NF4 passive chipset heatsink warning

All about them.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Pooh-Bah
Posts: 41
Joined: Sat Apr 02, 2005 9:26 am
Location: Los Angeles

NF4 passive chipset heatsink warning

Post by Pooh-Bah » Tue May 31, 2005 1:29 pm

I've read on several forums including this one about people replacing their active cooling on the Nforce4 with a passive chipset heatsink like the blue or gold Zalman's. We have a whole thread on it here so it's a big issue for many people. I suspect that many people are now running their motherboards with passive chipset cooling believing that their chipset temps are much lower than they actually are.

Many people report temp readings before and after the change. I suspect that the majority of these temp reports are based on windows-based readings from Motherboard Monitor or SpeedFan, and they are NOT based on an accurate reading of the chipset temps.

I have a raytek heat gun and I can accurately read the temp on my NF4 chipset. This temp is NOT reported by speedfan anywhere. My chipset with just the gold Zalman reaches 62C under a full load within minutes at a 223mhz bus speed. At this temperature I experience stuttering performance but no crashes. I have not let it run for long like this so I don't know what the stabilized temp is under a full load.

It is my opinion that the Nf4 cannot be cooled by a passive heatsink alone in many systems. If you don't overclock, or you rarely run under a full load, or you have incredibly good air circulation in the chipset area, maybe this doesn't apply to you.

Can anybody confirm that Speedfan or Motherboard Monitor actually reads the chipset temps?

Post your REAL chipset temps here if you know them.

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Post by m0002a » Tue May 31, 2005 1:45 pm

Most of the OEM heatsinks that are installed on Nforce4 chipsets are incredibly small, especially if you subtract the area taken up by the fan (just the heatsink itself). I am talking about a 10-1 difference in surface area compared to a Zalman. So it is not hard to imagine that a passive Zalman heatsink can cool the chip about as well as a tiny heatsink with a tiny fan. If you have before and after numbers, I would like to see them.

But for the record, I do agree that one should have some air blowing on the chipset heatsink, either from a nearby video cooler, or another case fan that is blowing some air in the direction of the chipset cooler (perhaps zip tied to the hard drive cage). But I do not agree that the fan must sit on top of the heatsink or that the airflow must be significant. I also believe that blue Zalman is bigger than the gold one you used.

I am not sure if you were overclocking, but if you were, then I would not recommend doing that. Any tests conducted should at least include a test running at stock speeds.
Last edited by m0002a on Tue May 31, 2005 2:06 pm, edited 1 time in total.

lenny
Patron of SPCR
Posts: 1642
Joined: Wed May 28, 2003 10:50 am
Location: Somewhere out there

Post by lenny » Tue May 31, 2005 1:56 pm

Pooh, can you post before and after temperatures from both SpeedFan and your Raytek? Did you test overclocking with stock NB cooler?

Pooh-Bah
Posts: 41
Joined: Sat Apr 02, 2005 9:26 am
Location: Los Angeles

Post by Pooh-Bah » Tue May 31, 2005 2:06 pm

Lenny - I don't have any temp readings with the stock cooler. I didn't test it until I had already removed that ridiculously loud stock cooler. Speedfan doesn't report the chipset temp on my motherboard.

Thunder
Posts: 168
Joined: Sat Jan 01, 2005 5:06 pm
Location: Bay Area Ca.

Post by Thunder » Tue May 31, 2005 2:10 pm

My ECS KN1 stays under 49c with the stock removed and just over 50 while gaming but speed fan dosent even find my chipset (or its way off) I just use my fan controler.

lenny
Patron of SPCR
Posts: 1642
Joined: Wed May 28, 2003 10:50 am
Location: Somewhere out there

Post by lenny » Tue May 31, 2005 2:18 pm

Pooh-Bah wrote:Lenny - I don't have any temp readings with the stock cooler. I didn't test it until I had already removed that ridiculously loud stock cooler. Speedfan doesn't report the chipset temp on my motherboard.
I see we're alike - mod first and ask questions later :-)

I usually mod while waiting for the rest of the compoments to arrive, which might be a while if I haven't ordered them yet. And once a system is up and running, I'm too dependent on using it to actually go back and mod it again.

