Performance figures for new core on 4xXeon

darkone · 09-13-2003, 05:20 PM

--- clip
[00:13] <_Jarod> wow
[00:13] <_Jarod> d1
[00:13] <_Jarod> finally HARDDISK speed
[00:13] <_Jarod>

[00:13] <dark0n3> harddisk speed?-)
[00:13] <_Jarod> yes
[00:13] <_Jarod> 40-100mbyte/s
[00:13] *** Mouton is now known as Mouton[O]
[00:13] <_Jarod> and not 3-6 :P
[00:14] <dark0n3> wtf are you running it with?-)
[00:14] <ADDiCT_> what was holding you back ?
[00:14] <dark0n3> 100mbyts/sec sounds crazy
[00:14] *** Joins: ephx (~a*****@just.h4ck.it)
[00:14] <dark0n3> mbytes/sec
[00:14] *** Parts: ephx (~a*****@just.h4ck.it)
[00:14] <dark0n3> mbits/sec = more reasonable
[00:14] <nVision> it's quite joyful
[00:14] <Zer0Racer> yeah millibytes

[00:14] *** dReam` is now known as dRm\l0st
[00:14] <_Jarod> scsi 15000
[00:14] <dark0n3> uhh
[00:14] <dark0n3> crazy
[00:14] <_Jarod> 4x xeon

[00:14] <dark0n3> but i'm gald

[00:14] <dark0n3> glad
[00:14] * freak007| ist away (Freak007 afk - BNC on)
[00:14] <dark0n3> as i said
[00:14] <_Jarod> yeah me too!
--- clip
[00:18] <EiQ> it does 30meg/s here on single pc

[00:18] <[RT]> D1 - I will benchmark for you on dual Xeons on gigabit with EMC storage on Monday
[00:18] <dark0n3> but like i promised
[00:19] <_Jarod> Transferred: test.avi 730,454,016 bytes in 11.52 (63,407,47 KBps)
--- clip
[00:25] <_Jarod> Transferred: test2.avi 730,454,016 bytes in 9.32 (78,374,89 KBps)

I guess these figures should satisfy even the most demanding users

Now I'm certain, that new core was really worth all the trouble... (and it was a lot of trouble to get it working..)

ADDiCT · 09-13-2003, 05:22 PM

nice work d1

(i wish i had the hardware to get those speed

)

darkone · 09-13-2003, 05:29 PM

Ok.. that's actually nothing, here's performance for two concurrent transfers (remember that both daemon & client are running on the SAME computer, so actual performance is nearly twice as much!!!)

[01:08] <_Jarod> 4690092032 bytes received in 78.34Seconds 60037.02Kbytes/sec.
[01:08] <_Jarod> 4690092032 bytes received in 82.57Seconds 56801.41Kbytes/sec.
[01:08] <_Jarod> http://www.seagate.com/support/disc/...t373453lw.html
[01:11] <_Jarod> stresstested at 130mbyte/s, so that about correct

Btw. I think I forgot to say this: Thank you, jarod

(I hope you aren't getting spammed due this

)

thewarden · 09-14-2003, 01:22 AM

Quote:

Originally posted by ADDiCT
nice work d1
(i wish i had the hardware to get those speed )

You can at a fraction of the price ---- SATA !!!

WesternDigital 36GB S-ATA 10k RPM drives

richto · 09-15-2003, 07:46 AM

IBM X360 Quad Xeon connected to local EMC Clarion fibre channel storage as server , with Suse Linux 8 Server as the client via an IBM blade centre Gigabit switch I get transfer of 87.5 MB / Sec on a 100 MB file, single transfer to /dev/null

That's without enabling Jumbo frames on the Gigabit link - im going to try that next.

richto · 09-15-2003, 12:47 PM

Using Jumbo Frames (9000) I get 111 MB / Sec, repeatable performance, single file transfer, file cached in server memory (4GB), downloaded to DEV/NULL on Linux client.

Wow.

Mavke · 09-16-2003, 05:16 AM

Nice performance on ioFTPD :-)
I will have to upgrade mine to beta5 shortly and then I can test it as well...

Flare2 · 10-13-2003, 08:50 PM

I have some 8x2.6GHZ P4 Xeons machines that you can stresstest on.

10 of those machines clusternoded together. I'm sure the results would be quite interesting.

if you're interested let me know.