BSEL Mod on a socket P explained with photos

Hi, thank for share. i have AS 4315 too but it comes with CELERON 550 @2.0GHz 133FSB. i tried to BSEL pinmode it to 166MHz and IT DONE.
now i try increasing it to 200Mhz. But now,it comes with BSOD after POST. Sometimes BSOD comes 5 seconds after i logged on.
i am running it with dual chanel RAM 1G Nanya PC5300,and 512M V-GEN PC5300

 

so this is a no-go on intel chipsets, like the pm45?

 

Thanks for the guide.
Tried with T4200, socket P with Nvidia chip set
VID5 connect Vcc didn't boot (-0.4V)
VID4 connect Vcc worked (-0.2V)

Should have tried VID5-Vcc + VID3-Vss.

RMclock voltage 0.975 for all frequency, core temps below 40c (ambient 29-30c). However, in linux temps stay above 44 .

 

I´ve an as 5315 celeron M550 2 ghz, with 166mhz mod works great, absolute no problem. With 200 mhz (800mhz fsb) mod works good, the problem i´ve is bsod only when i launch cinebench r10, that uses opengl, in directx games works great but with no aceleration at all(in 166mhz mode i´ve 10% better performance, in 200mhz the same , i think that its related to the memory speed that its the same at 166mhz and in 200mhz)
And sometimes i´ve minor problems with another programs (shut downs suddenly) that uses fullscreen and a lot of cpu. No overheat problems at all at 3ghz.
What could be the problem? memory tests runs without errors. I don´t think that its the cpu because with stress tests i´ve no problem neither. Could be the chipset? http://valid.canardpc.com/show_oc.php?id=777614

 

Hmmm.. can anybody confirm if this undervolting method works on t7200?

 

yeah, i think bsod comes because my cpu very hot,
in 166mhz cpu idle arround 50c, NFSCarbon load: 75c
in 200mhz i can't measure it. i just touch the heatzink near cpu and its temp increase very fast.

normally, cpu voltage: 1.325V
maybe i should do undervolt. because intel say that celeron 550 works in 1.1-1.25V.
Any suggestion?

 

Could it possibly be your ram becoming unstable?

You can use SPDtool or thaiphoon burner to flash different timmings and a different frequency onto the ram.

 

The ram won't be a problem. the bios will use a different frequency for the CPU and for the RAM

Undervolting will work. FSB mod as you said is a no-go.

Is 0.975 the lowest vcore you can access to in RMclock?

VID5-Vcc + VID3-Vss could wourk, but you are likely to lose access to most Vcores. In my opinion ViD4 is the best option.

yes I think your assumption regarding the memory is right. the memory frequency doesn't change with the mod.

regarding the BSOD, my guess is that your cpu is not getting enough voltage @3Ghz.
I did the mod with a celeron M 530 on an Acer 4315. Result 2.6Ghz. the M 530 was stable under orthos and when I played NFS most wanted.

I have been using a T7200 in my gateway MX3101b laptop since last summer. I successfully pin moded it to run at 2.4Ghz (fsb 200). I removed the mod because under load my CPU was reaching 90c.
I tried to undervolt my T7200 but for some reason I failled. My laptop didn't like the mod. If I lock the multipliers in RMclock to use only one multiplier at the time, my laptop is stable, but if I use all of them my laptop crashes. It crashes at the exact moment when it goes from one multiplier to the next one.

What I'm trying to say is try it. If it doesn't work just remove the wire you've have used for the mod.

Actually I find it weared that you CPU uses 1.325v.
Temperature wise you're OK. The Acer 4315 I was refering to earlier belong to a friend. After moding the Celeron M 530, its owner decided to go for a T7300. The T7300 reaches 76c when stressed with orthos.

 

Naton, thanks for your answer. After some tests i´ve concluded that the cpu needs just very little more voltage. With 3ghz mode in celeron m550 sometimes i`ve bsod (i don´t know why but cinebench its the program that makes bsod instantly, but in cpu stress test usually i dont get bsod.(no overheating,69º max temp) Rarely bsod with high cpu programs. Memtest shows no errors.
Changing cpu vid will help??? i dont know the cpu voltage, everest reports 1.337, rmclock 1,225 hehe I´must change a lot of vids pins to raise a little voltage. Anyway the performance increases only the cpu. Chipset, memory and video doesnt improves anything in 200mhz pinmod compared to 166mhz mod. Also increases opengl speed because uses cpu on the intel x3100.

