Zalman CNPS7000 to Artic Cooling Freezer 64 Pro

grahamjenks

grahamjenks

grahamjenks

Associate
Joined
8 Oct 2005
Posts
604
Basically is it worth upgrading from the zalman to the artic?

Bearing in mind i currently run the zalman with the fan controller at low speed (mainly for queitness) - will changing to the freezer pro give me better temps - without greater noise (could I use the zalman fan mate with the artic?).

This is mainly to allow me either to potentially overclock a little - or just generally have lower temps throughout my case.

I'm guessing some of the benefit i'd get was from using the thermal compound thats pre applied to the artic rather than the stuff zalman supplied that I applied.
 
The Arctic Freezer 64 and the Thermaltake Big Typhoon are allegedly very efficient. The Big Typhoon is said to be highly efficient at dissipating heat at high load although the Freezer is supposedly not that far behind it. I will probably go for the Big Typhoon although I don't know how noisy the fan will be to me or if it needs to be run at max speed to achieve good results.

I wonder if there will be much difference between them when there is a lot of heat due to highish vcore (~1.6v?).
 
I've just tested the Freezer 64 with a 64-bit Sempron 2800+ and I'm impressed. It's fan is running at ~2400 RPM (very quiet to my ears) and the 2800+ is at 2.4Ghz (8 x 300) with 1.54v. Not that it needs such a high vcore as it can do that speed on default volts but I wanted to create as much heat as possible.

I'm currently using BIOS 1.5 on my Neo Platinum which maxes out the vcore at 1.54v although I can get 1.6v by using BIOS 2.1. I switched to 1.5 just to see if I had any extra options because I knew it wouldn't be able to recognise this E6 sempron but alas the default vcore option isn't available like it is when I use a mobile AMD 64 CPU (would have allowed me to use 1.65v and hopefully get 3Ghz stable). It's weird with the 2.1BIOS as the vcore adjustments are backwards so +10% = +3.3%, +8.3% = 5% etc.

Here's a breakdown of the temperatures:

Idle:

CPU: 29c
NB: 27c
AUX: 26c
HD: 23c

I replaced the heatsink on the NB hence why I get such a good NB temperature. :)

Max load:

I run 2 x Prime95 on priority 1(1 doing small FFTs and the other doing in-place large FFTs) and S&M CPU (FPU) on 100% load.

CPU: 45c to 49c (fluctuates probably because the stress created by S&M changes)
NB: 27c
AUX: 27c
HD: 23c

I have the Freezer 64 pointing directly at the PSU fan (I'm using an Enermax 285w SFX PSU in a Antec Lanboy case) so the hot air from the CPU is being removed instantly and as you can see from the other temps it's working really well! The PSU fan is temperature controlled and goes from ~1400 RPM to ~2200 RPM quite quickly.

What's even more surprising about the Freezer 64 performance is the CPU drops almost instantly to idle temperature when I stop the Prime95s and S&M.

Also room temperature is probably around 18c or so as I don't have any heating on and the window is open although it doesn't feel very cold and I'm only wearing a t-shirt. :)

By comparison in the same case with same everything but using a 3400+ DTR with a SLK-948U and 92mm fan on high speed (noisy) with the 3400+ at 2.4Ghz (8 x 300 again) but with 1.63v the system freezes up requiring a reboot after about 20 mins of the above max load testing. Reason appears to be because the CPU temperature goes beyond 65c so I suspect the CPU just can't handle such heat and obviously the SLK-948U isn't up to the task of dissipating the heat fast enough to keep the temperatures low enough for it to be stable. Also the other temperatures go quite a bit higher and no doubt it's because the fan on the SLK-948U is blowing down onto the heatsink compared to the Freezer 64 which blows sideways into the heatsink.

Also for comparison the CPU at default vcore (1.4v) and 2.4 Ghz (8 x 300) the max temperature is only 33c to 38c when running it at max load.

So yeah this Freezer 64 is incredibly efficient but saying that I can't see it being able to cool a CPU running at 1.6v+ below 60c so I guess I will have to try the Thermaltake Big Typhoon to see if it can do it and maybe also try the Zalman.

I guess it's possible that none of the air coolers will cool a CPU with 1.6v well enough at max load and is probably why a lot of people using air cooling on here tend not to go beyond 1.55v.

I type too much don't I? :D :o
 
Cheers for that - lots of excellent info there - espically if I start overclocking - although looking at my bios for my quite old biostar k8vha pro - I dont have much control over vcore.

Might consider the freezer pro espically as it seems so efficient.
 
I decided to redo the testing and this time take screenshots of the testing so I can post evidence of what I've talked about above. I'm currently online via my Linux PC but will get the test machine hooked up to the network so I can upload them in a bit.

I also flashed back to the 2.1 BIOS and did more testing and took a couple more screenshots. The CPU at 1.6v idles around 35c and hits 57c at max load. One thing though, a hot day will likely increase the temps by around 5c to 10c and would most likely result in the CPU overheating and crashing/freezing.

