Annual Update to "CPUs ranked by Noise/Heat"

Want to talk about one of the articles in SPCR? Here's the forum for you.
MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Annual Update to "CPUs ranked by Noise/Heat"

Post by MikeC » Sat Mar 06, 2004 2:58 pm

Didn't mean to go this long -- one day short of a year! -- but holy spinning fans, how time files! :shock:

Minor changes to body text, main changes in table highlighted in blue. CPUs ranked by Noise/Heat

Seal
Posts: 522
Joined: Sat Oct 25, 2003 4:39 am
Location: Uk

Post by Seal » Sat Mar 06, 2004 5:42 pm

Thats a brilliant page, good work. I love the fact that the mobile barton 2500 is in there too :D. I just bought one last week and theyre amazing!

clive
Posts: 47
Joined: Mon Jan 26, 2004 3:27 pm
Location: Sydney Australia

Post by clive » Sun Mar 07, 2004 4:31 am

Mike - a very good resource and greatly appreciated.

Not sure if this is the right forum but is there a definitive guide to acceptable temperatures and as importantly how these are measured.

If not would it be possible to have a guide to measuring temps - both cpu and case both by inbuilt mobo sensors and ones that can be user positioned. Perhaps there could be an adjustment/calibration for each mobo?

I note for similar cpu's a significant range of reported temps which I suspect are not measured with the same degree of accuracy, as there is often insufficient information to make a judgement, and therefore may lead to a false sense of achievement (or lack thereof). I have seen reports of similar cases and cpu's with a range of temps that would be outside statistical range and measuring error.

Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Post by Ralf Hutter » Sun Mar 07, 2004 5:54 am

Hmm, isn't it interesting how all of the newly released CPU's are found in the bottom half (the "loud, hard to cool" section) of the rankings. :)

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun Mar 07, 2004 10:14 am

clive wrote:Mike - a very good resource and greatly appreciated.

Not sure if this is the right forum but is there a definitive guide to acceptable temperatures and as importantly how these are measured.

If not would it be possible to have a guide to measuring temps - both cpu and case both by inbuilt mobo sensors and ones that can be user positioned. Perhaps there could be an adjustment/calibration for each mobo?
Page 2 of SPCR's Unique Heatsink Testing Methodology has a lengthy discussion on this topic. The bottom line: ALL the current methods available for casual monitoring of CPU temps are merely rough guides. You can play it safe, by trying to keep the indicated temps low, or play it a bit more risky by using the onset of system instability to establish the practical high temp you should allow your CPU to go.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun Mar 07, 2004 10:15 am

BTW, the CPU page was updated again this morn with info on Pentium-M Centrino left off inadvertently the other day.

xarope
Posts: 97
Joined: Sat May 03, 2003 8:16 pm

Post by xarope » Sun Mar 07, 2004 11:33 pm

Hmm, I'm not sure I would agree with the rankings of the newer CPUs e.g. the Athlon64s - no idea about the opterons, FXs etc since I don't own one, but I do own a 3200+ and that thing is way cooler than my Athlon XP 2500+. So even though it has a higher watt rating (isn't that kindof misleading, that's just the highest right not the actual rating?), I would consider it easier to cool than an equivalently clocked XP.

or is it the IHS that screws temp. monitorings up?!?

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Mar 08, 2004 12:28 am

xarope -- notice the deliberate wide range of rating for the Athlong 64s: 3-5. WHy? Depends how the system will be used. If doing 100% CPU usage all the time, then let's face it -- it's a 90W CPU, it's like a P4-3.4. But if it is idling most of the time, then it's much closer to a 5, Cool & Quiet makes a huge impact.

Athlon 64 FX-51 doesn't have Cool & Quiet.

Jake
Posts: 15
Joined: Sun Aug 11, 2002 3:26 pm

Post by Jake » Mon Mar 08, 2004 3:51 am

The information for Pentium M can't be correct. My Acer C111TCi Centrino powered by a Pentium M 1GHz uses just 14W with the screen turned on, 10W with screen off, as verified by a power meter I boght. That's at 'normal usage', not using CPU burn software or anything like that. And that's for the whole machine, HDD, motherboard, USB components (wireless keyboard/mouse), WiFi etc.

