Re: Re: Illusory Multicore Support in 5.2?
- To: mathgroup at smc.vnet.net
- Subject: [mg66056] Re: [mg66046] Re: Illusory Multicore Support in 5.2?
- From: George Woodrow III <georgevw3 at mac.com>
- Date: Fri, 28 Apr 2006 06:32:32 -0400 (EDT)
- References: <200604260838.EAA02742@smc.vnet.net> <email@example.com> <firstname.lastname@example.org> <200604270935.FAA03741@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
Using Mathematica 5.2.2 on OS X 10.4.6 PowerMac QuadCore. I haev 8 Gbytes RAM I ran Activity monitor to see the CPU usage graphically. I executed the following in 1 cell so that all the computations would be passed to the kernel at the same time: N[Ï?, 100000000]; // Timing N[Ïµ, 100000000]; // Timing N[Sqrt, 100000000]; // Timing (doing just 1000000 places returned too fast -- pi in 3.4 seconds and the others much faster. I asked for 100 times the decimal places so the program would run for a while.) I would expect one thread per computation -- they are independent. Instead, I get 100% of one CPU all the time. The CPU switches from time to time -- probably the OS resolving usage/heat issues. The Activity monitor reposts MathKernel using 3 threads and 99.9% of the CPU -- the max is 400%. I can get more than 200% of CPU usage running iTunes's visualliser with all the options cranked up, or Sims 2. There appears to be no problem with the OS distributing a multi- threaded app among the processors. The good news is that there is no slowdown from the rest of my apps: the OS distributes the usage among running apps very well. the Bad News is that Mathematica seems to be using only one CPU at a time. george On Apr 27, 2006, at 5:35 AM, Chris Chiasson wrote: > Do the posters in this thread happen to be running a 32 bit version of > Mathematica? >