jackal2513
09-02-03, 06:13 PM
okay, further to my post yesterday ive been tweaking a lot and some of the reults are pretty interesting.
just to recap my new system is: eheim 1250 > Thermochill 120.1 > WW P4 3.06 > MAZE 4 GPU > RES
push fan = papst 120mm @ 7v
pull fan =akasa 80mm @ 7v
here are the tests:
(ambient temp at intake is always approx 22-23.5 C)
SYSTEM AS ABOVE:
PRIME 95 x50 tests
water: 32.9
cpu: 47 C
idle:
water: 28.7
cpu: 38 C
FIRST I PUT BOTH FANS AT 12V giving a temp drop of 1 C:
Papst 120mm @ 12v, Akasa nebula @ full volume
water: 27.5
cpu: 37 C
NEXT, REMOVED THE PULL 80mm FAN, similar results, obviously it was not having much effect really. Tried a sunon 80mm with similar results:
Papst 120mm @ 12v, no pull fan
water: 27.8
cpu: 37 C
NEXT UP REMOVED THE GPU BLOCK, giving a 1 C drop in water and cpu:
Papst 120mm @ 12v, no pull fan & GPU GeForce block removed (again)
water: 26.7
cpu: 36 C
THEN PERFORMED x50 PRIME 95 TESTS and then allowed the sytem to settle back to idle giving a further temp drop:
x50 prime 95 tests
water: 29.9
cpu: 43/44 oscillating (a high 43 C)
back to rest:
water: 26.1
cpu: 35 C (a high 35)
FINALLY AS A LAST TEST I SET THE PAPST FAN BACK TO 7v with the gpu still disabled, this gave a 1.5C water increase and 2C cpu increase:
Papst 120mm @ 7v, no pull fan & GPU GeForce block removed
papst now 7v:
water: 27.4
cpu: 37 C (a high 37)
some conclusions:
1. the thermochill rad obviously works real bad with a low cfm papst @ 7v. Increasing to 12v effects up to a 2C water and CPU drop. The noise is sadly not bearable for me so i'll have to try different fan combos to get the same airflow but lower noise.
2. Having the papst at 12v with higer cfm meant that the system recovered from a stress test a lot quicker.
3. Disabling the GPU again has a marked effect on the system, giving in one test almost a 2C water drop and a 2C cpu drop. The prime test does not affect graphics so with the gpu in this loop heavy gaming might have an even more detrimental effect on load temps. I might consider moving the gpu to my 2nd NB/HD WC loop when its ready to be installed.
4. with the papst @12v and the gpu removed the load temp was far more stable.
5. 35 idle, 44 load is more like it. At least ive proved to myself that everything is reasonably well set up and i havent wasted money. I think now i understand how people here report such good temps..... they use x2 shrouded fans with huge CFM. Even at 12v teh papst doesnt oush very much. Who knows, if i had x2 80cfm 120mm fans in a sandwich round the rad i could maybe get to around 32 idle/ 41 load ?
6. looking back at old records, although similar water temps gave the same CPU temps for both WC systems when idle, its clear that the LRWW performs much better under load than the old innovatek setup. Having now got figures for teh papst @ 12v and the gpu removed in the new rig i now have a good direct comparison with the old rig:
old rig/new rig:
water: 26.0 / 26.1
CPU: 36 / 35
AMB: 22.7 / 22.5
as you can see idle si the same. But then after only just x10 prime95 tests (in the new rig with GPU disabled and papst @ 12v the temp is only 41 or 42 after x10 tests):
old rig x10 tests / new rig x10 tests / new rig x50 tests
water: 27.8 / ? / 29.9
CPU: 44 / 41-42 C/ low 44
you can see that even though the water was a lot hotter after another 30 prime tests, the cpu temp was the same - this demonstrates the WW+1250 quite nicely. If i'd let the prime tests in teh old rig run to x50 based on my memory i reckon the cpu would have been at the 37/38C mark.
The 2 systems are comparable as well cos they both had the papst fan running at 12v, they both had the gpu removed. The main difference was that the old rig had a second rad so i can also conclude that the thermochill with teh 12v papst is as efficient if not better than the old BIMicro2 + MaxXpert combined.
