Jump to content
Sign in to follow this  
Romir

Supermicro C2SBA+II ATX G33 775, 6 ICH9R Sata, Intel 82566 Gig-e

8 posts in this topic Last Reply

Recommended Posts

"Works."  ;)

 

http://www.supermicro.com/products/motherboard/Core2Duo/G33/C2SBA+II.cfm

 

I didn't want to buy this, but I had to find out if the G33 chipset was the most efficient for socket 775 ATX builds. It has the lowest idle and load power requirements and the meager three phase cpu power is great for low idle numbers.

 

My initial findings certainly aren't disappointing.

 

34w idle with a Celeron 440, 1gb dimm, no fans, and a Seasonic S12 430w. 37w a with gig-e connection. An e7400 idled 1 watt higher which shows the 65nm Celeron holds up against the new 45nm chips. (Edit: A bios change now has the e7400 idling 2w lower than the Celeron.)

 

I'm going to move my array over to it and will have more details later.

Share this post


Link to post

"Works."  ;)

 

http://www.supermicro.com/products/motherboard/Core2Duo/G33/C2SBA+II.cfm

 

I didn't want to buy this, but I had to find out if the G33 chipset was the most efficient for socket 775 ATX builds. It has the lowest idle and load power requirements and the meager three phase cpu power is great for low idle numbers.

 

My initial findings certainly aren't disappointing.

 

34w idle with a Celeron 440, 1gb dimm, no fans, and a Seasonic S12 430w. 37w a with gig-e connection. An e7400 idled 1 watt higher which shows the 65nm Celeron holds up against the new 45nm chips.

 

I'm going to move my array over to it and will have more details later.

 

Once we get some more information I will added it to the Hardware list.

Share this post


Link to post

Great numbers! Thanks for the information.

 

You'd probably find the following link interesting. http://ark.intel.com/cpu.aspx?groupID=37212&code=e5200 Scroll down a bit to "Valid Chipset Combinations" and click on "System TDP" to sort the list. Having seen that earlier would've saved me quite a bit of time.

 

I have my array up and running outside of a case right now.  I'm not sure if I want to keep using my Antec P180 case, its such a cable mess. The bare drives laid out are only 7c over the ambient temperature while spun up and idle.

 

I think I found another another cpu power saving setting because the e5200 idle results were a few watts lower than the other cpus.

 

Seasonic S12 430w, gigabit ethernet plugged in, and NO FANS:

34w e5200 1gb dimm

35w e5200 2gb dimm

38w e5200 2gb dimm x2

42.5w e5200 4gb and 6 WD10EADS spun down

61w e5200 4gb & 6 WD10EADS spun up and idle

85w e5200 4gb & 6 WD10EADS during parity check

93w peak seen when stopping the array while the drives are spun down. The drives have to spin up to flush and sync the drive caches. It sounded like 4 drives spun up at once followed by the last 2. I don't know if the Power Up In Standby jumper helped here or not. Its use after the initial system boot confuses me.

 

Regarding the PUIS jumper, on startup the motherboard thinks the drives have failed because they don't report all their information to the bios. I had the set the bios to continue booting on all errors. The drives do spin up one by one while the kernel detects them. They're detected and my arrays working but the following happens with each drive.

Apr 11 06:41:28 Tower kernel: ata5.00: qc timeout (cmd 0xef)
Apr 11 06:41:28 Tower kernel: ata5.00: failed to IDENTIFY (SPINUP failed, err_mask=0x4)

 

One last thing. Adding my 3 platter 7200rpm cache drive and a Syba pci-e controller changes the all spun down consumption to 46w, 52w with only the cache drive spun up, and 72w with the green drives up and idle.

 

I'd run parity check overnight but the system time was off and the mover started. It'll have to wait until tomorrow when its finished. The scythe ninja heatsink isn't clamped down, only has residue thermal paste on it, and is fanless but the cpu cores are only 45c. It should be safe overnight.

 

Aside from a 80+ power supply of debatable size, I think I'm finished with this build, and it's quite a success.

 

Share this post


Link to post

While I had the system open on my desk I decided to get the idle/load consumption of the cpus I had on hand.

 

Idle load power consumption testing in Windows on a 20gb 2.5" sata hard drive. Seasonic S12 430w as before, passively cooled cpu with no fans attached, gig-e connected, and a usb keyboard + mouse.

 

Celeron 440 2.0ghz, 1MB L2, 800 FSB, 65nm Conroe.

1.10 VID @ min 6x multi (1.2ghz). 1.2625 VID @ max 10 (2.0ghz).

38w idle

58w prime95 load

 

e3200 2.4ghz, 1MB L2, 800mhz FSB, 45nm Wolfdale.

1.10 VID @ min 6x multi (1.2ghz). 1.2625 VID @ max 12 (2.4ghz). Poor load vid sample.

36w idle with both cores enabled

64w prime95 load with both cores enabled

 

e5200 2.5ghz, 2MB L2, 800mhz FSB, 45nm Wolfdale.

1.15 VID @ min 6x multi (1.2ghz). 1.25 VID @ max 12.5 (2.5ghz). Poor idle vid sample.

36w idle with one core disabled in bios

50w prime95 load with one core disabled in bios

36w idle with both cores enabled

64w prime95 load with both cores enabled

 

e7300* 2.66ghz, 3MB L2, 1066 FSB, 45nm Wolfdale.

1.10 VID @ min 6x multi (1.6ghz). 1.25 VID @ max 10 (2.66ghz).

36w idle with both cores enabled

69w prime95 load with both cores enabled

*The older bios is running my e7400 with the e7300 10x max multiplier.

