williedigital Posted October 9, 2005 Share Posted October 9, 2005 I haven't been able to find really good answers for this stuff... 1. Does OSx86 recognize hyperthreaded and dual core cpu's? If so, do they show up in activity monitor, etc as such? 2. Does the machine take advantage of them? Can you "feel" the effects of dual cpu's? Subjective or xbench marks would be cool... 3. I've heard that hyperthreading and single thread cpu's benchmark the same in xbench (much as they do on windows benchmarking), but that on pc's you can "feel" the difference with hyperthreading from a subjective end user multi-tasking vantage. Is the same the case for within osx? thanks for any info... Link to comment Share on other sites More sharing options...
stryder Posted October 10, 2005 Share Posted October 10, 2005 I haven't been able to find really good answers for this stuff... 1. Does OSx86 recognize hyperthreaded and dual core cpu's? If so, do they show up in activity monitor, etc as such? 2. Does the machine take advantage of them? Can you "feel" the effects of dual cpu's? Subjective or xbench marks would be cool... 3. I've heard that hyperthreading and single thread cpu's benchmark the same in xbench (much as they do on windows benchmarking), but that on pc's you can "feel" the difference with hyperthreading from a subjective end user multi-tasking vantage. Is the same the case for within osx? thanks for any info... Um... ever hear of a search?!? http://forum.osx86project.org/index.php?sh...=hyperthreading Link to comment Share on other sites More sharing options...
Miaplacidus Posted October 11, 2005 Share Posted October 11, 2005 I haven't been able to find really good answers for this stuff... 1. Does OSx86 recognize hyperthreaded and dual core cpu's? If so, do they show up in activity monitor, etc as such? 2. Does the machine take advantage of them? Can you "feel" the effects of dual cpu's? Subjective or xbench marks would be cool... 3. I've heard that hyperthreading and single thread cpu's benchmark the same in xbench (much as they do on windows benchmarking), but that on pc's you can "feel" the difference with hyperthreading from a subjective end user multi-tasking vantage. Is the same the case for within osx? thanks for any info... Hi there, I tested dual-core multiprocessing very informally on my system by launching simultaneously two CPU-intensive processes (bzip2 compression of large-ish files) and monitoring progress with top and Activity Monitor (which shows two CPU usage panels). The system is based on Athlon X2 (so no hyperthreading, specs are here).While the processes were running, top showed each process taking up >90% of cpu, and Activity Monitor showed continuous user cpu cycles on both panels. I then ran a single compression process, and verified that the time it took the process to run was substantially the same as that with another process running. So both cores are definitely recognized and engaged in multiprocessing. PS Per thrunner's suggestion I ran Xbench on the machine and posted the result on the Xbench thread. Compared with single-core Athlon 64s I'm getting a far superior Thread Test result - as would be expected for a dual core machine. Link to comment Share on other sites More sharing options...
williedigital Posted October 11, 2005 Author Share Posted October 11, 2005 sorry, i did a search for dual-core, but not hyperthreading. thanks for all the info guys. Link to comment Share on other sites More sharing options...
would_pay_for_mac_on_x86 Posted October 25, 2005 Share Posted October 25, 2005 I then ran a single compression process, and verified that the time it took the process to run was substantially the same as that with another process running. So both cores are definitely recognized and engaged in multiprocessing. very nice....do you know how to direct each job to a core? i remember i could do this in solaris but i don't know how onder os x. so TIME COMPLETED for 2 jobs on 2 CPUS = TIME COMPLETED for 1 job on 1 CPU? you were able to verify that the latter 1 job was only running on 1 CPU? Link to comment Share on other sites More sharing options...
Miaplacidus Posted October 26, 2005 Share Posted October 26, 2005 very nice....do you know how to direct each job to a core? i remember i could do this in solaris but i don't know how onder os x. No, I don't actually. On a dual-processor Powermac, it is possible to switch on and off one of the processors from System Preferences -> Hardware. More about this below.... so TIME COMPLETED for 2 jobs on 2 CPUS = TIME COMPLETED for 1 job on 1 CPU? Yes. That's what I observed. you were able to verify that the latter 1 job was only running on 1 CPU? No, and in fact Activity Monitor showed activity on both cores even when just one compression process was running; however it also indicated that each core was less than fully engaged. Top, on the other hand, showed just one process taking up 90+% CPU time. BTW, when I ran two compression processes at the same time, both cores were shown to be fully engaged. The same thing happens when I run a compression process on a dual-processor Mac. These's activity on both CPUs, but neither CPU is used fully at any given time. Similarly, top shows a single process taking up more than 90% of CPU time (sometimes it actually exceeds 100% slightly). Now, as I mentioned above, it is possible to turn off one of the CPUs on the Mac; and when I do this and run a single compression process, the active CPU is close to 100% engaged. There is no difference in the execution time of a single compression process, whether one or two processors are activated. So I'm pretty sure only a single process is running, which however gets shuttled between the two processors/cores. This is most certainly done by a scheduler in order to even out the load between the two. Link to comment Share on other sites More sharing options...
would_pay_for_mac_on_x86 Posted October 26, 2005 Share Posted October 26, 2005 thanks, very interesting... Link to comment Share on other sites More sharing options...
Recommended Posts