This is a multi-part message in MIME format. --------------080107030703040906020402 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Memtest 3.1 is not working with some newer boards/CPU's, so we switched to using memtest+ 1.2 prebuilt bootable iso from http://www.memtest.org/. The bootable version eliminated the OS inconsistencies, down side is you have to reboot the system to run the test. Our MSI K8D Master has 6 mem slots and all are populated with 512 MB Reg ECC DDR 266. BIOS is set to Interleaving = auto, Node Interleaving = Auto, Burst Length = 8 Beats, Memclock Mode = Auto. Have not run Streams on this system, but will. Thanks for the reply. Doug >On Mon, 2004-06-28 at 14:32, Doug Shubert wrote: > > >>Hello Joe, >> >>Your numbers look very linear to ours. >>I would expect a 45%-50% performance >>increase based on CPU speed (240 vs 246). >> >>Other factors could be in memory speeds and OS tuning. >>Our Opteron is running DDR Reg ECC 266 memory. >> >>If you have a chance to run memtest86+, I would be interested >>in the memory speed of a 246 system. >> >> > >Ok. Are you using a particular memtest? I found 3.1 was needed for >Opteron. I am using DDR 333 memory. Lets also check the layout of your >ram. Is it all piled on one processing node? If so, you want to set it >up as interleaved. I get streams of about 2GB/s per CPU with streams >triad. > >I am in the middle of some other stuff now, so I'll try to get the >memtest86 data later tonight (unless you meant a different memtest...) > > > >>ioinformatics.org/mailman/listinfo/bioclusters >> >> --------------080107030703040906020402 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <html> <head> <meta http-equiv="Content-Type" content="text/html;charset=ISO-8859-1"> <title></title> </head> <body text="#000000" bgcolor="#ffffff"> Memtest 3.1 is not working with some newer boards/CPU's, so we <br> switched to using memtest+ 1.2 prebuilt bootable iso from <a class="moz-txt-link-freetext" href="http://www.memtest.org/">http://www.memtest.org/</a>.<br> <br> The bootable version eliminated the OS inconsistencies, down side is you have to reboot <br> the system to run the test.<br> <br> Our MSI K8D Master has 6 mem slots and all are populated with 512 MB Reg ECC DDR 266.<br> BIOS is set to Interleaving = auto, Node Interleaving = Auto, Burst Length = 8 Beats,<br> Memclock Mode = Auto. <br> <br> Have not run Streams on this system, but will.<br> <br> Thanks for the reply.<br> Doug<br> <blockquote type="cite" cite="mid1088447930.17261.117.camel@protein.scalableinformatics.com"> <pre wrap="">On Mon, 2004-06-28 at 14:32, Doug Shubert wrote: </pre> <blockquote type="cite"> <pre wrap="">Hello Joe, Your numbers look very linear to ours. I would expect a 45%-50% performance increase based on CPU speed (240 vs 246). Other factors could be in memory speeds and OS tuning. Our Opteron is running DDR Reg ECC 266 memory. If you have a chance to run memtest86+, I would be interested in the memory speed of a 246 system. </pre> </blockquote> <pre wrap=""><!----> Ok. Are you using a particular memtest? I found 3.1 was needed for Opteron. I am using DDR 333 memory. Lets also check the layout of your ram. Is it all piled on one processing node? If so, you want to set it up as interleaved. I get streams of about 2GB/s per CPU with streams triad. I am in the middle of some other stuff now, so I'll try to get the memtest86 data later tonight (unless you meant a different memtest...) </pre> <blockquote type="cite"> <pre wrap="">ioinformatics.org/mailman/listinfo/bioclusters </pre> </blockquote> </blockquote> </body> </html> --------------080107030703040906020402--