Will test that situation soon by putting my heating on and seeing how hot I can make it run. :)
 
Here we go with the screenshots of my Freezer 64 testing. The following screenshots are when running at 2.4Ghz (8 x 300) with 1.54v:

Idle:



Load after 1 minute:



Load after 5 minutes:

 
Last edited:
Load after 10 minutes:



Load after 30 minutes:



30 seconds idle after the above load testing:



The main thing to look at is the graph SensorsView provides which gives a clear indication of the reported temperatures. Also notice the SYS, AUX and HD temperatures remain stable throughout!

The speed of the SYS fan is being reported wrongly at times and no doubt it's because of a bug in MSI's Cell thing (their CoreCentre software is even worse where sometimes it reports the CPU speed at 8 x 600 or more so I uninstalled it). My guess is it's sometimes doubling the reported RPM so just ignore that. :)
 
Last edited:
Here is the screenshots when running at 1.6v with the 2.1 BIOS:

Load after 15 mins:



30 seconds idle after the above load testing:



Also notice how with BIOS 1.5 it doesn't know what CPU is installed and CPU-Z reports it as Model Unknown but then with BIOS 2.1 it knows and CPU-Z gets it right.

Weird huh. :)

:D
 
Last edited:
TRy a utility called occt for benchmarking, it'll give you all these graphs in one.

Whats your superpi time can I ask?
 
Yeah if I had a dual core setup and was testing for speed I'd probably use OCCT. I'm just testing the maximum heat with the Freezer 64 and having a bit of fun overclocking with the Sempron 2800+.

SuperPI 1M with everything at defaults (memory defaults to DDR333 and the CPU is at 8 x 200 = 1.6Ghz) is 1 minute and 2 seconds. At 2.4Ghz (8 x 300) with the memory at DDR480 it's 41 seconds. As you probably know though Super PI is heavily dependant on ram speed and timings as well as the level 2 cache so no matter what this Sempron will overclock to it'll always be really bad at Super PI.

I left it running overnight at 2664Mhz (8 x 333) with 1.54v. Maximum temperature was 55c and it locked up sometime while I was sleeping. I've no idea why it locked up but no doubt it's probably because of the motherboard.

I rebooted it and got the dreaded detecting raid problem so it wouldn't even boot properly until I reset the cmos and reset everything. It's also got the low vcore problem where the vcore is defaulting to 1.1v and the max I can configure it to is only 1.36v. This MSI K8N Neo Platinum is really weird like that, I guess it's why it's considered one of the worst socket 754 boards for overclocking, so I think it's time to get a better board. :)
 
Ive got a sempron too, 2.6ghz is about the limit it seems. Iam just using a stock heatsink though and didnt even bother with paste as I was hoping to use a cnps too.
Iam just messing about with this rig but I managed 342x7 last night but I was slightly too ambitious and 350 HT locked it up and I havent managed to get it that high again, clockgen just locks up now when I try raising it above 9x250.

My PI time is 38 secs but your right about it not being its thing, Iam not sure what would be more approiate I'll try 3dmark03 in a bit

Its a bit too much like voodoo compared to my old socket A overclocking for me, maybe its the paste but Iam getting a 775 soon anyway
 
You can change the multiplier on your board in the BIOS or only using a utility in Windows? Also what board you using mate?

I'm hoping to get the DFI NF3 250Gb soon and hopefully I'll have better luck than I've had with the K8N Neo Platinum. Now on every boot it's doing 1.1v for the Sempron and I'm having to use +10% which somehow gets it to 1.4v. Kinda strange I guess but doesn't seem anything I can do about it. :o

And yeah I know what you mean about it being voodoo. These AMD 64 based CPUs really are temperamental especially when using a high HTT. :)
 
Last edited:
I heard they were bad except the new ones which have more cache. Iam not sure how important cache is for games, since all the repetitive code is on the gpu?

str said:
You can change the multiplier on your board in the BIOS or only using a utility in Windows? Also what board you using mate?
Click to expand...


A utility I found for altering the multiplier downwards in windows is CBId which shows in google. It gives you options on everything, some very obscure memory timing stuff as well as the normal timings which is usefull for dual SPD chips.

The board am using temporally is an AK89 Nvidia 150 with a tiny 600HTT. I havent managed to get that 340HT again but I got a superpi of 37 seconds by just sticking to the stock multipler

377976wo.jpg
 
Last edited:
Fulcrum said:
I heard they were bad except the new ones which have more cache. Iam not sure how important cache is for games, since all the repetitive code is on the gpu?
Click to expand...

The old Celerons with 256kb or 128kb cache were completely rubbish at most things (including superpi). The new 65nm Celeron has 512kb cache and performs much better (although it has a few architectural tweaks as well).
 
Fulcrum, my E6 Sempron 2800+ only has T-Sensor showing in CBID so I don't get to do much with it (ie. no multiplier change, voltage change, etc) in Windows. :(

Also, my motherboard is definately faulty, it sometimes alternates between 1.36v and 1.6v (when set to 1.6v in the BIOS). Most of time it's at 1.36v so I'm definately in need of a new motherboard.
 
You must log in or register to reply here.
Back
Top Bottom