As I've said before on these boards, this notebook (doubles as a rather-pointless tablet pc) is a life saver for those looking for a silent solution. Its just so damn quiet!!! I bought it for around £1,300 half a year or so ago, should be able to get it for well under £1,000 now, or even less second hand. And its fast, too. No problems with it whatsoever. Highly recommended.

sgtpokey
*Lifetime Patron*
Posts: 301
Joined: Sat Dec 14, 2002 11:29 pm
Location: Dublin, CA / Liverpool UK

Post by sgtpokey » Mon Mar 08, 2004 6:39 am

Yeah, I'm wondering about the Pentium M rating too.

I've got a bunch of laptops, a pIII 700 mhz coppermine, a Pentium M 1 ghz, and a pentium M 1.4 ghz tablet pc.

From a non-scientific standpoint, the cpu area of the pIII gets way warmer (and the fan turns on) than either Pentium M. The Pentium M 1.4 runs fanlessly for everything save graphics intensive things.

[ as an aside, my wife has a gateway m275 tablet for her med-school stuff and at least in that context tablet pc's are very useful :) and quiet!! ]

* oh and also the Pentium M 1ghz has no fan whatsoever... but I thought that went without saying since a good proportion of of PEntium M 1ghz notebook designs are fanless

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Mar 08, 2004 7:54 am

Jake & sgtpokey --

The info about the Pentium-M is correct. It comes from Intel's own tech docs about the P-M, with cross-checks of all the updates to the original spec. A 1.7 does produce 30W and a 1.2 IS rated at 20.8. The Speed Step technology tends to lower real power consumption, however.

I did have it ranked too low. Studied the wattages of the other CPUs and moved the P-M one notch to 7-8 instead of 6-7.

lacsap
Friend of SPCR
Posts: 84
Joined: Mon May 05, 2003 11:29 am
Location: Montreal, Canada

Post by lacsap » Wed Mar 10, 2004 7:57 am

How about P4-M? You didn't include them in your rankings. And they can fit on a S-478 mobo.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Wed Mar 10, 2004 8:07 am

lacsap wrote:How about P4-M? You didn't include them in your rankings. And they can fit on a S-478 mobo.
True. I did mention them in the P-M listing. The truth is that the P4-M's power ratings are almost identical to the P4 anyway, which is why I didn't bother. They have to be more $ and there's not much advantage, your can underclock most P4s anyway. It seemed like a lot of work to put the thing in there with all the different speeds -- would have been another 3 rows to be added... Convince me otherwise, I listen to reason. :)

Jake
Posts: 15
Joined: Sun Aug 11, 2002 3:26 pm

Post by Jake » Wed Mar 10, 2004 8:11 am

Mike - in my opinion, the Pentium M should be given a ranking of "7-10", or broken down into two separate categories. The 1GHz ULV is easily worth a 10 (if the C-3 gets one). Compared to the Via C-3 it dissipates less power and runs much faster in real-world applications (this is comparing my 1GHz ULV against a 933MHz C-3). The result is a notebook (and hopefully, in the future, desktops too) which runs silently and easily fast enough, and also perfect for energy-saving enthusiasts such as myself. Okay, so it can't be bough separately at the moment, and hence results in more expensive systems. But to my knowledge, the scores aren't based upon price but their usefulness in silent systems.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Wed Mar 10, 2004 8:16 am

Jake -- I am willing to bow to your first-hand experience... but can anyone else corroborate Jake's comments?

Jake
Posts: 15
Joined: Sun Aug 11, 2002 3:26 pm

Post by Jake » Wed Mar 10, 2004 8:33 am

"The Centrino processor chip will ship in three power/performance versions – the standard, low-voltage (LV), and ultra low-voltage (ULV) versions. The standard chip is expected to operate under 1 watt power consumption (compared to 1.5W for PIII-M and around 2.5W for PII-M) during normal use. The ULV version should nominally consume less than .5 watts! The thermal design power (TDP), or essentially the maximum power dissipation running software that activates much of the chip's circuitry, is less than 25 watts for the standard chip, less than13 watts for LV, and less than 8 watts for ULV."

