LinkExchange

  LinkExchange Member
  The
  OS/2 Warp
  Benchmark Centre
 
 
  The following bugs are known and I'm working on a solution to them. Watch this space....

 
  • Fixed 0.9.4e. SYS3175 saving HTML output when an unknown controller was found
  • Fixed 0.9.4e. HTML output code had incorrect syntax
  • Fixed 0.9.4e. FLOPS test was reporting rubbish
  • Fixed 0.9.4e. On Aurora beta and maybe GA I issued a message about Warp version because I screwed up
  • Fixed 0.9.4d. On German and Spanish systems I issued a message about Warp version because I wasn't checking for the right output from VER /R
  • Fixed 0.9.4d. Sometimes a lockup would happen if a music CD was in the drive
  • Fixed 0.9.4d. Corrected the code for recognising SCSI adapters that couldn't handle more than 16MB RAM
  • Fixed 0.9.4d. Corrected some problems with swapfile recognition
  • Fixed 0.9.4c. Some hard disks were still not performing the new disk I/O tests introduced in Sysbench 0.9.4. The symptoms were that some tests reported throughput of 63.999MB/sec or ---.---. This happened for disks that have 32 sectors/track.
  • Fixed 0.9.4b. Sysbench will not run on machines without OS2CDROM.DMD loaded - this includes machines without a CD drive at all!
  • Fixed 0.9.4b. Systems running the original version of OS/2 Warp 3 (revision 8.162) seem to throw a TRAP 000D when Sysbench 0.9.4 is invoked. This appears to be a bug in the version of RESOURCE.SYS that was shipped with this release. Replacing that file in \OS2\BOOT with the copy from Warp Connect dated two months later solves the TRAP. All versions of RESOURCE.SYS that I've tested with will work except the one from the original version of Warp 3. You can also bypass this TRAP by using the /nodetect switch to invoke Sysbench. I've added a check for this specific revision level and have forced the /nodetect switch.
  • Fixed 0.9.4b. Some systems fail the hardware detection and Sysbench thinks there are disks attached to controllers that cannot be identified. If your system hits this bug then the main display wil show this in the list of disks where it will say things like "Disk -1-1" and "Disk -1-2". When this problem happens it is not possible to save a set of HTML results as Sysbench takes a SYS3175.
  • Fixed 0.9.4b. There's a problem with systems containing SCSI controllers that are driven by the AHA152X.ADD driver. This driver erroneously reports the controller itself as a disk device to the OS/2 Resource Manager and Sysbench takes a SYS3175 when this happens. Using /nodetect will bypass this error if you have this driver installed.
  • Fixed 0.9.4a. Some hard disks were not performing the new disk I/O tests introduced in Sysbench 0.9.4. The symptoms were that some tests reported throughput of 63.999MB/sec or ---.---. This happened for disks that have 32 sectors/track.
 


Back to The OS/2 Benchmark Centre