Open the case and direct a large fan at the NB - that way you can determine if stuttering is due to high NB temps.

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Post by m0002a » Tue May 31, 2005 2:20 pm

Pooh-Bah wrote:Lenny - I don't have any temp readings with the stock cooler. I didn't test it until I had already removed that ridiculously loud stock cooler. Speedfan doesn't report the chipset temp on my motherboard.
I don't know of anyone who has NF4 chipset temps from the mb or software readings. Some people have placed temp sensors on the chipset.

Please answer the questions about the over-clocking.

Pooh-Bah
Posts: 41
Joined: Sat Apr 02, 2005 9:26 am
Location: Los Angeles

Post by Pooh-Bah » Tue May 31, 2005 2:28 pm

Thunder - how did you get those temps? If you used a temp probe, have you tested or calibrated it?

m0002a - my temp readings are from the top of the heatsink using a laser temp gun at a full load (Prime95+playing a video) with a 223mhz bus speed. I haven't check the passive temps at the stock bus speed since I don't plan to ever run this system at a 200mhz bus speed. Like I said in my post, I think that totally passive cooling with a typical chipset heatsink will work only under optimal conditions (including no overclocking).

Lenny - my system runs flawlessly with only a tiny bit of airflow over the chipset heatsink. The stuttering was with no airflow and I read about 62C during the stuttering. It only take a little airflow to lower the temps significantly.
Last edited by Pooh-Bah on Tue May 31, 2005 2:33 pm, edited 1 time in total.

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Post by m0002a » Tue May 31, 2005 2:32 pm

Pooh-Bah wrote:m0002a - my temps are at a full load (Prime95) with a 223mhz bus speed.
I am not an over-clocker so I don't know what that means. Are you, or are you not, over-clocking the system bus (and by how much)?

Pooh-Bah
Posts: 41
Joined: Sat Apr 02, 2005 9:26 am
Location: Los Angeles

Post by Pooh-Bah » Tue May 31, 2005 2:34 pm

m0002a - yes, that is overclocked by about 10%.

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Post by m0002a » Tue May 31, 2005 2:35 pm

Pooh-Bah wrote:m0002a - yes, that is overclocked by about 10%.
Well, if you overclock the system bus, then I would expect those kind of temps and potential instability. I would be interested to see the temps you get without over-clocking.

Thunder
Posts: 168
Joined: Sat Jan 01, 2005 5:06 pm
Location: Bay Area Ca.

Post by Thunder » Tue May 31, 2005 2:45 pm

My temps are from a flat surface temp probe mounted right next to the chip, and I also tested with a second probe on the HS. I have had no problems at all so it must be running relitivly cool because its still stable. I do worry about how hot its getting though.

Image
thats my setup

Pooh-Bah
Posts: 41
Joined: Sat Apr 02, 2005 9:26 am
Location: Los Angeles

Post by Pooh-Bah » Tue May 31, 2005 2:57 pm

m0002a - I'd like to focus on the temps issue and not get sidetracked with the overclocking issue. True, overclocking does raise the temp of the chipset and i should have tested the temp with both the stock heatsink and with the new heatsink at stock bus speeds.

My point for this thread is to get people thinking about their chipset temps if they choose to replace the stock fan. If you have an Nforce chipset with a passive fan, try to get an accurate reading running the Prime95 torture test and post your temp here. I'm not sure how most people can accurately read their chipset temps since few people have a temp gun like mine. A properly calibrated probe can work well too but you have to worry about it having good contact with the heatsink and/or chipset.

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Post by m0002a » Tue May 31, 2005 3:58 pm

Pooh-Bah wrote:m0002a - I'd like to focus on the temps issue and not get sidetracked with the overclocking issue. True, overclocking does raise the temp of the chipset and i should have tested the temp with both the stock heatsink and with the new heatsink at stock bus speeds.

