FSB downclock mod on the intel GL960 and GL40 --- useful info for PLL Modders

someone please !
Show on the picture GL960 -- FSB 200 to 166 mod.

 

Why do you want to go from 200 to 166. The GL960 supports 200 already!?

 

Any one with 800MHz will do. I installed a T7300 on a GL40 based laptop. That's said despite the fact that the T7300 has 4 times the T4300 cache the performance of both processors is in pair.

Do you really need to upgrade the CPU? What do you use your laptop for? gaming ? video editing?

Sometimes adding more RAM or putting a faster HDD is a better investment than upgrading the CPU.

 

i' try to experience, because something wrong with CPU
i describe problem here t4200 suspicious
BSel 2 1 0 | Frequency
---- L H L | 200
---- L H H | 166
in order to change fsb from 200 to 166 probably need connect bsel[0] to vcc. I try this with accordance to picture below. where shown two version of connection, represented by gary and orange line.
but they not bring success.

 

there is a possibility that windows stills think that you're using the t2310 instead of the t4200. Please go to the 'device manager' and check that the T4200 is showing under the 'processor' section. If you see t2310 instead, select it and press delete to delete it. This will force Windows to rescan the hardware for CPU change and upload the right drivers for the t4200. You might have to repeat the above process a couple of time. Also you'll have to restart the laptop after you've pressed delete (on the keyboard).

Yes you're correct, to go from LHL to LHH you have to connect BSel[0] to the closest Vcc, but that will force your t4200 to operate at 1.66GHz instead of 2GHz. That's not too much of an improvement compared to the t2310 at 1.46GHz.

If the T4200 is displayed correctly in the 'device manager' and if re-installing windows does not fix your problems, I would advise you to sell the T4200 and use the money to buy a T7300 instead. Performance-wise, the T7300 is slightly faster than the T4200, but the T4200 is much cooler though.

 

thank you for respond.
i followed by your advice about delete cpu from system.
i do it in safe mod.
then restart, and windows detect unnamed cpu by intel brend.
with this it was loading in normal mode. cpu was detected as t4200, and no freeze, Also windows shutdown goes normal, and power off work correct.

Before that I used the system with an incorrect shutdown, and probably windows had a little crash, that led to such troble, but why both OS behaved identically and simultaneously.. ( win_XP and win7 ).

Also yesterday, before read this messege, i I tried experience whith cpu t4200 I had a some succeed with starting windows in normal mod.
But fan not working.
CPU warmed up to 60 C , , after that fan activated only one sec and rotate at full speed, then shutdown.
And also yesterday, was a moment, even when the BIOS fails to boot. i.e. when power button pushed, fan rotated only one or two sec, and automaticaly shutdown.
maybe some trouble exist in bios too ?!

I have also ordered a t7500 two days earlier, because I do not expect that t4200 will work again.

 

Good to hear that you fixed your laptop.

Regarding the T7500, last summer I helped a friend upgrade his CPU. We had the choice between the T7500 and T4300. Performance wise the T7500 and T4300 are about the same but the T4300 is much cooler, and undervolt a lot. Although we had both CPUs after taking advice from the forum members we opted for the T4300. I think we managed to set all the Vcores in RMclock to 0.95v.

If it's not too late you should cancel the order for the T7500 and get T4400 or T4500 or a T6*00.

 

unfortunately it was already shipped by ebay seller. probably no way for cancellation.
i saw mobile cpu's rating and they have following position at present day.
134 - T4500
135 - T7500
although t7500 has a 4 mb L2 cache, but he don't have any advantage over t4500 which have only 1 mb l2 cache. probably the reason is a transistor count or lithography of processor.
else some word about non correct shutdown. The problem was fixed after turn off automatic power management of universal usb hub.

 

I'm using the scores from CPU Mark

T7500: 1253
T4200: 1271
T4300: 1436
T4500: 1581

As you can see the T7500 is in pair with the T4200

 

The T4500 is a cooler running CPU. Only advantages the T7500 would have is IDA mode (so can be dual-IDA overclocked to another 200Mhz on a Dell system or one with modified bios) as well as SuperLFM mode so can run half the FSB to save power running at a lower Vcore. The same features are available in the premium Penryns like The P8xxx and T9xx. Otherwise we see they are pretty much on par when running stock frequencies.

