![]() | ||
|
|
General Liquid/Water Cooling Discussion For discussion about Full Cooling System kits, or general cooling topics. Keep specific cooling items like pumps, radiators, etc... in their specific forums. |
![]() |
Thread Tools |
![]() |
#1 |
Cooling Neophyte
Join Date: Jul 2003
Location: Sussex, UK
Posts: 23
|
![]()
Well, after a good few months i'm now back to the watercooling fold, and yesterday I finished setting my rig up (with help of dad).
Gave it a good leak test and made sure all was running sweet and then took it upstairs to test. Anyway, to cut a long story short i'm not getting the temps I thought I would be getting. Last night it was hovering around 40-42c idle (socket temps) at 11*235 @ 1.9v, and it was the same or slightly lower before when I had my SK-7 and SmartFan II on. Ambient temps were around 25-26c (85-90F-ish) which, while not a/c standards, isn't too bad, and shouldn't cause such a rise in temps. My rig is...: Cascade CPU block Maze 3 GPU block Thermochill 120.3 6x Evercool 120mm (80cfm) fans in push/pull config Swiftech MCP600 Rev.2 Criticool Waterplant reservoir. 1/2 ID tubing. As you can see, in theory it should be a pretty top-notch and unrestrictive setup, and I wasn't expecting temps to get over mid-thirties centigrade. My rad is mounted on a frame with my pump and reservoir, all external to the case, and my direction goes PUMP>RAD>CASCADE>MAZE3>RES. The thing is, I also have a Eheim 1260 (600gph) that I can setup to test, and seeing as my Maze 3 GPU block didn't yield a significantly better overclock on my 9800PRO over my Arctic VGA cooler i'm thinking of taking it off and just running the CPU block on it's own. Presumably this should also lower temps a tad as it won't have to cope with the GPU heat. Basically, ye cooling watercooling PRO's, I need some serious help on optimizing this rig, i've spent a lot of money on it and I want it running smoothly and most importantly cooly. All help is deeply appreciated... ![]() ![]()
__________________
The hardest thing to do is argue with someone who is undoubtedly wrong, yet wholly convinced that they are right... - Richdog ------------------------------------------------------------------------ M-2500+ @ 11*235 @ 1.9v DFI Ultra B 1GB Corsair v1.1 w. BH-5 Sapphire 9800PRO w. Samsung 2.8ns Cascade, Maze 9700 GPU block, Thermochill 120.3, Swiftech MCP600 Rev.2, 6x Evercool 80CFM fans (push/pull), 1/2 ID Clearflex60. Arctic Silver 5 GlobalWIN SAF520 Last edited by Richdog; 04-17-2004 at 03:56 AM. |
![]() |
![]() |
![]() |
#2 |
Cooling Neophyte
Join Date: Oct 2001
Location: maui, hawaii
Posts: 52
|
![]()
sorry but at those voltage's with 2 blocks i think your temps are just fine.
there is a point where no matter how many radiators you have you just wont get much cooler. you've hit it, and then some. you can try mounten that block again to see if it isn't cocked or nothing, but those temps look about right. at 1.9v's with 28c ambient air i get around 48c's with a maze2 and a black ice prime.
__________________
Cool with the best or burn like the rest. |
![]() |
![]() |
![]() |
#3 |
Cooling Neophyte
Join Date: Jul 2003
Location: Sussex, UK
Posts: 23
|
![]()
So you don't think i'd see any noticeable difference if I was running the Eheim 1260 instead of the MCP600, and I removed the Maze3 GPU block? I would have thought without the heat the GPU gives out, and with a high power pump like the 1260 in a low-restriction system with a Cascade CPU block that the temps would go down quite a bit. This not so?
