help with watercooling setup/temps

[redgoo]

Hi All,

I installed my watercooling system this weekend, but I am concerned about the CPU temps.

Current setup: MCP 600 -> TDX Nozzle #4 -> Maze 4 GPU -> Black Ice Xtreme 120 (Radbox with 120x25mm Enermax Push + Coolshroud with 120x38mm Panaflo Pull) -> Resevoir. There is a little under 5 feet tubing total.

Temperatures:
CPU Idle (1.4v): 39
CPU Load (1.4v): 51

GPU Idle: 44
GPU Load: 46-47

The CPU temp. difference between idle and load seems pretty high to me. I'm thinking that maybe the mount is bad, but I have remounted it and got similiar results. I'm still not confident that it is mounted properly though. Anyone have pics of a proper TDX/939 mount?

If it isn't the CPU mount, I'm guessing the BIX 120 may not be enough. However, the idle/load temperature of my GPU does not vary much. Also, the temps only change by about 1C depending on whether I have both fans on full 12v, or greatly reduced in speed.

Do these temps seem normal for my setup? What upgrade would give me the best performance increase? Any insight would be appreciated! Thanks in advance.

[idleuser]

dont worry your neo2 is throwing it off.. my neo2 has different temps everytime I reboot!! But I seriously would consider changing that BIX for a BIP2 or even a weapon/heatercore shround.

[Jupiler]

I had the same temps on my Neo2.
Setup is about the same as yours.

Res > DD12V-D4 pump > HTFX-2 dual rad with 2*120mm in pull > DD TDX nozzle #1

Still running same setup, but on DFI SLI-D, temps are around 39° load right now.

[MaxxxRacer]

the neo2 cant read temps for a dam so dont worry about that.. but your gpu temps which are usually more accurate are a bit high. its probably becusae of your small rad and high power cpu and gpu.

[redgoo]

Thanks for the info!

I remounted the TDX again, and this time I got 37C Idle and 45C Load. which I think is a pretty good improvement. Hopefully it is not the NEO2 playing games, lol. I'm going to try it one more time and then I think that's the best I'll be able to get out of this setup.

I've remounted the Maze4 a few times and I always get around 44C. What is a good temp?

It's weird that whether or not I have the fans all the way up or turned down low, the temps are close to the same. Is that a sign I'm at the limits of the rad? I hate the BIX, it has paint flakes missing in a ton of spots as well as some nice scratches, and I've been trying to be careful.

I should have a Weapon Heatercore on the way soon :) Now to figure out how to mount it externally. It's probably too heavy to hang off the radbox, lol. Any ideas on what kind of temp gains I should expect out of that?

[bimmer01]

temps on the MSI neo2 sucks.
My old Asus p4c800-e read 32C load with air cooling.
and my msi neo2 platinum, reads 45 load with water cooling.

go figure.

[Ancient_1]

I will bet tho that your MSI board is much closer to actual CPU temps than your Asus board was. The P4C800 reads horribly low. When I had mine in my main case it showed idle temps lower than my water temp and at load it showed cpu temps at or a degree or so higher than my water temps and that is far from possible with a 2.4c @ 3500+ and 1.65v. Realisticly my cpu temps should of been in the 15~20° range over my water temps.

[bimmer01]

i was told, MSI temps are inaccurate. On the other hand, asus boards have fixed their problems with the temp reading issues.

[Ancient_1]

Well I can tell you the P4C isnt fixed. And if you think that a load temp of 32° for a modern cpu is realistic (unless using chilled water/air/pelts/phase change) you are sadly mistaken. The older Pentium 3s and the Pentium Ms could realisticly have load temps that low with good cooling but I really doubt you will see A64s and P4s with that kind of load temps without some kind of chilled cooling. Removing 80+ watts from such small dies is a real tough thing to do with conventional cooling. The two TIM joints themselves will add 5° to the temps alone.

[bimmer01]

True, you make a good point. But I had a XP-90 + 92mm Tornado fans blowing at FULL speed, and it cooled my P4 3.0c rig @ 32-35 load and 29 idle.

But seems to me that my A64 rig gets hotter that my old P4 3.0c rig.

Also, to prove a point. My MSI Neo2 reads my CPU temps as 40-45 idle and 50 MAX. That is very inaccurate. And MBM5 reads my Case temps as 18C
So thats:
40C idle
50C max
18C Case temp.

Does that make any sense? No, MSI has bad temp reading.

[Ancient_1]

I am not saying your MSI is right, I just know from personal experience the P4C800s read terribly low. With my other 2.4 (@ 3.4 1.65v) in my P4C with the SP-94 and 80mm Tornado I got load temps in the low to mid 40s which showed about the same to a couple degrees cooler than my Max3 with my good 2.4 @ 3500+ and 1.65v. I know when I switched from air to water I got a drop of 12~15° drop in temps from mthe SP-94 to my water setup.

As far as the AMD running warmer it is possible but I would guess they are close to the same heat output (I dont know the AMDs specs but the 3.0 is in the area of 80watts). What I really think is because the AMD reads higher it is just giving you the impression that it is hotter.

It would be nice if a person could trust that the readings they get from thier MBs were close to accurate but sadly that is not the case.

The only reason I kind of picked on you is its one of my main gripes on how everyone on these forums seem to think that the temps given to them are accurate when they get low readings and are inaccurate when the readings are on the high side. Or the fact they can compare thier temps to someone elses and would be meaningfull. One thing that usually is true is the same MB with the same bios usually read the temps close to the same, but that is not always true either.

Or the other thing too that can get to me is people thinking they have accurate temp readings when they use a proble from something like a digidoc or other similar temp reading instument. There is no place they can be placed that will get close to the actual cpu temps (directly behind the die if insulated would be close but still low). Then also is the questionable accuracy of the instrument itself (if I remember correctly I have around 10° difference is reported temps from the different probes when placed together at the same place and temp on my DigiDoc5)

I have a MSI 875P board that seemed to read temps in the area I would of expected when new, but then I flashed the bios to one for prescott compatibility and they went to very unrealistic readings also, I was getting load temps in the teens with a prescott clocked in the area of 3.7.

Piont is there is no MB you can say is near an accurate way to read cpu temps or to compare with others.

[redgoo]

I remounted the TDX again and temps dropped another degree or two, so I'm happy for now. I'm sure the Neo2 isn't reading the exact temps, but it is at least somewhat helpful to judge temps relatively.