The Ultimate Pentium-m Pinmod/Voltmod Guide

[smekta3]

One more question, how much variation should be in the clock after the overclock? My cpu fluctuates from quite a bit, heres a log of .1 second intervals,
19:04:47, 1731.63 MHz, (-0.0%)
19:04:47, 1806.62 MHz, (0.0%)
19:04:47, 1949.77 MHz, (100.0%)
19:04:47, 1850.38 MHz, (0.0%)
19:04:47, 1762.16 MHz, (0.0%)
19:04:47, 2014.61 MHz, (-0.0%)
19:04:47, 1764.74 MHz, (0.0%)
19:04:47, 2097.92 MHz, (0.0%)
19:04:47, 1791.40 MHz, (100.0%)
19:04:47, 1808.98 MHz, (0.0%)
19:04:47, 2044.83 MHz, (-0.0%)
19:04:47, 1819.80 MHz, (-0.0%)
Is this normal, since i havent checked before the pin mod?

 

[Como]

tell me how you checked, ive got a 1.80 @ 1.80, my laptop will run it at 2.4 but its a little short on voltage... ill be vmodding it soon as i get a desktop up, but it will probably be another week so i can check the variation... just dont know how.

 

[smekta3]

http://cpu.rightmark.org/products/rmgotcha.shtml and just change the check period.

 

[Como]

idle 1.80:
started: 04/14/06, 10:16:01 (MM timer is used)
10:16:02, 255.39 MHz, (5.9%)
10:16:03, 209.42 MHz, (0.0%)
10:16:04, 256.23 MHz, (4.7%)
10:16:05, 180.77 MHz, (3.1%)
10:16:06, 240.12 MHz, (3.1%)
10:16:07, 188.80 MHz, (3.1%)
10:16:08, 203.75 MHz, (3.1%)
10:16:09, 239.52 MHz, (3.1%)
10:16:10, 207.90 MHz, (3.1%)
10:16:11, 207.30 MHz, (0.0%)
10:16:12, 239.21 MHz, (3.1%)

LOAD 1.80:
started: 04/14/06, 10:17:24 (MM timer is used)
10:17:25, 1719.72 MHz, (11.4%)
10:17:26, 1676.47 MHz, (89.1%)
10:17:27, 1652.77 MHz, (89.1%)
10:17:28, 1649.78 MHz, (81.3%)
10:17:29, 1758.20 MHz, (87.5%)
10:17:30, 1709.47 MHz, (85.9%)
10:17:31, 1635.90 MHz, (84.4%)
10:17:32, 1690.13 MHz, (84.4%)
10:17:33, 1719.31 MHz, (85.9%)
10:17:34, 1435.24 MHz, (54.7%)
10:17:35, 1603.30 MHz, (87.5%)
stopped: 04/14/06, 10:17:36


doesnt appear it can read my cpu speed at all. notebook hardware control reports a steady 1.80 all the time because its plugged in. lets turn on powersaving and see.

1.80 w/ speedstep, idle

started: 04/14/06, 10:19:46 (MM timer is used)
started: 04/14/06, 10:22:07 (MM timer is used)
10:22:08, 199.67 MHz, (0.0%)
10:22:09, 226.27 MHz, (35.9%)
10:22:10, 372.03 MHz, (17.2%)
10:22:11, 199.53 MHz, (21.9%)
10:22:12, 199.46 MHz, (20.3%)
10:22:13, 199.53 MHz, (17.2%)
10:22:14, 199.46 MHz, (9.4%)
10:22:15, 199.52 MHz, (25.0%)
10:22:16, 199.45 MHz, (21.9%)
stopped: 04/14/06, 10:22:17


