[Competition] MAX CPU performance with MIN temperature rise

The forum for non-component-related silent pc discussions.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

[Competition] MAX CPU performance with MIN temperature rise

Post by maalitehdas » Mon Feb 22, 2010 9:48 am

Everybody who have a chance to run Zlib, CPU stress test and have access to CPU temp readings, please post here some speed results of your system. Idea of this thread is to
- compare different cooling systems and CPU/MB combinations
- have some discussion around subject "speed against temperature"

Competition is divided in 5 groups. Each post with required information will be updated to right group (quote box) of this first post. Information of each group conditions can be found in those result boxes.

Required information:

- ambient temperature during the tests
- CPU idle temperature (hottest core) after having computer on for 2h
- CPU load temperature (hottest core) after 30min 100% stress test for CPU only
- MB/CPU information
- system cooling information
GROUP 1, 10 fastest totally passive systems: wrote: [nick] [ambient/idle/30min load temperature (C)] [CPU Zlib result (kB/s]]
1. N-K, 21C/37C/50C. Result 32584 kB/s with +29C
2. [no results posted]
3. [no results posted]
4. [no results posted]
5. [no results posted]
6. [no results posted]
7. [no results posted]
8. [no results posted]
9. [no results posted]
10. [no results posted]
GROUP 2: 10 fastest partially passive systems. Chamber where CPU is located must be passive cooled. Active cooling for PSU is allowed if PSU is located in its own chamber. wrote: [nick] [ambient/idle/30min load temperature (C)] [CPU Zlib result (kB/s]]
1. maalitehdas, 20C/34C/40C, Result 22709 kB/s with +20C
2. [no results posted]
3. [no results posted]
4. [no results posted]
5. [no results posted]
6. [no results posted]
7. [no results posted]
8. [no results posted]
9. [no results posted]
10. [no results posted]
GROUP 3: 10 fastest air cooled systems. Fans are the only active cooling parts allowed. wrote: [nick] [ambient/idle/30min load temperature (C)] [CPU Zlib result (kB/s]]
1. [no results posted]
2. [no results posted]
3. [no results posted]
4. [no results posted]
5. [no results posted]
6. [no results posted]
7. [no results posted]
8. [no results posted]
9. [no results posted]
10. [no results posted]
GROUP 4: 10 fastest water cooled systems. Water pump is the only active cooling part allowed. PSU fan is allowed if PSU is located in its own chamber. wrote: [nick] [ambient/idle/30min load temperature (C)] [CPU Zlib result (kB/s]]
1. [no results posted]
2. [no results posted]
3. [no results posted]
4. [no results posted]
5. [no results posted]
6. [no results posted]
7. [no results posted]
8. [no results posted]
9. [no results posted]
10. [no results posted]
GROUP 5: 10 fastest otherwise cooled systems. Here are the results of those who doesn't meet the conditions of any other group. wrote: [nick] [ambient/idle/30min load temperature (C)] [CPU Zlib result (kB/s]]
1. [no results posted]
2. [no results posted]
3. [no results posted]
4. [no results posted]
5. [no results posted]
6. [no results posted]
7. [no results posted]
8. [no results posted]
9. [no results posted]
10. [no results posted]
Last edited by maalitehdas on Thu Jul 22, 2010 9:36 am, edited 5 times in total.

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Mon Feb 22, 2010 10:26 am

To start with:

ZLib result: 2x Core 2 Duo E8400 1800 MHz Abit IX38 QuadGT X38 Dual DDR2-800 5-5-5-15 CR2 22709 kB/s
temperature (ambient): 20C
temperature (idle): 34C
temperature (after 30min 100% CPU load): 40C

CPU cooling: Scythe Ninja Cu, no fan
Intake: P182 front vent, no fan, no ducting
Exhaust: P182 top vent, no fan, ducted by 12cm x 1,2m tube on top of case
Other info: E8400 underclocked (3,0GHz - > 1,8GHz) and undervolted (1,125V -> 0,810V)

not too fast but quite cool i think.

