stay away from bartons!

Cooling Processors quietly

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
jpsa
Posts: 145
Joined: Tue Feb 11, 2003 6:15 am

stay away from bartons!

Post by jpsa » Sun May 09, 2004 7:22 pm

well... its not that dramatic, it's just that I was used to my xp2000+ heat and now I got a xp2500 and this thing is HOT :evil:
even my xp2000 with oc to 1920at 1.8volt doesn't get nearly as hot as this barton at default 1833 and 1.65volt.
If you must get a barton get a mobile, I learned the hard way :(
well back to the xp2000 for me :roll:

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Sun May 09, 2004 7:42 pm

I wonder if there's not something else in the mix.

Wattage for your XP2000= 60.3 watts
Wattage for a Barton 2500= 68.3 watts

8 watts shouldn't make a huge difference. 3-4°'s max, certainly not enough to make you switch back to a slower proc.

Wattage for your XP2000@1920Mhz & 1.8v= 82.7 watts

If your Barton is running hotter than your OC'd 2000 did...you've got something funky going on.....try remounting the HS.

chylld
Posts: 1413
Joined: Fri Oct 10, 2003 4:45 pm
Location: Sydney, Australia
Contact:

Post by chylld » Sun May 09, 2004 7:47 pm

i think it's a common thing with bartons. all the bartons i've had (all 2500's) ran extremely hot. the thing is, the temperature doesn't rise much once you put load on them...

1398342003
Posts: 324
Joined: Wed May 07, 2003 10:35 pm
Location: Surrey, B,C

Post by 1398342003 » Sun May 09, 2004 8:28 pm

Another thing to consider is that the temp sensor on either proc could be flaky.

Pjotor
*Lifetime Patron*
Posts: 485
Joined: Mon Jun 30, 2003 9:06 pm
Location: Uppsala, Sweden

Post by Pjotor » Sun May 09, 2004 9:52 pm

If your Barton is running hotter than your OC'd 2000 did...you've got something funky going on.....try remounting the HS.
That was my first thought, too. You didn't state any temperatures in your post -- some folks are more temperature sensitive than others. What are your temps at idle and load, and which settings do you use for Vcore, FSB, and multiplier?

1911user
Posts: 109
Joined: Sat Mar 13, 2004 12:08 pm
Location: Oklahoma

Post by 1911user » Sun May 09, 2004 10:03 pm

For a comparison: I stopped folding on rig#2 and the temps dropped from 55C to 47C after a few minutes. The CPU is an XP barton overclocked to 2.25GHz (13.5x167) with a cpu voltage of 1.68v. It is cooled using a vantec aeroflow with the HS fan turned down to 4300 rpm. The aeroflow draws cool air from a side-panel mounted duct direct to the HS fan. The heat dissipation when 100% loaded should be around 87 watts.

EDIT: After cleaning out the dust bunnies from the aeroflow, the temps swing from 44-50C from idle in win2K to 100% load (folding). Moral of the story, keep the cpu heatsink clean.
Last edited by 1911user on Wed May 12, 2004 4:53 pm, edited 1 time in total.

silvervarg
Posts: 1283
Joined: Wed Sep 03, 2003 1:35 am
Location: Sweden, Linkoping

Post by silvervarg » Mon May 10, 2004 1:57 am

Just as Chylld already stated and also shown in a number of good review the temps of the barton at idle is quite high, but it does not raise much under load.
What really matters most is worst case IMHO, since you need to design the cooling to be able to handle worst case. That is why Prime95 is a very handy tool.
Also I always consider how much headroom I have from the manufacturers specs. Since the Barton can take a lot higher temps than the P4 can I see no problem with letting the Barton up to higher temps than I would like to bring a P4.

