Hi, First - let me point that there is a "bug" in the test/stress.cxx, in the stress8 function. It has a hard-encoded path to $ROOTSYS/test subdirectory ( in a couple of places ). This is WRONG. It should be "./" instead. Everyone should be able to run tests, for example, by simply copying the $ROOTSYS/test/* into his own subdirectory ( even under another name ). That is - it should find/make the libEvent library in the CURRENT working subdirectory. Second - nobody commented my results ( that I posted two days ago ), but I think I would like to ask some questions : 1. Rene - your "tests' table" says nothing, in some cases, about the CPU type and speed - for example -> dec server 4100 - means nothing for me - especially today - you can buy alpha-based machines from different manufacturers and different processor speeds ( I have here an offer - CPU 21164A, speeds 533/566/600 MHz ) - so I'd like to ask you to add these informations to your "tests' table" 2. The same concerns the number of CPUs ( if more then one ) 3. The same concerns the HD type ( IDE, UDMA, SCSI, UW SCSI, UW2 SCSI, ... ) The HD type is possibly quite an important point - I compared my results ( 2*PII/450, UDMA HD ) with "pcbrun" ( ?x?*PII/400, ?TYPE? HD ). In some numbers I can clearly see the cpu speed ratio 450/400, but there are tests where your ( "pcbrun" ) machine is better ! Especially E0-write, E1-write. This suggests that either you have a much better HD ( ?x? SCSI ? ), or that the 2.2.x Linux kernel is much better then the 2.0.36. I'd like to clear this for me ( in case it's the kernel, I'll need to "upgrade" ). Thanks in advance, Jacek. BTW. During some E0/E1 tests ( If I remember well - "write" tests ) my machine "froze" for a longer time ( I mean I wasn't able to move the mouse pointer, type characters, ... ). I almost thought I needed to press reset ;-( ... but now :-) ... I think the machine was simply so BUSY with writing to the local HD. Is this possible ? I have never met something like this before. Jacek.
This archive was generated by hypermail 2b29 : Tue Jan 04 2000 - 00:43:30 MET