1.80 w/speedstep load
started: 04/14/06, 10:21:25 (MM timer is used)
10:21:26, 1715.29 MHz, (0.0%)
10:21:27, 1757.14 MHz, (92.2%)
10:21:28, 1790.28 MHz, (82.8%)
10:21:29, 1718.03 MHz, (87.5%)
10:21:30, 1756.23 MHz, (82.8%)
10:21:31, 1711.86 MHz, (65.6%)
10:21:32, 1638.70 MHz, (100.0%)
10:21:33, 1795.54 MHz, (65.6%)
10:21:34, 1595.55 MHz, (100.0%)
10:21:35, 1776.74 MHz, (70.3%)
10:21:36, 1454.95 MHz, (67.2%)
10:21:37, 1634.95 MHz, (81.2%)
stopped: 04/14/06, 10:21:37


i really dont know. it cant even read my cpu speed properly, so i dont trust it. looks like its reading from windows, which also has a tendancy to report oddball numbers.

 

[smekta3]

Hmm, when i did it i opened up a program, for example prime doing TT,and i set it to 2 sec check period, i get completly stable,

11:20:51, 1858.25 MHz, (9.7%)
11:20:53, 1861.25 MHz, (7.7%)
11:20:55, 1862.06 MHz, (5.0%)
11:20:57, 1862.00 MHz, (4.9%)
11:20:59, 1861.86 MHz, (4.0%)
11:21:01, 1862.14 MHz, (5.9%)
11:21:03, 1862.28 MHz, (13.5%)
11:21:05, 1861.94 MHz, (5.9%)
11:21:07, 1861.79 MHz, (5.9%)
11:21:09, 1862.24 MHz, (5.1%)
11:21:11, 1861.94 MHz, (6.1%)
11:21:13, 1862.20 MHz, (8.2%)
11:21:15, 1861.74 MHz, (10.1%)

Than as i change it to .1 sec interval, i get

11:23:26, 1862.56 MHz, (0.0%)
11:23:26, 1775.24 MHz, (100.0%)
11:23:26, 2002.17 MHz, (68.8%)
11:23:26, 1791.21 MHz, (0.0%)
11:23:26, 1816.83 MHz, (50.4%)
11:23:26, 1901.73 MHz, (0.0%)
11:23:26, 1816.82 MHz, (37.9%)
11:23:26, 1904.67 MHz, (0.0%)
11:23:26, 2019.33 MHz, (29.2%)
11:23:26, 1825.47 MHz, (45.4%)

 

[ozzlo]

my cpu will fluxuiate +- 5mhz on idle(800mhz) and on full load(2.26ghz) +- 10mhz

One thing I noticed is that you guys are using RMclock... I stopped using RMclock because I noticed that the program was not stable for me. RM clock downclocks the cpu kindof differently than the NHC does as rmclock does not change the actual multipliers of the cpu.

as for getting lots of LARGE FREQUENCY FLUXUIATIONS... you must remember that the speedstep in your processor can change the frequency hundreds if not thousands of times per second if it needs to... although this is normally not the case of how it operates there are a couple of things that you can do if this bugs you guys...

there are settings in NHC where you can make it so that it takes longer to switch frequencies but this will make your computer laggy so I don't reccomend it... the max load and min load settings can be modified to make the processor wait longer before ramping up/down in speed also there is also a delay that you can add...

also if you are taking a speed from windows,,,, windows calculates an average speed over an interval and well it just really sucks at it... I only trust NHC for current speed display (I don't even trust CPUZ) if your taking a long interval over a second long you end up with a average frequency over all the frequencies the processor switched through which is my guess why you guys are ending up with oddball frequencies.

also RMclock program has that ability to reduce your speed even further but it's not reducing your actual speed but it's actually (in a way) disabling your processor for very small intervals of time which depending on when you take your reading can make your speed vary wildly...

If you guys want a true fluxuiation figures set up your multipliers in NHC so ONLY 1 frequency can be used (set the min and max multiplier to the same number) This way speedstep can't jump up and down on you and you can get true readings of your cpu fluxuiations... actually the way I measured my fluxuiations is I just opened up NHC and looked at the frequency that it spits out for a couple minuiets...


all in all it's allright if your cpu fluxuiates a little bit, It dosen't hurt anything, that's why my guide sugguested adding a .016v on top of the voltages you found to be stable to compensate for such fluxuiations...