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Tue Feb 23, 2010 9:03 am

Ch'mon, are there people with totally passive motherboard cooling? (PSU fan allowed if PSU is in its own chamber)
Last edited by maalitehdas on Wed Feb 24, 2010 11:47 am, edited 1 time in total.

OddSilence
Posts: 65
Joined: Wed Feb 03, 2010 5:03 pm
Location: Portugal, EU

Post by OddSilence » Tue Feb 23, 2010 7:36 pm

Sorry, maalitehdas, I might be wrong (i don't have the most knowlege/experience), but I don't agree with saying a computer with 1 fan (even if it is in PSu) is "totally passive".. :?

Even if it could be considered that way, how can you compare a "1-fan passive" computer with a "really completely passive" computer? Would the results of that comparison be fair?

Nevertheless, nice idea to compare results of different passive systems. I'd like to see the results.

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Tue Feb 23, 2010 10:01 pm

What about chipset temps, they're very important in fanless system. Components around the cpu area (VRMs), North/South bridge, etc.
I think the mobo is usually the first thing to go due to excessive temp.

I'd also suggest standardising the way temps are measured.

In my view, a system with a PSU fan should not participate. Although this depends on your goal.

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Wed Feb 24, 2010 12:37 am

I like it when topic authors include a detailed description of their goals/objectives in the first message body. Subject title is not enough.

A competition to do this? I think what we need is a site or a sub-site of SPCR where "passive" pc enthusiasts hang out. All passive info would be consolidated in this sub-site.

xan_user
*Lifetime Patron*
Posts: 2269
Joined: Sun May 21, 2006 9:09 am
Location: Northern California.

Post by xan_user » Wed Feb 24, 2010 6:49 am

PlanetOfTheApes wrote:What about chipset temps
and what about hdd temps...

/or is this for ssd only?

Do i still get a "one fan" waiver if my PSU doesn't have one? (pico) -Cause with my ambient temps regularly exceeding 90*f i have no hope of going fanless year around.

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Wed Feb 24, 2010 11:35 am

Good and important questions, thank you. "Passive" stands for passive, no fans, but since almost everybody have a fan in their PSU i suppose we get some more answers by allowing that one. Personally i have a fan in the PSU, but in P182 it's in its own chamber so the fan doesn't cool the main department. I hope to get results from those builds which are passive at least that far.

Not all of us have access to all the temperature sensors in their MB, counting only CPU at this point is fare since almost everybody can access that temperature in their BIOS. Competition is mainly for CPU performance (memory quality does effect to ZLib result a bit too) so it's reasonable that it's the one that counts. In most fast systems it's also the hottest component. Personally I have both GPU and North Bridge running warmer than my CPU.
PlanetOfTheApes wrote: I'd also suggest standardising the way temps are measured.
Standardising would be great, but it would demand people to set their ambient temps and air conditionings to exactly same level. I suppose we get nice directional results without standards and at least some interesting discussion about passive setups with different components in varying cases and circumstances.

Waiting for some results!

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Wed Feb 24, 2010 12:21 pm

Instead of rules, some of you might find some instructions useful:

1. Make sure no fans are running around your MB (passive cooled system). Don't shut off your fans unless you know what you're doing!
2. Let your computer stay on for a couple of hours (IDLE)
3. Check your room temperature and your CPU temperature
4. Run a CPU stress test for 30 mins. I'm using Everest software, but there are a lot of others as good
5. Check your CPU temperature after 30 mins, test still running
6. Use some software for CPU performance measurement. Everest have nice benchmarks in it
7. Post here the temperatures, the test result and some information of your system and its cooling.

There's no need to follow these instructions exactly but it does help us all in comparison of the results.

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Wed Feb 24, 2010 7:48 pm

Objective? [Competition] Looking for max CPU performance, passive build

Is this an attempt to find a passive "reference system" we can use to compare temps/performance against.

I'd prefer a volunteer build a "reference system", after a careful component selection process. Don't expect any help from SPCR, they're not interested.


Update continued...
maalitehdas wrote:....some interesting discussion about passive setups with different components in varying cases and circumstances.
I think what is being proposed is a general passive system discussion thread and a place to post test results.


The SPCR passive pc community lacks direction and focus. At the moment everyone is doing "their own thing", which is not productive.

