Gigabyte GA-G33M-S2
Core 2 Quad Q6600 G0, VID=1.275V – 1.05V in Bios(0.976V in CPU-Z under load)
2x1GB OCZ DDR2-900 4-4-3-14 at 667MHz 5-5-5-13
Silverstone ST230NF – 300W fanless
Samsung HD400LJ – 400GB SATA
Samsung SH-S183A – DVD+-RW
Scythe Ninja+ (Rev A) with Scythe S-FLEX SFF21E at 630 RPM

Idle - 1.6GHz, 266x6 – 58.5W (1.05V), 66.5W (Stock, 1.120V); Delta – 8W
Prime95 - 2.4GHz, 266x9 – 107W (1.05V), 141W (Stock, 1.184V); Delta – 34W

Over-clocked – RAM 833MHz 5-5-5-13

Idle - 2.0GHz, 333x6 - 67W (1.12V); Delta from 2.4GHz, 1.05V – 8.5W
Prime95 – 3.0GHz, 333x9 - 157W (1.275V); Delta from 2.4GHz, 1.05V – 50W

Notes:
Load - Prime95 x4 (Large FFTs)

Nice build. Quiet, fast, economic on power. The only thing it lacks is a video card, but I take it you don't play games

I measured power consumption using a real-world application as Prime95 isn’t typical. I chose the H.264 encoding benchmark that graysky posted on these forums and recorded 92W with all 4 cores fully loaded.

Here’s the cooling data which uses the same hardware as above but with these differences:

Antec P182
Samsung HD501LJ
Case fan (parallel to CPU) – Nexus 120mm @ 540 RPM
CPU fan - Scythe S-FLEX SFF21E @ 410 RPM
Hard Drive fan – Nexus 80mm with Zalman Fanmate 2 at lowest setting.

Ambient – 22C
Idle – 31/31/33/36C
Prime95 (Large FFTs, 2 hours) – 47/51/51/51
Delta – 16/18/18/15

NorthBridge – idle/maximum – 22/40C
Hard drive – idle/maximum – 32/37C

The temperatures blew me away more than the power consumption. Tjunction for this CPU is 100C so there’s nearly 50 degrees of headroom and that’s with the CPU fan at only 410 RPM. I’ll test the temperatures at 3GHz later on.

As you would imagine from the components used this system is very quiet and the noise doesn’t ramp at all under load.

Notes:
The drive bay in the main chamber was removed and the Samsung HD501LJ was placed on a foam acoustic tile (3cm thick) located where the drive bay would have been. The Nexus 80mm fan was attached to the tile with an elastic band and located about 1cm from the drive. I’d like to experiment with angling the fan so that more of the air passes over the surface of the drive.
Speedfan was used to control the speed of the two 120mm fans and Everest was used to monitor the fan speeds.

Does that motherboard have HDMI?

Thanks,
Chris

No, but the GA-G33M-S2H does and also adds DVI, optical S/PDif and eSATA but loses the COM and parallel ports. Apart from the ports the boards are just about identical in terms of features and power consumption.

Two differences between the -S2H and the other two mATX boards (-S2 and -DS2R) are that the -S2H uses the RTL8110SC chip for Gb ethernet, which uses a PCI interface (it can't technically handle the Gb bandwidth), while the other two use the PCIe-based RTL8111B, and that the -S2H uses 12 lanes of PCIe to provide its built-in DVI and HDMI outputs (the main PCIe slot is only 4x). Neither of those is particularly important for most people, but someone who may want to add in a high-performance video card, but wants DVI in the meantime, has to make some compromises, as does someone looking for top networking performance.

One question i have is about whether these boards can boot from a PATA drive, considering that the interface is provided by a 3rd-party chip, especially if there is also a SATA drive attached. Has anyone tried this or seen an option in the BIOS for it? I would most likely boot from a CF card on the PATA port if i ever decided to get one of these in spite of its energy gluttony.

Oops, I’d forgotten about those differences as they didn’t influence my purchasing decision. Gigabyte usually include a diagram in the user manual which shows how everything hangs together in terms of connectivity; that’s how I noticed the PCI NIC. It seemed an odd decision but the lack of PCIe lanes explains why they chose that route.

The BIOS doesn’t offer an option to boot specifically from the PATA port which isn’t surprising. I don’t have a PATA drive to test it with either.

Rats. Well, it's good to know, anyway.



Funny thing. Googling GA-g33m-s2 review gets a page of pricefinders, and then this link with the fragment where you say that it is an energy hog in a review.

There are different reviews all over the place. My plans changed and i don't need a system right away, so i'll just chill and see what is known by the time i have to buy. It seems like power consumption comparisons have become commonplace in enthusiast reviews, so hopefully we'll see more manufacturers paying attention to that themselves. Apparently ASUS has actually revised their faulty boards to cut off 10-20 watts, putting them in the neighborhood of the Gigabytes.

