At 04 Oct 2001 19:21:35 -0700, Josh Green wrote: > > On Thu, 2001-10-04 at 08:58, Takashi Iwai wrote: > > At 02 Oct 2001 14:25:03 -0700, > > Josh Green wrote: > > > > > > On Tue, 2001-10-02 at 10:33, Takashi Iwai wrote: > > > > Hi, > > > > > > > > i'm trying to run latency test for ALSA9 (based on Benno's latency > > > > suite) on my kernel to measure the latency with preemption and LL > > > > kernel patches. > > > > however, the test program hangs up almost always in the middle during > > > > disk write tests of 500 MB. > > > > I first though that this is related with VM and happens when the > > > > memory becomes full. But it's not true. This happens even if there > > > > are enough memory left unused. > > > > > > > > This appears only on 2.4.10 kernel. Looks like 2.4.9 kernel works > > > > fine. > > > > > > > > Could somebody test the program too? Of course, be prepare for hang > > > > up before you try :) > > > > > > > > The latency test program can be found at: > > > > http://www.alsa-project.org/~iwai/latencytest-0.42+alsa.tar.bz2 > > > > > > > > > > > > ciao, > > > > > > > > Takashi > > > > > > > > > > What are you using for your LL patch? It looks like Andrew Morton's > > > patch is only up to 2.4.9. I've applied his 2.4.9 patch to 2.4.10, but > > > there are a lot of rejects which I'm currently fixing up to the best of > > > my knowledge. > > > > It's my hacking version (applied not to vanilla but SuSE's kernel). > > The kernel itself works flawlessly, but obviously it doesn't work > > as expected during heavy disk write. I experienced periodical peaks > > up to 6 msec. The disk read and proc load are quite ok. > > > > It's interesting to compare the results between LL and PE patches. > > Usually PE kernel shows much better responce than LL kernel on my > > tests. The latency is less than 100 usec, while the LL kernel has > > usually 200-500 msec usecs latency. > > But both kernels have almos same level peaks in disk write/copy... > > > > I'll check this problem futhre when I have time.. > > > > > > ciao, > > > > Takashi > > > > Aggh.. I just ran a whole bunch of ALSA latencytests with various > drivers/kernels, all with fairly bad results :( I even went back to my > older kernel (2.4.5 with Andrew Morton's patch) with my AWE 32 card, > which I had good results with before, and got like 5 to 8ms spikes. I'm > not sure what the heck is going on with my machine now, but its > certainly not good.
I feel that your old results were too good. But the new results seem also too bad compared with mine. Or any hardware changes? hdparm, APM, etc.. > Are there any programs that report what piece of > code is causing big spikes? I haven't tested the 2.4.10 kernel yet, and > if I did it would be bare without any patches (as I don't know of any up > to date with 2.4.10). I did realize that the 2.4.9 kernel with LL patch > is worse than the 2.4.5 one I was using. This might be because I > compiled it with Mandrake's gcc a la 2.96. As far as I've experienced, the VM of 2.4.9 tends to have higher latency under heavy disk loads. On 2.4.10 the VM was changed much and became better. I discussed with Andrea about this problem, and he will check the relevant part. Hopefully any patch will come to 2.4.11. > I also noticed that the sound heard with your latencytest is not very > clean, it has quite a lot of "fuzz" in it. I ran Benno's latencytest > program and it sounded much cleaner. Perhaps.. I have not heared sounds during tests, as I always chose "quiet" output. > Your program also prints out > "overrun!" with the ALSA driver, which is probably just for debugging > right? Yes, it's just a debugging output when actually snd_pcm_writei() returns the EPIPE error. Please ignore. > I thought of a few improvements that could be done with latencytest. It > would be nice if the machine configuration could be generated in the > toplevel html file (kernel version/build info, CPU, memory, sound card > info, etc). Sometimes its easy to forget the configuration when testing > various kernels/drivers/sound cards. Yes. I put the condition in html file manually, but this should be automated of course. > By the way I'm posting my results to http://www.c0nfusion.org/~josh/ if > you want to have a look. The only hardware that has changed in my > machine since my previous tests (http://www.resonance.org/~josh/) is I > added an SB Live!. I think I upgraded to Mandrake 8 between then, so > most of the software is probably different. Ah, the last test looks really fantastic.. I wish in near future we get it back! ciao, Takashi _______________________________________________ Alsa-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/alsa-devel
