Re: 3Com 905C very slow (400Kbps)

Bogdan Costescu wrote:

>On Wed, 4 Jun 2003, Florian Hackenberger wrote:
>
>  
>
>>Well, I experienced an improved performance over the hub with the 
>>realtek card, compared to the 3Com device.
>>    
>>
>
>What do you call "improved" ? How do you measure it ?
>
Disk to disk copy via NFS. Ok, that is useless since we found out that 
the P75 disks are slow.

>  
>
>>What do you mean with media mismatch?
>>    
>>
>
>Forcing at least one of the partners (card and switch) to a specific speed 
>and duplex setting instead of letting both ends on auto/nway. Check the 
>archives of this list... 
>
>  
>
>>If you compare the throughput colums, you'll notice that the P75 is MUCH 
>>MUCH slower than the P4:
>>    
>>
>
>Well, it's something like 7-8 years difference...
>
>  
>
>>So here are the results from netperf P4=server
>>87380  16384  16384    10.00     33.46
>>P75=server
>> 87380  16384  16384    10.01      51.01
>>    
>>
I meant this table:
This is the printout from netperf (P75):
TCP STREAM TEST to localhost
87380  16384  16384    10.00      89.99

This is from my P4
TCP STREAM TEST to localhost
87380  16384  16384    9.99     3964.88

I think the difference is not that small. I think the problem is related 
to both the IDE Controler of the P75, as well as slow PCI transfers.

>Well, 33 Mbit/s ~= 4 Mbyte/s which is much better than any figure you 
>showed until now. While it's still far from the ~90Mbit/s theoretical 
>maximum, it's probably because the system itself is slow. I do not have 
>experience with such slow computers and PCI cards as to what they are able 
>to sustain, as I wasn't doing networking stuff at that time...
>
>  
>
>> DMA modes:  mdma0 mdma1 *mdma2
>>    
>>
>
>If this drive is implicated in any transfer, it's probably slow. You could 
>take it out of the measurements with something like copying a file from 
>the P4 computer to /dev/null on the P-75 computer (assuming that P4 is NFS 
>server and P-75 is NFS client) with something like:
>
>time cp /file/from/NFS/mounted/system /dev/null
>
>This will involve the disk on the P4 computer, but not the one on the P-75 
>computer.
>  
>
Wow...I archive 2.1 MB/sec (ca. 45MB / 21sec)...ok, I believe your 
theory about the disk beeing slow, but still...that is not 100Mbs. Ok, 
so will go and buy a new IDE-Controller and will about get 2MB/s.

>  
>
>>But even with the crossover cable the transfer rate is under 700Kbps, 
>>even without collisions and errors.
>>    
>>
>
>It depends on what you call transfer rate: if you talk about disk-on-P4 to 
>disk-on-P-75 then you might be limited by the disk speed, in which case 
>you might want to put a better IDE controller in the P-75 computer.
>
>  
>
I've tried copying the same file twice. The first time through the hub, 
the second time through crossover cable. Since I've copyied the file 
once before, it seems to be in the cache of the computer reading it from 
the disk. Through the hub it took 13 secs and with crossover 14 secs for 
45MB. I don't know if the NFS client caches files too, but if not, it 
would mean that the hub cannot be the problem.

Regards.

-- 
Florian Hackenberger
student @
University of Technology
Graz, Austria
[email protected]


_______________________________________________
vortex mailing list
[email protected]
http://www.scyld.com/mailman/listinfo/vortex