This is why I choosed a Barton 2500+ for my passively cooled computer.
If I would have choosed again today I would still have picked a Barton. Perhaps a mobile.

jpsa
Posts: 145
Joined: Tue Feb 11, 2003 6:15 am

Post by jpsa » Mon May 10, 2004 6:12 am

well, as I work with speedfan my temps are always the same, I just choose 50C and the prog handles the rest. With the xp2000 was at ~1000rpm at full load, with the barton the fan is at ~2500rpm at full load... that makes a great diference in noise as you can guess

jpsa
Posts: 145
Joined: Tue Feb 11, 2003 6:15 am

Post by jpsa » Mon May 10, 2004 6:14 am

I'm guessing that this cpu reports higher temps than the older one...
the sensor inside the cpu is a diode or a thermistor?

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Mon May 10, 2004 6:41 am

silvervarg wrote:Just as Chylld already stated and also shown in a number of good review the temps of the barton at idle is quite high, but it does not raise much under load.
What really matters most is worst case IMHO, since you need to design the cooling to be able to handle worst case. That is why Prime95 is a very handy tool.
Also I always consider how much headroom I have from the manufacturers specs. Since the Barton can take a lot higher temps than the P4 can I see no problem with letting the Barton up to higher temps than I would like to bring a P4.

This is why I choosed a Barton 2500+ for my passively cooled computer.
If I would have choosed again today I would still have picked a Barton. Perhaps a mobile.
How high a temperature can the Barton 2500 take?

alglove
Posts: 363
Joined: Fri Feb 06, 2004 11:21 am
Location: Houston, TX, USA

Post by alglove » Mon May 10, 2004 12:26 pm

JVM wrote:How high a temperature can the Barton 2500 take?
The desktop Barton can take a die temperature of up to 85 degC. If you want to take a look at the spec yourself, go to

http://www.amd.com/us-en/Processors/Tec ... 48,00.html

Choose the "Athlon XP Processor Model 10 Data Sheet"
http://www.amd.com/us-en/assets/content ... /26237.PDF

and take a look at Chapter 12, "Ordering Information" on page 79. This tells you have to break down the CPU numering code.

Athlon XP-M mobile processors are not categorized as desktop processors, so there is a different specification that pertains to them.

POLIST8
*Lifetime Patron*
Posts: 703
Joined: Mon Jul 14, 2003 12:59 pm
Location: Madison, WI USA
Contact:

Post by POLIST8 » Mon May 10, 2004 12:42 pm

Is your system unstable? Does Windows Crash? Does your computer or other internal components start on fire?

If no, Keep the Barton. It is by far superior.

AZBrandon
Friend of SPCR
Posts: 867
Joined: Sun Mar 21, 2004 5:47 pm
Location: Phoenix, AZ

Post by AZBrandon » Mon May 10, 2004 12:52 pm

alglove wrote:The desktop Barton can take a die temperature of up to 85 degC. If you want to take a look at the spec yourself, go to
Isn't that specification at stock voltage and clock speed? From everything I've read, as you increase voltage and clockspeed, the temperature at which you run in to instability decreases. So if you can do 2.0Ghz at 1.65v and 85C, once you up it to 2.3Ghz and 1.8v it may only be stable up to 65-70C or so.

Anyway, for those of you seeing very little temperature drop at idle, you should really look in to downloading CPU Idle. I have a kill-a-watt hooked up to my Barton powered unit so I can see the power draw instantly. Since I have a mobile Barton, it's unlocked and I've run everything from 1.55v at 3x100 (300Mhz) to 1.76v at 13x178 (2314Mhz). I've seen power draw anywhere from 39 watts at the plug for the whole system at idle up to 133 watts at max voltage and overclock running CPUBurn.

