How to Enable Intel Dynamic Acceleration (IDA) on Both Cores of a Core 2 Duo

I think TS 2.99 does this automatically for you now.

I also have a D830 with a T8100 and it's happy with 1.05 V in Dual IDA too.

 

meh, just the programmer. :laugh:

As for fulltime IDA it was just an accident waiting to happen and besides yourself there was only one other person who showed any interest in it over the course of ~3 months and that was for a quad with DDA (Dual Dynamic Acceleration). AFAIK he got it working but disappeared so quickly I couldn't get any feedback about it. I think if it wasn't for your commitment to providing great software then probably DualIDA would be little known about today so really you deserve all the credit and appreciation for doing that.

Regarding veloscaper's CPU, I can not see any problems with the CPU registers or why it should not work with TS. The only thing I could think that might be different from TS and the program I gave him to test is forcing idle states so possibly it could be the higher idle states are not used by the OS and hence no IDA. Other than that idk.

 

Hi guys,

I detected some issues with Phoenix BIOS when trying to rebuild the BIOS image (for dual-IDA mod). Please visit post #256 for more information.

 

How can i use this modded bios file. I have same laptop and bios, i think it will work on my laptop to.

 

Nevermind, i figured it out and is working now.

Thanks to everyone

 

FWIW, my 8530p w/ P9600 cpu is working in dual IDA mode with TS. Not sure why but it just started to work today.

 

hi guys has anyone with p8400 on hp dv5 has successfully modded bios to enable IDA on both cores as currently EIST is grayed out in mine?
thanks.

 

dualIDA bios for the DV5 doesn't exist as yet. You could be the first to mod the DV5 bios. Just follow kizwan's excellent instructions here.

 

^i will look into it,thanks by the way however i have heard that enabling slfm cuts the frequency in half but in my case both cpuz and throttlestop are showing 266 x 8.5 so have i understood something wrong?

 

farsang: If you are using ThrottleStop to control your CPU, you will need to use the Power Saver feature and have SLFM checked before you will see the bus speed get cut in half.

If you are using Windows to control your CPU, you need to make sure the Control Panel - Power Options - Minimum processor state is set to a low number like 5% to give your CPU a chance to idle down and go into SLFM mode. If you want ThrottleStop to be in control then set this to 100%.

 

And if you do figure out how to do that, see if you can enable the advanced menu in the BIOS. Would love to have the option of selecting IDE legacy mode.

 

HP Pavilion dv5-1000 series Entertainment Notebook PC (Intel Processors)

WLAN whitelist check "removed" + SLIC 2.1 + no_EIST

 

I've been running this for a couple of days in dual ida mode on an m1330. t7500 but today the cpu maxed out and everything slowed down. I rebooted and even that was really slow so i thought i'd damaged it. Eventually it returned to normal and seems fine now. Could this have been caused by throttlestop? I read of the risks involved but i thought it was relatively safe

 

What exactly does that mean? What sort of load were you running and how high did the core temperature get? I've never heard of an issue like this before while using ThrottleStop and after you reboot, everything should have been back to normal. You might have a software problem that isn't related to your CPU.

Turn on the ThrottleStop Log File option so next time this happens, at least you will have a record of what went on.

What options did you have checked in ThrottleStop. Many Dells have clock modulation throttling issues so its possible that the CPU got too hot and triggered that. Post a screen shot of how you have ThrottleStop setup so I can have a look.

 

The whole system became very slow and unresponsive. Opening a window or menu would cause the cpu to jump to 100%. This was with nothing running as i thought initially it could be due to video i had been streaming but it continued upon reboot. Core temp was around 52c and the gpu was around 77c when idle. I think i have a log so i'll upload that and a screenshot. Cheers.

 

Here's the screen shot:

I'm having trouble uploading the log but if you tell me what info you need i'll cut and paste it here.

 

ThrottleStop 2.99.7 has some improved Dual IDA features.

http://forum.notebookreview.com/har...et-upgrades/531329-throttlestop-guide-37.html

tjphillips: Most Dell laptops use one of two types of throttling which can be triggered by core temperature but can also be triggered by excessive power consumption in some models. I don't know everything there is to know about your model off hand but it is very likely that you triggered this type of throttling. Once this has been triggered, it can last for quite a while, even after the core temperature returns to normal. Some laptops took something like 15 minutes before they were allowed to return to full performance.

