lkml.org 
[lkml]   [2001]   [Oct]   [10]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
Subjectseq i/o kills all kernels
I just did some more testing with 2.4.10-ac10, 2.4.10-ac10-eatcache, and
2.4.11. Fast sequential i/o on any of these kernels makes the system
stop doing anything else.

On 2.4.10-ac10 and 2.4.10-ac10-eatcache, the performance is the same. I
start a mmap i/o process that reads and writes 16MB blocks in a 1GB file.
This proceeds at about 20MB/s while nothing else is running. XMMS is
undisturbed by this activity. Then, I start dd if=/dev/zero of=bigfile
bs=32k count=32768. This proceeds at 40MB/s but XMMS stops completely and
the mmap i/o rate at the same time fals to about 100KB/s. Essentially no
other process makes progress while the dd is running. The system is
unusable.

On 2.4.11, XMMS skips[1] but the mmap i/o continues to make progress at
about 5MB/s. The dd proceeds as well, slightly slower than under
2.4.10-ac10[-eatcache]. Interactive behavior is abyssmal.

The machine is 1.4GHz Athlon, 256MB main memory, software RAID-1 on
aic7xxx.

No, I haven't tried preemptible kernel yet.

-jwb

[1] Skipping is a gentle term. XMMS plays 10ms bursts every 5-10 seconds.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:17    [W:0.023 / U:0.200 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site