Benchmarking Utility (video, CPU, disk, and memory).
Program is distributed as ZIP package: download to temporary directory and unpack to destination folder. Do NOT use the shipped copy of DIVE.DLL. You should search for the latest PCI devices file. Runs on OS/2 Warp 3 or higher. MMOS/2 must be installed.
See below for download link(s).
Following ones are the download links for manual installation:
Sysbench v. 0.95d (30/10/2017, Henrik Harmsen) | Readme/What's new |
Changes in Sysbench 0.9.5c
1) problem with more than 10 physical disk drives.
Changes in Sysbench 0.9.5a
1) Seems problem with more than 20 devices still wasn't fixed
Changes in Sysbench 0.9.5
1) Changed font used at 1600x1200 and above to System VIO, same as all other
resolutions. For language versions that don't have a System VIO font, we now
try to use Courier which is also a fixed width font. Let's hope it exists eh!
2) Corrected a problem in cpu identification that meant that coppermine cpus
with 256KB of 2nd level cache were not read from the table correctly.
3) Same problem for cpus with 512KB cache...
4) Corrected a problem that meant that the final graphics test sometimes
returned a negative value and didn't display a result.
5) Reduced the amount of data queried from the PCI configuration space in
order to lessen the chances of hanging the entire system when weird
devices (mainly Tekram SCSI controllers) are found. No idea if it'll help...
6) After talking with Michal Necasek and the guys at SciTech, they suggested
that the overall graphics score should be calculated in a different manner.
Done. Because of this change, we get 0.9.5 so don't compare numbers
against older releases please.
7) Fixed crash when an i/o error occurred during the disk test
e.g. removeable media drive with no media.
8) File I/O test now detects cachesize in use for HPFS, HPFS386 and JFS
and uses files of twice the largest cachesize to try to avoid the
entire file fitting in cache and artificially inflating the results.
Don't compare the File I/O test results from 0.9.5 with any from
previous releases, they measure different things.
9) FLOPS test result would wrap on fast machines (where fast is > 3.06GHz P4).
Fixed until they get 10 times as fast again - so about 2014 if Moore's law
holds!
10) fixed divide by zero error during file i/o test
11) added new Intel and AMD processors to cpuid.dat and code in pmcpu3a.asm
to decode various new cache flags
12) Stray sysbench.ini files written if you change the default File I/O test drive
13) Fixed crash when more than 20 PCI devices on a machine
14) Removed window border from fullscreen graphics tests
Changes in Sysbench 0.9.4g
See file chngs094.txt for all 0.9.4 series and previous changes |
hobbes.nmsu.edu/download/pub/os2/util/benchmark/sysb095d.zip | ||
Sysbench v. 0.91 (source code, 30/10/2017) | Readme/What's new |
Changes in Sysbench 0.9.5c
1) problem with more than 10 physical disk drives.
Changes in Sysbench 0.9.5a
1) Seems problem with more than 20 devices still wasn't fixed
Changes in Sysbench 0.9.5
1) Changed font used at 1600x1200 and above to System VIO, same as all other
resolutions. For language versions that don't have a System VIO font, we now
try to use Courier which is also a fixed width font. Let's hope it exists eh!
2) Corrected a problem in cpu identification that meant that coppermine cpus
with 256KB of 2nd level cache were not read from the table correctly.
3) Same problem for cpus with 512KB cache...
4) Corrected a problem that meant that the final graphics test sometimes
returned a negative value and didn't display a result.
5) Reduced the amount of data queried from the PCI configuration space in
order to lessen the chances of hanging the entire system when weird
devices (mainly Tekram SCSI controllers) are found. No idea if it'll help...
6) After talking with Michal Necasek and the guys at SciTech, they suggested
that the overall graphics score should be calculated in a different manner.
Done. Because of this change, we get 0.9.5 so don't compare numbers
against older releases please.
7) Fixed crash when an i/o error occurred during the disk test
e.g. removeable media drive with no media.
8) File I/O test now detects cachesize in use for HPFS, HPFS386 and JFS
and uses files of twice the largest cachesize to try to avoid the
entire file fitting in cache and artificially inflating the results.
Don't compare the File I/O test results from 0.9.5 with any from
previous releases, they measure different things.
9) FLOPS test result would wrap on fast machines (where fast is > 3.06GHz P4).
Fixed until they get 10 times as fast again - so about 2014 if Moore's law
holds!
