Folding Forum

mdk777 wrote:update for OpenMM 6 in the works?

Yep. Soon.

I'd post a benchmark for 780Ti, but it's on a Linux PC. Will there ever be a Linux FAHBench, or would that introduce incompatibilities in the comparisons?

David

davidcoton wrote:I'd post a benchmark for 780Ti, but it's on a Linux PC. Will there ever be a Linux FAHBench, or would that introduce incompatibilities in the comparisons?

David

Yep. Soon.

Lenovo T440p Nvidia GT730M CPU/MEM 720/1000Mhz 335.23 driver Win8.1

OpenCL Explicit SP 7.2005 ns/day
OpenCL Implicit SP 34.9985 ns/day

Hello!

Is this any good?


p.s. shall I run it to verify accuracy again?
p-p.s. the benchmark does not seem to run on my CPU...

Could you please explain what you mean by "is this any good?".

The FAHBench gives you a rough idea (in ns/day) of how your GPU will perform when folding FahCore_17 WUs. However, just because you can successfully run it, doesn't mean that your GPU is 100% stable since future Projects may be more stressful then FAHBench.

IIRC, on Intel CPU, you need to install OpenCL run-time package.

EDIT -> Please note that while ns/day can be used for comparison, it may change from protein to protein. Thus, care has to be taken to ensure that the comparison is valid.

Windows 7 Ent, AMD Oland HD 8570, 14.4 Cat driver

Explicit Solvent SP: 8.435 ns/d
Implicit Solvent SP: 27.252 ns/d

Explicit Solvent DP: .731 ns/d
Implicit Solvent DP: 1.64 ns/d

Intel i7-4770 @ 3.40 GHz (stock)
Intel OpenCL Runtime 14.1 (32 bit only)
Windows 7 Ent 64 bit

Explicit Solvent SP: 3.491 ns/d
Implicit Solvent SP: 4.205 ns/d

Explicit Solvent DP: 2.551 ns/d
Implicit Solvent DP: 2.865 ns/d

FAH Benchmark v1.2.0

EVGA 750 TI SC +165 GPU offset +410 Mem offset = Single Implicit 117.998 ns/day

EVGA 750 TI SC +165 GPU offset +410 Mem offset = Single Explicit 29.0371 ns/day

Spongebob25 -> Are those values OpenCL or CUDA?

Spongebob25 wrote:FAH Benchmark v1.2.0

EVGA 750 TI SC +165 GPU offset +410 Mem offset = Single Implicit 117.998 ns/day

EVGA 750 TI SC +165 GPU offset +410 Mem offset = Single Explicit 29.0371 ns/day

Would be curious if the numbers change with no Mem Offset.

Here are my numbers for OpenCL

Asus GTX 750 Ti (non-overclocked version) under WinXP (337.88 drivers)
1072 MHz GPU Clock - 1150 MHz boost
1350 MHz Memory Clock

Explicit: 27.4239 ns/day
Implicit: 99.0701 ns/day

JimF wrote:Here are my numbers for OpenCL

Asus GTX 750 Ti (non-overclocked version) under WinXP (337.88 drivers)
1072 MHz GPU Clock - 1150 MHz boost
1350 MHz Memory Clock

Explicit: 27.4239 ns/day
Implicit: 99.0701 ns/day

Am I correct in assuming that it is for Single Precision and not Double Precision?

PantherX wrote:Spongebob25 -> Are those values OpenCL or CUDA?

Sorry. They are for OpenCL .

PantherX wrote:

JimF wrote:Here are my numbers for OpenCL

Asus GTX 750 Ti (non-overclocked version) under WinXP (337.88 drivers)
1072 MHz GPU Clock - 1150 MHz boost
1350 MHz Memory Clock

Explicit: 27.4239 ns/day
Implicit: 99.0701 ns/day

Am I correct in assuming that it is for Single Precision and not Double Precision?

Yes.