If you want to read a thorough example of this issue, check out tinkerdude's paper about this issue. It's a huge 24+ MB download but explains exactly what happens during a throttling episode. It motivated me to write ThrottleStop.

throttlegate.pdf

He had a lot to say before he quietly disappeared from the scene.

When small windows are sluggish to open or drag around your screen, that is usually a sign of this throttling problem. CPU performance can be instantly reduced to as little as 10% of its rated power. That's why it takes 100% of the CPU in the Task Manager to get your CPU to do any meaningful amount of work. The C0% in ThrottleStop is a much better load meter when a CPU is being throttled like this.

If you have a log file, look in there for signs of throttling. The CKMOD and CHIPM columns in the log file show the two types of clock modulation throttling that Dell likes to use. These columns should both be at 100% unless the heatsink falls off but in some models, Dell was using throttling to control maximum power consumption.

Power consumption in many laptops went up and up as faster CPUs and GPUs were used but the power adapter rating was not keeping up with this. The XPS 1645 that originally shipped with a Core i7-720QM and an ATI 4670 GPU only had a pathetic 90 watt adapter. Dell has finally come to their senses so this is rarely an issue for the 2010/2011 line up but before this, numerous models had issues going back a couple of years and some laptops were never properly fixed.

One way to reduce power consumption on your CPU is to lower the core voltage. That can help reduce heat and power consumption significantly, especially in the older CPUs like you have. If you go too far, this can cause your computer to lock up and you will need to reboot so don't be working on anything important when adjusting this. With each step down in voltage, do some testing with Prime95 or a similar program to make sure your computer is 100% stable before trying to reduce the core voltage some more. Lowering heat and power consumption is a good way to avoid both causes of throttling. Head over to the ThrottleStop Guide thread in my sig if you have any further throttling specific questions. Dual IDA looks like it is working correctly on your CPU. Just keep in mind that you need to lower the VID voltage before you enter Dual IDA mode.

 

Hey Uncleweb been trying your throttle stop on my M6500 i7 940xm 2.33 should run at 3.+ but rarely seems to get past 3.1/2 on a x24multiplier - have 8gig of ram and a raid 0. Could really doing with sticking some life into this, I know its a laptop but still! do a bit of FSX and its ok but not what is should be!, any advice? something I noticed is that the core voltage is greyed out in cpu-z?

 

I'm back after using throttlestop extensively for a few weeks. The problem hasn't returned since. I'm not sure what the cause was but it sounds like the built in throttling was triggered from your description.

I use my system for audio and was about to upgrade my processor but i was having a recurring issue where the cpu would spike randomly which i think was due to the speed throttling. Since using throttlestop i'm no longer having this issue. Cpu peformance is much better and cpu temps are negligible.

I've set up profiles for performance and internet and my system is running better than ever. I would have needed to upgrade at the time i discovered this neat little app so thanks, you've saved me the cost of an upgrade. Plus there is the added performance gains when i do upgrade.

This is possibly the best piece of software i've come across for audio on laptops. Rep added for a great piece of software! Cheers.

 

tjphillips: I'm happy to hear that everything is working great now. Some of Dell's various throttling schemes can be extremely frustrating when you don't know what's going on or how to prevent your laptop from slowing down to a crawl.

holgate: Click on TRL in ThrottleStop and open up the Turbo Ratio Limits window. What's hiding in there. The M17x-R2 sets these all to 25 when using a 940XM. You can also click on the TPL button and crank up the Turbo Power Limits. The 940XM can be significantly overclocked using ThrottleStop but you have to keep an eye on the core temperature which is usually the limiting factor. Do some testing with wPrime or a similar benchmark testing program. As you increase the ThrottleStop settings I mentioned, you should see a significant increase in your CPU performance as shown by some improved wPrime times.

 

New entry by Lenovo added to the dual-IDA C2D bios list:

Great work by middleton +rep.

 

Hi kizwan.

I have some comments on your "EIST Unlocked Example" guide.


