(CoolBook Author's original comments)
"I just wanted to let you to know that CoolBook is ready for downloading. CoolBook is a free program that displays the processor frequency, voltage and temperature of the CoreDuo CPU. It will currently work with MacBook and MacBookPro (CoreDuo ) computers.
For a $10US license, the program will allow the user to manually set the voltage and frequency of the CPU. This can dramatically reduce the heat dissipation.
To download go to http://www.coolbook.se.
(I asked if it also had fan speed control/adjustments-Mike)
No, maybe in a future version. There is no need to change the fan speed if you run the CPU at a low voltage.
The main reason for making this program was to reduce the fan speed, and it does so by reducing the power consumption of the CPU.
(I asked about the range/limits of the adjustments in the registered version.)
There are many different ways of reducing the heat. One way is to lock the processor in a lower frequency mode. When I use my MacBook on battery, i like to run it at 1000MHz and 0.9500 V (the original voltage-frequency "pair"). This way I can use it on my lap, and for a longer period of time than with the original regulation. Sure it is not as fast, but a dual 1GHz CoreDuo is still faster than last years laptops.

The CPU has a built in upper limit for frequency and voltage, so you can't damage it. You can crash it however, but if you are prepared with no documents unsaved, you should be fine. But this is mostly a feature for the more experienced users. The manual should clearly state this.
The regulation of my MacBook will change the frequency from around 1GHz to about 1.8GHz. At the same time the voltage will go from 0.9500V to 1.2625V. This is done because a higher voltage will permit a "faster" current flow, and lower switching times.
If you calculate the power consumption of a CPU, you will find in the formula that the voltage affects the power consumption more than the frequency will. The formula contains frequency*(voltage)^2. So a small decrease in the voltage will actually make a very large impact on the power consumption. (This came up in the past with CPU upgrades that had adjustable voltage settings.)

Raising the RPM of the fans may be a way to cool your Mac, but it will cost extra power and noise.
Best regards, Magnus L."

If any readers try this utility, let me know what you think of it. (I don't own a MacBook/Pro, but some readers have used fan adjustment utils like Fan Control or smc FanControl for cooler running without manually adjusting CPU speeds/voltages. My gut feeling initially was I'd prefer the fan adjustments rather than manually adjusting voltages/cpu speeds.

CoolBook User Reports/Comments: (later replies first)
(I welcome other user feedback, including comparison tests.)

(added Nov. 27th, 2006, later revised per Wayne's request)
"I downloaded, installed, and paid for CoolBook. Although it does let you set frequency and voltage, it's not an unqualified improvement over the stock SpeedStep driver, which automatically sets frequency and voltage based on CPU load. There are tradeoffs: - You have to manually set frequency and voltage, and each time you do this, you need to enter your password. - Although it dramatically reduces temperature at high CPU loads, under low load, you're forced to choose between the following: - - - Slightly increased temperature (about 2C in my tests). - - - Lower performance ceiling for bursts of CPU activity. (eg., speed fixed at 1.00GHz instead of changing to 1.83GHz when needed.)

I emailed Magnus (the author of Coolbook) about my concern, and he said that, when under AC power, I should select the maximum frequency (for me, 1.83GHz) and run it at a reduced voltage. Suppose the minimum voltage at 1.83GHz for my computer is 1.1V; according to Magnus, I should fix the voltage here. With the default SpeedStep driver, when my computer has low load (like when I'm just writing or reading something), the frequency and voltage go between 1002MHz at 0.950V and 1336MHz at 1.075V. Under these circumstances, the SpeedStep voltage is lower than what CoolBook would do. So CoolBook would actually make my computer hotter!

I told him that this is a reasonable expectation of the program, given the way it is marketed:
A.) It reduces the temperature of the computer when under high load.
B.) Under no circumstance does it INCREASE the temperature of the computer.

The program succeeds brilliantly at (A), but it fails at (B).

It turns out that for me, the lowest stable voltage for 1.83GHz is 1.000V. Under low-load use, my computer idles at 57C according to smcFanControl (note: CoolBook gives a reading about 7C lower) with these settings. The SpeedStep driver runs the CPU at 1.00GHz and 0.95V under low load, and here my computer idles at 55C. So CoolBook actually makes my computer 2C hotter when there's low CPU load! It's not a scalding temperature or anything, but it does go against the advertised claims of CoolBook. In email correspondence, Magnus says that he thinks it's working exactly as advertised, but I have to disagree. Never is it mentioned that the program can make your computer hotter.

