[BENCH MARK] 'tiobench' results comparision for 2.5.43 kernel and 2.4.19

[BENCH MARK] 'tiobench' results comparision for 2.5.43 kernel and 2.4.19

[Siva Koti Reddy]

[-- Attachment #1: Type: text/plain, Size: 3395 bytes --]

Hai...
    Here is the benchmark results of tiobench for 2.5.43 kernel and 2.4.19
kernel. There is a  drastic performance degradation in Read operations of
2.5.43 kernel as compared to 2.4.19 kernel though there is a little
improvement in write operations. Any cooments..?

Details of Test Machine:
=================
Hardware: 400Mhz  Celeron with 64Mb Ram with ext2 file system.

************************************************************************
linux-2.5.43
Tiotest results for 4 concurrent io threads:
,----------------------------------------------------------------------.
| Item                  | Time     | Rate         | Usr CPU  | Sys CPU |
+-----------------------+----------+--------------+----------+---------+
| Write          40 MBs |    8.9 s |   4.509 MB/s |   0.4 %  |   6.2 % |
| Random Write   16 MBs |   19.5 s |   0.802 MB/s |   0.0 %  |   1.4 % |
| Read           40 MBs |    7.9 s |   5.090 MB/s |   0.3 %  |   4.9 % |
| Random Read    16 MBs |   37.3 s |   0.419 MB/s |   0.1 %  |   1.2 % |
`----------------------------------------------------------------------'
Tiotest latency results:
,-------------------------------------------------------------------------.
| Item         | Average latency | Maximum latency | % >2 sec | % >10 sec |
+--------------+-----------------+-----------------+----------+-----------+
| Write        |        2.144 ms |     3340.874 ms |  0.02930 |   0.00000 |
| Random Write |        7.420 ms |     8241.095 ms |  0.10000 |   0.00000 |
| Read         |        2.526 ms |      534.798 ms |  0.00000 |   0.00000 |
| Random Read  |       35.709 ms |      131.234 ms |  0.00000 |   0.00000 |
|--------------+-----------------+-----------------+----------+-----------|
| Total        |        7.737 ms |     8241.095 ms |  0.02458 |   0.00000 |
`--------------+-----------------+-----------------+----------+-----------'


*********************************************************************
linux-2.4.19
Tiotest results for 4 concurrent io threads:
,----------------------------------------------------------------------.
| Item                  | Time     | Rate         | Usr CPU  | Sys CPU |
+-----------------------+----------+--------------+----------+---------+
| Write          40 MBs |    9.7 s |   4.135 MB/s |   0.3 %  |   3.9 % |
| Random Write   16 MBs |   27.3 s |   0.573 MB/s |   0.0 %  |   0.4 % |
| Read           40 MBs |    1.7 s |  23.597 MB/s |   0.6 %  |  14.2 % |
| Random Read    16 MBs |    4.0 s |   3.879 MB/s |   0.7 %  |   4.5 % |
`----------------------------------------------------------------------'
Tiotest latency results:
,-------------------------------------------------------------------------.
| Item         | Average latency | Maximum latency | % >2 sec | % >10 sec |
+--------------+-----------------+-----------------+----------+-----------+
| Write        |        0.618 ms |     1449.211 ms |  0.00000 |   0.00000 |
| Random Write |        0.100 ms |       61.018 ms |  0.00000 |   0.00000 |
| Read         |        0.158 ms |       59.913 ms |  0.00000 |   0.00000 |
| Random Read  |        1.561 ms |      484.705 ms |  0.00000 |   0.00000 |
|--------------+-----------------+-----------------+----------+-----------|
| Total        |        0.512 ms |     1449.211 ms |  0.00000 |   0.00000 |
`--------------+-----------------+-----------------+----------+-----------'


Rgds,
Siva



[-- Attachment #2: Wipro_Disclaimer.txt --]
[-- Type: text/plain, Size: 522 bytes --]

**************************Disclaimer**************************************************    
 
 Information contained in this E-MAIL being proprietary to Wipro Limited is 'privileged' 
and 'confidential' and intended for use only by the individual or entity to which it is 
addressed. You are notified that any use, copying or dissemination of the information 
contained in the E-MAIL in any manner whatsoever is strictly prohibited.

****************************************************************************************

Re: [BENCH MARK] 'tiobench' results comparision for 2.5.43 kernel and 2.4.19

[Andrew Morton]

Siva Koti Reddy wrote:
> 
> Hai...
>     Here is the benchmark results of tiobench for 2.5.43 kernel and 2.4.19
> kernel. There is a  drastic performance degradation in Read operations of
> 2.5.43 kernel as compared to 2.4.19 kernel though there is a little
> improvement in write operations. Any cooments..?
> 

The tiobench threaded read phase is very sensitive to the
interworking between readahead and the disk scheduler.  2.4
gets this right, and 2.5 does not.

2.5 seeks between the files with a granularity equal to the
readahead window size.  2.4 seeks betwen the files at a
granularity equal to the elevator request expiry count.

So yup, it's known.  Haven't investigated it yet.