1) First of all: You make a fundamental error providing a code like this in the guide:

Code:

0000XXXX B9A0010000	mov ecx, 0x1a0
0000XXXX 48		dec eax
0000XXXX 0FBAE814	bts eax, 0x14
0000XXXX 48		dec eax

Actually it's a 64-bit code and therefore it must be disassembled as follows:

Code:

0000XXXX B9A0010000	mov ecx, 0x1a0
0000XXXX 480FBAE814	bts rax, 0x14
0000XXXX 488BD0		mov rdx, rax

You are a very lucky man because patching BIOS according to wrong disassembled code is very dangerous procedure.


2) I can't understand why you recommend to patch a code like this:

Code:

00003CCF B9A0010000	mov ecx, 0x1a0
00003CD4 48		dec eax
00003CD5 0FBAE814	bts eax, 0x14
00003CD9 48		dec eax

in that manner:

Code:

00003CCF B9A0010000	mov ecx, 0x1a0
00003CD4 90		nop
00003CD5 90		nop
00003CD6 90		nop
00003CD7 90		nop
00003CD8 90		nop
00003CD9 48		dec eax

Can't you patch just one byte instead of five? For example:

Code:

0000XXXX B9A0010000	mov ecx, 0x1a0
0000XXXX 480FBAF014	btr rax, 0x14
0000XXXX 488BD0		mov rdx, rax

3) Recently I made dual-IDA patches for nine Lenovo ThinkPads (see here model numbers). We need to modify BIOSCOD5.ROM. Here is the code:

Code:

seg000:4264 0F A8                                   push    gs
seg000:4266 66 50                                   push    eax
seg000:4268 66 53                                   push    ebx
seg000:426A 66 51                                   push    ecx
seg000:426C 66 52                                   push    edx
seg000:426E 68 00 F0                                push    0F000h
seg000:4271 0F A9                                   pop     gs
seg000:4273 66 B9 17 00 00 00                       mov     ecx, 17h
seg000:4279 0F 32                                   rdmsr
seg000:427B 66 F7 C2 00 00 04 00                    test    edx, 40000h
seg000:4282 74 6D                                   jz      short loc_42F1
seg000:4284 66 B8 01 00 00 00                       mov     eax, 1
seg000:428A 0F A2                                   cpuid
seg000:428C 66 F7 C1 80 00 00 00                    test    ecx, 80h
seg000:4293 75 23                                   jnz     short loc_42B8
seg000:4295 66 B9 A0 01 00 00                       mov     ecx, 1A0h
seg000:429B 0F 32                                   rdmsr
seg000:429D 66 25 FF 7F FE FF                       and     eax, 0FFFE7FFFh
seg000:42A3 66 0D 00 00 10 00                       or      eax, 100000h
seg000:42A9 0F 30                                   wrmsr
seg000:42AB 65 83 0E 2F 33 04                       or      word ptr gs:332Fh, 4
seg000:42B1 9A B1 27 00 F0                          call    far ptr 0F000h:27B1h
seg000:42B6 EB 39                                   jmp     short loc_42F1
seg000:42B8 66 B9 94 01 00 00                       mov     ecx, 194h
seg000:42BE 0F 32                                   rdmsr
seg000:42C0 66 A9 00 00 02 00                       test    eax, 20000h
seg000:42C6 75 29                                   jnz     short loc_42F1
seg000:42C8 9A B1 27 00 F0                          call    far ptr 0F000h:27B1h
seg000:42CD 66 BB 00 04 10 00                       mov     ebx, 100400h
seg000:42D3 B8 0F 03                                mov     ax, 30Fh
seg000:42D6 9A 37 3C 00 F0                          call    far ptr 0F000h:3C37h
seg000:42DB 74 07                                   jz      short loc_42E4
seg000:42DD 66 81 CB 00 00 01 00                    or      ebx, 10000h
seg000:42E4 66 B9 A0 01 00 00                       mov     ecx, 1A0h
seg000:42EA 0F 32                                   rdmsr
seg000:42EC 66 0B C3                                or      eax, ebx
seg000:42EF 0F 30                                   wrmsr
seg000:42F1 66 5A                                   pop     edx
seg000:42F3 66 59                                   pop     ecx
seg000:42F5 66 5B                                   pop     ebx
seg000:42F7 66 58                                   pop     eax
seg000:42F9 0F A9                                   pop     gs
seg000:42FB CB                                      retf