The real advantage the 45nm T4500 offers over the T7500 is overclocking. A 200->266Mhz PLL overclock presents no problems to the cooler running T4xxx/T6xxx CPUs whereas the 65nm T7500 will start hitting thermal limits earlier and may not be able to handle such an overclock.

 

Some days ago I bought a T8300 800MHz CPU and its running with my GL40-chipset. Virtualization also works fine. Thank you all.

 

Hi to all, guys.
You have a very good thread here. Let me ask a question. I didn't understand yet: is x9100 works on 1066MHz with PM965 or not? I have T9300 in my Asus M50Sa right now and want to upgrade it without dealing with any hardware mods. I've already run with 8Gb RAM on it.
Thanks in any advance.

 

A X9100 works on some PM965 systems but requires a 266->200Mhz BSEL pinmod as documented on the first post for it to work. It will then boot at 2.1Ghz instead of 2.8Ghz. Once the system has booted into Windows, you use Throttlestop to unlock multipliers where you may be able to overclock up to 4.2Ghz or the thermal limits of your system, whichever you encounter first.

The E0-stepping X9100 run cooler than the C0-stepping X9100 or OEM x9000 CPUs plus X9100 CPUs tend to be cheaper than X9000 ones. If I was wanting to max out a PM965 system then I'd be tracking down a E0-stepping X9100 cpu.

 

nando4, thx for replay. Actually I'm interested in running x9100 at stock frequency. I read many forums and someone claimed that Penryns can run at 266 MHz in some laptops with PM965. I'm gonna try it. I've already found that my Asus has only one FSB/DRAM ratio - 3:5 (Bios limitation, as I suppose). I tested these combinations of RAM:
1. 2x4Gb 800Mhz - they work at 333,5MHz (5-5-5-15);
2. 1x4Gb 800Mhz + 1x2Gb 667Mhz - the same 333,5MHz (5-5-5-15);
3. 1x4Gb 800Mhz + 1x2Gb 800Mhz - 333,5MHz (5-5-5-15).
So, I think that ratio is locked in Bios (I can change nothing regarding FSB/multi in Bios). And I know there is a small chance that 1066MHz Penryn will work at 266MHz bus speed without any mod. And since the ratio is 3:5 I suppose that my 800MHz RAM will operate at 443MHz approximately.
Most of people I asked before didn't try to install 8Gb RAM. But I googled a little, found the same PM965 in another laptops and successful stories of installation of 6Gb. So, I installed 8Gb and it works good. Since Intel officially said that 4Gb RAM is max for this chipset I have no reasons to believe them. I know that they changed the specification of PM965 a couple of years ago. That's why I'm gonna check whether Penryn will work at stock bus speed in PM965 or not. I know that it's possible in theory and I know that I've never seen FSB/DRAM ratio 1:2 (or another) in Asus laptops. Only 3:5. So, I'll receive my P7350 for tests next week and try it. There are only 2 cases possible:
1. It will work at 266MHz (1066 FSB).
2. It will work at 200Mhz (800 FSB as current T9300 do) and lower frequency.
I'll post the results.

 

Well, my laptop didn't even start with P7350 installed. There was no POST sounds at all. Nothing. The power is on and silence. Maybe used CPU is bad or maybe it means that my PM965 can't work with 1066Mhz Penryns. I don't know actually.

 

You need to do a BSEL pin mod to force the CPU to run with a 200mhz FSB, then it will work.

 

thx, moral hazard
I'm not sure what to choose: x9100 with BSEL or x9000 (T9500)? They are almost the same price at ebay. I usually use setfsb and standard cooling system with little add-on of some heatsinks. My current T9300 run with +40-50C at idle and up to +70 at high load. So, I want the best replacement for it.
Maybe my BIOS don't support new 1066MHz Penryns? I got the latest BIOS (ver.205) for my asus.
Can I just add some CPU microcode data info into my original BIOS for support new Penryns? Or it's weird? Asus m50 series has several models, two of them are on PM965 Santa Rosa Refresh and 4 of them - on PM45 Cantiga (native 1066 Penryn support). So, can I just copy some CPU related strings from original BIOS of Cantiga model and add it to my BIOS?

