AMD Processors
Decrease font size
Increase font size
Topic Title: Retail HSFs
Topic Summary:
Created On: 10/07/2003 03:27 AM
Status: Read Only
Linear : Threading : Single : Branch
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
 10/07/2003 03:27 AM
User is offline View Users Profile Print this message

Author Icon

Posts: 23
Joined: 10/07/2003

I just thought I'd comment on the retail HSFs included with AMD's CPUs.

Up until you started shipping that Ajigo AL/CU hybrid cooler with the Athlon XPs (I think the 2200s may come with them... I know the 2500s do) all the retail heatsinks you've used have been totally worthless. I know that they should provide "adequate" cooling, but that's only in ideal situations. Consider, for a moment, the average computer user's PC.

They turn it on, and expect it to work. They never open it, they never clean it, they never bother to check to see if the CPU fan is still functioning properly, or if it's on it's way out. They just turn the thing on, and expect it to work. The last AMD retail heatsink I used (before I bought the 2500s in my dually) was the Dynatron skived-fin aluminum thing that came with the Palomino XPs. My dad bought a 2100+ system with the retail heatsink in it. In a brand new system (no dust or anything to impact the temps.) the CPU was running at 55*C under full load, as reported by Asus's PC Probe software and an A7V-333. The case had more than adequate ventillation, so that wasn't the issue. The issue was simply that the HSF couldn't handle the 2100. I realize the maximum die temperature for the palominos was something like 75*C, but 55*C is hot enough to make me very, very uncomfortable. Especially given that this was in a brand new system. If you take apart a system that hasn't been opened in 6 months or a year, and has been used more-or-less every day, you'll find that the heatsink on the CPU is inevitably clogged by dust. Dust is a very, very, VERY poor conductor of heat. It's also not good for fan bearings. If a cooler is only marginal to begin with, and you then force it to deal with that kind of dust buildup, the outcome is not going to be a good one. I'd guess that if my dad's system had been allowed to run for a year without a good cleaning, the CPU temperature at the end of that time would probably be approaching 65*C, if not more. That's simply too hot for me to feel comfortable with.

It's not just an issue of the heatsinks, either. The fans are a good part of the problem. I realize that you have to take space constraints into mind, but I know people who insisted on using your retail heatsinks and have fans that are starting to fail on computers that are 1-2 years old. The smaller the fan, the faster it has to spin to move a given amount of air. Higher RPMs = more noise and faster bearing wear.

So, basically, I'd like to see aluminum/copper hybrid heatsinks with 80x25mm, low-medium speed fans. The Sunon KD1208PTB2-6 is a perfect candidate, in fact. 39CFM @ 32dba. It's quiet, reliable, and moves a lot more air than any reasonably quiet 60mm fan could ever hope to move. Ideally, you'd combine that with a Thermalright SK-7 as the standard HSF, but ~$30/heatsink is a little much for retail CPUs... So, how about something like the Coolermaster HAC-V81 with a fixed-speed fan? The lower fin density would allow it to function more efficiently with a given buildup of dust, and even with a Vantec Stealth (or equivalent) 80mm fan, it should offer more than enough cooling for any current Athlon XP CPU.

On a side note, I retapped the two Ajigo heatsinks I have that came with my 2500s, and put 60x25mm 40cfm Delta fans on them. They're noisy, but they perform very, very well for heatsinks that use 60mm fans.

Also, I said that all of the retail heatsinks except these Ajigos are worthless. That's actually not quite true... I put a 14cfm silent Coolermaster 60mm fan on the dynatron skived-fin aluminum heatsink that came with the 1700 Palomino I bought a while ago, and it kept the 1.3GHz Celeron in my print server <110*F under full load , even with very minimal case ventillation (120mm Enermax intake, 2 80mm Vantec Stealth + 1 120mm Enermax exhaust fans, all on a Vantec Nexus fan controller, and all set at minimum speed, plus a dual-fan Antec PS)

Finally, since I mentioned I've got a Celeron running my print server, I should probably explain why... I'm a student, so my computers are in the my bedroom. I've got a dual 2500 system, a 2.51GHz 1800 system, the Celeron system, and a 2.4GHz P4 Sager laptop, all in the same room. When I was setting up a print server, I wanted something I could leave on all the time, but which would run quiet enough that I could still sleep. The only way to do this with an Athlon XP would be water cooling, which isn't an option (although it would've been nice, since I already had 3 extra AXPs and 4 extra socket A motherboards sitting around...). So, I went with a Celeron because the P3 still costs more than I want to spend, and the Celeron/P3 is the only really cool-running cpu that has any kind of power whatsoever (the Via C3 runs as cool as it does because it has as much CPU power as your average ice cube).