Pay attention to address 42A3. You can see a well-known OR instruction (or eax, 100000h). But if you patch it, it will give you nothing because this code is executed only if CPU doesn't support EIST.
To disable setting EIST lock bit, we need to modify this mov instruction:

Code:

seg000:42CD 66 BB 00 04 10 00                       mov     ebx, 100400h

in the following manner:

Code:

seg000:42CD 66 BB 00 04 00 00                       mov     ebx, 400h

I don't know how you can describe this situation in the guide. The number 100400h may vary in different BIOS'es. Moreover: there can be other ways to set EIST lock bit. In such cases only analysis of code can help to determine a correct place for the patch.


4) You wrote that you had problems in patching Lenovo IdeaPad Y450 BIOS. To avoid this error "Module is too big" you need to have some skills in optimizing x86 assembler code. Don't hope this is rare case and only IdeaPad Y450 BIOS is affected. I faced this error many times and it has never been the reason not to make a patch. For example I met "Module is too big" message when working on nine ThinkPad dual-IDA patches.

I don't know what to advise you, there's no universal recipe. For example Y450 patch looks like this:

Source:

Code:

seg000:000000000000168F B9 A0 01 00 00                          mov     ecx, 1A0h
seg000:0000000000001694 48 0F BA E8 14                          bts     rax, 14h
seg000:0000000000001699 48 8B D0                                mov     rdx, rax
seg000:000000000000169C E8 A7 6D 00 00                          call    near ptr 8448h

seg000:0000000000002697 B9 A0 01 00 00                          mov     ecx, 1A0h
seg000:000000000000269C 48 0F BA E8 14                          bts     rax, 14h
seg000:00000000000026A1 48 8B D0                                mov     rdx, rax
seg000:00000000000026A4 E8 9F 5D 00 00                          call    near ptr 8448h

Modified:

Code:

seg000:000000000000168F B9 A0 01 00 00                          mov     ecx, 1A0h
seg000:0000000000001694 48 0F BA E8 14                          bts     rax, 14h
seg000:0000000000001699 48 8B D0                                mov     rdx, rax
seg000:000000000000169C 90                                      nop
seg000:000000000000169D 90                                      nop
seg000:000000000000169E 90                                      nop
seg000:000000000000169F 90                                      nop
seg000:00000000000016A0 90                                      nop

seg000:0000000000002697 B9 A0 01 00 00                          mov     ecx, 1A0h
seg000:000000000000269C 48 0F BA E8 14                          bts     rax, 14h
seg000:00000000000026A1 48 8B D0                                mov     rdx, rax
seg000:00000000000026A4 90                                      nop
seg000:00000000000026A5 90                                      nop
seg000:00000000000026A6 90                                      nop
seg000:00000000000026A7 90                                      nop
seg000:00000000000026A8 90                                      nop

There was no "Module is too big" message during rebuilding Y450 BIOS.

 

First of all, I'm not an assembly language guru & what I wrote in the guide are base on my testing & experiment on my own notebook & then later expanded to another notebooks (nando4 volunteered to test it). Since I can recover my notebook, I can do countless test on it. It's also based on the solution I found at that time.

Based on the information I gathered at that time, I need to find the ecx & eax register. That's why I disassembled it in 32bit mode. And I did tested this myself & it worked.

My learning pace is a bit slow. The first method is based on existing solution at that time & I tested it myself to proved it does work. Later, I use the second method which change "bts" to "btr" instruction. Yeah, I forgot to update the document.