 

Moral Hazard pointed out that you must do the 266->200Mhz BSEL pinmod as covered on the first page of this thread to have any chance of the 1066Mhz FSB P7350 CPU working in your 800Mhz FSB PM965 system. Did you do that? The CPU doesn't automatically downclock from 1066Mhz to 800Mhz for you.

If you can get the P7350 working with the 200Mhz BSEL pinmod then the highest performance upgrade possible would be a E0-stepping X9100. If not, then a X9000 is the highest performer. The X9100 can run up to 20 degrees cooler than a X9000 at full load at same clocks. That means the X9100 can overclocked further before hitting Tjmax thermal limits. HDX9000 owner's have X9100 CPUs running at to 4.2Ghz whereas X9000 would go to a max of 3.8-4.0Ghz.

 

nando4, here is it I suppose. I used aluminum scotch covered by usual black scotch instead of copper wire.
So, I need x9100 SLGE7 or SLB48 next, right?

High temp of P7350 goes from not accurate heatpipe connection as I think. I'll try to correct it. Or can be P7350 so hot?

Also, can someone tell me is it normal behaviour of intel speedstep? 2693 Mhz and 13,5 multiplier? It didn't raise higher than 12x before I start experiments with different setfsb and clockgens... And now I can see 13,5x sometimes.

 

Oh good. Your test confirms a 1066Mhz Penryn can boot and run OK when BSEL pinmodded to 200Mhz operation. So you could now consider a X9100 E0-stepping for highest performance.

Your screengrab is of a T9300. They run at x12.5 and 13.5 in IDA mode. IDA mode allows an extra core to run at one full multiplier faster. There are some dual-IDA bios that allow both codes to run that extra half or full multiplier, depending on the CPU. Since you can overclock your system using setfsb, another option would be a T9300/T9500 + dual-IDA bios + setfsb overclock.

 

Thanks.

I used setfsb several times, full stable CPU frequency was ~2880MHz. After it - BSOD. After 3-4 BSODs I noticed that CPU-Z shows me 13,5x sometimes. Previously I saw only 12x as maximum (I did screenshots). The BIOS or CPU-Z versions didn't changed during the experiment, but version of setfsb did (I tried several versions and found the best one that works with my ICS9LPR363 correctly). So, this 13,5x is rather strange for me. I also have n270 Atom running at 2200 Mhz (thanks to setbsb) and OC potential of T9300 (from 2500 to 2800) is not so good as in case of Atom (1600 - 2200). So, I think I'll try to get x9100. Thanks for the help nando4.

UP: T9900 SLGEE vs X9100 E0 - pros/cons? Does it have the same OC potential?

 

Hi kaza,

Thanks for sharing your experience with us. I've updated the first post to show that this mod works on the Asus M50Sa.

Did you by any chance overclocked the P7350 after applying the pin mod to it?

 

Hi naton,
no, sorry. I didn't overclock it because it's running pretty hot (70C at idle). And I need to return it today (I bought it with money back).
About multipliers of T9300: I found that they initial depend of modes of Power4Gear Extreme (Asus software for power modes). For example, the max perfomance mode boost voltage of CPU up to 1,225 and multiplier up to 13,5. But there are a lot of submodes of this mode. For example, I can choose min perfomance = 0 and max perfomance = 100% for this parent "max perfomance mode". Interesting. I never knew that soft from Asus even affect something.
Power4Gear Extreme thus can affect on working of intel's Speedstep (lower multiplier fixed at 12 in "max perfomance mode").

 

70c is very hot considering that it's a P series CPU. Also I've notice from the screenshot that you posted before that there is nearly 10c difference in temperature between both cores.

Are sure that the heat sink was making proper contact with the CPU?

 

Yes, the heatsink was fixed (original cooling system of laptop). I used new Arctic Cooling MX-3 paste, but I didn't move the heatsink on the CPU surface. Just accurate pressed and that's all. I did't because I was to carefully test it and return to the seller. I've already received my money back. So, maybe there was no proper adjusted contact between CPU and copper pad or CPU was just kind a "buggy" itself.

