On Wed, Dec 03, 2008 at 11:55:18AM +1100, Roger E wrote: > > > Giso Grimm wrote: > > > > Recently I measured the CPU load and battery run time of some low-delay > > audio DSP on an Asus Eee PC 701 (Intel Celeron M, 630 MHz), and on an > > Acer Aspire one (Intel Atom 1.6 GHz). The time spent in the CPU was > > pretty much the same for both systems, and the battery runtime was > > between 3 and 3 1/2 hours on the Asus (Celeron) and 2 1/2 hours on the > > Acer (Atom). Here is the data: > > > > algo1 > > Asus: 56.4% CPU, 3h14' battery > > Acer: 51.0% CPU, 2h28' battery > > P4: 20.5% CPU > > algo2 > > Asus: 49.8% CPU, 3h16' battery > > Acer: 47.5% CPU, 2h09' battery > > P4: 16.5% CPU > > algo3 > > Asus: 36.5% CPU, 3h22' battery > > Acer: 35.0% CPU, 2h27' battery > > P4: 13.0% CPU > > > > jackd > > Asus: 6.0% CPU > > Acer: 4.8% CPU > > P4: 2.2% CPU > > > > Sound card IRQ handler > > Asus: 4.5% CPU > > Acer: 4.2% CPU > > P4: 1.0% CPU (network interface with netjack) > > > > P4 is Intel Pentium 4 @ 3 GHz. I know this is no perfect benchmarking, > > but gives an idea of the processor performance. > > > > > > Giso > > > Did the Acer have HDD or SSD? Could make the difference in battery time. > or different battery capacity? > I ran super_pi 20 to test basic CPU number crunching on my Core2Duo > E6400 and EeePC900 (Sidux on both systems).:- > EeePC - 71secs > C2D - 18.5secs Awesome statistic work! Is there some kind of a JACKstone metric that would be useful for measuring Linux RT audio performance in some kind of pseudo-standardized way? It seems like a lot of the metrics are kind of fuzzy "I was able to run Ardour with 16 tracks for an hour with no Xruns", or some such. In my case, I'm interested only in softsynth CPU capability with low (128us/3periods) Ingo-RT latency on a USB interface. -ken _______________________________________________ Linux-audio-user mailing list Linux-audio-user@xxxxxxxxxxxxxxxxxxxx http://lists.linuxaudio.org/mailman/listinfo/linux-audio-user
