Possible RAID card problem (heard that one before?)
Using SLES 8, patched up to SP3 and beyond on a Rioworks HDAMA board with dual Opteron 242 and 4GB Ram. Disk is a 3ware 7500-8 with 5x200GB disks set up as a RAID5 with one redundant disk. Mainboard BIOS flashed up, brand new firmware on the controller, running the newest driver version (compiled for SMP x86-64 and gently slid into the module tree and initrd). I'm seeing real performance issues, though. With a 2GB file, and no competition for resources: |scandium:~ # time cp testfile testfile2 | |real 0m26.034s |user 0m0.570s |sys 0m5.980s lspci -v is interesting: |03:02.0 RAID bus controller: 3ware Inc 3ware 7000-series ATA-RAID (rev 01) | Subsystem: 3ware Inc 3ware 7000-series ATA-RAID | Flags: bus master, medium devsel, latency 72, IRQ 30 | I/O ports at 3000 [size=16] | Memory at fd200000 (32-bit, non-prefetchable) [size=16] | Memory at fd800000 (32-bit, non-prefetchable) [size=8M] | Expansion ROM at <unassigned> [disabled] [size=64K] | Capabilities: [40] Power Management version 1 It's in a 64bit-port, and it's alone on the #03 bus, which is the PCI-X bus. It _shold_ be reporting 64bit, but it isn't. Any ideas on how I proceed with this? And yes, it reported exactly the same before I compiled my own driver for it. That's why I did it... :/ Bjørn -- Bjørn Tore Sund Phone: (+47) 555-84894 Stupidity is like a System administrator Fax: (+47) 555-89672 fractal; universal and Math. Department Mobile: (+47) 918 68075 infinitely repetitive. University of Bergen VIP: 81724 teknisk@mi.uib.no Email: bjornts@mi.uib.no http://www.mi.uib.no/
lspci -v is interesting: |03:02.0 RAID bus controller: 3ware Inc 3ware 7000-series ATA-RAID (rev 01) | Subsystem: 3ware Inc 3ware 7000-series ATA-RAID | Flags: bus master, medium devsel, latency 72, IRQ 30 | I/O ports at 3000 [size=16] | Memory at fd200000 (32-bit, non-prefetchable) [size=16] | Memory at fd800000 (32-bit, non-prefetchable) [size=8M] | Expansion ROM at <unassigned> [disabled] [size=64K] | Capabilities: [40] Power Management version 1 It's in a 64bit-port, and it's alone on the #03 bus, which is the PCI-X bus. It _shold_ be reporting 64bit, but it isn't. Any ideas on how I proceed with this?
what do you mean? that 32bit, nonprefetch up there means its mapped into 32bit memory space, since fd200000 is a 32bit memory address. that has nothing to do with 64bit pci/x bus/slot or anything the like.
And yes, it reported exactly the same before I compiled my own driver for it. That's why I did it... :/
what did u expect? i dont really see your problem so far. and i dont see any other relevant 32bit or 64bit pci related info in your lspci posting. cheers, andy
On Wed, 11 Feb 2004, Andreas Bittner wrote:
what do you mean? that 32bit, nonprefetch up there means its mapped into 32bit memory space, since fd200000 is a 32bit memory address. that has nothing to do with 64bit pci/x bus/slot or anything the like.
Thank you, this is useful. On various other cards I see information in the "Capabilities" part about PCI-X and 64bit, but not with this one.
And yes, it reported exactly the same before I compiled my own driver for it. That's why I did it... :/
what did u expect? i dont really see your problem so far. and i dont see any other relevant 32bit or 64bit pci related info in your lspci posting.
I'm seeing performance well below what I expected but am fumbling in the dark for why. Thank you for pointing out that what I thought was an indication of what the problem is, isn't. Thus I'm still wondering. Bjørn -- Bjørn Tore Sund Phone: (+47) 555-84894 Stupidity is like a System administrator Fax: (+47) 555-89672 fractal; universal and Math. Department Mobile: (+47) 918 68075 infinitely repetitive. University of Bergen VIP: 81724 teknisk@mi.uib.no Email: bjornts@mi.uib.no http://www.mi.uib.no/
participants (2)
-
Andreas Bittner -
Bjorn Tore Sund