I mostly mod Acer & Sony BIOSes. I has only mod one Toshiba's BIOS (Qosmio X300). All of 'em works. I know different manufacturer use different method to set EIST Lock bit. So far Acer, Sony, Clevo, Packard Bell, HP, one Toshiba Qosmio X300 & some Chinese-brand notebooks use standard method to set EIST Lock bit. At least that's what I found so far & I said "standard" because all of 'em using the same "pattern" to set the EIST Lock bit. Usually I found EIST lock bit in these modules:-

• MOD_5100.ROM (Phoenix)
• F7731B4C-58A2-4DF4-8980-5645D39ECE58.ff (Phoenix)
• F7731B4C-58A2-4DF4-8980-5645D39ECE58_X_XXX.ROM (Insyde)

Except Acer Aspire 5920G which have EIST Lock bit set in BIOSCOD03.ROM module, in addition to MOD_5100.ROM module.

The guide is based on the result I got during my experiments. It's not something I guess & just wrote about it. I know the methods I use is not one for all solution. I was always thought (hopping) it is common sense to everyone that anything related to BIOS is dangerous (BIOS flash & BIOS mods). That's why whenever a person ask me to mod their BIOS, I always remind them about the risk & only continue if they know how to recover from bad BIOS. It has been long time since somebody tried my modded BIOS & bricked their computer. The success is not based on luck, no sir, it is based on my effort to experiment it myself first before making the solution available to the public. I bricked my notebook countless time in the process. I know many people can do this but not many who willing to wrote one simple document (step-by-step guide) & share it.

I do appreciated your comments & inputs except the "lucky" part (I am a normal person who have feelings & I'm not afraid to said it out loud. ) Since you have better grasp on the assembly language, I welcomed you to perfects/corrects the guide.

Thank you,
kizwan

 

kizwan, I didn't mean to offend you in any way. I wrote the previous post just to help you make the guide better.

And what about "Module is too big" error? I would recommend to mention in the guide the following: If "Module is too big" error message is displayed after changing bts to btr, then we can modify the code like so:

Code:

seg000:000000000000168F B9 A0 01 00 00                          mov     ecx, 1A0h
seg000:0000000000001694 48 0F BA E8 14                          bts     rax, 14h
seg000:0000000000001699 48 8B D0                                mov     rdx, rax
seg000:000000000000169C 90                                      nop
seg000:000000000000169D 90                                      nop
seg000:000000000000169E 90                                      nop
seg000:000000000000169F 90                                      nop
seg000:00000000000016A0 90                                      nop

seg000:0000000000002697 B9 A0 01 00 00                          mov     ecx, 1A0h
seg000:000000000000269C 48 0F BA E8 14                          bts     rax, 14h
seg000:00000000000026A1 48 8B D0                                mov     rdx, rax
seg000:00000000000026A4 90                                      nop
seg000:00000000000026A5 90                                      nop
seg000:00000000000026A6 90                                      nop
seg000:00000000000026A7 90                                      nop
seg000:00000000000026A8 90                                      nop

or even so:

Code:

seg000:000000000000168F B9 A0 01 00 00                          mov     ecx, 1A0h
seg000:0000000000001694 48 0F BA E8 14                          bts     rax, 14h
seg000:0000000000001699 90                                      nop
seg000:000000000000169A 90                                      nop
seg000:000000000000169B 90                                      nop
seg000:000000000000169C 90                                      nop
seg000:000000000000169D 90                                      nop
seg000:000000000000169E 90                                      nop
seg000:000000000000169F 90                                      nop
seg000:00000000000016A0 90                                      nop

seg000:0000000000002697 B9 A0 01 00 00                          mov     ecx, 1A0h
seg000:000000000000269C 48 0F BA E8 14                          bts     rax, 14h
seg000:00000000000026A1 90                                      nop
seg000:00000000000026A1 90                                      nop
seg000:00000000000026A1 90                                      nop
seg000:00000000000026A4 90                                      nop
seg000:00000000000026A5 90                                      nop
seg000:00000000000026A6 90                                      nop
seg000:00000000000026A7 90                                      nop
seg000:00000000000026A8 90                                      nop

We can change to nop even these instructions

Code:

seg000:000000000000168F B9 A0 01 00 00                          mov     ecx, 1A0h
seg000:0000000000001694 48 0F BA E8 14                          bts     rax, 14h

seg000:0000000000002697 B9 A0 01 00 00                          mov     ecx, 1A0h
seg000:000000000000269C 48 0F BA E8 14                          bts     rax, 14h

if the previous changes are not enough.

