OC & Modding AMD A64 Heatsink roundup Q2 2005 @Madshrimps

J

JediMaster

Disciple
A comprehensive heatsink roundup @madshrimps dated May 15, 2005 !
(mods: please delete if this a repost)

source: http://www.madshrimps.be/?action=getarticle&articID=224
In this massive Athlon 64 heatsink roundup we compare 17 cooling solutions from different manufactures including Scythe, TTIC, Arctic Cooling, Thermaltake, Titan, Thermalright, Zalman and Coolermaster.

....

Previously tested but included for reference (links lead to detailed overview of these heatsinks done in previous reviews):
# Coolermaster Hyper6
# Coolermaster Hyper48
# Thermalright XP-120
# Thermalright XP-90
# Zalman CNPS7000CU
# Zalman CNPS7700CU
# Stock A64 Cooler

New contenders:
# Scythe SCNJ-1000 Ninja
# Scythe SHOGUN Heatlane
# Titan Vanessa-L
# Titan Vanessa-S
# ThermalTake Sonic Tower
# TTIC NPH K8-1
# Arctic Cooling Freezer64
# Scythe Kamaboko
# Scythe KamabokoZ
# Thermalright XP-90C
Click to expand...

Nice to see the performances classified with respect to the same fans and at low, medium and high CFMs ! This makes it easy to observe the isolated effect of CFM across heatsinks.

In particular, note the Thermalright Interclub graph for comparisons of xp-90, xp-90c, and xp-120.

If those numbers are to be believed, looks like Thermalright has some competition from TT SonicTower, CoolerMaster Hyper 6 and Scythe. But xp-90 may still be a good choice all factors considered.

And what about xp-120, eh ? Hmmm...
 
I really don't know how they are getting such high temps. Something is screwed up! I'm basically on the same setup... Same board as well to rule out diffs of temp monitors. (other than the fact that its a clawhammer oc'ed to 2.160GHz) and my load temps are around 39 with a thermaltake 92mm fan. Do rmr that this is the indian summer and its in a room with no airconditioning.

Idle:


Burn:
 
his methods are very confusing. I am sure he must have confused himself.
I mean rather than putting fan at 30%-50%-100% changing fan etc etc what he should have done is use a 1 standard fan which gives less than 50db noise on the heatsinks that allows you to change fans. If the mounting size is different then he should have used a fan with similar cfm and run the damn thing at full speed.

I cant makeout the damn thing from that table. Its too confusing and too many fans used.
 
Well,actually I agree with their methods.K7 burn is much more cpu intensive than any app like prime95/F@H etc...
I am sure that unless u have a very good heatsink , ur temp will cross 60 withing 10 mins of K7 burn.
Chaos use K7 burn(cpuburn) and then see.Though u have thermalright,u temps will still increase a bit.
 
funkymonkey said:
his methods are very confusing. I am sure he must have confused himself.
I mean rather than putting fan at 30%-50%-100% changing fan etc etc what he should have done is use a 1 standard fan which gives less than 50db noise on the heatsinks that allows you to change fans. If the mounting size is different then he should have used a fan with similar cfm and run the damn thing at full speed.

I cant makeout the damn thing from that table. Its too confusing and too many fans used.
Click to expand...

:) !
1 standard fan at full speed ?
There is a >10 page thread here in these forums titled "Best Fan for use with ThermalRight XP90C". And note in there the nvnews link Masky gave on the comparison between Enermax and Sunon at low and high speeds and nvnews' comment:

"While the Sunon's CFM rating is apparently less than the Enermax at full speed; it seems to put out a more forceful stream of air than the Enermax.."

Tells you that folks are interested and find value in such studies !

Yep, the last graph in the madshrimps article looks a l'il too overcrowded. If that's confusing, then you could look at the preceding graphs for individual fans - Papst, Panaflo, and Tornado and simply ignore the datapoints for 50% ratings. But to neglect the whole study would be like "throwing the baby along with the bathwater" unless, of course, there is some reason for not believing in the numbers which is a different story altogether.

Generally it is of interest to examine the efficiency of cooling with respect to:
1. effect of variations in CFM of a single fan bcoz the deltatT drop could be nonlinear
2. the effect of different fans at the same CFM

The CFM rating is a gross characterization of a fan - it doesn't tell you anything at all about the local profiles of the velocity components at the fan outlet, i.e., the distribution of axial, radial, and azimuthal components. And different fans because of their specific structural designs may have different profiles impacting the cooling efficiency. For instance, a fan may have a relatively pronounced radial velocity component resulting in relatively more of the flow being directed away from the heatsink and thereby possibly impairing the cooling efficiency (see nvnews' comment above between Enermax and Sunon). Or, it could have a much bigger hub/dead-center.

Well, the review is new and TBH I haven't examined the numbers thoroughly. But clearly a database of heatsink and matching fan characteristics is very useful.
 
Chaos said:
Tried cpuburn for 5 mins. No diff... The max temp was 39 :)
Click to expand...

Just guessing...could the difference be because of different monitoring softwares - the madshrimps article uses speedfan. And should there such a discrepancy, then I'd think that the "absolute temperatures" may not be meaningfully compared.

I'd also think that for the same ambient conditions, CPU, motherboard, and monitoring software, the relative differences in absolute temps of different heatsinks can be compared. And I think that this is prolly the reason why even if the absolute temps in that study may not look correct, the relative temp differences between xp-90 and xp-90c appear to be reasonable.

And if the monitoring software over/underestimates the temp, then T_load - T_ambient will not be correct. But (T_load - T_ambient) of Heatsink_1 - (T_load - T_ambient) of Heatsink_2 can be sensibly compared because the T_ambient cancels out. In other words, for the same set of conditions and sw/hw, you could follow the trend in absolute T_Load for different heatsinks.

just a thought...though I'm still surprised by xp-120 results... :)
 
JediMaster said:
Just guessing...could the difference be because of different monitoring softwares - the madshrimps article uses speedfan. And should there such a discrepancy, then I'd think that the "absolute temperatures" may not be meaningfully compared.

I'd also think that for the same ambient conditions, CPU, motherboard, and monitoring software, the relative differences in absolute temps of different heatsinks can be compared. And I think that this is prolly the reason why even if the absolute temps in that study may not look correct, the relative temp differences between xp-90 and xp-90c appear to be reasonable.

And if the monitoring software over/underestimates the temp, then T_load - T_ambient will not be correct. But (T_load - T_ambient) of Heatsink_1 - (T_load - T_ambient) of Heatsink_2 can be sensibly compared because the T_ambient cancels out. In other words, for the same set of conditions and sw/hw, you could follow the trend in absolute T_Load for different heatsinks.

just a thought...though I'm still surprised by xp-120 results... :)
Click to expand...

Yeah even I have that feeling. Maybe he used a rather poorly ventilated case. The absolute temperatures depend a lot upon case airflow. The same HSF can be 20 degrees hotter if there is no fresh air flowing over it. Along with that another thing that can make a tremendous difference is the thermal compound used for mounting the HSF. I saw a 6 degree drop in idle temps from the same. You can rule out the issue of software as on my pc, speed fan and smartguardian report exactly the same temps. Also its not a thermal probe calibration issue as I mentioned earlier since I have the same motherboard.
 
Back
Top