If he changes it so that you can keep the automatic frequency adjustment, but set the voltage at each frequency, then CoolBook will be more useful -- it will never make your computer hotter than the default settings. Otherwise, I consider the program in some ways a downgrade, because it will increase the computer's temperature most of the time, given my usage profile. I emailed him and said that if such a change isn't made, I would like a refund. He says that I should have read the documentation before buying, which I did. It's just that the documentation never says that the voltage is fixed, and it never says that Coolbook can lead to greater power consumption and heat, when used in ordinary circumstances.

Be warned about CoolBook: For low-load use, it will result in either a hotter computer, or one with a lower performance ceiling, depending on how you adjust the settings.

About stability testing: I strongly suggest using memtest for mac and/ or Glucas (the mersenne prime search client for Mac) to stress your computer and test for errors. (Reader FYI - Here's the home page for Memtest OS X command line utility and there's a 'Rember' GUI for it also.-Mike) At some voltages, these programs detected errors, but the computer appeared to run along fine -- the computer could be building up little memory errors that you don't notice until, say, you get filesystem corruption.
I had to hack the Makefile for Glucas to get it to compile on my computer running 10.4.8, so it may be easier to do the following: run memtest in one terminal window, and in another run 'yes > /dev/null' to make sure both CPUs are fully occupied.
-Winston "

(added Nov. 20th, 2006 edited per reader request, updated with more comments/test notes)
"I bought coolbook cause i wanted to test it... I'm still testing it more but my first test was with coolbook at stock settings. (1837@1.2625V), I put fancontrol on 55C lower and 75 high threshold. I put the base speed of the fan back too stock 1000rpm.

I ran 2 instances of yes > /dev/null too let the macbook heat up and ran this test for 10 minutes all run on the adapter.

After 10 minutes i noted the fan speed and its was on 4300rpm. More info on the temps from menu-temp below:

Menu-Temperature Report (stock 1.2625v setting)
======================
Running 2 instances of yes > /dev/null with normal cpu settings. 1837mhz @ 1.2625 V on adapter.
Fan speed on full load around 4400 with current settings of fan control.
Idle temp after 5min cool down: 51 c according 2 menutemp.

Temperature Report (stock 1.2625v setting)
----------------------
Current temperature: 67.00°C
Statistics for last ten minutes:
- Minimum: 46.00°C, Average: 61.02°C, Maximum: 72.00°C
Overall statistics:
- Minimum: 46.00°C, Average: 66.21°C, Maximum: 72.00°C

Frequency Report (stock 1.2625v setting)
----------------------
Current frequency: 1.83 GHz
Statistics for last ten minutes:
- Minimum: 1.83 GHz, Average: 1.83 GHz, Maximum: 1.83 GHz
Overall statistics:
- Minimum: 1.83 GHz, Average: 1.83 GHz, Maximum: 1.83 GHz
Running for 10 minutes, 16 seconds

Now next i set the cpu to 1837mhz @ 1.000 V and run the-same tests again. Here is the report from menutemperature.

MenuTemperature Report (using 1.000v setting)
======================
Running 2 instances of yes > /dev/null with coolbook cpu settings. 1837mhz @ 1.0000 V on adapter.
Fan speed on full load around 3200 with current settings of fan control.
Idle temp after 5min cool down: 49°c according to menutemp.

Temperature Report (using 1.000v setting)
----------------------
Current temperature: 63.00°C
Statistics for last ten minutes:
- Minimum: 58.96°C, Average: 58.98°C, Maximum: 64.00°C
Overall statistics:
- Minimum: 51.00°C, Average: 62.82°C, Maximum: 64.00°C

Frequency Report: (using 1.000v setting)
----------------------
Current frequency: 1.83 GHz
Statistics for last ten minutes:
- Minimum: 1.83 GHz, Average: 1.83 GHz, Maximum: 1.83 GHz
Overall statistics:
- Minimum: 1.83 GHz, Average: 1.83 GHz, Maximum: 1.83 GHz
Running for 10 minutes, 4 seconds

As you can see from this test, the max temp is lowered by 8°C but more importantly for me, the fan is running at a 1200rpm lower. That means less sound.