7. i'll also try mounting the WW again when ive got more time.
just to recap my new system is: eheim 1250 > Thermochill 120.1 > WW P4 3.06 > MAZE 4 GPU > RES
push fan = papst 120mm @ 7v
pull fan =akasa 80mm @ 7v
here are the tests:
(ambient temp at intake is always approx 22-23.5 C)
SYSTEM AS ABOVE:
PRIME 95 x50 tests
water: 32.9
cpu: 47 C
idle:
water: 28.7
cpu: 38 C
FIRST I PUT BOTH FANS AT 12V giving a temp drop of 1 C:
Papst 120mm @ 12v, Akasa nebula @ full volume
water: 27.5
cpu: 37 C
NEXT, REMOVED THE PULL 80mm FAN, similar results, obviously it was not having much effect really. Tried a sunon 80mm with similar results:
Papst 120mm @ 12v, no pull fan
water: 27.8
cpu: 37 C
NEXT UP REMOVED THE GPU BLOCK, giving a 1 C drop in water and cpu:
Papst 120mm @ 12v, no pull fan & GPU GeForce block removed (again)
water: 26.7
cpu: 36 C
THEN PERFORMED x50 PRIME 95 TESTS and then allowed the sytem to settle back to idle giving a further temp drop:
x50 prime 95 tests
water: 29.9
cpu: 43/44 oscillating (a high 43 C)
back to rest:
water: 26.1
cpu: 35 C (a high 35)
FINALLY AS A LAST TEST I SET THE PAPST FAN BACK TO 7v with the gpu still disabled, this gave a 1.5C water increase and 2C cpu increase:
Papst 120mm @ 7v, no pull fan & GPU GeForce block removed
papst now 7v:
water: 27.4
cpu: 37 C (a high 37)
some conclusions:
1. the thermochill rad obviously works real bad with a low cfm papst @ 7v. Increasing to 12v effects up to a 2C water and CPU drop. The noise is sadly not bearable for me so i'll have to try different fan combos to get the same airflow but lower noise.
2. Having the papst at 12v with higer cfm meant that the system recovered from a stress test a lot quicker.
3. Disabling the GPU again has a marked effect on the system, giving in one test almost a 2C water drop and a 2C cpu drop. The prime test does not affect graphics so with the gpu in this loop heavy gaming might have an even more detrimental effect on load temps. I might consider moving the gpu to my 2nd NB/HD WC loop when its ready to be installed.
4. with the papst @12v and the gpu removed the load temp was far more stable.
5. 35 idle, 44 load is more like it. At least ive proved to myself that everything is reasonably well set up and i havent wasted money. I think now i understand how people here report such good temps..... they use x2 shrouded fans with huge CFM. Even at 12v teh papst doesnt oush very much. Who knows, if i had x2 80cfm 120mm fans in a sandwich round the rad i could maybe get to around 32 idle/ 41 load ?
6. looking back at old records, although similar water temps gave the same CPU temps for both WC systems when idle, its clear that the LRWW performs much better under load than the old innovatek setup. Having now got figures for teh papst @ 12v and the gpu removed in the new rig i now have a good direct comparison with the old rig:
old rig/new rig:
water: 26.0 / 26.1
CPU: 36 / 35
AMB: 22.7 / 22.5
as you can see idle si the same. But then after only just x10 prime95 tests (in the new rig with GPU disabled and papst @ 12v the temp is only 41 or 42 after x10 tests):
old rig x10 tests / new rig x10 tests / new rig x50 tests
water: 27.8 / ? / 29.9
CPU: 44 / 41-42 C/ low 44
you can see that even though the water was a lot hotter after another 30 prime tests, the cpu temp was the same - this demonstrates the WW+1250 quite nicely. If i'd let the prime tests in teh old rig run to x50 based on my memory i reckon the cpu would have been at the 37/38C mark.
The 2 systems are comparable as well cos they both had the papst fan running at 12v, they both had the gpu removed. The main difference was that the old rig had a second rad so i can also conclude that the thermochill with teh 12v papst is as efficient if not better than the old BIMicro2 + MaxXpert combined.
7. i'll also try mounting the WW again when ive got more time.