Subject: Re: [linux-audio-dev] hdr disk throughput
From: David Jander (djander_AT_alumnos.utfsm.cl)
Date: Tue Mar 14 2000 - 14:25:20 EST
On Mon, 13 Mar 2000, Andrew Clausen wrote:
>Paul Barton-Davis wrote:
>> i should remember to "run cc first, open mouth later".
>
>;-)
>
>> i tried your program and got 14.2MB/sec throughput. I also noticed
>> that your buffer size was 1MB, which means that my initial concern
>> about the gap between the read/write locations is irrelevant. Thanks a
>> lot for checking this out. Now I have to work out a buffering scheme
>> that works :)
>
Sorry to disturb your testing, but I found something strange with your programm:
I tried it On the following system:
CPU : AMD-K5 116 MHz, 64 Mb RAM, kernel 2.2.12, and
two Quantum Fireball HD's (5.1 and 6.4 Gb IDE drives) with software RAID0
The various tests I made yielded some strange results:
As root, I do "./overwrite testfile" on a non-raid partition of the first
drive, and get:
Total time: 22 seconds
Throughput: 5.818182 Mb/s
Also as root, the same thing on the other drive (non raid partition) yields:
Total time: 18 seconds
Throughput: 7.111111 Mb/s (this drive is newer :-)
Now, on a software RAID0 partition (both drives together):
Total time: 13 seconds
Throughput: 9.846154 Mb/s (way too cool for this poor system)
Until now you can get a hint about software RAID0 performance :-)
And my test machine is still very poor performing (AMD-K5+IDE disks) !!
The really weired thing arises when I use this program as regular user (non
root, non-raid), PLEASE CHECK THIS:
Total time: 6 seconds
Throughput: 21.333333 Mb/s (??????)
What's possibly going on here ? I'll have to figure that out. But it means,
your little hack is not always reliable. The file (aprox. 134 Mb) is created in
that short time although.
yours,
----------------------------------
David Jander J.
Electronics Student, UTFSM, Chile
E-Mail : djander_AT_alumnos.utfsm.cl
This archive was generated by hypermail 2b28 : Tue Mar 14 2000 - 22:22:08 EST