Differences

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

--- cluster:119 [2013/09/25 12:39] – [matlab.sub] hmeij
+++ cluster:119 [2021/06/17 19:32] (current) – hmeij07
@@ Line 1: / Line 1: @@
 \\
 **[[cluster:0|Back]]**
+==== Submitting GPU Jobs ====
 Please plenty of time between multiple GPU job submissions.  Like minutes.
-Jobs need to be submitted to the scheduler on host sharptail itself for now and will be dispatched to nodes n33-n37 in queue mwgpu. They can also be submitted from host greentail but remember any output will be in shraptail's /home.
+Jobs need to be submitted to the scheduler via cottontail to queues mwgpu, amber128, exx96.
- --- //[[hmeij@wesleyan.edu|Meij, Henk]] 2013/09/25 08:33//
+This page is old, the gpu resource ''gpu4'' should be used, a more recent page can be found [[cluster:173|K20 Redo Usage]]. Although there might some useful information on this page explaining gpu jobs.
+ --- //[[hmeij@wesleyan.edu|Henk]] 2021/06/17 15:29//
+**Articles**
+  * [[http://www.pgroup.com/lit/articles/insider/v5n2a1.htm]] Tesla vs. Xeon Phi vs. Radeon: A Compiler Writer's Perspective
+  * [[http://www.pgroup.com/lit/articles/insider/v5n2a5.htm]] Calling CUDA Fortran kernels from MATLAB
-==== Submitting GPU Jobs ====
@@ Line 46: / Line 54: @@
 </code>
-With ''gpu-info'' we can view our running job.  ''gpu-info'' and ''gpu-free'' are available [[http://ambermd.org/gpus/]] (I had to hard code my GPU string information as they came in at 02,03,82&83, you can use deviceQuery to find them).
+With ''gpu-info'' we can view our running job.  ''gpu-info'' and ''gpu-free'' are available <del>[[http://ambermd.org/gpus/]]</del> [[http://ambermd.org/gpus12/#Running]](I had to hard code my GPU string information as they came in at 02,03,82&83, you can use deviceQuery to find them).
 <code>
@@ Line 60: / Line 68: @@
        Tesla K20m      21 C            0 %
 ====================================================
+[hmeij@sharptail sharptail]$ ssh n33 gpu-free
+,3,0
 </code>
@@ Line 138: / Line 151: @@
 /cm/shared/apps/cuda50/toolkit/5.0.35/open64/lib:/cm/shared/apps/cuda50/toolkit/5.0.35/extras/CUPTI/lib:\
 /cm/shared/apps/mvapich2/gcc/64/1.6/lib
-# signal that this is charmrun/namd job
 ## leave sufficient time between job submissions (30-60 secs)
@@ Line 176: / Line 189: @@
 # NAMD
-# from greentail we need to recreate module env
+# signal that this is charmrun/namd job
 export CHARMRUN=1
 # stage the data
@@ Line 434: / Line 447: @@
 exit $?
+</code>
+===== elim code =====
+<code>
+#!/usr/bin/perl
+while (1) {
+        $gpu = 0;
+        $log = '';
+        if (-e "/usr/local/bin/gpu-info" ) {
+                $tmp = `/usr/local/bin/gpu-info | egrep "Tesla K20"`;
+                @tmp = split(/\n/,$tmp);
+                foreach $i (0..$#tmp) {
+                        ($a,$b,$c,$d,$e,$f,$g) = split(/\s+/,$tmp[$i]);
+                        if ( $f == 0 ) { $gpu = $gpu + 1; }
+                        #print "$a $f $gpu\n";
+                        $log .= "$f,";
+                }
+        }
+        # nr_of_args name1 value1
+        $string = "1 gpu $gpu";
+        $h = `hostname`; chop($h);
+        $d = `date +%m/%d/%y_%H:%M:%S`; chop($d);
+        foreach $i ('n33','n34','n35','n36','n37') {
+                if ( "$h" eq "$i" ) {
+                        `echo "$d,$log" >> /share/apps/logs/$h.gpu.log`;
+                }
+        }
+        # you need the \n to flush -hmeij
+        # you also need the space before the line feed -hmeij
+        print "$string \n";
+        # or use
+        #syswrite(OUT,$string,1);
+        # smaller than specified in lsf.shared
+        sleep 10;
+}