My point for this thread is to get people thinking about their chipset temps if they choose to replace the stock fan. If you have an Nforce chipset with a passive fan, try to get an accurate reading running the Prime95 torture test and post your temp here. I'm not sure how most people can accurately read their chipset temps since few people have a temp gun like mine. A properly calibrated probe can work well too but you have to worry about it having good contact with the heatsink and/or chipset.
I think that if someone is going to make the claim that a passive Zalman will not work on a NF4, then they have the responsibility of full disclosure regarding overclocking of the system bus. To claim that such disclosure is off-topic is ridiculous.

I don't think it is too late for you to take readings at stock speeds, since it does not require any hardware changes.

I don't agree with your assessment of the accuracy of temp probe attached to the heatsink vs. your temp gun, especially when you are trying to get a reading from a multi-dimensional surface.

Personally, I have touched the blue Zalman on my NF4 with my hand, and it is much cooler than the Northbridge passive OEM heatsink on a Soltek 890Pro-939 board I previously had (before I RMA'ed it).

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

Post by Rusty075 » Tue May 31, 2005 4:23 pm

Considering the large number of people who are running passive NB coolers on their NF4's, and the very cases of reported instability, I say that Pooh is grossly generalizing based soley on his limited experience.

It's just as likely that your board is overly sensitive. NB's are rated to much higher temps than 62°. (Intel states their NB max temp as being 99°!) Perhaps your particular motherboard is just a poor overclocker.

Overclocking and overvolting makes components exponentially more sensitive to tmeperature. A stock speed chip will run just fine at temperatures that often choke an OC'd one. No magic there, that's just the way it is.

To rephrase Pooh's statement more accurately; "If you're going to overclock the northbridge, be aware that you will need more cooling than someone who is not overclocking"

ddrueding1
Posts: 419
Joined: Sun Sep 19, 2004 1:05 pm
Location: Palo Alto, CA

Post by ddrueding1 » Tue May 31, 2005 4:29 pm

m0002a wrote:I think that if someone is going to make the claim that a passive Zalman will not work on a NF4, then they have the responsibility of full disclosure regarding overclocking of the system bus. To claim that such disclosure is off-topic is ridiculous.
To his defense, it was stated in the original post not only that he was overclocking, but by how much. I would assume that most (obviously not all) enthusiasts would be aware that the FSB on a NF4 is 200Mhz stock; and that a 223Mhz FSB implys ~10% OC.
m0002a wrote:I don't think it is too late for you to take readings at stock speeds, since it does not require any hardware changes.
Agreed, though a nicer tone might be appropriate.
m0002a wrote:I don't agree with your assessment of the accuracy of temp probe attached to the heatsink vs. your temp gun, especially when you are trying to get a reading from a multi-dimensional surface.

Personally, I have touched the blue Zalman on my NF4 with my hand, and it is much cooler than the Northbridge passive OEM heatsink on a Soltek 890Pro-939 board I previously had (before I RMA'ed it).
I can't comment on this as I have no knowledge or experience to back it, but I must comment that my NF4 motherboard came stock with a very small passive heatsink and it ran fine at 100% sitting open on the workbench for weeks with no direct airflow.

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Post by m0002a » Tue May 31, 2005 5:01 pm

ddrueding1 wrote:To his defense, it was stated in the original post not only that he was overclocking, but by how much. I would assume that most (obviously not all) enthusiasts would be aware that the FSB on a NF4 is 200Mhz stock; and that a 223Mhz FSB implys ~10% OC.
This is not an "enthusiast" forum. It is a silent computing forum. That is why questions not relating to quiet computing and/or heat are considered off-topic.

I would bet that most of the people who read this forum (not just those who post here) don't know what the NF4 stock bus speed is. A substantial number of people who read this forum and who are concerned about replacing the stock HSF on an NF4 board, are not interested in over-clocking, or at least would be willing to not over-clock if they could run the NF4 chip with passive cooling.

Natronomonas
Posts: 73
Joined: Wed Jan 12, 2005 1:07 am
Location: Melbourne, Australia
Contact:

Post by Natronomonas » Tue May 31, 2005 5:56 pm

"that is why questions not relating to quiet computing and/or heat are considered off-topic."