 

That is good idea. (I didn't comment about it in my previous post, did I? ). We can just nop the "call near ptr 8448h" because it is no longer needed.

I agree, we can just nop the "mov" & "bts" instruction too if the module still too big to reintegrate.

I'll update the guide as soon as possible. Thank you.

 

I has updated the " EIST Unlocked Example" guide per middleton's suggestion. The new revision can be found at the same place which is at post #256.

 

I'm having problems with the cygwin side of this procedure. I got past the phantom carriage return (hidden '\r') issue I spent 2 hours on, but now my cygwin won't recognize the ndisasm command. Is there some package that I need to install for cygwin to know what this command is?

Or someone could just mod my bios for me!
http://ftp.compal.com/asp/driver_dnd/index.htm
(mine is the JHL90)

My laptop does have a bios recovery mode btw so bricking it should be impossible.

 

I found the package btw. It's called NASM (network assembler).

I also found the eist disable code segments. The corresponding bios file is a .ff, not a .ROM. What program should I use to edit the file?

EDIT2:
Figured it all out I think. Time to flash and hope I don't have to make a recovery flash drive on another pc

EDIT3:
VICTORY! Finally can do 3ghz+ fully stable, ram was holding me back before. You can now add sager np2096/ compal JHL90 to the list of moddable bios's. I did not have to use the crisis recovery btw.

i picked up this thing's top end CPU off of ebay for under 150 including shipping (yay for having an outdated pc so I can get low prices!). Playing around with that should be interesting.

Things to note:
This system ended up using a WPH file, not a .bin or .ROM for its bios image. Also, I had to modify a .ff file, not a .ROM. There were 2 eist disable entries. The .ff file changed actually followed your Compaq 2530P example completely (same exact DASM and offsets). I also didn't have to change anything to "nop". Final note: god this was an involved process.

 

thank you first for your excellent work. i'm hp(campaq) notebook user, the type is cq45-203tx, the current bios im's using came from ftp://ftp://ftp.hp.com/pub/softpaq/sp46501-47000/sp46656.exe,this bios is insyde and already updated to slic2.1 officially, could you help to remove the EIST lock in such way i could use dual ida overclock with ThrottleStop。thank you in advance.

 

Good work. Can you post the modded dual-IDA np2096/JHL90 bios somewhere? Then I can add it to the dual-IDA C2D bios list.

 

Bios is here:
np2096/jhl90 EIST unlocked bios

Drag and drop those into a bootable USB and the system will flash automatically.

 

My e8435 arrived. This CPU was actually meant for low power apple desktops, but also works in notebooks at the same speed as the t9900, but 1/3 the price. It turns out that the e8435 does NOT have an IDA multiplier, but 11.5 is high enough and the BIOS mod I did was worth it anyway since I can disable EIST to keep the CPU stuck at an 11x multiplier without having to keep throttlestop open. With EIST enabled, it reverts to default immediately after throttlestop closes.

Oh and this thing can do 3.5ghz fully stable! Well, games are fully stable. If I ran orthos at anything higher than 3.2 it'd go over 100C and hopefully catostrophic thermal protection would kick in. Cooling needs work.

 

error of post, sorry

 

As you also made this request on MDL forum, your bios mod is ready to download HERE

regards

 

This mod is for Core 2 based CPUs only.

 

Haven't logged into these forums since 2007 apparently. Anyway, I wanted to drop in and say how awesome this is. My Dell Studio XPS 1640 runs at 2.66 on command!

Obviously, a five-minute P95 test does not a stable overclock make, and since I don't have a 130w adapter (yet) I won't be pushing it until then.

Side note: Does AS5 "go bad" after being in storage? I have a tube and an more than willing to use it to hold down thermals.

 

Wondering if anyone has dealt with a t61p.

I can't seem to get it to dual-ida.

CPU:T7700

Installed newest dual-IDA firmware.
Turned off Speedstep in BIOS.
Booted into Win7 64bit and was at 6x.
Turning on ThrottleStop took me to my normal 12x, where 13x is dualIDA.

