Gigabyte DES - motherboard power supply management

All about them.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

CA_Steve
Moderator
Posts: 7492
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Gigabyte DES - motherboard power supply management

Post by CA_Steve » Thu Jan 10, 2008 9:00 am

Gigabyte website info

Tweaktown short article

Kinda cool.

wim
Posts: 777
Joined: Wed Apr 28, 2004 5:16 am
Location: canberra, australia

Post by wim » Thu Jan 10, 2008 3:56 pm

interesting, if i'm not misunderstood the new boards with an 'E' in the model name are pretty much the same hardware as the old boards but with a newer bios installed?
and for those that don’t like flashing lights on their motherboard there’s also the option to switch them off in this utility
hoorayyyyy :)

for the gui display, what does it actually mean..i don't quite understand the picture here... how can you have a total power saving over some period of time, in Watts instead of joules or say kWh ?? :?

Shemale
Posts: 24
Joined: Tue Jan 01, 2008 11:25 pm
Location: România

Post by Shemale » Thu Jan 10, 2008 10:38 pm

switching of power phases depending on CPU workload
I hear something similar is used in laptops. Is a good idea, but must be implemented in bios, not with some application.

On x48 mainboard things look good: http://www.tweaktown.com/reviews/1271/1 ... index.html
But at idle difference is too big to be only a power phases tweak. Hmmm, 52W difference. And 47W difference at full load ? :shock:

line
*Lifetime Patron*
Posts: 338
Joined: Wed Sep 28, 2005 10:07 pm
Location: Israel

Post by line » Sat Jan 12, 2008 9:55 pm


mcb
Posts: 32
Joined: Sat Aug 24, 2002 10:24 pm
Location: San Francisco

Post by mcb » Sat Jan 12, 2008 10:41 pm

Here's another test: http://www.hardspell.com/english/doc/sh ... ws_id=2295

If I read their results correctly:

With C1E, EIST, VID Auto, but DES off
Idle: CPU 10.09W, System 102W
Load: CPU 36W, System 136W

With C1E, EIST, VID Auto, DES on
Idle: CPU 5.5W, System 99W
Load: CPU 29.39W, System 130W

They also tested other configurations. I'm not sure how the TweakTown testing methodology differs but radically different results there. Both use the X48 Gigabyte motherboard though with two significant differences:

CPU: QX6700 (overclocked to 3GHz) vs. E8500
graphics card: 8800GTS 640MB vs. NVIDIA FX 330 (a low-end part).

I assume the graphics card contribution would not be affected by DES.

MiKeLezZ
Posts: 110
Joined: Sun Feb 20, 2005 8:00 am
Location: ITALY
Contact:

Post by MiKeLezZ » Wed Jan 16, 2008 4:50 pm

mcb wrote:Here's another test: http://www.hardspell.com/english/doc/sh ... ws_id=2295

With C1E, EIST, VID Auto, but DES off
Idle: CPU 10.09W, System 102W
Load: CPU 36W, System 136W

With C1E, EIST, VID Auto, DES on
Idle: CPU 5.5W, System 99W
Load: CPU 29.39W, System 130W
3,1GHz@1,1V, C1E off, EIST off, DES off
idle: 10,25 W (12P)
load: 31,85 W (12P)

3,1GHz@1,1V, C1E off, EIST off, DES on
idle: 7,00 W (4P)
load: 29,40 W (8P)

3,1GHz@1,1V, C1E on, EIST on, DES on
idle: 5,50 W (4P)
load: 29,40 W (8P)

mcb
Posts: 32
Joined: Sat Aug 24, 2002 10:24 pm
Location: San Francisco

Post by mcb » Thu Jan 17, 2008 9:38 pm

TweakTown has an updated article on DES here: http://www.tweaktown.com/articles/1275/1.

In part, it reads:

"The test system consisted of the aforementioned GIGABYTE P35-DS3P rev 2.1 motherboard, a Core 2 Duo E6700, 2GB of DDR2 800 memory, a Radeon X1950 Pro, four SATA hard drives, an Ageia PhysX card, an X-Fi Elite Pro, two optical drives and three 120mm fans..."

"After about two days of uptime the system used had saved close to 400W of power, not much in the big scheme of things, but if you have a PC up and running 24h a day then you can save a lot of power in the long run. The actual number for this system was about 189W a day, which turns out to be close to 69,000W 69kWh a year, since power is normally measured in kWh...."

The article (and apparently the DES UI) confuses power with energy. I assume this means they observed a 189W-hr energy savings over 24hrs or 8W power reduction on average.

lm
Friend of SPCR
Posts: 1251
Joined: Wed Dec 17, 2003 6:14 am
Location: Finland

Post by lm » Fri Jan 18, 2008 1:14 am

Something like this applied to PSUs would mean we could have a high maximum capacity PSU that is efficient throughout it's whole range, right?

moogaloo
Posts: 18
Joined: Mon Dec 17, 2007 8:52 am

Post by moogaloo » Fri Jan 18, 2008 5:10 am

I would be interested to see if there was a difference in power consumption in the new X48 Motherboard and the old P35 motherboards with the bios / utility update.

