[ofa-general] IB card manufacturer

[sebastien dugue]

  Hi Jason,

  a few things you could try to determine who the manufacturer is:

  - 'mstflint -d 07:00.0 q' will give you some info, especially the HCA GUIDs.
    The 24 most significant bits of the board GUIDs gives you the manufacturer.
    For example: 
	- 0002c9 Mellanox
	- 0008f1 Voltaire

  - try to run 'mstflint -d 07:00.0 dc'. The PS_INFO, ADAPTER and IB sections
    could give some more hints.



  Hope this helps.

  Sebastien.



On Tue, 2 Jun 2009 21:03:13 -0400
Jason Hill <hilljj at ornl.gov> wrote:

> All,
> 
> Being somewhat cautious I'd like to see if anyone can point me at the correct 
> manufacturer info for an Infiniband card that we bought through a cluster vendor
> (i.e. it was in the box when we got it). I need to upgrade the firmware to 
> attempt to isolate an issue we're seeing with several nodes that have this 
> combination:
>  
>  ==> /sys/class/infiniband/mthca0/board_id <== 
>  INT0020000001
> 
>  ==> /sys/class/infiniband/mthca0/fw_ver <==
>  1.1.0
>  
>  ==> /sys/class/infiniband/mthca0/hw_rev <==
>  a0
>  
>  ==> /sys/class/infiniband/mthca0/hca_type <==
>  MT25204
> 
> ~$ lspci -vvv
> <snip>
> 
> 07:00.0 InfiniBand: Mellanox Technologies MT25204 [InfiniHost III Lx HCA] (rev 20)
>         Subsystem: Mellanox Technologies MT25204 [InfiniHost III Lx HCA]
>         Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- 
>         Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR-
>         Latency: 0, Cache Line Size: 64 bytes
>         Interrupt: pin A routed to IRQ 50
>         Region 0: Memory at feb00000 (64-bit, non-prefetchable) [size=1M]
>         Region 2: Memory at fd800000 (64-bit, prefetchable) [size=8M]
>         Capabilities: <access denied>
> 
> On the card itself it has :
> 
> Part Number: D83914 
> Serial Number: MEPT702000093
> And the card has a tall bracket, but the vendor could have swapped that.
> 
> Google has turned up nothing for the Board ID and part number/serial number. 
> 
> To me it looks kind of like a Cisco card, so I was thinking it might be a 
> Topspin. We bought the cluster in 2007.
> 
> Would the right thing be to install the MemFree MHGS18-XT firmware, or try a 
> revision of the MHGS18-XTC?
> 
> Thanks for the time and thoughts.
>