__________________
The hardest thing to do is argue with someone who is undoubtedly wrong, yet wholly convinced that they are right... - Richdog ------------------------------------------------------------------------ M-2500+ @ 11*235 @ 1.9v DFI Ultra B 1GB Corsair v1.1 w. BH-5 Sapphire 9800PRO w. Samsung 2.8ns Cascade, Maze 9700 GPU block, Thermochill 120.3, Swiftech MCP600 Rev.2, 6x Evercool 80CFM fans (push/pull), 1/2 ID Clearflex60. Arctic Silver 5 GlobalWIN SAF520 |
![]() |
![]() |
![]() |
#4 |
Cooling Neophyte
Join Date: May 2002
Location: Denmark
Posts: 31
|
![]()
As varsis says - the high Vcore could be an explanation. Nore likely though, is that the temps are monitored by the socket-probe which is notoriously unreliable. Buy a temp-probe with thermometer (senfu or other brand will do), and get it positioned next to the core. that way you will get much more precise readings.
__________________
AMD XP1700@2400MHz (1.925Vcore (11x200), 1.5GB DDR3200, Epox 8RDA3+, Seagate Barracuda IV 80GB, Seagate 7200.2 120GB, Seagate 7200.2 160GB, WW ClearTop (CPU) Twinplex GF4 (GPU) w/ 2x120mm on heatercore from Accord '92 and Eheim 1048. To be further upgraded! (RAID 5, HD/PSU/NB cooling etc.) |
![]() |
![]() |
![]() |
#5 |
Cooling Savant
Join Date: Nov 2003
Location: Suffolk, UK
Posts: 234
|
![]()
It would be worse with a 1260 (its not a good pump due to high power consumption)
Removing the GPU block may help a little tho |
![]() |
![]() |
![]() |
#6 |
Cooling Neophyte
Join Date: Mar 2003
Location: Austria
Posts: 12
|
![]()
I agree with previously said. You results are very realistic with your setup. Plus your socket sensor is very unreliable (mein also). I always get a little bit different temps so i bought a flat sensor thermometer. Much better and more constant temps.
You should know that your ThermoChill likes very strong fans, like Deltas or with similar CFM. Your Evercools (12 or 7 Volts???) maybe are just not strong enough. I have 120mm Titans (same fan as Evercool 120mm) and with my rad I get similar temps as you regardless on 7 or 12 volts. If you put strong fans you will improve your temps a little but not too much. You may try to put a fan 10cm from the socket blowing to the waterblock as I did, my temps sunk 6-7°C so I'm happy now :-) Which bios do you use, main is bios18. Did your socket sensor touch the back of the CPU? Last edited by deltawing11; 04-17-2004 at 01:49 PM. |
![]() |
![]() |
![]() |
#7 | |
Cooling Savant
Join Date: Feb 2002
Location: Dayton, Ohio
Posts: 836
|
![]() Quote:
|
|
![]() |
![]() |
![]() |
#8 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
![]() Quote:
|
|
![]() |
![]() |
![]() |
#9 |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
![]()
Another thing to consider, and this is especially true of the Abit NF7-S boards, is that socket air-flow can seriously alter the temperature readings seen. The presence/absence of a decent 80mm fan blowing over the socket/mosfet area can be as much as 6C reported temperature differences, just due to the air-flow alone cooling down the very hot mosfet area, and also cooling down the socket region which affects the reading of the thermal probe.
With a HSF you get the air-blast off the fan that provides that effect. With water-cooling all of that effect is gone, and this is one of the biggest reasons for why people who rely on the in-socket thermistor readings can be less-than-impressed with the switch-over. |
![]() |
![]() |
![]() |
#10 |
Cooling Neophyte
Join Date: Jul 2003
Location: Sussex, UK
Posts: 23
|
![]()
OK stupidly, I forgot to update my sig specs form the last time I visited, so as you can see I have the DFI, and mobile 2500+ etc.
Yes Cathar, my SK-7 and SFII managed my current overclock PRIME95 stable, so I was surprised that watercooling hasn't in fact increased my OC. As someone suggested, i've stuck a big (yet quiet) house fan next to my case and temps have dropped a lot, I might keep it there, even though I didn't really want to have to use it, lol. So Cathar, in your opinion if I lost the GPU block and added the Eheim 1260 I wouldn't see a decrease in temps? I'm not sure how much the Thermochill 120.3 and the cascade adds to resistance in the loop, but I thought with 1/2ID tubing the flow would be far more with the Eheim 1260, as opposed to the MCP600 which is best for high-resistance rigs, is it not?
