In comp.os.linux.networking, you wrote:
> Peter T. Breuer <ptb@oboe.it.uc3m.es> wrote:
>> William Park <opengeometry@yahoo.ca> wrote:
>>>> Keeg wrote:
>>>>
>>>>> At what speed (and maybe other factors, like memory?) does a gigabit
>>>>> ethernet NIC card become pointless?
>>>>
>>>> A standard PCI bus is barely able to keep up with a busy 100 Mb NIC. It
>>>> hasn't a hope of keeping up with Gigabit, no matter what CPU you have.
>
>>> Wrong. 1Gigabit and 100Mbit translates to 100MB/s and 10MB/s (Byte per
>>> second), respectively, in real life. PCI bus has 133MB/s theoretical
>>> limit, but max out at about 100MB/s in real life.
>
>> Well, that's 32 bit, 33MHz PCI, no? I have 64 bit, 66MHz PCI myself!
>> And 2 of them per board.
>
>> (server boards)
>
>> I have reports of people measuring 60MB/s across GE using my ENBD
>> device. That must be on a fast bus, as the NIC and the scsi disks
>> both need that bandwidth, and both are on PCI.
>
> The best I could get out of 64bit PCI bus, using SAN FC cards, is about
> 100 MBytes/s (rw), guess the real bottleneck are the disks, many cheapo
> IDE stuff isn't even able to cope with 100 Mb/s LAN running at full
> speed. GB NICs only make really sense if your I/O can handle it...
Or if you are running a supercomputing centre, and the jobs your group
run are primarily CPU intensive MPI jobs rather than disk bound.
-- TimC -- http://astronomy.swin.edu.au/staff/tconnors/double value; /* or your money back! */ short changed; /* so triple your money back! */ -- Larry Wall in cons.c from the perl source code - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Jan 07 2003 - 22:00:17 EST