I haven't tested speed increase or decreases yet. And at this time im testing battery life with max cpu speed @ 1.000V vs stock...
(he later wrote)
i did my test again yesterday to comfirm my initial report. The biggest difference for me is that the fan RPM is 1200 to 1400rpm less over stock. This is running the CPU at 1.83Ghz constant.

I also did a battery test yesterday. My test was on 1.83 Ghz @1.0125V, the battery would last me 3:43 until my MB would go in sleep mode. This was while doing some light browsing, running safari and with airport on. I also ran a test with my CPU @ 1.0Ghz@0.9500v locked but according to minibatterylogger the (estimated) time would almost be the same as on 1.83Ghz@1.0125V.(there should be more runtime at 1GHz/.95v than 1.83/1.125v in reality, although other things like the backlight, drive activity, etc. also affect runtime.-Mike)

Just ran cinebench and these are my results:

CINEBENCH 9.5
****************************************************
1.83ghz @ 1.0125V

Processor : MacBook CD
MHz : 1837
Number of CPUs : 2 (core duo)
Operating System : 10.4.8
Graphics Card : GMA950
****************************************************
Rendering (Single CPU): 283 CB-CPU
Rendering (Multiple CPU): 530 CB-CPU
Multiprocessor Speedup: 1.88

Shading (CINEMA 4D) : 332 CB-GFX
Shading (OpenGL Software Lighting) : 1187 CB-GFX
Shading (OpenGL Hardware Lighting) : 1216 CB-GFX
OpenGL Speedup: 3.66
****************************************************
1.83ghz @ 1.2625V
Rendering (Single CPU): 283 CB-CPU
Rendering (Multiple CPU): 529 CB-CPU
Multiprocessor Speedup: 1.87

Shading (CINEMA 4D) : 332 CB-GFX
Shading (OpenGL Software Lighting) : 1137 CB-GFX
Shading (OpenGL Hardware Lighting) : 1172 CB-GFX
OpenGL Speedup: 3.53
****************************************************

Note that after running cinebench using 1.83Ghz@ stock Voltage, the coolbook controller showed the cpu speed was 1002mhz. This was after the cinebench test was finished. Dont know if this is a bug in coolbook controller, cause my minibatterylogger shows a constant 1.83ghz.

BTW this is not the apple stock speedstep setting vs coolbook setting. Just a coolbook 1.83 @ 1.0125V VS 1.83 @ stock voltage using coolbook.
-Ernesto J."

(added Nov. 17th)
"I just used Coolbook. It works very stable no problems here. I have a Macbook 1,83Mhz and i wanted to use it comfortably on my laps. So I set the voltage to 0,95(the lowest) and using the smcfan utility I set the fan to the max setting(6200rpm) . After a while the temperature doing office work,internet,etc went to 26 degrees C. I was pretty amazed! But for my disappointment I realized that even at 26C ,you can't work for a very long time because you still feel a little uncomfortable. (26C?? Must be a typo as that's not much over typical ambient room temperature.-Mike)
Soon I will write a mini review in my site. It's in Greek ,but soon it will go in English too. http://www.fousterthoughts.blogspot.com/
Sorry for my English but I'm in a bit of hurry right now....... :-)
-ASKi "

Seems a lot of users of Coolbook go by nicknames.

The author replied to some previous comments

(added Nov. 17th)
"I read your new posts (below) about CoolBook, and I just wanted to explain how the default "Speed Stepping" works.
The default Apple driver is ramping up frequency and voltage depending on load. So if the CPU is being used at full performance (like in a Cinebench run) it will be running at the maximum frequency. The reader who commented about the down throttling, probably just observed the natural down clocking when the benchmark was finished. This regulation is very fast.

There is however a theoretical possibility to get a very small performance boost using CoolBook, and that is if you fix the frequency at maximum. The default driver throttles down when the computer is idle, requiring some extra time to reach full performance at load.

