Quote:
Originally Posted by Thaipo
Awesome writeup. Will be trying this within the next week or so (hopefully today).
My specs are as follows:
Acer Aspire 5920g
t5250 (1.5) Merom Core CPU
4gb of zSkill 4-4-4-8 RAM
8600M GT gpu
The PLL with the data sheet is 9LPRS365BGLF. Will most likely start with method 1 in much the same manner and go from there.
Attempts to use setFSB have just given me a bit of a head scratching; changing values within the main sliders and hitting 'Set FSB' force the sliders back to their original positions. Attempting to change settings in Customize, although stick after 'Set FSB', didn't appear to have any effect on the system (checked with rivaTuner, cpu-z, OCCT, and just maxing out all values expecting a BSOD, to no avail.)
As an added note, within the setFSB utility you can see the smBus read out under Diagnosis. Is this the same as described in method 0 for linux machines?
(ie: my read out:
40 00 08 95 D3 03 00 05 00 44 00 00 02 00 07 07 )
Can this be reliably used to check for the TME variable? |
Success is had. Although the diagnosis from setFSB still shows the 9th register as 44, I've confirmed with CPU-ID FSB overclocking successful with SetFSB.
Again, Acer 5920g with 1.5ghz CPU, successful FSB overclock using method 1 of the PLL mod on an ICS9LPRS365BGLF.
Two things to note here: 1) I had to take the motherboard *completely* out of the plastic casings. Was lucky to be able to leave the cpu and gpu heatsinks and fans in place. The PLL chip is directly underneath the plastic bridge beside the hard drive.
2)In order to get that 10k resistance, there are two resistors coming out from pin 4 on this board. After finally giving up on re-soldering the second resistor back on to the board, I ran a wire straight from where the resistor would contact the board, to the ground of the ac plug. The only issue after that was getting the wire in a position that wouldn't get knocked around on reassembly.
Final setup is pin4->resistor1->where resistor2 would make first contact to board-> wire-> ac ground. *Do note here that in order for this to work, I had to make sure there was a solid connection between the AC ground and the ground for the battery. I'm not sure if this would be equal in other boards, but I would hope so.
*OC Results for the t5250 (base is 1500mhz):
Instability when trying to reach above 1626mhz
1607 seems to be the sweet spot (336/760) -- a couple added points to 340 didn't really do anything substantial that's worth the added stress.
The Bus speed is thus clocked from 166.7 to 178.7 (FSB: 333.3 to 357.4)
3dMark06 scores (just the CPU overclocked):
Vanilla Undervolted:
____3070 (1290, 1158, 1253)
Overclock @ 1607mhz:
____3205 (1398, 1265, 1358)
CPU @ 1607, undervolted to .95v, PCIe bus @ 115.7(38.6mhz):
____3314 (1387, 1256, 1349) (best & most stable so far)
***A quick note. After a couple of weeks of use, 1607 became unstable. Down to 1601 for continued testing.***
I managed to get a score of 3338 pushing the PCIe to 117.7mhz, but got artifacts after running OCCT for 15 minutes.
To give an in-game perspective, TF2 tends to run a bit low, even when setting the direct x level to 8.1. After running just the CPU up, it's more than able to keep stable, with a lot less FPS hiccups.
Any suggestions on where/how I could tweak the GPU or CPU to squeeze out a bit more performance?