Full article at http://www.extremetech.com/article2/0,3 ... 228,00.asp

As I said, the Acer laptop/tablet consumes 10W without the screen on mains power, under normal usage (web surfing, email client, mp3s, IM, MySQL/Apache servers runnng locally). My laptop's tech specs say it has a 26Wh hour battery, and a 3 hour battery life. That would therefore be based upon a total laptop power dissipation of 8.67W, screen and all. In reality, I find it lasts for around 2.5 hours on the road (around 10W).

Any way you look at it, it definitely dissipates less than 10W of power, and deserves (in its 1GHz ULV format) at least as good a grade as teh C-3. The C3 was too slow for me. The Pentium-M definitely isn't. And I'm not an Intel fan per say - I've tried Transmeta Crusoes twice before, found them too slow too. And I've also tried the 933MHz C3 - again, much slower.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Wed Mar 10, 2004 8:55 am

OK, the 7-10 rank is in.

Also, here's a link to one of the very few P-M desktop boards available: Endura LS855 by Radisys I believe the MSRP is ~$350. Then you have to find someone who will sell you a P-M, at what price, who knows?

epohh
Posts: 1
Joined: Wed Mar 10, 2004 9:23 am

Post by epohh » Wed Mar 10, 2004 9:28 am

Athlon64 + Cool n Quiet can do 800 MHz 1,3V 35W
66w @ 1800 MHz (1,4V)

Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Post by Ralf Hutter » Wed Mar 10, 2004 11:49 am

MikeC wrote:Jake -- I am willing to bow to your first-hand experience... but can anyone else corroborate Jake's comments?
Boy, I sure wish I could. I'd love to be able to say I've had some experience with one. Maybe someday someone'll actually release an m-ATX mobo to the general public...
MikeC wrote:Then you have to find someone who will sell you a P-M, at what price, who knows?
Well, among others, how about good 'ol Mwave.com, about 10 miles away from where I live. Banias CPUs from 1.4GHz to 1.7GHz ranging in price from $218 to $437. How many do you want?

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Wed Mar 10, 2004 3:06 pm

epohh wrote:Athlon64 + Cool n Quiet can do 800 MHz 1,3V 35W
66w @ 1800 MHz (1,4V)
yeah but as soon as you have any real apps or a game going, 89W, here we come! This is why it's given a range rating of 3-5.

sthayashi
*Lifetime Patron*
Posts: 3214
Joined: Wed Nov 12, 2003 10:06 am
Location: Pittsburgh, PA

Post by sthayashi » Wed Mar 10, 2004 5:20 pm

This link deserves it's own sticky post. Bookmark it.

http://users.erols.com/chare/elec.htm
(although I just looked now, and it doesn't answer your questions).

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Wed Mar 10, 2004 8:08 pm

Radeonman wrote:Do you guys have datasheets for the various speed bumps and core types just lying around? I can't find crap for the differences between a 2.4C or a 2.6B, an opteron 140 vs an opteron 142, etc. Maybe I just have poor sources.
You gotta sort throught the Intel & AMD tech specs & docs.

ariman
Posts: 16
Joined: Mon Mar 01, 2004 2:58 am
Location: FIN

Post by ariman » Mon Mar 15, 2004 6:30 am

lacsap wrote:How about P4-M? You didn't include them in your rankings. And they can fit on a S-478 mobo.
Dont work on Abit 865/875, Asus boards 865/875 needs bios 1012 and older. Most 845 should work tought You have always multiplier 12x no matter whats original speed.

Mobile Celerons should remain original multiplier, thats what I can verify tomorrow since My MC 2,4G SL75J arrives.

sneaker
Posts: 133
Joined: Tue Dec 24, 2002 8:37 pm
Location: Australia

Post by sneaker » Tue Mar 30, 2004 1:59 pm

MikeC wrote:yeah but as soon as you have any real apps or a game going, 89W, here we come! This is why it's given a range rating of 3-5.
I can't help but think the Athlon 64 is a little underappreciated being shoved all the way down the bottom of the list. Its 89W peak certainly won't impress SFF fans who take the "how hard is this to cool in a 5-inch square box" perspective, but many of us are still tower users who (as the vid card forum will attest) desire good performance in the latest games and power for things like video processing. As much as we'd all like to see high performance sub-30W Pentium Ms take over the desktop, the choice at this end of the market at the moment is between an Athlon 64 or >3GHz Pentium 4.