 

0.975 is the lowest vcore I can access in RMclock with mobile CPU selected. Any other way to lower the vcore?

 

Thankx for the answer Naton. I guess nothing will be sure until I go ahead and test it.

 

I have a T4200 cpu (45nm penryn), but my voltage choice in rmclock or cpugenie is very limited - only voltages betwen 1,1V and 1,15V are selectable. Yes, thats total of 5 voltages - all the settings in programs are correct (ticked mobile cpu,...).

Before i go and open my laptop to vid mod, i want to contemplate my options:
- vid4 mod: will lower voltage by 0.2V, but only if I manually select voltage 1,1V (lowest possible voltage for me via rmclock), any higher selected voltage (in rmclock) and vid 4 mod wont have any effect. Since only the lowest multi is using 1,1V as default, i shouldn have any trubles getting into windows with this mod... But question is, if i can select 1.1V (-0,2V) on my max multi (10X) will I still have stability....? If not, I will set my max multi on 1,125V.

- vid4 mod+ vid3 mod will lower voltage by 0,3V, which is certanly too much for the max multi (0,8V at 2000mhz), but still I shouldnt have any problems getting into windows (only 6X multi is using 1,1v - 0,3V=0,8V at 1200mhz). Then I will try set as many multipliers as possible on 1,1V (-0,3V) in rmclock. Those who wont be stable, will work on 1,125V.

Any other suggestions? Its a shame that my options are so limmited... damn you intel

 

here is an option, overclock the cpu. Then the high voltage will come in handy.

let me know if you want some help overclocking.

 

I thought about OC, but i have a intel GM45 chipset notebook, which support higher FSB than 800, but if i pinmod to higer fsb, im stuck with 6X multi. Also i couldnt find any sutable clockgen, so the only option would be to do a PLL mod, which looks pretty complicated and my soldering skills arent any good + its a new notebook and i dont want to destroy it (at least not right away )

 

Can you at least open up the notebook and find the PLL.
If you take photos of the motherboard I will help you spot it.

Also, I did a PLL mod on my notebook and it's fine.

Now you don't have to do it, but it's good just to look at the PLL and think about it.

As you can see in my post here:
http://forum.notebookreview.com/showthread.php?t=393027

You may get lucky, so the only thing you would have to do is move a resistor like I did.

Anyway, I think it's worth just looking at the PLL, even if you dont mod it now, you may want to mod it in 2 years.

 

Update: I did vid4 mod, its stable on all multis - even 10x200@0,9V.

Is there any way to confirm mod is a success, except lower temp? Coretemp is showing the same voltage before and after the mod, but I did one test:
- setting voltage at 1.1V (cpugenie) - 0,2V (real voltage 0,9V), start temp 50 degrees and running stress test for 1 minute -> final temperature is 65 degrees
- seting voltage at 1,125V (real voltage 1,125V), start temp 50 degrees and running stres test for 1 minute -> final temperature 80 degrees

OH .. I forgot to plug in cpu fan connector, thats why temperatures are so high. Disasembling notebook is a real pain in the - I had to get off every single part to get to the cpu I dont think I will be doing that again - cpu will be a little hot without a fan (but at 0,9V at 2000mhz the temp shouldnt exceed 80 - tmax is 105). It idles at 40 (0,9V@1200mhz). I will post pics of my notebooks mobo later, maybe moral hazard will spoot the name of PLL.

Edit: pics of notebook motherboard: http://fantek.shrani.si/?cq71 (sorry for large size and bad quality)

 

Hi
I don't see PLL placed on this side of mobo. Ther'a no ics looking like clock generator. It usually have 64 pins(squared or rectangular) and oscilator nearby. Can You provide photos of other side of mobo? I think that's the place where we should looking for.

 

Actually I think I see an RTM PLL in the photos, but they are taking too long to load for me right now. I will check when I get home.

 

I think i killed my cpu My fault, I am so stupid

Insted of doing vid4+vid3 mod, i done vid3+vid5 mod (i just realised that). Is the speedsteep working from the start (before going into windows?) - if it is ... than my cpu got a full 1,5V, but is that enough to kill it? I got my desktop 45nm celeron working at 1.6V 24/7...