If you read one of the end posts on that thread I report my own testing of the G33M-S2H which found the power consumption to be very competitive; the Tomshardware review seems to be an anomaly. You can achieve 55W at idle without trying and under 50W if you undervolt with an E2140. That’s with 400GB SATA, 2GB DDR2-667, DVD-RW and with a Silverstone ST30NF which isn’t especially efficient at low loads.

How much power does this save you compared to say a P35 part and a 8500GT? 10 watts?

Thanks,
Chris

A G33 board utilising the IGP should consume around 32W less than a P35 board using a 8500GT with all other things being equal. A G33 board utilising the same card will also consume about 32W more as well.

I'm pretty amazed by this...though, I do have a bit more power demanding setup.

x2 4000+
2x hard drives(raptor+500gb WD)
x1950pro
matx nvidia 7025
4x1gb DDR-800
1xoptical drive
I've tested the system with a PicoPSU and a FSP Zen 400W. Both are approximately the same for overall efficiency sake. Undervolted to 1.05v @ stock speed, the system draws 109-111W while folding. Idle is around 85W. If I subtract ~40W for the extra hard drive+the idling X1950, that's down to 71W under load. Now, with two more cores, you only draw 20-30W more...quite impressive. If the intel motherboards and the Q6600 weren't so expensive, I would have gone that route.

Yeah, i have read the thread. It was just funny that your comment about it being a power hog was at the top of google for the mobo.

I have seen several reviews of Gigabyte products (mobos and vid cards) and they almost universally consume more power than the comparable models from other OEMs. I wish it wasn't so, but sadly, Gigabyte seems to have some power hog issues. That's not to say that, objectively speaking, one can't get great results with them, though.

With regards to PATA booting I have just successfully installed Vista Business 64 from a SATA DVD-RW drive onto a Samsung SP0812N IDE drive on my G33M-DS2R.
So PATA booting can definitely be done.

I pulled the trigger on the deal for the Solo case and Earthwatts 380 from newegg for less than $50 combined today. It seems like one of the Gigabyte boards is still the best way to go, so i browsed through the manual and it says that the PATA devices are counted before the SATA devices in the physical disk ordering. If and when i get one of the boards all report on my success with CF as an OS drive and a GP 1TB as the storage.

So what exactly are the power requirements for the Q6600 then? Is a standard 380W 80plus rated supply going to be enough?

Yep, my NSK3400 (with non 80-plus 380w PSU) was more than sufficient.

Also for info my current server setup with 4*500GB WD, 4*250GB Hitachi, 1*80GB Samsung and 1*40GB 2.5" Toshiba HDD along with 4*1GB DDR2 RAM, DVD-RW, G33M-DS2R with onboard VGA etc. and Q6600 @ 1.1v @ 2.66GHz uses only ~175w (peeking at ~220w when all the drives spin up) from the wall whilst running 2*F@H SMP under Vista 64.

2 instances of F@H? Wouldn't it be just as efficient to run 1?? The SMP client uses 4 processing cores by default.

Amazingly, no!
I found this little program http://distributed.org.ua/forum/index.php?showtopic=1149 which I've been testing out on XP 32 and Vista 64. So long as you have a quad core it enables you to run two simultaneous instances of F@H SMP and, somehow, the average PPD increases by up to ~40%. My XP 32 Q6600 @ 2.66GHz has gone from ~2,400 PPD to 2 * 1,600 PPD using the 2653 project. My new G0 Vista 64 Q6600 is showing similar gains but Vista seems a little flakier with SMP as if you turn the client off it trashes all the checkpoints done and starts again from the beginning!

As a way of boosting SPCR's flagging PPD score this definitely helps! I think I might post another thread in the F@H forum to highlight this.

If im not mistaken the 4core WU's get less points...

Really? I haven't seen any difference between identical projects on different cored machines.

Somehow I don't believe it. Running two instances should definitely be slower, because you'll be running two threads on each cpu. Guess it wasn't loading the cpu properly with just 1 client.

Yep, the client although intended for 4 cores (so it says somewhere I believe on the F@H site) mustn't have fully utilised them.
And from the Affinity Changer url
So, it looks like it claims to improve perfromace on regular dual core machines too. Most odd, but good.

I guess I'm still confused as to why this is necessary, and I don't have the time right now to read about affinity and it's impact on the F@H client. I just know that I've got 4 F@H threads, and they're loading both cores a combined 100%...not sure how I could get any more performance out of it. But F@H centric tools are a plus, nice to see people working on tweaks.

I suppose it isn't really necessary but if you've got a machine that can handle the load why not simply install the program, run a second SMP client and get a free 30-40% performance boost.
Saves spending money on a QX9650 etc. to get the same effect.

I wonder whether Stanford would actually endorse/incorporate this kind of tool themselves?

Hey, this info belongs over on the Folding Subject. Also - Do you know anyone else doing this. If I recall correctly some people were running SMP with P4's that have VT, but I don't have an averager PPD for them.
And how many watts are you Q6600's using?

It is. Here.
However nobody seems interested in that thread
I'll try to continue this conversation over there.