Tried checking Start DualIDA and Profile1 in options.
Tried adding dualida=1 to config file.
Tried right-clicking whitespace and choosing start dualida.
Tried Set Multiplier 13x
Tried turning EIST on/off or switching to other profiles and back.
Tried putting a load on the CPU.
Power settings in win7 set to min/max 100%.

Any ideas would be appreciated.


http://i.imgur.com/BPsTs.jpg
Thanks.

 

Hi everyone.
Thx for great tool unclewebb.
I'm wondering if there is possible to mod Asus M50SA to unlock EIST. Cause now it's grayed out in ThrottleStop and my cpu definitely support this feature (T9300). Is there any guide to mod AMI BIOS?

 

Strmtrper6: I don't understand any of this but I think IDA mode might be disabled within your CPU. Dufus might have a trick to get this working for you but he hasn't been around for a while. Here's some more information about this.

http://forum.notebookreview.com/win...tions-reports-suggestions-discussions-30.html

dontcha: Try contacting nando4. He has accumulated a large number of bios versions for different models with EIST unlocked.

 

The dual-IDA capable modded bios and a link to Kizwan's document explaining how to mod your bios to unlock EIST for dual-IDA functionality are at http://forum.notebookreview.com/win...ida-both-cores-core-2-duo-15.html#post6646598 .

 

It would be nice if there was anything about modding AMI BIOS. At least after some searching I know that I need other tool than used in tutorial, cause the one from tutorial can't do antything with AMI BIOS - it's for Phoenix/Award/Insyde only.

So now I have the proper tool (al least I think that AMI SLIC MOD is the right one) and I stuck, cause I don't know if anything of Kizwan's guide is suitable for totally different kind of BIOS.

 

Just wanted to drop in and confirm that DualIDA works on an HP EliteBook 8530w/T9600/FX770M.



Getting almost 3Ghz is pretty snazzy! Now for more RAM, IC7 TIM (on its way ), possibly an x9100 upgrade, and maybe a GPU upgrade somehow.

ETA: Does the whitelist remove BIOS locks for only WLAN cards or does it remove locks for GPUs as well? A 9800GTm might fit in here...
Edit2: For some reason I cannot adjust the voltage. DualIDA runs at 1.175V which bugs me, because I was able to get 10.5x running 100% stable at 1.06V

 

I have a dell studio 1555 with a p8600 and I can't disable EIST using ThrottleStop it's grayed out. Am I doing something wrong?

EDIT: Whoops my bad. /facepalm
I figured out what I did wrong. I needed to turn it off in the bios. >.<

 

Hi everyone,

I was wondering whether it would be possible to DISABLE IDA completely...

Also, the MSR Tool does not work on this laptop, I change the values, but they are back to the default after clicking Read MSR again to verify - is it locked or is it just the tool that doesn't work (I have Windows 7 64 bit)?

Thank you!

 

The VID voltage has to be set before you go into Dual IDA mode. When EIST is disabled, the FID and VID adjusters in ThrottleStop don't work anymore.

Many bits in many MSRs are locked. Let me know what MSR you are trying to modify. When the EIST lock bit is set, you can't unlock it with the MSR Tool. The CPU ignores changes to locked bits.

 

Thanks unclewebb for your advice!
I found out how to disable EIST and/or IDA by modding the BIOS, so I edited this post as it's not related to enabling Dual IDA.

 

Dual-IDA working on Linux

Briefly, I was able to engage dual-IDA on my 2530P while running Linux using specific codes for my L9400 (non-IDA=x7/1.86Ghz, IDA=x8/2.13Ghz) in the spoiler improving a prime95 run from 197.098ms to 173.248ms (12%)/

Code:

[Worker #1 Jul 19 11:58] Timing 10 iterations of 8192K FFT length.  Best time: 196.833 ms., avg time: 197.098 ms.
[Worker #1 Jul 19 12:05] Timing 10 iterations of 8192K FFT length.  Best time: 173.023 ms., avg time: 173.248 ms.