 

Update: I bought x9100 ($103). Here is the result. It's in idle. CPU hits 75C upon the high load.
And I get BSOD after 3074MHz. Maybe chipset is bottleneck? RAM memory is designed for 800MHz, so 840-860 is almost nothing for RAM...

 

A great price for a X9100. Use Throttlestop to overclock the X9100 using unlocked multipliers and higher voltage up until you hit thermal limits at full load. A HDX9000 users was able to get it to 4.2Ghz.

 

nando4 thx. I have some questions:
1. Can I use Windows Sleep mode while Throttlestop is running? Laptop just shut down when I press the power button while Trottlestop is running. And that's all. If I swith off the battery and CPU and plug them in again the laptop stars. But I can't find the way to go into Sleep mode without these actions...
2. I got BSOD after 3500MHZ using Trottlestop (16 or 17x, 1.2 VID). Which options are optimal for this x9100? When I got BSOD the temperature on both CPUs was about 50C, so, it was ok. FSB is 800 - ok. RAM is 333 - ok. So, why I get BSOD? Thanks in advance.

 

1.2V is not enough which is why you are BSODing. Increase the voltage in Throttlestop against your chosen multiplier. With the higher voltage ensure the CPU is under 100 degrees at full load. I'd say aim for 90-93 degrees max to get some level of safety margin in.

 

Thanks. I'll try some experiments. First, I need to replace AC/DC Adaptor. I have 90W now and I think it's too small. My current stable (=safe) configuration is 16x199,50 @ 1,2375. But after 3500MHz system is quite unstable. Also, without Trottlestop (cool soft), only via SetFSB I raised FSB almost up to 1066. I wonder does it means that PM965 technically can work on 1066MHz? And are there any benefits of using TrottleStop + SetFSB in pair?

 

Thank you, naton, and unclewebb! Successfully running X9100 E0 (QHBQ though, but oh well ) on my 5920G Hello 3,6GHz. Pics later (crappy connection).

 

Congrats mate! Make sure you keep that sig updated. Plus I wanna pics of the fan job

 

Thanks, mate. I was waiting to see how stable the system is, both through stability programs and under actual useful load (games) Have to iron down some things as well, hopefully today I'll upload some pics.

 

Hi guys,

I've updated the 1st page to reflect your successful stories with the X9100.

Keep it up

 

Thanks, naton

Here are the pics:



This was the initial setup. Well, kinda. The top one (-0.2v) wasn't there in the first run, in order to be sure that there is nothing else to cause problems. This setup was for easier PnP - I can remove wires, throw a different CPU etc.



The final setup, it cost me a good 2 hours, but it's well worth it. Wires are completely inside the socket



Up'n'running

Forgot to take pics of the CPU, but I'll take some in the near future.

Also thank you to RickiBerlin, without him I wouldn't even know that it's possible and that X9100 E0 runs THAT much cooler

Since it's related to other tool, here's what it's capable of.

The higher clocks wont be possible without my Cooling overhaul.

 

Let me take you to strawberry fields...
bum

 

intel Core 2 Quad cpu's have a 45 tdp,
gl960 designed to 35 tdp.. isn't it ?
so. when we apply bsel mod - 266 to 200 for c2q cpu.
Then, what about TPD, when we do this mod, ?

 

There is a lot of confusion in the above. There is no realationship between DTP and chipset (i.e. the GL960). Knowing the DTP of a CPU is good as it helps estimating if the current cooling system is or is not enough to cool down a CPU. For example, a cooling system design to cool down a CPU with a DTP of 25W will have a very hard time cooling down a CPU that has a DTP of 45W.

Would the Quad work with the GL960?
It is hard to tell. the GL45 uses an FSB of 266, but not all motherboards equipped with the GL45 support quad. A side from the chipset, a motherboard needs to be designed to support quad cores CPUs. I mean that the motherboard needs to have the necessary circuitry to supply enough power to the quad for it to work. Also, the motherboard BIOS needs to support quad cores processors. If the motherboard or the BIOS are not engineered to support quad, then the quad CPU will not work even if the chipset supports it.