Genuine problem?

If yes, should (or does) the above proposal address this problem?
Last edited by PlanetOfTheApes on Thu Feb 25, 2010 7:13 am, edited 2 times in total.

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Thu Feb 25, 2010 6:45 am

Passive system education and misconceptions

Argument 1: Passive builds are only for people that have the time, energy and skills to build such systems.

Counter: Passive reference systems and recommended components would invalidate this argument. Same weak argument could have been made against silent pc enthusiasts before SPCR site was created.

Argument 2: Passive systems tend to fail prematurely due to excessive heat

Counter: Only true when little thought is given to thermals (i.e. component selection and tuning). Same argument can be made against overclockers. Again, reference systems and recommended components would be used to offset the risk of excessive heat.

Argument 3: Well executed fan based systems are almost inaudible in most situations, are cheaper and have greater performance potential, so why bother going passive.

Counter: Because it's a hobby, period. No sale, then go find another hobby.

Argument 4: "Passive pc enthusiasts think they're better than us."

Counter: If you believe this, you need help and I mean professional help.

(work in progress) to be continued....

Feel free contribute or critque

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Mon Mar 01, 2010 1:30 pm

After one week not a single result posted, too bad :( Instead of that, it's nice to read some very interesting arguments so thank you :o

I suppose most people writing to this forum have either a silent fan build or a WC build, so I suppose I'm not getting much results later on either... Let's move on with the discussion about passive ideas and experiences. A lot of us have a passive cooled GPU, some have passive cooled CPU, some are doing fine without case fans. Purpose of passive system is total silence, but not many of us need that (including me. For me it's a hobby, so I'm totally in argument 3 category in previous mail by PlanetOfTheApes). It's always enough if PC sound vanishes to ambient noise. So let's start with that. Try these nice tests I recently found, they will tell you how many dB's difference your own ear is able to notify. Then use these formulas and tables to calculate how many dB's your system should have at max (for it to totally disappear in ambient noise). Couple of examples: If your ear can make a difference in 1 dB sound levels you need a system 5 dB quieter than your ambient. If your ear manages only 2dB differences, it's enough for you to have a 2dB quieter system (still compared to your ambient). With those formulas you can also predict your systems total sound level if you know a sound level for each of your components (please find them by searching through SPCR and other test reviews, don't believe in manufacturers specs).

Another interesting thing that I've run into has something to do with room acoustics. This is worth a thought especially if you're using +14cm fans at slow speeds. Most of the big fans have a peak in sound level somewhere between 60Hz and 200Hz - and almost every room in the world have standing waves in exactly those frequencies. Setting your system close to a wall (or against any other surface) will easily double it's sound level (adding 3dB), setting it wisely (considering acoustics) it's possible to drop the whole peak frequency in your sitting place! You can check your room's standing waves by listening to this and walking around the room. It's also possible to calculate the ideal locations for noise source / listening place.

Well, that is just one special solution after your build is "as passive as possible" (= APAP in my later posts). There are many better secondary solutions to use first. A high performance PC is quite difficult - almost impossible - to build with passive cooling nowadays, so I suppose this and other secondary methods (acoustic plates, cool air ducting, putting computer to other room...) to reduce noise will be discussed a lot also here at SPCR forums. If acoustics will rise more discussion, i suppose we could open a new thread for that - here we could talk more about APAP. I'd like to hear about your opinions about highest safe temperature levels for each component? They do get hot easily in passive builds. Maybe there will be a day when I check how much heat a component can take before I buy it.

N-K
Posts: 54
Joined: Wed Nov 22, 2006 10:23 am
Location: Finland

Post by N-K » Thu Mar 18, 2010 1:14 pm

I have a totally passive build.
HFX mini case, heatpipes from cpu and gpu to the case.
It was the most horrible thing in my life to put these parts together.
Anyway, here's some results. Load tested with Everest.
(e5300 @ 2,6GHz & 1.05V, GT240, 4GB RAM, SSD)

Ambient temp: 21C
GPU idle: 36C
SSD temp: 38-42C
CPU Core1 idle: 33C
CPU Core2 idle: 37C
CPU Core1 load: 46C
CPU Core2 load: 50C

CPU Zlib 32584 kB/s
CPU Queen 10762

What do I win?

