A question for Athlon 64 mobo owners: heat generation?
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
-
- Patron of SPCR
- Posts: 946
- Joined: Sat Sep 06, 2003 9:57 pm
- Location: Berkeley, CA, USA
- Contact:
A question for Athlon 64 mobo owners: heat generation?
I recently became the proud owner of an Athlon 64 socket 939 system, using this motherboard (Gigabyte GA-K8NSNXP-939)
Notice the fans on northbridge and the power conversion circuitry.
I assumed the power conversion fan was there for show, and removed it almost immediately after getting the board. Imagine my surprise when, under prime95 load, the radar temp gun measured the external body of the now-fanless metal shroud at over 200F! Holy.. crap? Evidently the fan wasn't on there just for show. This is literally too hot for me to touch, and although I am tolerant of rather high temps, this made me nervous.
The northbridge also got quite hot, which surprised me.. I thought the memory controller was on the chip die and therefore this chip would have less to do, eg, run cooler? The heatsink I mounted on there was an old 486 model, and it also got uncomfortable to touch under prime95 load, although not nearly as hot as the power circuitry heatsink.
Defeated, I ended up putting both fans back (well, I used a different model fan I salvaged from another board on the NB) and undervolted them both to 5v using a 3-wire conversion and a Zalman 4-pin to multiple 3-pin power splitter. Pretty quiet, but there's no way I could make these things run fanless. The diffference between "modest amount of directed airflow" and "totally passive" is rather large and hard to overcome.
Here's my question: is this normal for Athlon 64 boards? Should my power conversion circuitry, and the northbridge, be running this insanely hot under load?
Notice the fans on northbridge and the power conversion circuitry.
I assumed the power conversion fan was there for show, and removed it almost immediately after getting the board. Imagine my surprise when, under prime95 load, the radar temp gun measured the external body of the now-fanless metal shroud at over 200F! Holy.. crap? Evidently the fan wasn't on there just for show. This is literally too hot for me to touch, and although I am tolerant of rather high temps, this made me nervous.
The northbridge also got quite hot, which surprised me.. I thought the memory controller was on the chip die and therefore this chip would have less to do, eg, run cooler? The heatsink I mounted on there was an old 486 model, and it also got uncomfortable to touch under prime95 load, although not nearly as hot as the power circuitry heatsink.
Defeated, I ended up putting both fans back (well, I used a different model fan I salvaged from another board on the NB) and undervolted them both to 5v using a 3-wire conversion and a Zalman 4-pin to multiple 3-pin power splitter. Pretty quiet, but there's no way I could make these things run fanless. The diffference between "modest amount of directed airflow" and "totally passive" is rather large and hard to overcome.
Here's my question: is this normal for Athlon 64 boards? Should my power conversion circuitry, and the northbridge, be running this insanely hot under load?
-
- Moderator
- Posts: 7681
- Joined: Mon Dec 09, 2002 7:11 pm
- Location: Maynard, MA, Eaarth
- Contact:
NO!
Hello:
This is definitely NOT normal for Athlon 64 mobos! There are many with passive northbridges, and only the Gigabyte (and maybe Abit?) have fans on the power circuitry -- why did you get this motherboard?!
Here's the AOpen:
Here's the MSI:
Here's the Asus:
Not one fan in sight...
This is definitely NOT normal for Athlon 64 mobos! There are many with passive northbridges, and only the Gigabyte (and maybe Abit?) have fans on the power circuitry -- why did you get this motherboard?!
Here's the AOpen:
Here's the MSI:
Here's the Asus:
Not one fan in sight...
-
- Patron of SPCR
- Posts: 946
- Joined: Sat Sep 06, 2003 9:57 pm
- Location: Berkeley, CA, USA
- Contact:
Hmm, well, one variable, this is the nForce3 chipset. So maybe its NB has more going on or is on a hotter chip process? The NB could probably be run passive with the larger zalman passive heatsink (I want to say nb32j?), my problem is, I assumed it would run relatively cool and perma-mounted a 486-style heatsink on it. Probably from looking at the same images you are..
However, the voltage regulation circuitry is harder to explain. Even with the 40mm fan running at 5v, the temp gun says that parts of the voltage heatsink range from 120f ~150f.
Dunno, I'm open to suggestions, but that voltage heatsink shroud gets CRAZY hot on my board without any directed airflow..
However, the voltage regulation circuitry is harder to explain. Even with the 40mm fan running at 5v, the temp gun says that parts of the voltage heatsink range from 120f ~150f.
Dunno, I'm open to suggestions, but that voltage heatsink shroud gets CRAZY hot on my board without any directed airflow..
-
- Moderator
- Posts: 7681
- Joined: Mon Dec 09, 2002 7:11 pm
- Location: Maynard, MA, Eaarth
- Contact:
-
- Patron of SPCR
- Posts: 946
- Joined: Sat Sep 06, 2003 9:57 pm
- Location: Berkeley, CA, USA
- Contact:
Well, with a decent sized 486-class pin-fin heatsink on the nForce 3 250 northbridge chip running totally passive, the laser temp gun says...
~55c idle, ~65c load.
So evidently "uncomfortable to touch" can be quantified by my Mk. I finger as "approximately 55-60c".
I found a page that referenced the Intel i875 technical datasheet, and evidently intel's 875 nb is officially rated to 99c (!). I can vouch for the fact that the NB on my my i875 mobo, with the larger Zalman passive heatsink on it, gets darn toasty.
I think I'm just going to suck it up, go totally passive, and assume those operating temps are within the normal operating range for the nForce3 250 northbridge. I wish I had waited and installed a larger passive heatsink, but given the images above.. that is, the typically crappy default heatsinks installed on these kinds of boards.. I assumed the NB was doing less work on the A64 due to the lack of the memory controller on the chip. Looks like that's not the case
To top it all off, the stock heatsink/fan was mega cheesy. I'm talking maybe 20 edge fins, tops, on a flat metal platform, with a small 40mm fan embedded on the top. And it was attached by a thermal adhesive pad making marginal contact with the nb chip, no thermal paste in sight! I don't think a whole lot of "cooling" was going on with the stock solution anyway...
~55c idle, ~65c load.
So evidently "uncomfortable to touch" can be quantified by my Mk. I finger as "approximately 55-60c".
I found a page that referenced the Intel i875 technical datasheet, and evidently intel's 875 nb is officially rated to 99c (!). I can vouch for the fact that the NB on my my i875 mobo, with the larger Zalman passive heatsink on it, gets darn toasty.
I think I'm just going to suck it up, go totally passive, and assume those operating temps are within the normal operating range for the nForce3 250 northbridge. I wish I had waited and installed a larger passive heatsink, but given the images above.. that is, the typically crappy default heatsinks installed on these kinds of boards.. I assumed the NB was doing less work on the A64 due to the lack of the memory controller on the chip. Looks like that's not the case
To top it all off, the stock heatsink/fan was mega cheesy. I'm talking maybe 20 edge fins, tops, on a flat metal platform, with a small 40mm fan embedded on the top. And it was attached by a thermal adhesive pad making marginal contact with the nb chip, no thermal paste in sight! I don't think a whole lot of "cooling" was going on with the stock solution anyway...