__________________
The hardest thing to do is argue with someone who is undoubtedly wrong, yet wholly convinced that they are right... - Richdog ------------------------------------------------------------------------ M-2500+ @ 11*235 @ 1.9v DFI Ultra B 1GB Corsair v1.1 w. BH-5 Sapphire 9800PRO w. Samsung 2.8ns Cascade, Maze 9700 GPU block, Thermochill 120.3, Swiftech MCP600 Rev.2, 6x Evercool 80CFM fans (push/pull), 1/2 ID Clearflex60. Arctic Silver 5 GlobalWIN SAF520 |
![]() |
![]() |
![]() |
#11 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
![]() Quote:
Regardless, more air-flow around the socket area will lower the reported temperatures for either board since both have the in-socket thermistor. Re: overclock. You're running at 11x235. How are you attempting to overclock further? Have you tried 13 x 200, and then start bumping up the FSB from there? 235FSB is fairly high for the NF2 boards, and in my experience, the higher the FSB, the lower the peak overclock attainable out of the CPU, as in the FSB tends to place a bit of an upper limit on how far the CPU can be pushed if the FSB is too high. Drop the FSB and raise the multiplier, and then see how it goes. If you can't because you have a locked CPU (perhaps a locked XP2500+ ??) then I'm afraid that you may be bmping into the limit of the FSB of the board/memory when coupled with the CPU's speed. |
|
![]() |
![]() |
![]() |
#12 |
Cooling Neophyte
Join Date: Jul 2003
Location: Sussex, UK
Posts: 23
|
![]()
Nope, as it says in my sig and in my other post, I have a mobile 2500+, so the locked multiplier isn't an issue. And the temps in my very first post were regarding my DFI, hence the part about me not changing my sig in a long time (3 months) to update my new specs.
The DFI Infinity and Ultra B boards will do 250Mhz FSB no sweat with the right RAM (BH-5, of which I have several sets, all tested to 245Mhz+), so with watercooling I am looking for a speed of 10.5*250 preferably, or 11*240 would also be acceptable. I've been overclocking for a long time, so I know what to expect out of my hardware, and I bought each part specifically for performance. I'm just interested in getting the most out of this watercooling setup first, and when I know its performing as it should, then i'll concentrate on tweaking my stuff to the max. So, once more (and I sincerely appreciate the replies Cathar, don't doubt me there) do you think losing the GPU block and replacing the MCP600 with the Eheim 1260 would make any noticeable difference in temps? BTW i've added a housefan to the side of my case and idle temps are now around 32c (reported socket temps) at 11*235 w. 1.85v and an ambient temp of around 25c. It's only because i'm a n00b to watercooling that i'm asking all the annoying questions, but rather ask and be a fool for a minute, than stay quiet and be a fool forever, eh? ![]()
__________________
The hardest thing to do is argue with someone who is undoubtedly wrong, yet wholly convinced that they are right... - Richdog ------------------------------------------------------------------------ M-2500+ @ 11*235 @ 1.9v DFI Ultra B 1GB Corsair v1.1 w. BH-5 Sapphire 9800PRO w. Samsung 2.8ns Cascade, Maze 9700 GPU block, Thermochill 120.3, Swiftech MCP600 Rev.2, 6x Evercool 80CFM fans (push/pull), 1/2 ID Clearflex60. Arctic Silver 5 GlobalWIN SAF520 |
![]() |
![]() |
![]() |
#13 |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
![]()
Rich, I appreciate that you are an experienced overclocker, but have you at least tried a lower FSB/higher mult to see if that helps you?
The other thing that water does do for you is to allow you to bump the CPU voltage further for better overclocking. The difference from 1.9v to 2.0v is typically around 100MHz or so in my experience with the AthlonXP's. Moving to the Eheim 1260 is a bad idea. No need to lose the GPU block - it won't make a major difference. It seems like your issues with "too hot being reported" are apparantly solved now with some air-flow about the socket. Try playing with your FSB/multiplier/voltages some more. In my experience with AthlonXP's, I've never been able to max out both CPU and FSB MHz at the same time. It's always been one or the other, or some middle ground between the two. |
![]() |
![]() |
![]() |
#14 |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
![]()
Regarding your overclocking issues, having now owned and experienced what a DFI LanParty UltraB has to offer, I can honestly say that it's the motherboard's totally crap CPU power regulation that is holding you back from pushing the CPU further.
