I got my new server :)

graudeejs · Jan 21, 2010

Some of you might probably notice, that yesterday, I wasn't here on forum as much as usually

Finally, yesterday, at about 14:00 localtime I got my hands on my new server

The only difference from that post was, i picked different PSU and Kingston ram 2x2GB

It took me few hours to assemble it [It's first time I completely assemble PC]

Then after 2-3 hours I insert FreeBSD AMD64 dvd. and boomer, it won't boot. Now after few hours of messing with bios configuration, and trying to boot from AMD64 dvd i decided to try i386. It worked.

Then I found out, that AMD64 dvd was damaged for some reason [It was fresh dvd], I burned another dvd with freebsd amd64 and vuala
It worked.

It seams that everything worked, even integrated sound card. I din't really tested sound, but if someone is interested in buying same motherboard, let me know, I will test it.

I Will post dmesg later

Right now I have to study a bit for tomorrow.

Also with this motherboard there was 1 interesting problem. FreeBSD DVD {bough i386 and amd64} won't boot if sata is set in achi mode.

So right now I'm using 2 bios profiles, 1 if I want to boot from DVD {this one turns ACHI off} and one for usual usage {turns ACHI off}

While disk names changes {ad2 to ada1 for example} zfs works without problem

I"m very happy
It recompiled word with make -j 8 buildworld or -j 10 {don't remember} in 25m without ccache. And later In bios i found some features, that would improve this time

Now they are turned on

dennylin93 · Jan 21, 2010

Using /etc/src.conf might speed up the process a little. By the way, do you have any pictures

?

graudeejs · Jan 21, 2010

I will post pictures later {tomorrow I'll get camera}

EDIT:
When I dd from /dev/random to zfs
zpool showed speed up to 250M/s. Compare to my desktop PC with max 35-40M/s

sixtydoses · Jan 21, 2010

killasmurf86 said:
I will post pictures later {tomorrow I'll get camera}

I was just asking to myself.. "where are the pictures?", before I saw this post

.

Congratulations

phoenix · Jan 21, 2010

To prevent the name change issues, you can use glabel(8) (for block device, MBR slice, and filesystem labels) or gpart() (for GPT partition lables). Then just reference everything by the label. That way, it doesn't matter what the underlying driver does or what the device node is called.

Also makes it easy to move drives from controller to controller.

Haven't played with GPT yet, but I now label everything with glabel. Even for simple things like CompactFlash-based gmirrors.

graudeejs · Jan 21, 2010

Yes, I used to do that when I was using UFS
Works fine with gpt as well, in fact I label my flash with glabel

Daisuke_Aramaki · Jan 21, 2010

Awesome. Post some pictures when you get hold of the camera mate.

graudeejs · Jan 22, 2010

here are pictures:
http://www.failiem.lv/list.php?i=cszmxh

I will post dmesg later... can't connect to server via ssh for some reason [will fix tomorrow]

Sorry, failiem.lv ignores exif data, that's why some pictures aren't rotated

graudeejs · Jan 23, 2010

attached dmesg

Up until yesterday Server was extremely quiet {only 2 fans, 1 for PC, 1 for PSU}. Yesterday I added 2x 90x90mm fans.

Later I will create simple fan controller, and these fans will be turned on only when necessary. Current temp inside box is 25-30 Celsius

vermaden · Jan 23, 2010

killasmurf86 said:
attached dmesg

Check your hardware mate:

Code:

[U]ad2: DMA limited to UDMA33, device found non-ATA66 cable[/U]
ad2: 476940MB <WDC WD5000AAKS-00V1A0 05.01D05> at ata1-master UDMA33

... because other (identitcal) two disks are detected correctly (in AHCI mode):

Code:

ada0 at ahcich2 bus 0 target 0 lun 0
ada0: <WDC WD5000AAKS-00V1A0 05.01D05> ATA/ATAPI-8 SATA 2.x device
ada0: 300.000MB/s transfers
ada0: 476940MB (976773168 512 byte sectors: 16H 63S/T 16383C)
ada0: Native Command Queueing enabled
ada1 at ahcich3 bus 0 target 0 lun 0
ada1: <WDC WD5000AAKS-00V1A0 05.01D05> ATA/ATAPI-8 SATA 2.x device
ada1: 300.000MB/s transfers
ada1: 476940MB (976773168 512 byte sectors: 16H 63S/T 16383C)
ada1: Native Command Queueing enabled

graudeejs · Jan 23, 2010

Hmm, it's allright now

Code:

ada0 at ahcich10 bus 0 target 0 lun 0
ada0: <WDC WD5000AAKS-00V1A0 05.01D05> ATA/ATAPI-8 SATA 2.x device
ada0: 300.000MB/s transfers
ada0: 476940MB (976773168 512 byte sectors: 16H 63S/T 16383C)
ada0: Native Command Queueing enabled
ada1 at ahcich11 bus 0 target 0 lun 0
ada1: <WDC WD5000AAKS-00V1A0 05.01D05> ATA/ATAPI-8 SATA 2.x device
ada1: 300.000MB/s transfers
ada1: 476940MB (976773168 512 byte sectors: 16H 63S/T 16383C)
ada1: Native Command Queueing enabled
ada2 at ahcich14 bus 0 target 0 lun 0
ada2: <WDC WD5000AAKS-00V1A0 05.01D05> ATA/ATAPI-8 SATA 2.x device
ada2: 300.000MB/s transfers
ada2: 476940MB (976773168 512 byte sectors: 16H 63S/T 16383C)
ada2: Native Command Queueing enabled

that dmesg was before I did some bios enhancing, and some SATA cable switching.... weird, well anyway it's ok now

Now each of drives is on it's master SATA

Graaf_van_Vlaanderen · Jan 23, 2010

Interesting machine...