For those power users who also value quietness and low power consumption, the Athlon 64 is a clear winner in that race. Such a user, who leaves his PC on 24/7 might average four hours a day on high-utilisation stuff (games, video encoding, etc.) with the system either idle or at low utilisation (web browsing, e-mail, word processing, playing Oggs, etc.) the rest of the time. For the current A64's with a 32W min P-state this gives us a TDP of 41.5W over the 24-hour period, and the revised 22W versions should be out any day now. While the system might be far from silent while gaming, the rest of the time it could be quieter than most of the old CPUs you've given a "6" score, courtesy of the superior heatsinks that are available for the socket 754/939.

Its lowly position and the "3-5" score range don't really convey the significance of these things.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Tue Aug 31, 2004 12:20 pm

This page somehow got replaced with Undervoltable Motherboards, and the backup for the file got lost somewhere in the server move a few weeks ago. It was rebuilt from a really old copy -- which probably is a bit out of date, so feel free to post corrections and new info for the page here.

pod03
Posts: 77
Joined: Mon Mar 24, 2003 12:36 pm
Location: UK

Underclocked Athlon 10?

Post by pod03 » Fri Oct 15, 2004 7:11 am

Mike

Given how easy it is to underclock and undervolt multiplier unlocked AthlonXPs (Durons? Semperons?, etc.) do you think that they should get a wider range of scores (ie 5-10). If so do you think it worth linking the relevant articles http://www.silentpcreview.com/article102-page1.html, http://www.silentpcreview.com/modules.p ... =69&page=1
and http://www.silentpcreview.com/article164-page1.html. This is particularly so as they still appear to offer the widest range of performance and heat output of any processor baring perhaps the Intel Pentium-M Centrino which is difficult to do in a desktop. See http://forums.silentpcreview.com/viewto ... highlight=
and http://forums.silentpcreview.com/viewto ... highlight= as an indication that the AthlonXP can be set to use less energy than the A64.

Perhaps your other CPU entries need to be adjusted to take their undervolting and underclocking potential into account. Alternatively a separate table that deals with this might be clearer.

Regards,

Mark

scdr
Posts: 336
Joined: Thu Oct 21, 2004 4:49 pm
Location: Upper left hand corner, USA

Might add an interesting upcoming, but unrated section

Post by scdr » Thu Oct 21, 2004 7:04 pm

It might be nice to add a promissing newcomers section to the "CPUs Ranked by Noise/Heat article." (Rather like in the heatsinks page, with the adenda for new and interesting items that haven't gotten reviewd.)
I am in the process of trying to sort out what to get to build a new system, which I would like to make as quiet as reasonable. The recomended lists make a good starting point for research, I found that several other CPUs seemed worth investigating that are not mentioned in the list.
e.g.

Newer Pentium M (Dothan) to 2GHz. (Similar to the Banias Pentium M, but larger cache, some higher clocks. Benchmarks seem to put speed equal to about P4 with 1.5x clock in desktop apps (e.g. Pentium M 2000 ~ A64 3000+ - e.g. Anand Tech))

Sempron 3100+ (Paris) - 62W TDP

Athlon 64 Mobile

Though keeping the CPU list up to date would be a job and a half, perhaps just appending a mention of new processors might be helpful. (Such entries could gradually be fleshed out as details/time became available, until they got promoted to full ranked list status.)

Thank you

jamesm
Posts: 185
Joined: Sat May 21, 2005 7:59 am
Location: California, USA

Post by jamesm » Mon May 30, 2005 9:11 am

the san diego core has a max power output of 67w

check this article:

http://en.wikipedia.org/wiki/Athlon_64#Models

clawhammer: 89W
newcastle: 89W
winchester: 67W
venice: 67w
san diego: 67w

Post Reply