 

[Brinnswift]

5. Decide on a speed that you want to get (remember after the pin mod it will be 33% faster) and remember you NEED A 400mhz processor with 2mb l2 cache.

This might sound stupid, but since it says you NEED a 2mb cache, this mod is impossible for a 1mb cache pentium-m? That's what I first thought, but then there is a user who posted sucessfully doing it with a 1.4 1mb cache cpu. Can someone explain?

 

[ozzlo]

Hmm nice typo...

You are correct sir, You can do the pinmod with a 1mb cache chip as far as I know (haven't personally done it but others claim to of)

Usually if you are looking to purchase your chip (if you don't allready have one) a 2mb cache chip and a 1mb cache chip are pretty much the same price, so obviously you would get the one that peforms better.

Sorry It took me so long to respond but I was swamped by school...

 

[Fluxburn]

You need the 915i chipset... that sucks I have the 855i chipset. So I am guessing no other way exists to modify clock speed on your laptop cpu then this mod. Unless someone can write binary for bios...

 

[Sir. BOBSONATOR]

Wait so it wont work on a i855PM?????

 

[ozzlo]

correct... 915 only

 

[ibrake ifry]

I have the i915gml chipset which doesn't have the 533 mhz memory controller...do you guys know of any way to increase the ram speed. Of course i have a dell bios so i can't modify anything!!! So far i am using rmma 3.65 to alter my timings to 3-2-2-5 which helped a little but to match the FSB would be sweet!

 

[Luie]

I have the i915gml chipset, too.
Maybe undervolt it to get more battery life?

 

[ozzlo]

Yes you can use Notebook Hardware Control to undervolt your laptop... The NHC website has a nice explanation of how to use the program.

 

[betterlife18]

Ok, so I've got a 735 (1.7Ghz/400fsb) and I tried the pin mod, it failed horribly

Even with a volt mod (chose the recommended voltage from the Excel spreadsheet thing)... when I power it up, the fan spins for about 5 seconds and then stops.

Help please!!!

 

[ozzlo]

Ok there's some things I need from you In order to help you out..

I need...

Cpu-z screenshot (if you took one before you started modding)
Which chipset your using
Make and Model laptop
Cpu sSpec# (found on the processor) Example of what the numbers are (left column) http://processorfinder.intel.com/List.aspx?ProcFam=942&sSpec=&OrdCode=
Which pins you are modding
photo of pinmod (if possible)


Sorry If I am asking alot but the more information you can provide the more I can help you out. Alot of pinmod failures can be the result of silly mistakes such as not tightening down the cpu socket after re-inserting the processor (I am guilty of that one myself).

 

[betterlife18]

Ok this is the before CPU-Z shot (and after too, lol, since I couldn't get it working):

It's an Acer Aspire 1642ZWLCi and has the 915GM chipset.

I modded C15 and C16 with speaker wire for the boost to 533MHz. After doing this and it failing, (fan came on for a few seconds, then died)... I moved on to the volt mod...

For a volt mod, I pinned F2 and G2 to obtain a 1.372 voltage as shown by the "Voltmod Interactive Guide" (XLS spreadsheet).

Same thing happened. Sorry I don't have pics of the mod itself. I ordered some OCZ 5+ Thermal Compound and it should be here this weekend (along with a 100GB Seagate Momentus 7200rpm drive ) I'm going to give the mod another try. To be honest with you, I think I remember tightening the ZIF socket ...

Any tips?

 

[ozzlo]

It's possible that you might need to mod VID 2 for 1.404 volts.

another mistake I've made is mistaking the top of the socket for the side of the socket (the mod was oriented 90 degrees off)... this caused exactly what you described, the pc refused to boot because the processor had one of it's data pins shorted and was failing post.

 

[betterlife18]

Hmmm... I'll have to check that out.

After all of that and if it doesn't work still, it is possible that the Acer Bios won't support the 533MHz bus?

 

[ozzlo]

a very very small possibility, make sure you have the latest bios. I'll look into some stuff over the weekend.