watercooling :)
Printable View
What waterblock are you using and where did you buy it?
:banana::banana::banana::banana:ty mount with old waterblock but it should work. My waterblock has a plate to push it down and it has a lot of holes and 2 of them work but the screws have to be a bit loose so they are not entirely vertical.
In the meantime I'm gonna look to build some better mounting. Not much blocks available huh and I don't wonna use the stock cooler.
http://users.pandora.be/dajefn/wcg.JPG
I'm gonna run it like this for a night and tomorrow it's tweakin time :D
Got myself a testing setup to see whether i7 is ready yet to make it into builds for my customers. i7 940, GA-EX58-Extreme, 3x2GB Compustocx DDR3-1600 low voltage sticks (actually I have 2 different sets with different chips, gonna try which one's better), 2x Compustocx 64GB SSD.
All parts slapped on the table for now, using air cooling (Noctua U12P with 2x P12 fans) I got the chip to run primestable at 4Ghz using 1,386V Vcore (well, 1 hour SmallFFT so it should be rather stable). This is a mediocre result for a 940 I was told, I hoped to achieve 4Ghz with less vcore but still... Passed 1 hour of prime blend as well, BCLK of 182 seems to be pretty easy on the setup, as I only had to raise QPI VTT and ICH Core by 0,1V.
Bios still has a few bugs, but overall it's working great and the performance at 4Ghz is really impressive. Gonna be watercooled of course.. temps are in the 75C range during prime, a tad too high for my taste.
Go for it guys!
100 Nehalems added to the team can do over 2.5 mil a day.:up:
Effectiveness of i7 HT on WCG projects other than HCC - ?
Over at WCG, http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=22295#192664, Community Advisor JmBoullier recently posted results of some experiments with HT on his 2.4 GHz P4, running HCC. He found a large throughput increase with HT.
Conventional wisdom was/is that on P4s, HT gives worse performance on WCG. Perhaps that only applies to projects other than HCC. He suggested that HT on the P4 may be so effective on HCC because HCC uses mainly integer calculations, while other projects use more floating-point.
MovieMan: Before lots of people who crunch projects other than HCC rush out and buy i7s, could you please try out the effect of the i7's HT on projects other than HCC, currently FAAH, HPF2 and NRW? No disappointments, I hope.
(Because of the variability of the length of FAAH WUs, the best way of comparing crunch speeds would be on points awarded per hour, after eliminating the WUs that had a quorum greater than 1. An average of about 10 WUs would do ... I suspect a bias of more pts/hr for longer WUs tho. Or do my data-snapshot + rerun-same-WUs trick as per Post #63 above.)
PS: I did some testing of XP-32 vs XP-64 on WCG projects, and have updated my previous post, #63 above http://www.xtremesystems.org/forums/...5&postcount=63. - Rick.
[Update, 16 Dec 08] Over at WCG, BE04642 in thread http://www.worldcommunitygrid.org/fo...ffset=0#195901 posted some data re HT with WCG on his P4s. He says that running a FP-intensive project like FAAH with an int-intensive one like HCC or Rice in the other thread works well. You'd have to do a lot of manual assignment of core affinities to achieve best pairing of WUs on an i7 quad tho. Mods to BOINC are needed.
Lookie lookie what was in the mail today:yepp:
http://i33.tinypic.com/8wf3tx.jpg
Sorry for the tease but for some reason it actually felt good to hurt the cc:p:
Whohaa nice . How high does it clock ? :rofl:
dunno yet, still waiting on mainboard to arrive.
For the people with Asus P6T boards : watch out for 'AUTO' settings and overclocking !!
If you leave everything on auto and start pushing the BCLK , the voltages raise automatically . I've caught my DRAM voltage at 1.66vdimm (:shakes:) , QPI/DRAM 1.3v , PLL voltage also very high IIRC.
There's a nice tool called 'turboV' , that's how I noticed.
Best to set these voltages manually before starting overclocking !
It's pretty difficult and confusing when you're new to core i7. But don't worry , here to help :)
That is, unfortunately, a bad habit that nearly all Asus boards have. I've caught both my P5E and my P5K Deluxe doing that.
Eller
My words exactly, that's what I always warn people about. Sadly, hardly anyone listens to me, they all love their asus "auto oc" mobos :rolleyes:
Even though these values are still quite harmless, especially compared to certain S775 mobos using 1,6V+ VTT on 45nm parts when left on auto :shakes:
jcool , I should mention i was doing very mild overclocks !! only 155 BCLK !!
Don't want to see what happens when you try 200 BCLK with everything on 'auto':eek:
Yeah it prolly goes to 1,7V QPI and 2V Vdimm at 200 BCLK :D
Gonna start crunching part-time with new hardware :) Core i7 @ 4,2 ghz and HT ON :D a real upgrade from an E6700 @ 3,4 hehe
client benchmark says:
Number of CPUs: 8
4002 floating point MIPS
10162 integers MIPS
http://img44.imagevenue.com/loc824/t..._122_824lo.jpg
Finally got my Watercooling back together, after a few minutes I had the above, voltage scaling seems a bit odd with I7, up to around 4 I had to bump it up pretty much linear, above 4 it needs volts like madness.
Yeah I noticed that as well, in fact most i7 CPUs seem to run 3,8 with very low volts but require a lot more to run 4ghz stable. tjelaw seems to have a very good CPU.
I don't see what is so odd with needing large jumps in voltage past a certain point.
I only have one question for you.........Does it Feel Smoother?:ROTF::D
Being 'smooth' is a Phenom feature.
Bloomfield is the opposite , raw brutal power :rofl:
Good article on Anand about i7 overclocking.
http://www.anandtech.com/cpuchipsets...oc.aspx?i=3502