Le Tue, Nov 13, 2007 at 08:49:24PM -0200, Nelson A. de Oliveira a écrit : > Hi again! > > On Nov 13, 2007 8:34 PM, Nelson A. de Oliveira <[EMAIL PROTECTED]> wrote: > > It builds fine if using -O0 > > Do you want to use -O0 temporarily and upload it? > > With -Os it also builds fine (-O1 fails) > -Os is better than nothing (-O0)
Hi Nelson, this is a great idea ! Here is a benchmark I just made with bHLH proteins: chouca〔source〕$ for i in dialign-t-O* ; do echo $i ; for j in 1 2 3 ; do ./$i ../conf test2.fasta | grep time ; done ; done dialign-t-O0 Total time: 23.690000 secs Total time: 23.590000 secs Total time: 23.590000 secs dialign-t-O1 Total time: 17.350000 secs Total time: 17.350000 secs Total time: 17.340000 secs dialign-t-O2 Total time: 16.750000 secs Total time: 16.790000 secs Total time: 16.760000 secs dialign-t-O3 Total time: 17.040000 secs Total time: 16.810000 secs Total time: 16.800000 secs dialign-t-Os Total time: 17.920000 secs Total time: 18.020000 secs Total time: 17.960000 secs chouca〔source〕$ ls -la dialign-t-O* -rwxr-xr-x 1 charles charles 88445 2007-11-14 12:59 dialign-t-O0 -rwxr-xr-x 1 charles charles 103629 2007-11-14 12:58 dialign-t-O1 -rwxr-xr-x 1 charles charles 105311 2007-11-14 12:57 dialign-t-O2 -rwxr-xr-x 1 charles charles 114863 2007-11-14 12:57 dialign-t-O3 -rwxr-xr-x 1 charles charles 84256 2007-11-14 12:58 dialign-t-Os Os is much better than O0, and the temporary loss of performance is in my opinion acceptable. I hope that we can make a few more serious benchmarks and decide which options of compliation to use before Lenny's release; for the moment I tend to override the upstream makefiles and use O2 everywhere. (And this benchmark seems to confirm that it is not a too bad idea). Have a nice day, -- Charles Plessy http://charles.plessy.org Wako, Saitama, Japan -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
