As stated in this thread ([Fr3ak] Biostar Tpower I45), I had trouble with the Biostar Tpower I45 to get past 600 MHz FSB. I couldn't even boot at 600 MHz into Windows. I tried 4 different E8500, one of them with my single stage, but the result was always the same. Crash between 550 and 595 MHz. I posted pictures of the BIOS and the max. selectable voltages in the thread mentioned above btw.
First, lets have a look at the board first what it looks like when it comes out of the box:
More pictures of the board can be found here:
http://extreme.pcgameshardware.de/at...wer-i45-01.jpg
For the weekend, we had a "little" bench session in Germany called AOCT which stands for AwardFabrik Overclocking Treffen (Meeting). We were between 20 and 30 people doing weird things and benching all over the place.
I didn't have any time for preparations, so I decided to try to break my current personal percentage OC value, which I set last year with a E2140 at a 151.51% OC. To reach that goal, I bough 3 E1200 from 3 different shops. What can I say: One CPU was worse than the other. Couldn't even get 2 CPUs running at the 456 MHz my E2140 runs on plain aircooling.
Ok, so what can I do to spend the rest of te 24 hours? First of all, I desiced to give the E2140 a try again, because I reached the previous (personal) record on P35 without a temp probe. That E2140 kept me busy for about 5 hours. After some time, I got the idea to "insulate" the NB with Armaflex and use dry ice on the stock NB, just for the fun of it. To my surprise, the FSB wall of my CPU increased and I was able to gain another 1.5%. Wow, what an archievment for 5 hours of benching, but it has been fun:
http://valid.x86-secret.com/show_oc.php?id=391707
Here is a picture of my ghetto NB cooling with dry ice:
For whatever reason, the CPU doesn't seem to scale with voltage and temperature din't even imporess it too. I was surprised seeing 516 MHz FSB with a multiplier of 6. I ran the E2140 with 2V which moved the coldbug to -145C, which was nice, but as I said, it was of no use.
After using dry ice:
Right, after benching with the E2140, I still had plenty of time and mainly LN2 left, so I decided to see what FSB I can reach with a E8500. I knew the CPU runs at 575 MHz on air with a X38, so there was hope to get at least past 600 MHz FSB. After the good results with the dry ice and one of the guys copying my method to bench with his Tpower with dry ice too, I decided to do some improvements to my previous idea: I did a better job at insulation everything from the start. The NB was pretty much completely sealed. I then made a funnel for the NB. Being known for stupid things, I now wanted to cool the NB with liquid nitrogen, using the stock NB cooler.
Now the fun could continue:
600 MHz FSB was kind of a piece of cake. It took me some time to try all kind of different values, but running the NB at -186.6C (according to the bad blue K-type probes) and +0.7V VNB / +0,65V VFSB made it all the way up to 673 MHz FSB. I could have probably gotten a little higher than that with more finetuning. I didn't touch DRAM timings at all, which might have been a stupid thing since I had it set to semi decent bench settings.
http://valid.x86-secret.com/show_oc.php?id=391714
http://valid.x86-secret.com/show_oc.php?id=391705
Whatever, I had a lot of fun, everybody thought I am totally nuts, so mission accomplished! The low NB temperature had some side effects, that went even worse as soon as the CPU started coldbugging. If you want to know what on a mainboard is connected to the northbridge, just freeze the NB with LN2 and you will find out! Everything that starts forming ice in seconds has traces to the NB
Enough of the talking: Here are the pictures:
Surprisingly, everything was still working fine - more or less. =)
Finally, I was able to make a longer post in this section again, time is killing me, I would still be busy haveing 48 hours days =/
The NB after the LN2 madness:
Bookmarks