Now I cant get past boot screen (compaq)- its barely visible (backlight off), no key is working (cant get into bios or anything), fan is working ok, drive also. I tries to remove 1 ram stick, remove hdd, dvd,... nothing helps.

Can it be something else wrong (not the cpu)?

Pics of wrong mod: http://www.shrani.si/f/1U/jF/3DgOcZfU/cpugonebad.jpg
http://www.shrani.si/f/L/Ly/3iWV3acm/img4084.jpg
Thats not blood from me cutting myself for beeing soo stupid, but its the nail polish for insulation purposes...

 

1. undo the mod.
2. remove CMOS battery for 5 mins.
3. Put it all back together.
4. Disconnect and power (power cord and battery) if plugged in.
5. Hold the power button for at least 30 secs.
6. Put battery and power cord in then try to boot again.

I doubt the CPU is dead, but it may act like that.

 

I tried that, but still no change.
I will try to get a secund socket p cpu somewhere, to test it in my laptop.

Edit: Its not the cpu, because i tried t5470 and it also doesnt work
I can see default copaq start screen, but its not backlit, very strange. If I try to hook up external screen via VGA, its no signal. I hope i didnt overcook graphic card (4500hd) which shares heatsink with cpu, and Cpu had more than 80 degrees (with no fan).

 

Just want to report that Bsel mod works with T4200 on a NVIDIA chipset running at 266 with all multipliers.
RMclock 1.13V for 2.66G (266x10) Used to need 0.975V to get 2G (200x10) now need over 1V to run at 1.86G (266x7). Still use 0.975 for 1.6G (266x6).

 

Hey Jeff,

What Nvidia chipset are you using?

I just pin modded a studio 14z laptop with T6500 and 730i chipset, but it doesn't work as it's still at 200fsb. Moreover the VID is always set at maximum now, as opposed to before varying between 0.925 and 1.125

Thanks

 

Mine is a MSI CR400
Chipsets: MCP79MVL

 

I'm not sure if this happened after the BSEL mod.
The ACPI temp is getting very high when full loaded. I stop orthos when it reached 89c. The core temp for the cpus are 71c.
What is the ACPI THRM refers to? Will reapplying better thermal paste help? Any suggestions?

Also, the multiplier seems to lock to either x6 or x10 with not intermediate stepping.

Jeff

 

ACPI THRM is a value based off of the temps of the devices in the thermal zone.

Depending on how your BIOS is coded, usually based on CPU temp.

That ACPI THRM value is the trigger for active cooling (fan) and passive cooling (throttling).

I can modify your dsdt to stop throttling if you want. Also there may be fan code, meaning you could possibly control the fan speeds.

I attached a guide + program to this post.

Let me know if you want to cap the ACPI THRM value in code, so you will not see it go too high anymore.

Another way to solve it is to keep your CPU cool, undervolting and paste. Even a copper mod.

 

Thanks, I did the VID4 mod already. I use cheap paste. But after the BSEL mod the ACPI temp is alarming. When full loaded the temperature just keep climbing up. The fan is at full speed but after a min the ACPI temp went up to 89c. The CPU core temps are at least 10c to 20c lower than the ACPI temp. Just not sure where it's sensing the temp.

I might get some good paste.

 

ACPI THRM is not a sensor value.

It's a calculation based on other temps, used as a trigger.

 

I have done a lot of voltmod/VID combinations with my T7100 and L7700 (Merom CPU's). It seems like the initializing boot up uses:
-specific boot-up voltage
-lowest multiplier
-possibly highest multiplier (not IDA)

Then at initializing Windows it seems to use:
-IDA (or some other high voltage like boot-up voltage yet again)
-highest multiplier, middle multipliers or SLFM depending on the BIOS power settings (at least one of these)

Either way most CPU's probably fail to boot with VID4 or VID5. VID5 especially would be very nice mod if it worked, it only lowers very high voltages to very low voltages, which would be useful for SLFM.

I tried to pass the boot problem by connecting one VID to another VID, instead of VID to VCC. This means that if either of connected VID would be active, the other one would be too. For example, VID5=VID1 would lower some very high voltages to very low voltages, but not all of them. With some combination you could pass the boot-up and it would be almost equally good mod as VID5 to VCC.