In the article it mentions that the X48 has 12 phases and the P35 only has 6 phases. I wonder how this compares?

MiKeLezZ
Posts: 110
Joined: Sun Feb 20, 2005 8:00 am
Location: ITALY
Contact:

Post by MiKeLezZ » Fri Jan 18, 2008 5:21 am

moogaloo wrote:I would be interested to see if there was a difference in power consumption in the new X48 Motherboard and the old P35 motherboards with the bios / utility update.

In the article it mentions that the X48 has 12 phases and the P35 only has 6 phases. I wonder how this compares?
X48 is an overclocked P35 so it consumes a little more.
Also, 12 phases consume more than 6 phases.
By the way, the largest differences are due to the CPU: new penryns have an impressive low power consumption, and so DES has less influence, compared to "old" Core 2 Duo.

TENAX
Posts: 26
Joined: Sat Jan 05, 2008 8:52 pm

Post by TENAX » Sat Jan 19, 2008 4:52 am

in reference to the notion that a bios update would enable this on a p35 board not currently supported, i'd like to see that information. i do know that i get a not compatible message when i try to run it on a p35 ds4 revision 2.0. sucks pretty big time that they didn't implement this feature by the 3rd revision of my board. on the other hand, i like to do some overclocking with my system and it doesn't sound like the 2 notions will go hand in hand very well, if at all.

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Sat Jan 19, 2008 10:27 am

TENAX wrote:in reference to the notion that a bios update would enable this on a p35 board not currently supported, i'd like to see that information. i do know that i get a not compatible message when i try to run it on a p35 ds4 revision 2.0. sucks pretty big time that they didn't implement this feature by the 3rd revision of my board. on the other hand, i like to do some overclocking with my system and it doesn't sound like the 2 notions will go hand in hand very well, if at all.
I’m under the impression that you need the latest revisions (2.1!) for this feature to work. If it doesn’t work when you over-clock the FSB but leave the VCore as standard then it has limited appeal to me. Otherwise it sounds promising.

TENAX
Posts: 26
Joined: Sat Jan 05, 2008 8:52 pm

Post by TENAX » Sat Jan 19, 2008 10:45 am

your correct..i can verify it does not work on 2.0. nice gimmick, but would need to save a heap of dollars to consider it worthwhile to change my board.

CA_Steve
Moderator
Posts: 7492
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Post by CA_Steve » Sat Jan 19, 2008 11:07 am

I'm just happy to see a mobo manufacturer trying something for energy management.

TENAX
Posts: 26
Joined: Sat Jan 05, 2008 8:52 pm

Post by TENAX » Sat Jan 19, 2008 11:10 am

tres true..and i agree with the review that if someone had 100 computers in there operation with these boards, the environmental savings could start to be signifigant. then again, if they didn't make boards we could overclock, that would be a sizeable saving as well:)

line
*Lifetime Patron*
Posts: 338
Joined: Wed Sep 28, 2005 10:07 pm
Location: Israel

Post by line » Sun Jan 20, 2008 9:24 am

TrustedReviews tested the six-phase GA-EP35-DS4 with Intel's QX9650 and came away with 5W savings at idle and 10W under load.

http://www.trustedreviews.com/motherboa ... P35-DS4/p1

By the way, this board has officially landed in my area. :)

wim
Posts: 777
Joined: Wed Apr 28, 2004 5:16 am
Location: canberra, australia

Post by wim » Sun Jan 20, 2008 4:00 pm

lm wrote:Something like this applied to PSUs would mean we could have a high maximum capacity PSU that is efficient throughout it's whole range, right?
hm, i thought most efficiency loss in psu happens at the AC -> DC rectification stage, so it's probably a very different problem to begin with.

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Mon Jan 21, 2008 12:23 pm

I imagine the power savings using the entry level boards would be smaller or negligible as don’t they use 3 phase power circuitry?
I’m thinking of upgrading my Gigabyte G33 board due to limited PCI/PCIe slots and am only interested in the EP35-DS3 or DS3R so I’m wondering whether to expect any power savings.

moogaloo
Posts: 18
Joined: Mon Dec 17, 2007 8:52 am

Post by moogaloo » Wed Jan 23, 2008 6:52 am

I don't seem to be able to find any Ga-EP35 or GA-P35 Rev 2.1 MBs for sale in the UK (or anywhere)

Anyone know where I can get one?

CA_Steve
Moderator
Posts: 7492
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Post by CA_Steve » Wed Jan 23, 2008 9:19 am


ghettojiggalo
Posts: 51
Joined: Thu Aug 09, 2007 8:28 pm
Location: Burbank, CA

Post by ghettojiggalo » Wed Jan 23, 2008 2:32 pm

CA_Steve wrote:Anand weighs in
seems like DES isn't all its marketed up to be. Using Easy Tune 5/changing Vcore in bios would net bigger energy savings.