Overclocking/underclocking are both very relevant in the context of quiet computing - underclocking as it makes in easier to quiet a PC, and overclocking as it makes it harder.
If you don't know bus speeds etc, should you be performing mods to the hardware?

On the topic of NB cooling, I think it's well established that even a little bit of airflow helps a lot - the "tiny" stock hs + fan probably cools a lot better than the entirely passive zalman.

The zalman should be substantially better with the same airflow as that provided by the stock NB fan though, due to the higher surface area available. Personally, I'd just use an 80mm @ 5v somewhere in the vicinity of the NB if I was worried, or use a CPU cooler/VGA cooler that moved some air over that area (your case airflow might be enough if configured correctly).

Pooh-Bah
Posts: 41
Joined: Sat Apr 02, 2005 9:26 am
Location: Los Angeles

Post by Pooh-Bah » Tue May 31, 2005 6:44 pm

Rusty075 - I'm not trying to claim that passive cooling is not effective. I'm only trying to bring up the point that many people are disillusioned if they think that their chipset was running at 35C with the stock heatsink, and 40C with the passive heatsink. The chipset is running a lot hotter than many people believe. I have seen many components run at 60C-75C without any problems. My personal setup exhibited some impact from the temps reached without a fan but it did not crash (as I originally posted).

M0002a,

You seem very touchy about this subject and you should really relax a bit. I’m not claiming anything and I’m sorry if what I’ve read and experienced myself does not agree with your personal experiences. I’m telling you what I’ve learned - that people are reading the wrong temperatures and that is what leads them to believe that their temps are lower than they actually are. Did I say that a passive Zalman will absolutely not work on an NF4 board? No. I said that it won’t work in many systems. Read the posts on various forums and you may come to the same conclusion.
m0002a wrote:This is not an "enthusiast" forum. It is a silent computing forum. That is why questions not relating to quiet computing and/or heat are considered off-topic.
Silent and overclocked are not mutually exclusive. Forums are for discussions. We are discussing something here. Feel free to disagree but please be nice about it.
m0002a wrote:I think that if someone is going to make the claim that a passive Zalman will not work on a NF4, then they have the responsibility of full disclosure regarding overclocking of the system bus. To claim that such disclosure is off-topic is ridiculous.
I’m not sure what else you want to know. If you look at my original post, I said the following: “My chipset with just the gold Zalman reaches 62C under a full load within minutes at a 223mhz bus speed.” I think that pretty much defines full disclosure. If you wanted more info, you could have sent a PM.

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Post by m0002a » Tue May 31, 2005 8:11 pm

Pooh-Bah wrote:M0002a,
You seem very touchy about this subject and you should really relax a bit. I’m not claiming anything and I’m sorry if what I’ve read and experienced myself does not agree with your personal experiences. I’m telling you what I’ve learned - that people are reading the wrong temperatures and that is what leads them to believe that their temps are lower than they actually are. Did I say that a passive Zalman will absolutely not work on an NF4 board? No. I said that it won’t work in many systems. Read the posts on various forums and you may come to the same conclusion.
As I said, I don't think most people who read this forum knew that the stock NF4 bus speed was 200 MHz and that you were over-clocking. Even some who may have known the stock bus speed is 200 MHz, may have missed the fact that you were overclocking, which I was trying to clarify. If I knew, I would not have asked.

I think if you had said that systems running stock NF4 bus speeds should have litle or no problem with a passive heatsink, and that systems running 10% over-clock on the system bus may not work if passively cooled, that would have been much more clear than saying “it won’t work in many systems.”
Pooh-Bah wrote:Silent and over-clocked are not mutually exclusive. Forums are for discussions. We are discussing something here. Feel free to disagree but please be nice about it.
I think I was very nice the first time I asked, maybe not so nice (and a bit touchy) the second time because I was told by someone else (not you) that I have no business asking, since almost all “enthusiasts” should know that 223 MHz is over-clocking.