 

e8400 C0 3.0ghz, 6MB L2, 1066mhz FSB, 45nm Wolfdale.

1.15 VID @ min 6x multi (2.0ghz). 1.25 VID @ max 9 (3.0ghz).

42.5w idle with both cores enabled

70w prime95 load with one core disabled in bios

90w prime95 load with both cores enabled

 

e6300 1.86ghz, 2MB L2, 1066 FSB, 65nm Conroe.

1.15 VID @ min 6x multi (1.6ghz). 1.325 VID @ max 7 (1.86ghz).

49w idle

80w prime95 load

 

Aren't Wolfdale's great?

Share this post


Link to post

I picked up a celeron e3200 to downgrade the cpu in my XMBC box (thanks GPU decoding) and had to test its power consumption.

 

Seasonic S12 430w, gigabit ethernet plugged in, and NO FANS:

34w e5200 1gb dimm

34w celeron e3200 1gb dimm

 

&

 

e3200 2.4ghz, 1MB L2, 800mhz FSB, 45nm Wolfdale.

1.10 VID @ min 6x multi (1.2ghz). 1.2625 VID @ max 12 (2.4ghz). Poor load vid sample.

36w idle with both cores enabled

64w prime95 load with both cores enabled

 

Those are the exact results as the my e5200, probably due to the higher voltage. I can't manually set it in this Supermicro board so we'll never know.

 

----------------

 

e8400 C0 update

 

Seasonic S12 430w, gigabit ethernet plugged in, and NO FANS:

34w e5200 1gb dimm

40w e8400 c0 1gb dimm

 

&

 

e8400 C0 3.0ghz, 6MB L2, 1066mhz FSB, 45nm Wolfdale.

1.15 VID @ min 6x multi (2.0ghz). 1.25 VID @ max 9 (3.0ghz).

42.5w idle with both cores enabled

70w prime95 load with one core disabled in bios

90w prime95 load with both cores enabled

Share this post


Link to post

I ordered ten WD10EADS drives for my backup sever so it's time for another power consumption test while getting the system ready. I installed all my controller cards for the build into this motherboard board and measured the watts used without drives or case fans.

 

System: Supermicro C2SBA+II, Celeron 440 with Intel HS/F, 2x 1GB DDR2 800 1.8v, SeaSonic 430w S12, 2x Adaptec 1430SA, RocketRaid 2300, Mascool 4 port PCI silicon image card (for port 19/20 and cache if wanted). Usb keyboard and gigabit lan plugged in like before. (A larger and more efficient power supply will be used down the road when the drive count increases.)

 

Idle wattage in unRAID: 53w  :o Having a 45nm dual core and only 1gb might bring in down to 50w even.

 

Parity sync speeds should be around 50 MB/s give or take 10 with the maximum 20 data drives. I plan on testing the speeds of the three controllers with drives only attached to them to determine the bottleneck. IIRC, the x1 slot was only good for 200 real world MB/s last time I tested a 1430sa in it.

 

Edit: Power consumption without the 4 add-on controllers is 40w idle.

Share this post


Link to post

The backup server build around this motherboard is now finished. Rsyning the files over ssh went smoothly so all that's left is moving it into the classic stacker case.

 

With the same setup as the previous post the power consumption was:

53w no drives, fans, backplanes, only the 4 add-in cards - gig-e connection etc

67w 10 WD10EADS spun down

93w spun up with heads parked

112w spun up without heads parked

128w during parity creation

194w boot up peak

 

With only 1 1430SA and no other add-in cards (10/10 ports used):

40w no add-in cards, backplanes, or case fan

58w 10 WD10EADS spun down

86w spun up with heads parked

103w spun up without heads parked

119w during parity sync

 

Same as above but with a SeaSonic S12 II 330w 80+ Bronze psu:

40w no add-in cards, backplanes, or case fan

53w 10 WD10EADS spun down (take that HP Mediasmart! or Drobo!)

81w spun up with heads parked

98w spun up without heads parked

113w during parity sync

179w boot up peak

 

One 120mm case exhaust fan was plugged in for the previous tests so the results are reflective of the final in place system. Using 1GB and a 45nm dual core should shave a few watts of those numbers. I'm really tempted to swap the cpu and see if the magic 50w idle number with ten drives is hit.

 

Parity creation performance tests with 3 platter WD10EADS drives:

2 on the PCI bus (1 parity, 1 data)

Current position:  19,564,776  (2.0%)

Estimated speed: 43,566 KB/sec

 

4 on the PCI bus

Current position:  11,747,268  (1.2%)

Estimated speed: 24,571 KB/sec

 

2 on PCI-E x1 Syba

Current position:  10,051,296  (1.0%)

Estimated speed: 59,537 KB/sec

 

4 on PCI-E x1 RocketRaid

Current position:  48,798,816  (4.9%)

Estimated speed: 49,539 KB/sec

 

4 on PCI-E x4 1430SA

Current position:  14,982,812  (1.5%)

Estimated speed: 57,195 KB/sec

 

4 on ICH9R

Current position:  21,424,784  (2.1%)

Estimated speed: 59,106 KB/sec

 

The RocketRaid 2300 bios kept restarting the machine when drives were connected to it and a 1430SA. Port 19 and 20 will have to sit on the PCI bus for now, along with 17 and 18. They won't be used for a long time so it isn't a big deal. Taking a full 20-24 hours to calculate parity on 2TB wouldn't be a deal breaker for a home backup server regardless.

 

Share this post


Link to post

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this