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

Bogdan Costescu wrote:

>On Wed, 4 Jun 2003, Florian Hackenberger wrote:
>
>  
>
>>I've brought the 3Com card back to the shop since it was expensive and 
>>didn't perform that well.
>>I'm now testing on the Realtek 8139 Card.
>>    
>>
>
>If the CPU/mainboard are the cause of the slow data rate, this will make 
>it even slower as the CPU will spend some time taking care of the data 
>transfer with RTL8139.
>
Well, I experienced an improved performance over the hub with the 
realtek card, compared to the 3Com device.

>
>  
>
>>  eth0:10851530   20130    0    0    0     0          0         0 
>>50981444   51197  377    0    0     1     754          0
>>    
>>
>
>Actually more carrier errors than collisions - cable problems or media 
>mismatch.
>
What do you mean with media mismatch?

>
>  
>
>>TCP STREAM TEST to localhost
>>    
>>
>
>Well, testing to localhost gives you an indication of the speed of the CPU 
>and memory; you need to test it between 2 computers so that their network 
>interfaces are used.
>
If you compare the throughput colums, you'll notice that the P75 is MUCH 
MUCH slower than the P4:
So I don't think that the cable or the hub is the problem. But'll check 
out the test between two computers later.

This is the printout from netperf (P75):
TCP STREAM TEST to localhost
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    10^6bits/sec

87380  16384  16384    10.00      89.99

This is from my P4
TCP STREAM TEST to localhost
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    10^6bits/sec

87380  16384  16384    9.99     3964.88

>
>  
>
>>    ide0: BM-DMA at 0xe800-0xe807, BIOS settings: hda:pio, hdb:pio
>>    ide1: BM-DMA at 0xe808-0xe80f, BIOS settings: hdc:pio, hdd:pio
>>hda: ST340016A, ATA DISK drive
>>    
>>
>
>Could you make sure that hda uses DMA mode instead of PIO (using hdparm) ? 
>In PIO mode the CPU is doing lots of work talking to the disk which 
>severely limits the data rate of the disk. This could explain the very 
>slow file transfer that you are seeing. This is the reason why I asked for 
>purely network based transfer (ttcp, netperf, etc.) where the disk is not 
>involved at all.
>
I do use hdparm on hda. The options I use are as following:
hdparm -d1 -D1 -M 128 -S 60
d1.....DMA on
D1.....SMART on
M 128..acoustics to maximum silence
S 60...spindown to 60 secs

So here are the results from netperf P4=server
TCP STREAM TEST to 192.168.123.127
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    10^6bits/sec

87380  16384  16384    10.00     33.46

P75=server
TCP STREAM TEST to 192.168.123.158
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    10^6bits/sec

 87380  16384  16384    10.01      51.01

Here is the output from hdparm -i /dev/hda

/dev/hda:

 Model=ST340016A, FwRev=3.19, SerialNo=3HSD2NT9
 Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
 RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4
 BuffType=unknown, BuffSize=2048kB, MaxMultSect=16, MultSect=16
 CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=78165360
 IORDY=on/off, tPIO={min:240,w/IORDY:120}, tDMA={min:120,rec:120}
 PIO modes:  pio0 pio1 pio2 pio3 pio4
 DMA modes:  mdma0 mdma1 *mdma2
 UDMA modes: udma0 udma1 udma2 udma3 udma4 udma5
 AdvancedPM=no WriteCache=enabled
 Drive conforms to: device does not report version:  1 2 3 4 5

>
>  
>
>>eth0: Setting 100mbps full-duplex based on auto-negotiated partner 
>>ability 45e1.
>>    
>>
>
>So autonegotiation does happen (if eth0 is the right interface) ! If 
>purely network load is still very slow and with errors, I'd suspect the 
>network cables.
>
>  
>
But even with the crossover cable the transfer rate is under 700Kbps, 
even without collisions and errors.

Regards!

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


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