 

I need an advise, guys. Here

What I did: I took BIOS from Asus G60VT (has support for gtx 260m), extract PCI Option ROM from it using AMI MMTOOL and insert it into my BIOS for Asus M50SA (has no support for gtx 260m). It seems that it's not enough
Does anyone have an idea? Gtx 260m is working, PCI-E slot is working (I can work with old ATI HD3650 installed in it). Should I change BIOS more correctly?

 

HI I HAVE A XPS m1530 and i would like to try the mod described in this introduction

This thread explains a BSEL Mod that will allow a latest generation of Core 2 Duo processor such as the T9900 to work in a laptop equipped with an old chipset such as the Intel GL960. The mod forces the latest generation of Core 2 Duo to operate with an FSB of 800MHz (4 x 200MHz) instead of its native 1066MHz (4 x 266MHz). Applying the PLL mod as explained by moral hazard here after applying the BSEL Mod should allow re-increasing the FSB back to 1066MHz.

IS THERE CLEAR INSTRUCTIONS ON HOW TO GET IT DONE ON MY DELL THE ONLY THING I HAVE DONE IS I TOOK IT APART ONCE, REMOVED MOST EVERY THING CLEANED THE FAN SO SCRAPING IS NOT A PROBLEM, AND MAY BE A LIST OF TOOLS AND WHAT I WILL NEED TO GET IT DONE .THANKS FOR THE HELP

 

Hello to u all you guys are great i have been read the threads and i would like to know if this will work i would like upgrade my m1530 by adding a 512 ssd hard drive 8 gb of ram they have CL@ 5 & 6 WHICH SHOULD I GET. IS THERE A BETTER CPU THAN THE P7350 FOR THIS TYPE OF MOD
AND OC BOTH MY GPU AND CPU I WOULD REALLY LIKE SOME HELP HERE GUYS SO I WOULD KNOW IF TO BUY THE STUFF OR NOT CAN I JUST GET A BETTER GPU AND OC THAT INSTEAD

 

Not sure if this matters to anyone but on my Dell Precision M6300 laptop I was testing if it would be compatible with an X9100 with a P8700 cpu from my job before I bought. It is but I did discover something on accident. You can make these cpu's run @ 133 mhz instead of 200 by simply moving the 2nd row that needs to be connected 5 pins in instead of 4. I accidentally put it in the wrong hole and didn't see and it booted but I noticed my speed was really low. Went in cpu z and saw I was at 133 mhz and then I took it apart and saw it was one over too many. Looking back at that chart in OP it makes sense. There is another BSEL pin next to the one for 200, that must be for 133.

So for even older boards you may be able to run penryn chips it sounds like.

 

I realise this thread is just over a year old now but I just made good use of it and got my Dell D830 to run an X9100 at 3GHz.



Unfortunately I didn't get the much preferred E0 stepping however I got the C0 at a good price to keep this machine running for a few more years

I was previously running a T7300 at 2GHz. Windows performance is up from 5.5 to 6.5 in CPU and Memory.

One core runs hotter than the other which sucks, may need to re-apply the thermal paste or get something better.

Thanks for the thread :thumbsup:

 

I got a successful case too.
My laptop is HP Compaq CQ61-421SA, with GL40 chipset, original CPU is Celeron T3100 1.9G
I bought P8700 and it originally doesn't work on my laptop.
I think it's due to GL40 is 800fsb and P8700 is 1066fsb.
I followed this thread and used a thin copper wire to short circuit the specified pin holes, and now it booted up and look ok to me.