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: [Competition] Looking for max CPU performance, passive b

Post by ces » Thu Mar 18, 2010 2:15 pm

maalitehdas wrote:Everybody with a totally PASSIVE build (no fans, no water pump, no any energy dependent cooling system), please post here some performance readings. One PSU fan is allowed if it's not in the same air space with motherboard.]
In your opinion, what chips and motherboards are good starting points for:
(a) a 775 socket passive system, and an
(b) 1156 socket passive system?

What board / chip combos are in the running?

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Fri Mar 19, 2010 4:54 am

I'd go for a combo that undervolts really well.

After undervolting everything in the bios. My system seems to run cooler, particularly the cpu at load.

As a guide, my i7 undervolts as follows (p95, blend 8 thread load):
0.720v at 1.8Ghz
0.752v at 2.0Ghz

At these voltage/speeds, the cpu power consumption was around 22w according to Gigabyte's EnergySaver utility.

I'd be interested to see how well the new i3 clarkdale combos undervolt. If it can't match an i7, it's FAIL in my view.

~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

Post by ~El~Jefe~ » Fri Mar 19, 2010 6:16 am

there is no reason to put side on a case if everything is passive. This is a complete waste of time to buy a nice case and put the doors on while having no fans. If you get that much dust in your case without any fan movement, I suggest cleaning your house on a regular basis.

e8400 has a lot of cache on it. There are less cache versions of that sort of chip. Cache uses up a lot of watts. I never buy psu's that have fans in them. They are the NOISIEST parts of the system. I have tried the "near silent" recommendations and they are noticable from 20 feet away. Nightjar it, no doors on case.

Also, to the above poster: you do realize that an overclocked e8400 uses less power than your undervolted i7 right? i7 is king of the furnace.

Also, rule until like 2012 or beyond: there are no low wattage quad chips. You need to be less than 45 watts on draw to even consider being passive in a build.

Another thing I would like to add... I would go for non-onboard video. I would get the lowest level 5000 series ati card. The reason is they do a heck of a lot more for a machine in offloading wattage, are built with quality in mind, and have very large heatsinks that arent stuck to the board.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Fri Mar 19, 2010 7:04 am

PlanetOfTheApes wrote:I'd go for a combo that undervolts really well.
What are those combos?
Last edited by ces on Fri Mar 19, 2010 7:20 am, edited 1 time in total.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Fri Mar 19, 2010 7:06 am

~El~Jefe~ wrote: I would go for non-onboard video. I would get the lowest level 5000 series ati card. The reason is they do a heck of a lot more for a machine in offloading wattage, are built with quality in mind, and have very large heatsinks that arent stuck to the board.
Please explain. Are you saying that a non-onboard video will use less power?

If so, what non-onboard cards have the best reputation for low power?

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Fri Mar 19, 2010 7:13 am

~El~Jefe~ wrote:there is no reason to put side on a case if everything is passive. This is a complete waste of time to buy a nice case and put the doors on while having no fans. If you get that much dust in your case without any fan movement, I suggest cleaning your house on a regular basis.
How about these then?

http://www.highspeedpc.com/Merchant2/me ... TopTechSTD

or this
http://www.geeky-gadgets.com/acrylic-cowboy-pc-case/
Put in a Thermalright HR-01+ and place the PSU so that its fan is gently sucking the air away from the HR-01+.

Or take a look at it's big brothers:
http://gadgets.boingboing.net/2008/07/2 ... cases.html

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Fri Mar 19, 2010 7:15 am

~El~Jefe~ wrote:e8400 has a lot of cache on it. There are less cache versions of that sort of chip. Cache uses up a lot of watts.
That would mean either an e3300 or possibly the lowest end Clarkdale. Correct?

How much of a performance hit would than mean from an e8200 or an i3-560?

How much of a wattage gain are we talking about in exchange for giving up that performance?