As I said earlier in this thread, this is a quiet computing forum, and not a PC enthusiast forum. Everyone has an inalienable right to quiet computing even if they don’t know jack about the inside of a computer. And PC enthusiasts who post on this forum ought to make a good faith effort to explain things clearly, because in this particular situation there are many people who are considering replacing an NF4 chipset HSF, but who have no interest in overclocking the system bus, and I don't think they should be discouraged from doing so.

As to what else I would like to know, I would like to know the temps of the Zalman gold HS with the NF4 running at stock speed. I think others would like to know also.

jackylman
Posts: 784
Joined: Sun May 22, 2005 8:13 am
Location: Pennsylvania, USA

Post by jackylman » Wed Jun 01, 2005 2:46 pm

Some questions for everyone here...

Which nforce4 chip are you using?
Does the SLI chip draw more power than Ultra or vanilla versions?
Will disabling unused peripherals/ports in the BIOS help decrease power draw?

Pooh-bah, your point is well-taken. The nForce4 has upped the ante by integrating North and South Bridges and making everything extremely fast. Mobo manufacturers should put a temp-diode in that chipset as well. Then we could have a MB chipset temp and a MB ambient temp.

Thanks, guys. Keep it civil :)

ChiBOY83
Posts: 116
Joined: Wed Jan 05, 2005 1:46 pm

Post by ChiBOY83 » Wed Jun 01, 2005 3:05 pm

Well, despite a bit of a broo-ha-ha developing for a few moments here, i think this is a topic that should be continued further, w/ more data etc.


Although everyone here is trying to get a quite PC, more people are trying to OC a PC, so it is only obvious that the two opposite camps are going to overlap.


..Personally, i will be getting a A8N-E mobo, and want to zalman the NB, and expect to mess around w/ some moderate OC'ing, so i will expect to have to insert a case fan to blow onto the HS.... SO id be very greatful if any OC'ers would give comments on OC'ing a NF4 setup w/ a Passive HS

Pooh-Bah
Posts: 41
Joined: Sat Apr 02, 2005 9:26 am
Location: Los Angeles

Post by Pooh-Bah » Wed Jun 01, 2005 3:47 pm

Thanks jackylman and ChiBOY83. I tend to agree with both of you.

I really like my NF4 and with the amount of noise generated by that hideous fan that most manufacturers install, there really isnt much choice but to swap it for something passive or semi-passive. Mine works perfect in a cramped and somewhat hot case with only a tiny bit of direct airflow.

I'm still hoping for more temp reports from this topic. I'd test mine again under various scenarios but it's an HTPC machine that isn't easy to test due to the way its installed and the lack of a monitor to see the BIOS screens.

andyb
Patron of SPCR
Posts: 3307
Joined: Wed Dec 15, 2004 12:00 pm
Location: Essex, England

Post by andyb » Wed Jun 01, 2005 4:21 pm

Being an Overclockcked NF3 owner with a low profile passive chipset heatsink I feel quite lucky that the NF3 is much cooler. However I would agree that a little more information would be appreciated by all parties concerned, myself included, who would consider an NF4 board in the future.

Please tell us the particular version of the NF4 (if it makes a difference), the board model and manufacturer, the idle temp while overclocked, the idle temp while at default clock, and the load temp at default clock, and your airflow setup. I certainly wouldnt ask you for temps with a foul 6,000 RPM fan on as that would involve lots of work.

This additional information would certainly be appreciated, and we would all gain from the use of your heat-gun, and this post.

BTW: Several people seemed to have missed the ORIGINAL point, so I will repeat, most if not all NF3 & NF4 boards either dont report temps at all or reports them incorrectly, my mobo does not report a temp.

Andy

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Post by m0002a » Wed Jun 01, 2005 5:51 pm

andyb wrote:BTW: Several people seemed to have missed the ORIGINAL point, so I will repeat, most if not all NF3 & NF4 boards either dont report temps at all or reports them incorrectly, my mobo does not report a temp.
Maybe I missed it, but I don't recall anyone saying that the NF4 board or chipset reported the temps. Some people have put temp sensors directly on the chip, which is the only place I have seen NF4 temps reported.