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Wed Jan 23, 2008 2:53 pm

CA_Steve wrote:Anand weighs in
It’s not that useful a review in that I doubt many SPCR readers are using an X48 board with an Extreme Edition CPU, PC3-14000 RAM and a kW power supply.

Overall I can’t say that this seems to offer much particularly as Gigabyte boards typically under-volt so well in the BIOS. If I do upgrade my G33 board I’ll test it against one of these new boards as a comparison. The lack of over-clocking makes it unattractive to me as I often buy a lower FSB speed chip and push it up to as close to 3GHz as I can using the stock voltage range.

drees
Posts: 157
Joined: Thu Jul 13, 2006 10:59 pm
Location: San Diego, CA
Contact:

Post by drees » Wed Jan 23, 2008 3:14 pm

smilingcrow wrote:
CA_Steve wrote:Anand weighs in
It’s not that useful a review in that I doubt many SPCR readers are using an X48 board with an Extreme Edition CPU, PC3-14000 RAM and a kW power supply.
No kidding. Anyone who is running a system which idles close to 200w is not going to bother with something that reduces power utilization by 5-10w.

Now if DES reduces power utilization by 5-10w on a "normal" system which idles between 50-100w, now we're talking something substantial.

mcv
Posts: 74
Joined: Sun Jan 06, 2008 11:26 pm

Post by mcv » Fri Jan 25, 2008 12:23 am

I don't mean to introduce a Slashdot meme here, but does it work on linux? As I understand, it comes with a nice piece of Windows software that lets you monitor your energy usage. I can't find anything about DES in relation to linux, however.

Scoop
Posts: 121
Joined: Sat Dec 08, 2007 1:28 am

Post by Scoop » Fri Jan 25, 2008 1:16 am

I was wondering about what they said about DES at Anandtech: "Intel's own power saving features known as C1E and EIST must be set to disabled before DES gains full control of altering processor core voltage on-the-fly. It is also important to point out that full operational use of DES is limited to stock (Auto) BIOS settings for both processor and memory busses."

But those Hardspell numbers seem to defy this as they got better numbers with everything ON.

mcb
Posts: 32
Joined: Sat Aug 24, 2002 10:24 pm
Location: San Francisco

Post by mcb » Sun Feb 03, 2008 9:43 pm

The last (as of Feb 3) response to the Anandtech article:

RE: EIST needs to be off? by Rajinder Gill, 10 days ago
Not really, these processors are now so efficient at idle that the core speed reduction does little to save extra power. Gigabyte have sorted out the EIST/C1E and DES issue in the latest BIOS. You can now have all 3 on simultaneously. Unlike the previous releases..

wim
Posts: 777
Joined: Wed Apr 28, 2004 5:16 am
Location: canberra, australia

Post by wim » Mon Feb 04, 2008 4:47 pm

DES on my GA-EP35C-DS3R does almost nothing. i'm measuring at the wall, and even with the highest level of voltage cut and speed throttle , it can only save 1 or 2 watt
i put this down to E8400 already being a pretty power efficient cpu. this DES technology seems to have missed the boat, would have been much more useful a few years ago when processors were eating lots of power
perhaps it would be better applied to GPU who are still power gluts, and idle more of the time too

Strid
Posts: 397
Joined: Tue Jan 08, 2008 5:09 am
Location: Copenhagen, Denmark

Post by Strid » Tue Feb 05, 2008 1:27 am

Seems like you need to run Windows and use proprietary software in order to activate DES properly. So that's a no-go for us Linux users. Same thing as Asus EPU - needs ASUS Windows software too. :(

mcv
Posts: 74
Joined: Sun Jan 06, 2008 11:26 pm

Post by mcv » Tue Feb 05, 2008 1:33 am

If DES does nothing in Linux, I guess I'd better just stick with the DS3L.

Still, I'm glad to see so many manufacturers pay more attention to power consumption these days.

martdj
Posts: 5
Joined: Mon Nov 13, 2006 7:59 am

Post by martdj » Tue Feb 12, 2008 7:46 am

wim wrote:DES on my GA-EP35C-DS3R does almost nothing. i'm measuring at the wall, and even with the highest level of voltage cut and speed throttle , it can only save 1 or 2 watt
i put this down to E8400 already being a pretty power efficient cpu. this DES technology seems to have missed the boat, would have been much more useful a few years ago when processors were eating lots of power
perhaps it would be better applied to GPU who are still power gluts, and idle more of the time too
Very interesting remark. I ordered a P35-DS3P with an E8400 recently. Still waiting for the cpu, but I received the mainboard and it's an revision 2.0. If I understand you correctly it wouldn't be worth the extra trouble and money to get it swapped for a rev 2.1 then. Or are others with a comparable setup seeing different results?

Post Reply