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Fri Mar 19, 2010 7:19 am

~El~Jefe~ wrote:there is no reason to put side on a case if everything is passive. This is a complete waste of time to buy a nice case and put the doors on while having no fans. If you get that much dust in your case without any fan movement, I suggest cleaning your house on a regular basis.
Take a look at this discussion:
viewtopic.php?t=58072&highlight=

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Fri Mar 19, 2010 7:36 am

~El~Jefe~ wrote:Also, rule until like 2012 or beyond: there are no low wattage quad chips. You need to be less than 45 watts on draw to even consider being passive in a build.
Take a look at this
Quad Q8400S undervolt 0.925V 24W one fan need to be removed
viewtopic.php?t=55084

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Fri Mar 19, 2010 8:47 am

ces wrote:
~El~Jefe~ wrote:Also, rule until like 2012 or beyond: there are no low wattage quad chips. You need to be less than 45 watts on draw to even consider being passive in a build.
Take a look at this
Quad Q8400S undervolt 0.925V 24W one fan need to be removed
viewtopic.php?t=55084
That's a great result for a quad. My i7 seems to be only drawing 22w at the same speed (i.e. 2Ghz), not bad for a "furnace".

To the "furnace" poster, I wasn't suggesting using an i7 to go passive, I was merely documenting the undervolting capabilities of my combo.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Fri Mar 19, 2010 9:29 am

PlanetOfTheApes wrote: My i7 seems to be only drawing 22w at the same speed (i.e. 2Ghz), not bad for a "furnace". To the "furnace" poster, I wasn't suggesting using an i7 to go passive, I was merely documenting the undervolting capabilities of my combo.
So what is the best chip/motherboard combo to do a low powered i7. Or if you were to do it a second time, how would you do it?

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Fri Mar 19, 2010 9:34 am

From Meato
Posted: Tue Sep 15, 2009 2:09 pm

Post subject:
Intel® G31 Express Chipset Max Idle Power 7.4W (worst case) TDP 15.5W

Only Intel Chipset I found more efficient..
Intel® Q43 Chipset (Tech docs)
Max Idle Power 5W
TDP 13W

Its on this Intel DQ43AP motherboard. I have no clue if the bios would let you underclock or adjust the multiplier at all.

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Fri Mar 19, 2010 11:28 am

ces wrote:
PlanetOfTheApes wrote: My i7 seems to be only drawing 22w at the same speed (i.e. 2Ghz), not bad for a "furnace". To the "furnace" poster, I wasn't suggesting using an i7 to go passive, I was merely documenting the undervolting capabilities of my combo.
So what is the best chip/motherboard combo to do a low powered i7. Or if you were to do it a second time, how would you do it?
The problem with the passive build community is that everyone does their own thing. There're no fully documented reference systems or recommended components for building a passive system. So I don't think anyone can answer your best chip/mobo question with any certainty.

I never set out to build a passive system with the i7, only a very quiet system which I think I have succeeded. If need be, I could easily use a single cpu fan to cool the entire system and replace the harddrive with an SSD and have a single moving part system, a 600 rpm cpu fan, which would be virtually inaudible once placed under a desk. I could go passive as an experiment but I don't see the benefit really.

The reason why I'm interested in passive systems is because I had a Zalman TNN500A case for many years and became addicted to it. I recently tried upgrading it with the i7, but couldn't pull it off due to the CPU block interfering with components around the mobo socket area. So I had to ditch it and build my current fan based system.

Just curious or is someone here really interested in building a passive system, if so, why?

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Fri Mar 19, 2010 2:28 pm

N-K wrote:I have a totally passive build.
HFX mini case, heatpipes from cpu and gpu to the case.
It was the most horrible thing in my life to put these parts together.
Anyway, here's some results. Load tested with Everest.
(e5300 @ 2,6GHz & 1.05V, GT240, 4GB RAM, SSD)

Ambient temp: 21C
GPU idle: 36C
SSD temp: 38-42C
CPU Core1 idle: 33C
CPU Core2 idle: 37C
CPU Core1 load: 46C
CPU Core2 load: 50C

CPU Zlib 32584 kB/s
CPU Queen 10762


What do I win?
Thanks for your result post, you beat me at least. I tried to get close to your temperatures, I ended up with 50C / 45C at CPU Cores (load temps), 31012 @ ZLib, 10125 @ CPU Queen. This result was reached with E8400 @ 2,4GHz and 0,9V.
I still hope to get more results from others too.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Sat Mar 20, 2010 6:29 am

Here is an interesting post from Mats
Post subject: TDP list for Intel chipsets
All numbers are maximum TDP taken from Intel.