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Post by m0002a » Wed Jun 01, 2005 5:57 pm

ChiBOY83 wrote:Personally, i will be getting a A8N-E mobo, and want to zalman the NB, and expect to mess around w/ some moderate OC'ing, so i will expect to have to insert a case fan to blow onto the HS.... SO id be very greatful if any OC'ers would give comments on OC'ing a NF4 setup w/ a Passive HS
Excuse my ignorance with regard to OC'ing, but I assume there is a difference between OC'ing the CPU and OC'ing the NF4 bus speed (and also the memory), and that these can be OC’ed independently.

The benefits of OC'ing the CPU are obvious, but what are the practical benefits of a desktop PC user OC'ing the NF4 bus speed? Is this just an exercise in benchmark testing, or are there real applications where the benefits of OC'ing the NF4 bus speed are even noticeable?

marteko
Posts: 3
Joined: Wed Jun 01, 2005 10:21 pm
Location: Belgium

nforce4 cooling

Post by marteko » Wed Jun 01, 2005 10:32 pm

Might be able to provide some info. My rig: asus a8n-sli deluxe in coolermaster stacker case with crossflow fan. I replaced the stock heatsink with a zalman nb47 blue passive cooler using arctic silver 5. I monotor the temperature with an analogue temp sensor (t-balancer)located on the bottom of the NB47, out of the air flow. Meaning that the real chip temperature must be even higher than my readings.
First remark: if your heatsink runs cool to the touch, there may be 2 explanations. First: it works with passive cooling, congratulations. Second: your heatsink isn't making good contact with the chip. Mine has good contact, since it runs very hot. I let it run without the cross flow fan and I chickened out at +60° Celsius, this is: no overclocking and idle. I could not see any stuttering or failure, but I wouldn't take any chances. But with the crossflow fan at its lowest settings, I get about 39°C. For me, this means that completely passive cooling is out of the question.

Eler
Posts: 14
Joined: Wed Jun 01, 2005 10:59 pm

Post by Eler » Wed Jun 01, 2005 11:09 pm

Well, some more data about the temps. I've got an a8n-sli and I replaced the noisy stock fan with the Blue Zalman one. Also My case is a Stacker without the crossflow fan. I attached a coolermaster temp probe to the heatsink when I changed it, so I don't have any temp measurements with the stock fan. Any way actual temeprature it's about 50ºC. This is with the frontal fan of the case blowing to the chipset at full speed, If i disconnect the fan, temp. rises to 62º. I hope this helps.

- Eler

P.S: I almost forgot, Asus replied me a mail about temperatures of the chipset, telling that 50-55º should be ok, but not above that.

ChrisH
Friend of SPCR
Posts: 65
Joined: Wed Oct 30, 2002 10:44 am
Location: Charlotte, NC USA - Go Panthers!

Post by ChrisH » Thu Jun 02, 2005 8:52 am

Do any NF4 motherboards allow you to undervolt the northbridge chip? That would significantly reduce the temperature, although it might hurt OC attempts.

Auz
Posts: 3
Joined: Thu Jun 02, 2005 2:55 pm

Post by Auz » Thu Jun 02, 2005 3:29 pm

Thought I would add my numbers:

I have a MSI K8N neo4 platinum (no sli), which I replaced the nb hsf to the blue zalman (nb47J) with some artic silver 5. I took temp reading with the stock HSF, but the fan being where it is, pretty much rules out any meaningful numbers as you have to measure from the edge - it was 32 deg C.

Image

With passive cooling only (except whatever draft is coming from the video cooler- which isnt much), at 100% load, it gets up to 45.1 C max (as measured with my thermocouple from the center of the heatsink)

Image

With a small 40MM fan placed on the side and running on the slowest speed that a zalman fanmate 2 controler will allow, I get about 34C.

A small fan does make a large difference from my tests. With the fan controler I cannot even hear it. However, I would think both these temperatures are acceptable. It is not overly hot to the touch with either setup - and I cannot imaging this being worse cooling than that tiny stock hsf.

-auz

Post Reply