Could you run 'flops' and 'ubench' from

Code:

/usr/ports/benchmarks

I'm curious about the CPU performance.

graudeejs · Jan 23, 2010

Code:

$ flops                                                                     

   FLOPS C Program (Double Precision), V2.0 18 Dec 1992

   Module     Error        RunTime      MFLOPS
                            (usec)
     1      4.0146e-13      0.0081   1721.5548
     2     -1.4166e-13      0.0091    769.4256
     3      4.7184e-14      0.0091   1875.5363
     4     -1.2557e-13      0.0099   1519.5352
     5     -1.3800e-13      0.0277   1048.0661
     6      3.2380e-13      0.0145   1993.6528
     7     -8.4583e-11      0.0265    452.9220
     8      3.4867e-13      0.0267   1122.6124

   Iterations      =  512000000
   NullTime (usec) =     0.0000
   MFLOPS(1)       =   953.2093
   MFLOPS(2)       =   867.2711
   MFLOPS(3)       =  1191.8184
   MFLOPS(4)       =  1511.5009

Code:

$ ubench                                                                    
Unix Benchmark Utility v.0.3
Copyright (C) July, 1999 PhysTech, Inc.
Author: Sergei Viznyuk <sv@phystech.com>
http://www.phystech.com/download/ubench.html
FreeBSD 8.0-RELEASE-p2 FreeBSD 8.0-RELEASE-p2 #0: Thu Jan 21 02:33:43 EET 2010     root@bsdroot.lv:/usr/obj/usr/src/sys/GENERIC amd64
Ubench CPU:   495485
Ubench MEM:   139070
--------------------
Ubench AVG:   31727

Graaf_van_Vlaanderen · Jan 24, 2010

killasmurf86 said:

Code:

$ flops                                                                     

   FLOPS C Program (Double Precision), V2.0 18 Dec 1992

   Module     Error        RunTime      MFLOPS
                            (usec)
     1      4.0146e-13      0.0081   1721.5548
     2     -1.4166e-13      0.0091    769.4256
     3      4.7184e-14      0.0091   1875.5363
     4     -1.2557e-13      0.0099   1519.5352
     5     -1.3800e-13      0.0277   1048.0661
     6      3.2380e-13      0.0145   1993.6528
     7     -8.4583e-11      0.0265    452.9220
     8      3.4867e-13      0.0267   1122.6124

   Iterations      =  512000000
   NullTime (usec) =     0.0000
   MFLOPS(1)       =   953.2093
   MFLOPS(2)       =   867.2711
   MFLOPS(3)       =  1191.8184
   MFLOPS(4)       =  1511.5009

Code:

$ ubench                                                                    
Unix Benchmark Utility v.0.3
Copyright (C) July, 1999 PhysTech, Inc.
Author: Sergei Viznyuk <sv@phystech.com>
http://www.phystech.com/download/ubench.html
FreeBSD 8.0-RELEASE-p2 FreeBSD 8.0-RELEASE-p2 #0: Thu Jan 21 02:33:43 EET 2010     root@bsdroot.lv:/usr/obj/usr/src/sys/GENERIC amd64
Ubench CPU:   495485
Ubench MEM:   139070
--------------------
Ubench AVG:   31727

I think you forgot to copy the last digit of Ubench AVG.
Here are some results of my HP DL320 G5 with a Xeon 3050 and
2GB of memory. I only did the 'flops' since I think 'ubench' is
behaving weird. It doesn't always seem to detect the right number of CPUs/cores.

Code:

FLOPS C Program (Double Precision), V2.0 18 Dec 1992                                             

   Module     Error        RunTime      MFLOPS
                            (usec)
     1      4.0146e-13      0.0146    960.5542
     2     -1.4166e-13      0.0165    425.5307
     3      4.7184e-14      0.0092   1850.3102
     4     -1.2557e-13      0.0084   1787.3292
     5     -1.3800e-13      0.0304    952.3951
     6      3.2380e-13      0.0160   1815.0393
     7     -8.4583e-11      0.0440    272.9022
     8      3.4867e-13      0.0306    981.8061

   Iterations      =  512000000
   NullTime (usec) =     0.0000
   MFLOPS(1)       =   568.6920
   MFLOPS(2)       =   597.3215
   MFLOPS(3)       =   953.5636
   MFLOPS(4)       =  1419.3555

This in fact only runs on one core. The second module tries to calculate 'pi' through iterations of Taylor series. All the other modules are numerical integrations. This is a good tool to see how fast those calculations are done. Lots of cache and very fast memory have limited effect on this test.

I got my new server :)

graudeejs

dennylin93

graudeejs

sixtydoses

phoenix

graudeejs

Daisuke_Aramaki

graudeejs

graudeejs

Attachments

vermaden

graudeejs

Attachments

Graaf_van_Vlaanderen

graudeejs

Graaf_van_Vlaanderen