IOH:

X58: 24.1 W


MCH:
(Note: Idle Power is based on a typical part in system booted to Windows OS with no background applications running.)


X48: 26.5 W, 12.3 W idle, with 333 MHz FSB
P45: 22 W, 9 W idle, with 333 MHz FSB
G45: 24 W, 9 W idle, with 333 MHz FSB
P43: 22 W, 9 W idle, with 333 MHz FSB
G43: 24 W, 9 W idle, with 333 MHz FSB

X38: 26.5 W, 12.3 W idle, with 333 MHz FSB
P35: 16 W, 5.9 W idle, with 333 MHz FSB
G35: 28 W, 11 W idle, with 333 MHz FSB
Q35: 15 W, 6.5 W idle, with 333 MHz FSB
G33: 14.5 W, 5.75 W, with 333 MHz FSB
P31: 15.5 W, 7.6 W idle, with 266 MHz FSB
G31: 15.5 W, 7.4 W idle, with 266 MHz FSB

G965: 28 W, 13 W idle
Q963: 28 W, 13 W idle
Q965: 28 W, 11 W idle
P965: 19 W, 10 W idle

975X: 13.5 W
955X: 13.5 W
945G: 22.2 W
945GZ: 22.2 W
945P: 15.2 W
945PL: 15.2 W

925XE: 13.3 W
925X: 12.3 W
915G: 16.3 W
915GV: 16.3 W
915GL: 16.3 W
910GL: 16.3 W

875P: 10.1 W
865G: 12.9 W
865GV: 12.6 W
865PE: 11.3 W
865P: 10.3 W

845GE: 6.3 W
845PE: 5.6 W

ICH:

ICH10: 4.5 W
ICH9: 4.0 W
ICH8: 4.1 W
ICH7: 3.3 W
ICH6: 3.8 W
ICH5: 2.4 W
ICH4: 2.2 W

Last edited by Mats on Thu Feb 12, 2009 3:49 pm; edited 6 times in total

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Sat Mar 20, 2010 6:35 am

Looks like a motherboard with a G33 or G31 chip set (or a 845GE chip set if you can find one), coupled with an ICH7, is the way to go for the smallest thermal envelope.

I guess the best CPU would be the e3300.

The only other major factor would be how well the MB Bios will let you underclock.

What boards are most flexible in underclocking?

jonj678
Posts: 20
Joined: Mon Jun 22, 2009 12:28 am
Location: England

Post by jonj678 » Sun May 09, 2010 4:51 pm

PlanetOfTheApes wrote: As a guide, my i7 undervolts as follows (p95, blend 8 thread load):
0.720v at 1.8Ghz
0.752v at 2.0Ghz

At these voltage/speeds, the cpu power consumption was around 22w according to Gigabyte's EnergySaver utility.
Power consumption of a microprocessor scales quadratically with voltage and linearly with frequency. This isn't true, but it's a fair approximation. A stock i7 920, at 2.66ghz, has a tdp (i.e. worst case power draw) of 130W. I'm going to assume a stock vid of 1.2V, as you didn't post yours and this is the right ball park for a 920 D0.

130*(0.72/1.2)^2*(1.8/2.66) = 32W. Pending current clamp data, this is a more reliable figure than the one gigabyte picked out of the air. I'm astonished to conclude that you're probably right, it really is using very little power.

The boards that allow fine voltage tuning are the ones targeted at overclockers. So it's the realm of dfi, asus, gigabyte, evga etc. My x58 UD5 will let me try to boot at 0.5V vcore if I want to. Intel boards are very unlikely to allow undervolting, and zotac have been particularly **** about this on their m-itx boards. Its rare, though not unheard of, for G31/G33 chips to be used in overclocking boards.

Post Reply