Driver 332.21 WHQL

[PantherX]

Nvidia has released Driver 332.21 WHQL so anyone here has tested it to see if the folding performance is fixed or not?

[pithos]

No visible improvement.

[heikosch]

No visible improvement.

We didn´t expect a "visible" improvement.
But please tell us how did you test the driver.

Heiko

[7im]

Yes, I expect them to visibly fix the loss of performance on the older K GPUs.

[heikosch]

Yes, I expect them to visibly fix the loss of performance on the older K GPUs.

I would like to call it "calculated" because folding proteins with OpenCL doesn´t produce a visible result on my screen.
So let us talk about TPF or PPD.

Heiko

[n_w95482]

I wish . I installed Windows 8.1 on my PC about a month ago and installed the 331.82 drivers. I was on a folding hiatus for a few weeks so I didn't test them immediately but they ran great in games, unlike all of the 32x.xx versions I tried. I started folding again a few days ago and noticed the PPD was MUCH lower than the 314.22 drivers I used prior - barely above 30k PPD vs 75-90k. I'm using a GTX 680.

I tried the 332.21 driver as well, no change. I tested all four driver versions in FAHBench, here's the results:

Explicit SP
314.22: 35.4306 ns/day
327.23: 34.8926 ns/day
331.82: 15.5557 ns/day
332.21: 15.6549 ns/day

Implicit SP:
314.22: 137.861 ns/day
327.23: 139.067 ns/day
331.82: 144.298 ns/day
332.21: 144.576 ns/day

Luckily 314.22 runs happily in Win8.1 .

Edit: I just tested the 327.23 drivers. Seems to be running fine so far. If PPD ends up being similar to 314.22, I'll keep them; otherwise I'll switch back.

Edit 2: I'm getting about 90k PPD with the 327.23 drivers, so I'll be hanging onto these .

[AZBrandon]

I tried the 332.21 driver as well, no change. I tested all four driver versions in FAHBench, here's the results:

Explicit SP
314.22: 35.4306 ns/day
327.23: 34.8926 ns/day
331.82: 15.5557 ns/day
332.21: 15.6549 ns/day

Implicit SP:
314.22: 137.861 ns/day
327.23: 139.067 ns/day
331.82: 144.298 ns/day
332.21: 144.576 ns/day

Nice! It's good to see all 4 of the recent sets of drivers benchmarked here. I find it interesting that they slightly improved implicit SP, but of course there is a catastrophic loss (and very relevant impact to F@H) on Explicit SP performance with the 33x series drivers.

[n_w95482]

Yeah, hopefully they get it resolved soon - as well as not sucking up a whole CPU thread to run OpenCL stuff. Those two things will make me very happy .

I'm still running 327.23 with zero issues so far. I'm not playing any brand-new games, so I should be fine for a while. I did recently run the built-in benchmarks for Bioshock Infinite and Tomb Raider, ran smooth and no obvious weirdness/artifacts. I'll just have to deal with my OCD-ness about updating drivers.

[kiore]

OK now there is 334.89.. anyone tried this and does it let fermi fold well finally?

[PantherX]

OK now there is 334.89.. anyone tried this and does it let fermi fold well finally?

The results are mixed(?) from what I can tell (viewtopic.php?f=80&t=25850).