Then I tried VID5 mod, but doing the connection after boot-up from outside the computer with insulated wire. This would be a nice mod too, but needs some work and the connection needs to be shut before every boot-up.

Both of these mods worked as expected, but they both crashed later after doing any tasks with computer (i.e not idling or using RMClock). VID to VID mod gave me different BSOD's and after-bootup-mod just crashed.

Naton or anyone, do you have any idea why the mod, with wire connection made after boot-up, crashes? If it works just like the regular mod and it doesn't crash immediately when the connection is made, why would it crash later? Maybe the CPU doesn't like long wires .

 

1st Sorry guys for not getting back to you sooner. I haven't been online a lot lately.

I don't think that everest is reporting the right voltage. 1.337v is to high even for a dual core cpu. I think the 1.225v is the right vcore. If you want to increase the vcore tou have only two choices, either 0.0125v or 0.1v. try this:

edit
see next page for the correct mod

My T7200 @ 2.4Ghz was reaching 90c with arctic silver 5. So your temps at 2.66Ghz are fine as they can't be any lower.
Now, for me 90c is a lot, I mean it is still safe for my T7200 but it is still a lot.

 

I'll try to answer you questions one by one
1- If you can access only values between 1.1v and 1.15v you should stay away from the volt mod.

2- If you want to decrease the heat use a good thermal compound instead such as arctic silver 5

 

to check the success of the mod do the following:
1- take a specific multiplier (e.g. x10)
2- let say that x10 is stable with a minimim vcore of 1v without mod
Selecting 1v after mod will crash the laptop
if the laptop crash, it means you did the mod properly.

 

Hi guys, last Saturday I did the VID4-VID1 mod on my XPS M1210 and I'm getting some mixed results, the problem I have is that some voltages that shouldn't be changed were modified.

For example, before the mod my T7200 was stable at [email protected] V, which is strange because according to the attached picture, it should remain unchanged. It's the same story with the other voltages, [email protected] V now is freezing. Maybe something is interfering with the VIDs which makes this mod not viable.

Thanks for your attention.

 

If you connected VID4 to VID1, those voltages you mentioned should remain unchanged, at least according to my tests. Did you insulate the wire, since VID4 pin hole is in between many other pin holes? I have used Scotch tape to trace the wire, sticking the tape onto socket where the wire is.

Did your mod freeze immediately at those voltages? Did the other voltages work as expected and did you notice lowered temperature at some? I have also tried VID to VID mods if you read my last post, but they have crashed later at any voltage for no reason. Have you read someone having this mod fully working? People in this thread have only connected multiple VID's to VCC or VSS simultaneously but not VID to VID (unless I missed something).

 

Hi zorror, I only used some inductor wire to connect the VIDs, making sure I only peeled off enough to make contact between those holes.

The temps do seem to match the new VIDs, e.g. 1.0125 V is 3 °C lower than 0.950 V at 6x with OCCT, so is 1.025 V. Both 1 V and 1.0375 V temps are higher than 0.950 V.

I think I'm the first to do this mod sort of successfully, maybe this problem is related to yours. The only case I read someone attempted the mod failed too, he was ivantagan from the RMClock forums http://forum.rightmark.org/topic.cgi?id=6:1455.

 

miguelcba21
there was a mistake in my last post regarding the volt mod that you can do. Below are the corrected mods:

 

1- Why did you connect VID4 to VID1? The right connections are:
VID + Vcc to lower the voltage
VID + Vss to rise the voltage

2- The values shown in your Excel picture seems to be wrong. When VID4 is connected to a Vcc this what happens:
0.9500v to 1.1000v become 0.7500v to 0.9000v
1.1125v to 1.3000v remain the same

3- If you T7200 is stable at 0.9500 before mod, after mod you should use 1.1125v to have a stable system. Same for x12. This is because with the mod you lose access to the vcores between 0.9125 and 1.1000v.
[email protected] V crash because 1.0375v after mod = 0.8375v

 

My guess would be that the intel graphics is OK. A lapt6op with a faulty graphics usually turns ON with either a black screen, or most often white screen or a screen with artifacts.
if you haven't fixed your laptop yet try the following (the order is different to the one described by moral hazard):
1- turn laptop off
2- unplug power cord
3- remove battery
4- remove any cpu mod
5- remove the cmos battery for few minutes (sometimes few seconds are enough, but to be in the safe side remove it for 10 minutes).
6- put everything back and test.