Typically, with the higher overclocks, it would only back off about 20-30 watts from peak usage to idle. Like if it drew 130 watts at CPUBurn, it would still draw 95 watts at idle. With CPU Idle installed it dropped to 45 watts at 1.76v @ 2300mhz. That's darn close to the same total draw running at 1.55v and 300Mhz. Obviously it won't do anything for you if your system is under load all the time, but I've been able to get away with running my mobile Barton at 12x170 @ 1.55v (2040Mhz) with the cheap HSF at 2200rpm and no case fan for mundane tasks like web surfing, email, etc. It usually takes very intense photoshopping to have enough CPU draw to bring the temps up over 47C or so and generally it spends it's idle time between 36 and 39C. Not bad for a 2Ghz Barton and no case fan running.

silvervarg
Posts: 1283
Joined: Wed Sep 03, 2003 1:35 am
Location: Sweden, Linkoping

Post by silvervarg » Tue May 11, 2004 10:52 am

jpsa:
well, as I work with speedfan my temps are always the same, I just choose 50C and the prog handles the rest. With the xp2000 was at ~1000rpm at full load, with the barton the fan is at ~2500rpm at full load... that makes a great diference in noise as you can guess
Depending on what motherboard you have you might want to try and reduce vcore, this helps to reduce CPU temp a lot.
I suggest that you reset speedfan to accept 60C. Assuming you have case temps of ~30C a change from 50C to 60C increases the allowed differential from 20 to 30C, so you should probably see fan speed drop by 1/3 of the RPM by this change alone.
Depending on what OS you are running you might want to run CPU idle or similar program to reduce idle temps even more.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Tue May 11, 2004 10:57 am

I don't think MBM states the Die temperature so should we add 10C to what MBM says?

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Tue May 11, 2004 11:25 am

JVM wrote:I don't think MBM states the Die temperature so should we add 10C to what MBM says?
MBM reports whatever your motherboard makes available to it, and whatever you have it configured to report.

For example, your A7V doesn't report diode temps at all, so your temps are from the socket. (actually your temps are Asus' wildly tweaked socket temps)

But if you had a A7N8X it could report either the socket, or the diode (or both, so you could compare them) depending on which sensor you set to read what.

wumpus
Patron of SPCR
Posts: 946
Joined: Sat Sep 06, 2003 9:57 pm
Location: Berkeley, CA, USA
Contact:

Post by wumpus » Tue May 11, 2004 12:34 pm

FWIW, I had a very similar experience. Going from 1800mhz non-Barton (256kb L2) to 1800mhz Barton (512kb L2) caused a SUBSTANTIAL increase in heat generation. Exact same clock speeds in both cases, just by happenstance of the upgrade.

Not too surprising, really, given the big jump in transistors (= heat) to populate that L2 cache.. I don't think there's anything else going on in a Barton vs. a T-Bred other than the increased L2?

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Tue May 11, 2004 2:40 pm

Rusty075 wrote:
JVM wrote:I don't think MBM states the Die temperature so should we add 10C to what MBM says?
MBM reports whatever your motherboard makes available to it, and whatever you have it configured to report.

For example, your A7V doesn't report diode temps at all, so your temps are from the socket. (actually your temps are Asus' wildly tweaked socket temps)

But if you had a A7N8X it could report either the socket, or the diode (or both, so you could compare them) depending on which sensor you set to read what.
My CPU temp is reported higher in Hardware Monitor than MBM. In any event, MBM is configured to read the CPU temperature as stated on the MBM website. Now, should I add 10C to what MBM says is the CPU temperature for Diode temperature?

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Tue May 11, 2004 3:04 pm

I would go with MBM over Asus' software. But in reality you have no way of knowing whether Asus just tweaked their software, or the hardware sensor circuitry as well.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Tue May 11, 2004 5:20 pm

Rusty075 wrote:I would go with MBM over Asus' software. But in reality you have no way of knowing whether Asus just tweaked their software, or the hardware sensor circuitry as well.
Thanks a lot... :)

So, what you are saying is I got a lousy MB?

aston
Posts: 139
Joined: Fri Jan 16, 2004 2:47 pm
Location: Victoria, BC

Post by aston » Tue May 11, 2004 5:24 pm

Have you tried letting it burn in for a while?

I'm not sure where I stand on the whole "burn-in" theory, but I do know that my Barton temps magically dropped by about 5C after a bit of usage.