My laptop has a P4 in it for one reason, and one reason only- there are no AMD-based gaming laptops. None. Zip. Zero. Nada. I bought the laptop in June of last year, at which time it was about 10 months ahead of everyone else in terms of CPU speed (the advantage of using a desktop CPU), and there were no really serious AMD laptops on the market at the time. It's gotten better since, but the best ones I've seen still only have 32MB of video memory, which is simply unacceptable for what I want to run. I'll be in the market for another laptop this coming summer, and if there is anything out there that's got cutting-edge graphics and an AMD CPU, I won't consider anything else.
 10/07/2003 08:49 PM
User is offline View Users Profile Print this message

Author Icon

Posts: 23
Joined: 10/07/2003

My mistake... the Palomino XPs do have a 90*C maximum die temp. I must've been thinking of something else...

Regardless, I keep hearing that the Athlons tend to get somewhat "flaky" above about 60*C... although I have no firsthand experience with this, and very little hard evidence of it, it comes up fairly often. I do have one case I can point to that I can think of off the top of my head, which is the review of the Aerocool Deep Impact HSF that Viper Lair recently did. The Deep Impact is a heatpipe-based heatsink, and they as part of their review, they ran it without a fan, basically to see why the temperature difference between the stock 14cfm fan and a 40cfm Delta was only 2-3*C. With passive cooling, the CPU they used in the test, a 1900+ @ 1.75v & 1.67GHz, stabilized @ 61*C. At that temperature, it idled fine but would not boot into windows. Relaxing the OC apparently did not help.

The review is here:' ">

However, I did notice they used an ABIT AT7 Max for the test, and I don't know how the AT7 reports CPU temps in the BIOS- that is, I don't know if 61*C is the in-socket temperature, or the CPU temperature as reported by the CPU's internal diode.

If it's the socket temperature, then the actual CPU temperature was probably around 76*C (In my experience, the difference between actual and in-socket temperatures is around 15*C).

As for why it makes me uncomfortable, it's basically a reliability issue. A smaller HSF generally has less cooling capacity than a larger HSF does. Less cooling capacity means less reserve capacity, which means that it's less well equipped to deal with varying environmental conditions.

If you think dust only raises temperatures by 10*C, if that, then you haven't seen some of the computers that I have. I have one friend who insisted on using the retail HSF, even after I told him that I didn't think it was a good idea. He's got a 1900+ Palomino, so it's got that dynatron skived aluminum HSF. I took that system apart about a year after he bought it, and that was the first time it had been opened since it was bought. The top of that heatsink was literally covered with dust. the vast majority of the air was going through the channel in the center of the heatsink that was left open for the spring clip. Because the fins on these newer heatsinks are spaced so closely, the dust had literally just built up on top of the fins, like it would on an air filter for a case fan or something... very little of it actually went into the fins themselves. I've taken apart older P2 systems where literally all of the space between the fins on the heatsink was packed with dust. So dust can have a major impact on the performance of a given HSF. And, since in many cases, the AMD retail HSF performance is marginal at best, making it deal with that kind of dust is just asking for trouble.

Beyond dust, many people still don't have air conditioning. If it's 90*F in a room, the intake air temperature for the CPU fan in a typical midtower (no intake, no exhaust except PSU fans) is going to be in the neighborhood of 100-120*F. To keep the CPU temperature below the absolute maximum allowable temperature IMO (150*F) would require no more than a 30*F delta t between the CPU temp. and the intake air temp. If you assume a worst-case scenario for an AMD-based system, the CPU would be either a 1.4 Tbird, 2100 Palomino, or 2200 TBred A, all of which put out a bit over 70w at most, as I recall.

Thermal resistance = delta t/input power; to handle one of those CPUs in those kind of conditions would require a thermal resistance of ~0.22*C/w to ~0.38*C/w, depending on intake temperature and assuming a 70w heat load. According to Dynatron's specs for what I THINK is the retail HSF that was included with the Palominos ( the thermal resistance is 0.49*C/w. Given a 100*F/38*C intake air temperature, a thermal resistance of exactly 0.49*C/w, and a heat load of exactly 70w, the CPU temperature would be 72.3*C, which is within it's design limitations, but which is hot enough to potentially cause stability problems.

It's simply a matter of preferring more headroom for varying conditions (dust, fan failure, use in extremely hot environments, etc.) than the stock AMD coolers have provided, with the exception of this Ajigo unit they're using now.
112018 users are registered to the AMD Processors forum.
There are currently 0 users logged in.

FuseTalk Hosting Executive Plan v3.2 - © 1999-2014 FuseTalk Inc. All rights reserved.

Contact AMD Terms and Conditions ©2007 Advanced Micro Devices, Inc. Privacy Trademark information