 

Thanks Naton, actually i was going to post because i´ve a doubt when i checked with celeron´s datasheet and the mods posted undervolted the processor instead of overvolt. I also read that the max voltage for celeron is 1.30V according to datasheet.
When you checked the celeron 530 that you overclocked with rmclock the voltage was 1.225? I think that maybe its variable within diferent celerons.
I´ll try soon, then i´ll tell you if i´succeed. (i must do it in total peace, hehe no mistakes are allowed, neither static electricity )
I can´t pass 58º core temp with 166 bsel mod, so i think that i´ve a good cooling to make overvolting and 200mhz bsel. What´s the increase of the voltage that you think it´s reasonable?

 

He connected VID4 to VID1 because he was trying to do another mod that has some advantages over connecting VID to VCC or VSS. I tried that too, see my post. Do you have any idea why these mods doesn't work properly (VID to VID mod, or the other mod I tried)?

 

If you tray VID + VID on a celeron it may work since celeron uses one vcore only. But with core duo / core2duo the value of any of the VIDs can be sometimes 1 and sometimes 0 depending on the multiplier selected, and depending on the vcore set by intel for that specific multiplier. In addition a VID = 0 acts like a Vss if I'm not wrong. For instance the T2060 I had uses 1.0250 at x6. This is VID 6543210 = 0100110. If I connect VID4 to VID2, the result would be VID4 = VID2 = 0, i.e. 1.0750v.

hope that the above makes sense

 

I don't remember the exact voltage but it was around 1.25v.

 

But in that case, the CPU would still remain stable because it'd only overvolt those voltages, something that it's not happening. If we assume that a VID = 1 overrides the other one with its value still doesn't explain why 0.95 V is not reacting in the same way before and after the mod. Both VID1 and 4 are 0.

 

But in that case, the CPU would still remain stable because it'd only overvolt those voltages, something that it's not happening. If we assume that a VID = 1 overrides the other one with its value still doesn't explain why 0.95 V is not reacting in the same way before and after the mod. Both VID1 and 4 are 0.

 

To be completely honest I've never tried to connect VIDs together. If VIDs are connected together to Vcc, their value will be 1. If they are connected to a Vss, their value will be 0.

Now from what I read on the web a while ago it seems that the VID with 0 will pull the VID(s) connected to it to 0 too.

Check this out: http://forum.rightmark.org/topic.cgi?id=6:1455

 

Naton, i ´ve interesting results. I tried +0,05 v pinmod (vid2) and the notebook run like a charm , but, i started benchmarking with superpi32 and temp rised very very fast and stabilized in 80º. I didn´t even wanted to try prime95. All cinebench and other problems were gone away.
Then i made vid1 mod, +0,0250 v and the computer runs fine, in superpi32 reachs 70º (superpi32 1m- 23"), prime95 81º in torture tests, but i get some errors in 8 minutes, i think its because overtemp. Maybe you could suggest me a better disipation solution.

 

If you want to keep the 200fsb it seems that you need the vid2 mod

I personaly don't like temperatures above 70c, though my T7200 with Arctic Silver 5 and undervolt, reaches 77 to 78c when stressed with orthos.

If the Celeron is like the Core Duo it will run fine until about 100c/105c.
To improve the situation try a better thermal past such as Arctic Silver 5 (AS5). One thing you should know AS5 needs some curring time (240h I think) to reach it maximum efficiency. Another excellent product is Arctic MX-3 (no curring time needed).

 

Thanks naton for the page, in fact it was one of the first pages I read about undervolting from about a year ago.

I'm going to revert the mod anyway, it seems to be causing some occasional BSODs .

 

I think you're getting the BSOD because you're connecting VIDs together. Try VID3 to Vcc, or VID4 to Vcc before you revert

 

Good News. The problem was that the gum pad of the chipset was preventing a full contact with the dissipator. The temp with vid 2 and 200 mhz bsel increases 3º respect of bsel 166 with no vid mod. And with original mod increases i think 6º. I really dont know because it varies with ambient temp a lot. Full throttle reachs 74-75º now with orthos, but when the weather becomes hotter, increases 4º more maybe i really didnt measure it.
Exellent guide Naton!