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Tue May 11, 2004 5:57 pm

JVM wrote:So, what you are saying is I got a lousy MB?
No, you've just got a board that will encourage you to not obsess about your temps. :wink:


The fact that Asus programs their boards to intentionally misrepresent the CPU temps, and the fact that they overvolt the CPU to ensure stability (instead of using quality design and components) combined with their lack of undervolting abilites, tends to remove them from my shopping list. But other than that, they're good motherboards. :lol:

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Wed May 12, 2004 5:43 am

Rusty075 wrote:
JVM wrote:So, what you are saying is I got a lousy MB?
No, you've just got a board that will encourage you to not obsess about your temps. :wink:


The fact that Asus programs their boards to intentionally misrepresent the CPU temps, and the fact that they overvolt the CPU to ensure stability (instead of using quality design and components) combined with their lack of undervolting abilites, tends to remove them from my shopping list. But other than that, they're good motherboards. :lol:
I don't have any stability problems and running VCore at manual setting of 1.65V

What brands do you like?

silvervarg
Posts: 1283
Joined: Wed Sep 03, 2003 1:35 am
Location: Sweden, Linkoping

Post by silvervarg » Wed May 12, 2004 9:23 am

Actually Asus have a reputation to be the most stable brand of motherboards on the market. I don't think they could have used low quality design or low quality components to achieve this long term reputation.
The fact that they often slightly overvolt the CPU does give even more stability, even if it gives a few other drawbacks. The overvolted CPU vcore is a bit hard to tell if it really is overvolted. The reading you get is either the set value in BIOS or the value measure by some cheap circuit that might give you the wrong value. Attaching a multimeter at the right point will give an idea of how far off the readings are.

Do you have one of the Asus boards with jumpers that you can use to undervolt the CPU? They are undocumented in the manual, but if you find jumpers on your motherboard that are undocumented you can find the information on how to set them with a few searches.

I know the question of what brands was aimed to Rusty, but I like to give you my view anyway:
If you want a board that undervolts you can check out the list in the recommended section that has all undervoltable motherboards listed.
Abit is probably the most widely used.
Aopen has a few good boards.
DFI Lanparty seems to be highly regarded.
Epox has boards that are rather good and have very good price.
There are more brands, but these are the most commonly used ones.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Wed May 12, 2004 9:50 am

silverarg, I don't see a need to undervolt as everything is very good at 1.65V set manually. Now that warmer weather is here, I could use a lower CPU temperature but so far is not critical.

jpsa
Posts: 145
Joined: Tue Feb 11, 2003 6:15 am

Post by jpsa » Thu May 13, 2004 8:48 pm

well, I'm keeping the barton afterall :P one more challenge on my case... think I'll try the default cooler with a 80mm fan, seems better than my akasa 821

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

Post by sneaker » Fri May 14, 2004 2:46 pm

AZBrandon wrote:Anyway, for those of you seeing very little temperature drop at idle, you should really look in to downloading CPU Idle.
Or VCool, which may do the same job and doesn't cost 30 Euros. I used Vcool Alpha 6 (with Cool Bit enabled, Halt Detect disabled) and it reduced my Barton 2500+'s idle temp from 62C to 50C. Still high by some standards, but I'm using a dusty old Zalman CNPS6000-AlCu with the fan in silent mode. I've had it up to 80C using burnk7 but decided not to push it any further :twisted:

Interestingly, it also reduced the "system" temp from 32C to 28C.

Note that it might not have any effect if your motherboard already supports bus disconnect and has it enabled.

trodas
Posts: 509
Joined: Sun Dec 14, 2003 6:21 am
Location: Czech republic
Contact:

Post by trodas » Mon May 17, 2004 7:53 am

Is nto better to let folding use the spare CPU cycles? :lol:
Yea, it raise the temp a lot, but hey, is for a good thing and some little more heat in room can't kill anyone :lol:

Just keep folding... Just keep folding... :)

Post Reply