VM tests on 2.4.14-pre3 and 2.4.13-ac3

VM tests on 2.4.14-pre3 and 2.4.13-ac3

Post by rwh.. » Mon, 29 Oct 2001 13:50:09



Summary:

2.4.14-pre3             The greased weasel is back
2.4.13-ac3              Inconsistent results on mtest01
2.4.13-ac3-freeswap     patch for 2.4.12-ac3 makes results consistent in mtest01

Test:   Run mtest01 and mmap001 from Linux Test Project.  
        Listen to long running (50+ minutes) mp3 sampled at 128k.

Default kernel parameters.

Light interactive use: lynx, 2 *x (console IRC client) sessions,
vmstat 8, iostat -d 10.  (52k link to net - mostly idle).

2.4.13-pre3 had much better interactive response than 2.4.13-ac3
during these tests.  No X.

Hardware:
AMD 1333 Athlon
512 MB RAM
1024 MB swap

mtest01
=======

2.4.14-pre3

mp3blaster did not skip.  Best time I ever recorded for this test
without mp3 skipping.  Previous "best" with no mp3 skips was:
2.4.13-pre3aa1-page-cluster=2  
Elapsed (wall clock) time:          47.878

Averages for 10 mtest01 runs
bytes allocated:                    1233859379
User time (seconds):                2.034
System time (seconds):              3.036
Elapsed (wall clock) time:          30.517
Percent of CPU this job got:        16.10
Major (requiring I/O) page faults:  107.0
Minor (reclaiming a frame) faults:  302029.3

2.4.13-ac3

Played about 90 seconds of mp3 during 200 second run.
747634688 - 1181745152 bytes allocated per iteration.  

Averages for 10 mtest01 runs
bytes allocated:                    894120755
User time (seconds):                1.499
System time (seconds):              2.744
Elapsed (wall clock) time:          20.646
Percent of CPU this job got:        20.40
Major (requiring I/O) page faults:  101.9
Minor (reclaiming a frame) faults:  218898.0

2.4.13-ac3-freeswap

Played about 120 seconds of mp3 during 350 second run.

Averages for 10 mtest01 runs
bytes allocated:                    1244449996
User time (seconds):                2.033
System time (seconds):              3.761
Elapsed (wall clock) time:          35.219
Percent of CPU this job got:        16.00
Major (requiring I/O) page faults:  114.1
Minor (reclaiming a frame) faults:  304611.1

mmap001 test
============

2.4.14-pre3

mp3 did not skip.

Average for 5 mmap001 runs
bytes allocated:                    2048000000
User time (seconds):                19.578
System time (seconds):              17.388
Elapsed (wall clock seconds) time:  171.45
Percent of CPU this job got:        21.20
Major (requiring I/O) page faults:  500158.8
Minor (reclaiming a frame) faults:  41.4

About 2-3 seconds of mp3 skip for entire run.

2.4.13-ac3
Average for 5 mmap001 runs
bytes allocated:                    2048000000
User time (seconds):                19.412
System time (seconds):              16.974
Elapsed (wall clock seconds) time:  151.49
Percent of CPU this job got:        23.40
Major (requiring I/O) page faults:  500180.2
Minor (reclaiming a frame) faults:  20.0

About 2-3 seconds of mp3 skip for entire run.

2.4.13-ac3-freeswap
Average for 5 mmap001 runs
bytes allocated:                    2048000000
User time (seconds):                19.260
System time (seconds):              17.456
Elapsed (wall clock seconds) time:  151.44
Percent of CPU this job got:        23.80
Major (requiring I/O) page faults:  500174.0
Minor (reclaiming a frame) faults:  20.0

runalltests.sh
==============

When executing runalltests.sh from LTP, 2.4.14-pre3 returns
FAIL on the personality01 tests (11 FAIL) whereas 2.4.13-ac3
returns PASS on the same tests.

--
Randy Hron

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://www.veryComputer.com/
Please read the FAQ at  http://www.veryComputer.com/

 
 
 

1. VM test comparison of 2.4.14-pre5, aa1, and 2.4.13-ac5-fs

Side note -- you cannot directly call this a vm vs vm test, not if you
are doing any significant amount of I/O. The -ac and Linus tree have
several significant changes in the queueing layer that makes this pretty
much and apples and oranges comparison.

--
Jens Axboe

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

2. Kernel version differences

3. Memory accounting problem in 2.4.13, 2.4.14pre, and possibly 2.4.14

4. fddi adapter error message

5. TCP Connections stuck in SYN_SENT state with 2.4.12, 2.4.13, 2.4.14

6. I dont like FVWM 95

7. OOPS on boot in 2.4.13-ac5 & 2.4.14-pre6

8. ethernet setup problem

9. VM test on 2.4.14pre3aa2 (compared to 2.4.14pre3aa1)

10. LINUX VM (2.4.14) vs FreeBSD VM in low memory machines

11. Unwanted Swapping in 2.4.14-pre8, no swapping in 2.4.14-pre6aa1

12. Oops on 2.4.14-pre3

13. 2.4.14-pre3 and umount