10) fixed divide by zero error during file i/o test
11) added new Intel and AMD processors to cpuid.dat and code in pmcpu3a.asm
to decode various new cache flags
12) Stray sysbench.ini files written if you change the default File I/O test drive
13) Fixed crash when more than 20 PCI devices on a machine
14) Removed window border from fullscreen graphics tests
Changes in Sysbench 0.9.4g
See file chngs094.txt for all 0.9.4 series and previous changes |
www.os2warp.org/Sysbench/sb091dsrc.zip | ||
Sysbench v. 0.95d (8/8/2008) | Readme/What's new |
Changes in Sysbench 0.9.5c
1) problem with more than 10 physical disk drives.
Changes in Sysbench 0.9.5a
1) Seems problem with more than 20 devices still wasn't fixed
Changes in Sysbench 0.9.5
1) Changed font used at 1600x1200 and above to System VIO, same as all other
resolutions. For language versions that don't have a System VIO font, we now
try to use Courier which is also a fixed width font. Let's hope it exists eh!
2) Corrected a problem in cpu identification that meant that coppermine cpus
with 256KB of 2nd level cache were not read from the table correctly.
3) Same problem for cpus with 512KB cache...
4) Corrected a problem that meant that the final graphics test sometimes
returned a negative value and didn't display a result.
5) Reduced the amount of data queried from the PCI configuration space in
order to lessen the chances of hanging the entire system when weird
devices (mainly Tekram SCSI controllers) are found. No idea if it'll help...
6) After talking with Michal Necasek and the guys at SciTech, they suggested
that the overall graphics score should be calculated in a different manner.
Done. Because of this change, we get 0.9.5 so don't compare numbers
against older releases please.
7) Fixed crash when an i/o error occurred during the disk test
e.g. removeable media drive with no media.
8) File I/O test now detects cachesize in use for HPFS, HPFS386 and JFS
and uses files of twice the largest cachesize to try to avoid the
entire file fitting in cache and artificially inflating the results.
Don't compare the File I/O test results from 0.9.5 with any from
previous releases, they measure different things.
9) FLOPS test result would wrap on fast machines (where fast is > 3.06GHz P4).
Fixed until they get 10 times as fast again - so about 2014 if Moore's law
holds!
10) fixed divide by zero error during file i/o test
11) added new Intel and AMD processors to cpuid.dat and code in pmcpu3a.asm
to decode various new cache flags
12) Stray sysbench.ini files written if you change the default File I/O test drive
13) Fixed crash when more than 20 PCI devices on a machine
14) Removed window border from fullscreen graphics tests
Changes in Sysbench 0.9.4g
See file chngs094.txt for all 0.9.4 series and previous changes |
www.os2warp.org/Sysbench/sysb095d.zip | ||
Sysbench v. 0.95c (19/5/2008) | Readme/What's new |
Changes in Sysbench 0.9.5c
1) problem with more than 10 physical disk drives.
Changes in Sysbench 0.9.5a
1) Seems problem with more than 20 devices still wasn't fixed
Changes in Sysbench 0.9.5
1) Changed font used at 1600x1200 and above to System VIO, same as all other
resolutions. For language versions that don't have a System VIO font, we now
try to use Courier which is also a fixed width font. Let's hope it exists eh!
2) Corrected a problem in cpu identification that meant that coppermine cpus
with 256KB of 2nd level cache were not read from the table correctly.
3) Same problem for cpus with 512KB cache...
4) Corrected a problem that meant that the final graphics test sometimes
returned a negative value and didn't display a result.
5) Reduced the amount of data queried from the PCI configuration space in
order to lessen the chances of hanging the entire system when weird
devices (mainly Tekram SCSI controllers) are found. No idea if it'll help...
6) After talking with Michal Necasek and the guys at SciTech, they suggested
that the overall graphics score should be calculated in a different manner.
Done. Because of this change, we get 0.9.5 so don't compare numbers
against older releases please.
7) Fixed crash when an i/o error occurred during the disk test
e.g. removeable media drive with no media.
8) File I/O test now detects cachesize in use for HPFS, HPFS386 and JFS
and uses files of twice the largest cachesize to try to avoid the
entire file fitting in cache and artificially inflating the results.
Don't compare the File I/O test results from 0.9.5 with any from
previous releases, they measure different things.
9) FLOPS test result would wrap on fast machines (where fast is > 3.06GHz P4).
Fixed until they get 10 times as fast again - so about 2014 if Moore's law
holds!
10) fixed divide by zero error during file i/o test
11) added new Intel and AMD processors to cpuid.dat and code in pmcpu3a.asm
to decode various new cache flags
12) Stray sysbench.ini files written if you change the default File I/O test drive
13) Fixed crash when more than 20 PCI devices on a machine
14) Removed window border from fullscreen graphics tests |
www.os2warp.org/Sysbench/sysb095c.zip |
This work is licensed under a Creative Commons Attribution 4.0 International License.
Add new comment