Spoiler

Code:

#############################
## CPU MSRS based off http://img223.imageshack.us/img223/6914/msr.png
##
## 0xCE: 7:0  - min VID
##       12:8 - absolute minimum FID (SLFM)
##       39:32 - absolute max VID
##       44:40 - absolute max FID (IDA) 
## 
## 0xEE: 28:28 - SuperLFM (1)
## 
## 0x198 (IA32_PERF_STATUS)
##      7:0 - Current VID
##      12:8 - Current FID
##      14:14 - HALF (1)
##      15:15 - SLFM (0)
## 
##       39:32 - current max VID
##       44:40 - current max FID
##       46:46 - HALF (1)
##       47:47 - SLFM (0)
##       55:48 - current min VID
##       60:56 - current min FID
## 
## 
## 0x199 (IA32_PERF_CTL)
##       7:0 - requested VID
##       12:8 - requested FID
##       14:14 - HALF (0)
##       15:15 - SLFM (0)
## 
## 0x1A0 (IA32_MISC_ENABLE)
##       16:16 - EIST (1)
##       20:20 - LOCK (0)
##       38:38 - IDA (0)
##############################
# rdmsr 0x198
rdmsr: open: No such file or directory
## Load msr module
# modprobe msr
# rdmsr 0x198
617071c06008611
## Set governor to performance
# echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor 
# echo performance > /sys/devices/system/cpu/cpu1/cpufreq/scaling_governor
## Check existing VID/FID of both codes
# rdmsr -p0 0x198
617071c0632071c
# rdmsr -p1 0x198
617082706000827
## Check the required VID/FID for both cores
# rdmsr -p0 0x199
827
# rdmsr -p1 0x199
827
## EIST is bit 16 of 0x1a0 - need to toggle that 
# rdmsr 0x1a0
1364872488
# wrmsr 0x1a0 0x1364862488
## Check if both cores are running x8
# rdmsr -p1 0x198 
617071c0600071c
## Toggle EIST again
# wrmsr 0x1a0 0x1364872488
# rdmsr -p0 0x198
617071c0632071c
# rdmsr -p1 0x198
617071c0632071c
# rdmsr -p1 0x198
827
## Toggle EIST again
# wrmsr 0x1a0 0x1364872488
# wrmsr 0x1a0 0x1364862488
# rdmsr -p1 0x198
617071c06000827
# rdmsr -p0 0x198
617071c06000827
## BOTH cores are running x8

Or see the more thoroughly explained example at http://forum.notebookreview.com/hp-...87-hp-2510p-owners-lounge-92.html#post8599758 .

It's easy for me to write a script that writes hardcoded values to my CPUs MSR to enable dual-IDA and undervolt the CPU. However I was wondering if anybody might want to write a Throttlestop for Linux wrapper that uses rdmsr/wrmsr? Would be handy to have a generic tool to set VID/FID/IDA/EIST/c-states/dual-IDA.

The exact same thing could be achieved on MacOS if there's a rdmsr/wrmsr equivalent utility.

 

I would LOVE this

Check it. These are all the same system. Look at the measured speed.
http://einstein.phys.uwm.edu/show_host_detail.php?hostid=4148884 Darwin 10.8.0
http://einstein.phys.uwm.edu/show_host_detail.php?hostid=4148417 Linux-2.6.36
http://einstein.phys.uwm.edu/show_host_detail.php?hostid=4180141 Win7x64

 

First of all, I never even knew the Core 2 Duo's had some form of Turbo. Anyway, this is really neat! I'm running a Dell Studio 1535 [BIOS Revision A06] with a T8100. Before I was running 1.2V at 2.1GHz with stepping, now I'm at 1.05V at 2.3GHz without stepping. I was able to run a TS Bench and got a 3 second improvement. Also, my Windows Experience Index went from 5.1 to 5.2. The only unfortunate thing is, that you have to redo this everytime you boot up your computer. From boot, it runs at a 6 multiplier, using this I get to a fixed 11.5. It's nice, but I'll probably end up deleting it and re-enabling SpeedStep and IDA in the BIOS.

Now if only I can find a way to OC my GPU a few MHz then I'd be excited to try out StarCraft II with the CPU/GPU performance boost.

 

Throttlestop can most certainly be setup to automate dual-IDA on startup and on resume-from-standby/hibernate. Save your desired TS AC and DC profile and put a link to TS in your startup folder.