tag:blogger.com,1999:blog-1386948037384435441.post8900392212168496708..comments2024-03-01T18:53:33.429-08:00Comments on Jeff Muizelaar: Trying out AVXJeff Muizelaarhttp://www.blogger.com/profile/17483047845050494642noreply@blogger.comBlogger5125tag:blogger.com,1999:blog-1386948037384435441.post-12612236963865895782011-01-10T08:37:53.165-08:002011-01-10T08:37:53.165-08:00jseward: ripping out the computation makes the loo...jseward: ripping out the computation makes the loop run in 30889 usecs.Jeff Muizelaarhttps://www.blogger.com/profile/17483047845050494642noreply@blogger.comtag:blogger.com,1999:blog-1386948037384435441.post-62623821525096315112011-01-10T07:38:59.200-08:002011-01-10T07:38:59.200-08:00jlebar: fixed.
jseward: The working set is proces...jlebar: fixed.<br /><br />jseward: The working set is processing about 50MB of linear data. This means the best data rate we're currently getting is about 690 MB/s which I expect is lower than the rate the machine can sustain. I'll rip out some of the computation to get a better idea of what the memory performance of the workload is when I get a chance.Jeff Muizelaarhttps://www.blogger.com/profile/17483047845050494642noreply@blogger.comtag:blogger.com,1999:blog-1386948037384435441.post-3946582987011405842011-01-10T02:41:01.355-08:002011-01-10T02:41:01.355-08:00Jeff, these fancy new insns are only
going to impr...Jeff, these fancy new insns are only<br />going to improve performance if it<br />isn't limited by some other factor,<br />particularly by the performance of the<br />memory system. Do you have any feel,<br />for the SSE code with the workloads<br />you're using, to what extent performance<br />is limited by the rate at which the<br />processor can dispatch SSE insns vs by<br />cache misses?jsewardnoreply@blogger.comtag:blogger.com,1999:blog-1386948037384435441.post-31440786027567581842011-01-09T14:14:43.110-08:002011-01-09T14:14:43.110-08:00You might also have mentioned that AVX can access ...You might also have mentioned that AVX can access unaligned data. This should make it much easier to use.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-1386948037384435441.post-64088696942309552002011-01-09T13:43:53.428-08:002011-01-09T13:43:53.428-08:00Jeff,
Can you fix the link to the presentation? ...Jeff,<br /><br />Can you fix the link to the presentation? It needs an http:// in front.jlebarhttps://www.blogger.com/profile/11889377730733853337noreply@blogger.com