Best I was able to achieve on my board, stable, was a whopping 300MHz lower than what I could achieve stable on the Abit NF7-S. Unstable I could get to within 80MHz of the NF7-S. The DFI's are great for high FSB shenanigans, no doubt about it, but for pushing the CPU hard they suck donkey balls. Last edited by Cathar; 04-24-2004 at 10:29 PM. |
![]() |
![]() |
![]() |
#15 |
Cooling Neophyte
Join Date: Jul 2003
Location: Sussex, UK
Posts: 23
|
![]()
Yes Cathar I totally agree! I sold a mobile Barton 2500+ to someone at AMDMB, and I had only been able to get it to 2550Mhz on this DFI, so wasn't particularly hapy with it. The guy who bought it stuck it in his NF7-S and on his first day he managed to get 2.7Ghz out of her.
![]() My currenct mobile 2500+ is even better, it boots into WindowsXP at 2.8Ghz, but running a SANDRA benchie freezes the mouse, nearly always a sign of not enough voltage. 2.72 Ghz should be easy with this CPU and watercooling, but the undervolting of both the Vdimm and Vcore is a serious thorn in my side. I have 3 pairs of 512MB (1GB DC sets) Corsair PC3500 v1.1 BH5's that are able to hit 245Mhz+ with 3.3-3.4Vdimm, but this board seems to constantly stay around 3.20-3.25 which gives me blue screens under stress. ![]() All in all I am not too happy with this board, a high FSB does not compensate for overall bad stability, cold-booting issues, and far lower overall Mhz. Besides, on a Vmodded NF7-S people are hitting 240Mhz-250MhzFSB and they have the option of up to 2.2Vcore and amazing stability. Not to mention SATA writes are much faster. ![]() DFI are said to be putting forth a new revision of the DFI Lanparty Ultra B though before the AthlonXP finally dies it'd death, most likely so as not to be remembered as having a piss-poor board out, apparently it will have better stability, voltage regulation, and more Vcore options. I think they should give current DFI owners a complimentary upgrade to be honest, and i'll push for that if and when it is released. On another note, the watercooling does seem to be working it's magic, I was expecting a bit too much in the first place I think. Idle temps of 33c @ 11*240, 2650Mhz, 1.95Vcore at an ambient of 26c are nothing to be sniffed at, and load only rises 5c or so. Only problem I have now is that it's going to be a scorchingly hot summer, and my bedroom is south-facing, lol, time to shell out for a mobile air conditioning unit methinks. :shrug: The MCP600 is a good little pump too, i'm very impressed with it's power and lack of noise, and it copes with the Cascade and Maze3 blocks without breaking a sweat, not to mention the humongous Thermochill 120.3 which must be a bitch to push water through. ![]() I'm waiting on that GPU block that you're supposed to be bringing out though, lol, I don't think the Maze3 GPU block is the be and end all, and hasn't helped my core OC all that much in comparison with my old Arctic Cooling VGA Silencer. Is that project still on hold indefinately then? ![]() Thanks for the replies I appreciate it, Regards, Rich. ![]()
__________________
The hardest thing to do is argue with someone who is undoubtedly wrong, yet wholly convinced that they are right... - Richdog ------------------------------------------------------------------------ M-2500+ @ 11*235 @ 1.9v DFI Ultra B 1GB Corsair v1.1 w. BH-5 Sapphire 9800PRO w. Samsung 2.8ns Cascade, Maze 9700 GPU block, Thermochill 120.3, Swiftech MCP600 Rev.2, 6x Evercool 80CFM fans (push/pull), 1/2 ID Clearflex60. Arctic Silver 5 GlobalWIN SAF520 |
![]() |
![]() |
![]() |
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
|
|