On Sun, 13 Nov 2011 16:54:53 +0100 Thomas Gstädtner <tho...@gstaedtner.net> said:
> On So 13 Nov 2011 16:29:02 CET, Gustavo Sverzut Barbieri wrote: ... > A cpufreq module is even worse, especially if it supports userspace > interaction. The linux kernels ondemand governor works good for many > years now, there are no more senseless slowdowns as there used to be actually... on a core i7 i have... you HAVE to manually change cpufreq governor when doing benchmarking. the kernels ondemand schedulers don't work well and keep the cpu at the highest clockrate. your benchmarks are all over the shop if you don't. on a pentium-m i have, if it clocks at 1ghz for more than 30sec it gets incredibly slow - most intel cpus self-throttle they DON'T drop clockrate. they just start throwing in lots of nops. if its throttling at t5 or t7 even at 1ghz its incredibly slower than at its 600mhz slot, so i have to keep it clocked at 600mhz or a flash anim or compile will make the system unusable. at 600mhz it never throttles. it has no fan. it's passively cooled. i had a p4 laptop from dell... that's why i wrote the temp module. once it got hot it too would start throttling. even if you kept its clockrate low. so once it got hot enough it was time to take it off your lap and maybe get a fan. at least you didn't have a mystery as to why it was happening with a nice little gauge. :) > some years back. Having a possibility to make the user interfere makes > him think it might be wise, e.g. setting it to "max performance" or > similar stupid decisions. -- ------------- Codito, ergo sum - "I code, therefore I am" -------------- The Rasterman (Carsten Haitzler) ras...@rasterman.com ------------------------------------------------------------------------------ RSA(R) Conference 2012 Save $700 by Nov 18 Register now http://p.sf.net/sfu/rsa-sfdev2dev1 _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