(Hewlett-Packard Compaq Presario CQ61 Notebook PC)
Summary
Number of Logical Processors 2
Number of Physical Processors 1
CPU #1 Intel Mobile Core 2 Duo P8700
CPU Name Intel(R) Core(TM)2 Duo CPU P8700 @ 2.53GHz
CPU Code Name Penryn
Vendor GenuineIntel
Number of Bits 64
Instruction Set MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, ET64, XD, VMX, SMX, EST
Platform Name Socket P (478)
Revision R0
Technology 45 nm
Original Clock 2533 MHz
Original System Clock 266 MHz
Original Multiplier 9.6
CPU Clock 1900 MHz
System Clock 199.5 MHz
FSB 798.0 MHz
Number of Cores 2
Core #1
Speed 598.5 MHz
Multiplier 3.00
Core #2
Speed 598.5 MHz
Multiplier 3.00
Virtual Technology Supported Yes
Hyper Threading Supported No
Cache
L1 Data Cache 2 x 32 KBytes
L1 Instructions Cache 2 x 32 KBytes
L2 Cache 3072 KBytes

 

Confirmed working with a Sony Vaio SZ7. Upgraded from T7500 to P8600. The system didn't boot until doing this FSB downclock mod, but now runs 10 degrees cooler and with a definite drop in power consumption.

Also did a hardware undervolt as described at http://forum.notebookreview.com/har...90696-bsel-mod-socket-p-explained-photos.html

 

Thank you, naton! As triturbo already proved - Special THX to you! :thumbsup:, the mod is working at an ACER 5920G for an Intel X9100 CPU. I use the mod he posted earlier ( #135). It was difficult for me - as an amateur - to mount the wires. The pictures of the sockets are always looking much bigger than the reality proves.

The X9100 CPU is starting at a level of 11.5 x 200 MHz = 2300 MHz. I can unlock the multiplier by throttlestop. Currently I am running the system without problems at 16.0 x 200 MHz = 3200 MHz with the help of this great tool. The voltage is 1.2000 V.

I have read something about the different methods to protect the other PINS from being shortened, using insulating tape and also heard about adding nail lacquer for isolation. My next idea was to use isolated wires of an electromagnetic coil. [...] Finally, I decided to add two small isolated wires and removed all filaments excepting one. I've connected the PINs as given in post one of this thread, but added also the voltmod connection of triturbo. The isolation looks maybe a bit too big, but it works fine and the PINs are fitting between the wires. - BUT: I connected another VCC PIN to VID[4] in comparision to triturbo. - I hope that makes no difference. You can decide, if my solution is okay, no guaranties, your risk, be careful:




Another question: How can I check, if the voltmod is working? - I've just added the wire as used by triturbo AND posted earlier by naton "connecting ViD4 to the closest Vcc to force it to be High (1) all the time": http://forum.notebookreview.com/har...-core-2-duo-pin-mod-part-2-a.html#post4663019. I do not even know, if that second wire is needed at all for a proper work of this setup.

EDIT: Some info about the temperatures can be found at:
UPGRADED the CPU to an Intel X9100 @ Acer 5920G

 

Congrats for the upgrade Which revision you got C0 or E0? Also, I used the second wire only after I checked the stable voltages at different multipliers. This way I know that, if (just an example) at x18 it was using 1,2750v, now it's going to be 1,0750v. I'll remove it though, as I think it cause some kind of interference or something and it's not working properly - means my idle temps are almost the same and the CPU is locked to the max multiplier (x11,5) most of the time (when I'm not using TS).

 

Thanks triturbo! Indeed, I was able to get one of the rare E0 CPUs. I mounted a second cooler today: X9100 @ Acer 5920G - Simple Cooling Mod. The CPU (and cooled GPU) is running stable at 3.2 GHz and a voltage of 1.2000 V (read by throttlestop). Does this mean, the real voltage with the wire is:
1.2000 V - 0.2000 V = 1.000 V, because of the second ViD4<-->Vcc wire OR does throttlesop still read the real and correct voltage?

Sorry, I don't want to open the system and remove the wire at the moment, to check that. I am just tooooo happy the system is working properly now. BTW: I always need to load throttlestop at start to get the x16 multiplier. Without throttlestop the multiplier is (max.) 11.5x200 _MHz _= _2300 _MHz. The given throttlestop 1.2 V setting is the lowest voltage for the multiplier of x16 to get a stable system with my CPU. What's the answer?

 

Question.
Do these boards have extreme cpu enabled in the bios?
I assume they do..

 

^ I guess so. I don't know what kind of microcode sets are in the BIOS.

Congrats for the find then, it's a rare one for sure What I meant by saying that it's locked at x11,5 is that it doesn't downclocks when in idle as it used to be.