mine has a release date of 11/19/08, i think its the one before the last.Quote:
Originally Posted by earthman
Printable View
mine has a release date of 11/19/08, i think its the one before the last.Quote:
Originally Posted by earthman
Quote:
Originally Posted by Astratuner
Anyone?:yepp:
I have had reasonable success with air but having done a fair amount of reading - I am thinking of moving to water. The oc'd i7's generate so much heat! My 965 is 4.1 prime stable 10+hrs on air but at a rediculas max temp of 97C, my config is in my signature below. I plan to raise the qpi v to see if I can run at a lower cpu v (1.55v used for 4.1) but I imagine my temps will stay pretty high. So, my question - if money is no object (but low noise is still desirable) - what is the best extreme cooling solution for hot i7s? Plan to use the machine 8-12 hours per day for two years then cpu can die. :D
For extreme cooling it would be a single stage but dont expect it to be low noise, maybe it can be modded to run quiter but im not that uptodate about phase change cooling. :p:
Water would be a possibility with a "monster rad" and 9 low rpm fans.
Its noticable for sure, especial in TF2. In crysis its less noticable. In Crysis it costs me ~2fps running boinc and crysis together but in TF2 its a lot more. The bad thing is that it considerably hurts the min fps, without boinc TF2 on a 32player server dont hits the 40s, with Boinc it dips down into the 20s :|...Quote:
Originally Posted by bowman
Thought i dont run WCG exclusive. I also run SETI with optimized clients and they make have use of the SSE untis.
Hornet331 - wow that's a radiator! I wonder if i can find that here in the states. What other components would be the best of the best: block, pump, ...
just get a simple 2x120 rad with a gtz, d5 and some quiet fans :)
A nice single stage would be ideal. Ask someone like Dimas what can be done.Quote:
Originally Posted by SteveRo
There's no need for a massive rad for just the cpu, it won't run much if any cooler than one half its size. A quad or triple rad like a PA120.3 would be more than enough. A GTZ block, I have the EK Supreme, works very well. Pumps, a DDC3.2 or a D5.Quote:
Originally Posted by SteveRo
I can attest that an i7 at 4.2ghz with HT at 1.36v load using a 2x120mm (Scythe fans at 1600rpm) with a GTZ dedicated for the cpu only is almost not enough.Quote:
Originally Posted by fornowagain
I'm definitely going the quad fan rad route in the near future.
It's amazing how hot they get compared to the last gen. That chip must be pushing 250w to worry a 120.2. A PA120.3 should be good for 425-450w with moderate fans. Quad's a safer bet. Mines on 2xPA120.3 the actual loop temp is no higher than a QX9650 same voltage/speed. Even with the EK the core temps aren't brilliant though, 15C hotter core than the QX with the same setup.Quote:
Originally Posted by ReverendMaynard
For ref:
http://img178.imageshack.us/img178/7...flowratfl9.jpghttp://img110.imageshack.us/img110/2...flowratrr8.jpg
It's odd. My ambient is around 22C, my loop is lean with no 90's or 45's, I just flushed my rad, cleaned everything meticulously, and my 4.2ghz HT on load using blend is cresting at +80C. You think it could be the Zerotherm TIM that I used? It worked great on my E8600 @ 4.9ghz 1.56v but now I'm skeptical. It seems the only thing that I can pinpoint as the weak link.
For water with i7 at 4.2+, HT on and with cpu v at 1.5 to 1.55v what should be an expected steady state cpu temp at full load?
On air at 4.1 - I am at 97C!
That high of voltage, with a livable ambient (between 19C and 24C) expect 80-90C.Quote:
Originally Posted by SteveRo
GTZ has a really nice mount, can't see it being that with a backplate. I mounted mine about 4 times (turned it 90, see if made any difference), the EK is kinda bowed. Had fairly good contact, doesn't go anywhere near the edges though. I wonder if these blocks are good enough. I'm guessing you've checked contact? Never used that TIM, tend to use MX2.Quote:
Originally Posted by ReverendMaynard
Thats a lot of volts even for water. Temps would be high.Quote:
Originally Posted by SteveRo
My set up is in the basement of an air conditioned house - I am thinking of getting the swiftech 220 apex single loop kit and find a good 4x radiator to go with, any thoughts?
It's volts that determine heat though, not clock right?Quote:
Originally Posted by ReverendMaynard
I'd say let's compare temps but have you done any testing with HT off yet, because i haven't done any with HT on (won't be using it)
Right now with [200x21] vcore 1.406v set in bios, 1.376v reported with droop during blend, 1.42v VTT and 1.78v vdimm i'm @ 52-55c on cores (all else stock). I listed vdimm and VTT because they no doubt contribute to cpu temp since they're all located on the cpu now.
With same settings but bios set vcore 1.49v, reported with droop @ 1.456v, i top out at 59/60c during blend, again HT off. Will do some testing with HT eventually to see how much temps rise out of curiosity.
my new one can handle higher uncore freq. :)
max. with my D9JNL, but i think it is the uncore limit :D
http://www.abload.de/img/ddr3-22203k1c.png
http://www.abload.de/img/memory-2304iq8u.png
http://valid.canardpc.com/show_oc.php?id=474139
1.344 - 1.360vcore, 1.375 VTT, 1.86v PLL and 1.66vdimm which is stupidly low for these clocks. I use HT on primarily because every benchmark that I've ran gets a boost from it. That's pretty much all I do with this rig; bench and game.Quote:
Originally Posted by Xello
Xello - yes HT off is much lower temps and high clocks (for me anyway) - see link in my sig.
I think I can do better with water but Reverend's temps (80+C) are troublesome.
Dude show some everest :D:up::yepp:Quote:
Originally Posted by Chri$ch
Same here with gaming, but one thing is for sure no game is going to benefit from an extra 4 cores yet! In fact i read somewhere that HT enabled was hampering performance in some games, no sure if true.Quote:
Originally Posted by ReverendMaynard
You get 80c+ temps with those volts? That's almost unbelievable, even with HT on that's 30c+ higher than me at those volts..... Someone told me HT will add 10-15c only?
Please don't buy a kit, they suck. That one will have an old GT block. Best to read up and ask lots of questions to get something more suited to your system.Quote:
Originally Posted by SteveRo
Its both, voltage and clock determin the power of a cpu.Quote:
Originally Posted by Xello
The Formula is P=C*V^2*F.
Since voltage is squared its the bigger factor then frequency. :yepp:
you know that the H20-220 is just a pre-fabbed kit with all of Swiftechs best components right? Hardly call it crap, and the new versions do come with the GTZ. The pump is best in class.Quote:
Originally Posted by fornowagain
Bench2die what do you have for video cards?
Because that average seems impossible.
I know it well enough. I mean generally, they don't usually match what you can get separately or it may not be entirely suitable. And if he doesn't know what he's buying there's plenty of stock around with the old Apogee. Personally I'd look at a PA120.3 (or Feser), DDC Ultra (Petras Top) with an impingement block like the EK Supreme. I tried a high flow D-Tek, gain a few degrees with the EK. Triple rad for a little more headroom with slower fans and the DDC 3.20 to match the more restrictive blocks, but a D5/MCP655 is fine.Quote:
Originally Posted by ReverendMaynard
I think the only thing that holds the H20-220 back is the rad for i7's but then again not allot of blokes out there are nailing 4.2ghz with HT on.Quote:
Originally Posted by fornowagain
I'll do some 4ghz, HT off tests when I get home and see what the deal is. Also, I'm changing this TIM. I have a feeling it's not up to the challenge.