DokuWiki

User Tools

Site Tools

Trace: • 83 • 159 • 167
cluster:182

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
cluster:182 [2019/08/13 12:37] – [Gromacs] hmeij07cluster:182 [2019/12/13 13:33] (current) – hmeij07
Line 2: Line 2:
 **[[cluster:0|Back]]** **[[cluster:0|Back]]**
  
 + 
 ==== P100 vs RTX 6000 & T4 ==== ==== P100 vs RTX 6000 & T4 ====
  
Line 54: Line 54:
 |  DPFP  |  5.21|  18.35|  0.75|  0.35|  1.29| |  DPFP  |  5.21|  18.35|  0.75|  0.35|  1.29|
 |  SXFP  |  11.82|  37.44|  17.05|  7.01|  18.91| |  SXFP  |  11.82|  37.44|  17.05|  7.01|  18.91|
-|  SFFP  |  11.91|  40.98|  9.92|  4.35|  16.22|+|  SPFP  |  11.91|  40.98|  9.92|  4.35|  16.22|
  
 Like last testing outcome, in the SFFP precision mode it is best to run four individual jobs, one per GPU (mpi=1, gpu=1). Best performance is the P100 at 47.64 vs the RTX at 39.69 ns/day per node. The T4 runs about 1/3 as fast and really falters in DPFP precision mode. But in SXFP (experimental) precision mode the T4 makes up in performance.  Like last testing outcome, in the SFFP precision mode it is best to run four individual jobs, one per GPU (mpi=1, gpu=1). Best performance is the P100 at 47.64 vs the RTX at 39.69 ns/day per node. The T4 runs about 1/3 as fast and really falters in DPFP precision mode. But in SXFP (experimental) precision mode the T4 makes up in performance. 
Line 90: Line 90:
 ==== Gromacs ==== ==== Gromacs ====
  
-Gromacs was build on each of the nodes locally letting it select the optimal CPU (AVX, SSE) and GPU accelerators. The ''cmake'' flag -DGMX_BUILD_OWN_FFTW=ON yields a mixed precision compilation which is recommended. Then we ran multidir options 01-04 on single GPU, and 01-08 and 01-16 on all 4 GPUs when possible.+Gromacs was build on each of the nodes locally letting it select the optimal CPU (AVX, SSE) and GPU accelerators. "GROMACS simulations are normally run in “mixed” floating-point precision, which is suited for the use of single precision in FFTW. " The ''cmake'' flag ''-DGMX_BUILD_OWN_FFTW=ON'' yields a mixed precision compilation which is recommended. Then we ran multidir options 01-04 on single GPU, and 01-08 and 01-16 on all 4 GPUs when possible.
  
  
Line 99: Line 99:
 |  Mixed  |  |  |  |  |  733|  gpu=4, 01-16  | |  Mixed  |  |  |  |  |  733|  gpu=4, 01-16  |
  
-The T4 is P100's equal in mixed precision performance. Add the wattage factor and you have a favorite.+The T4 is P100's equal in mixed precision performance. Add the wattage factor and you have a favorite. And GPU utilization was outstanding.
  
 +[heme@login1 gromacs-2018]$ ssh node9 ./gpu-info\\
 +id,name,temp.gpu,mem.used,mem.free,util.gpu,util.mem\\
 +0, Tesla T4, 66, 866 MiB, 14213 MiB, 98 %, 9 %\\
 +1, Tesla T4, 67, 866 MiB, 14213 MiB, 98 %, 9 %\\
 +2, Tesla T4, 66, 866 MiB, 14213 MiB, 99 %, 9 %\\
 +3, Tesla T4, 64, 866 MiB, 14213 MiB, 97 %, 9 %\\
 ==== Scripts ==== ==== Scripts ====
  
Line 191: Line 197:
 </code> </code>
  
-And GPU utilization was outstanding. 
  
-[heme@login1 gromacs-2018]$ ssh node9 ./gpu-info\\ 
-id,name,temp.gpu,mem.used,mem.free,util.gpu,util.mem\\ 
-0, Tesla T4, 66, 866 MiB, 14213 MiB, 98 %, 9 %\\ 
-1, Tesla T4, 67, 866 MiB, 14213 MiB, 98 %, 9 %\\ 
-2, Tesla T4, 66, 866 MiB, 14213 MiB, 99 %, 9 %\\ 
-3, Tesla T4, 64, 866 MiB, 14213 MiB, 97 %, 9 %\\ 
  
 \\ \\
 **[[cluster:0|Back]]** **[[cluster:0|Back]]**
cluster/182.1565699820.txt.gz · Last modified: by hmeij07