterburner and have it create a framerate + gpu load graph. Much better.

[ShinVector]

Feels like I getting around +10fps in DX9.
Noticeable improvement numbers wise...
Tried DX11 noticed crappy shuttering happens here and there when things happen in the screen.. Went back to DX9.

[Summon3r]

wow now running dx11 at max settings cept shadows to high, avg fps through 4 matches 57

[Smokeyjedi]

I have had to reduce the overclock on my AMP 760.....was causing instability in DX11 for me........so keeping core clock on my GTX760 under 1200 seems to solve this.....for me anyway.