Some CoolBook users claim their XBench scores are higher, and that the dock animation feels smoother. This is not verified, and not the real reason to use CoolBook. It may just be a simple "placebo effect".
Best regards
-Magnus L. "

I thought per apple's comments (* disclaimer on CPU speeds) that the CPU speed could be reduced depending on system thermal readings, etc. and therefore a cooler running system could have less throttling (which is what the first user claimed, although the cinebench scores were literally identical).

(added Nov. 16th, 11:30PM)
"You might like to check out this ArsTechnica thread regarding CoolBook:
http://episteme.arstechnica.com/eve/forums/a/tpc/f/8300945231/m/546008871831
The OP claimed better performance through zero throttling but the author of CoolBook (possibly also the OP) posted:
"I've been running countless Cinebench tests on my MacBook 1.83GHz. Unfortunately I can't get the results from my MacBook since I'm running some battery time tests on it. But I can assure you that they are at least in par with the original SpeedStepping results"
It's worth a read.
all the best
-Greg "

I wish I personally had a MacBook/MacBook Pro to confirm the claims on this - and the CB 9.5 scores from the full screenshot (not mentioned in mail, but shown in the screenshot) showed literally the same scores - yet the 'stock' settings had a 800MHz CPU throttle down. That doesn't make sense. (With the trimmed screenshots I didn't notice that originally.) I wish there was a trial version with full feature support before buying it - that way there would be more user feedback/test results on the claims.

(added Nov. 16th, 11:30PM)
I bought coolbook the day it was announced on your site. I own a 2.16 Macbook Pro Core Duo

The default voltage for 2.16 GHz is 1.225v, but the computer is still stable when using 1.0625v. This make a huge difference, at full load (dnetc), the fans used to run at more than 5000rpm and the temperature was more than 85°, now the fans stay at less than 3500rpm and temperature less than 80°. (degrees C I assume-Mike)

The lower voltage available (0.925v) is ok for frequencies up to 1.83 GHz.

I also ran some test when setting the speed at the lowest possible (1 GHz) on battery, with 0.925v, and it gives me about 20-30 min more battery life.

In order to really limit the CPU speed, you have to check "disable original driver" in the preferences. The computer still switch between frequencies, but only between 1 GHz, and the maximum that you specify.

I talked with the author to know it he plans to add more features:
- specify a voltage for each frequencies, and then led the speedstep driver choose the right frequency, with the voltage specified
- let the computer automaticly switch between battery ans power profiles when plugged and unplugges (by now you must restart or put the computer to sleep to allow the change)
- set a lower voltage than 0.925v when running at 1 GHz (but I was told it is impossible)

One interesting note, even if my 2.16 MPB is running at 1.225v by default, one of my friend's 1.83 GHz MBP is running at 1.250. It would be interesting to know it this is CPU-revision dependant, or frequency dependent, and also, what is the CPU core voltage of the new Core 2 Duo MBP.
Anyway, Coolbook is a cool utility, every macbook pro and macbook user should buy it.
(he later wrote)
I read the thread on Ars Technica and don't beleive how some people can be so sceptical. No I'm not the author of coolbook under an other name, maybe you remember me i'm the webmaster of the french site MacXtrem, and I was the guy who managed to overclock both aluminium powerbooks at 1.83 GHz, and titanium at 1.2 GHz.

Coolbook works very well, and I did a lot of testing, using benchmark software and power consomation monitoring, to see that coolbook does exactly what it pretends to do. I really think those guys at ars should grow up before claming that a program that works very well can't do what it is supposed to. By the way, coolbook author is a nice guy, he took the time to answer all my mails and my questions.
-Gamoul "

I don't have a MacBook or MBP to test this myself, but the results mentioned below were hard to believe. (No throttling of CPU speed and a lower fan speed even with a higher temperature. But literally identical CB 9.5 scores yet the stock settings had throttled down the CPU by 800MHz+ by the end of the test?)

Here's the first feedback from a reader to try CoolBook (registered version) the first day it was released.

(originally posted in Oct. 2006 news page, see below for update/tests)
" Hi Mike! I use a MacBook 1.83GHz, and have had issues with high fan speeds when running some apps. For example when I was playing a flash game (RX-8 drift at mazdausa.com) some days ago the CPU fan started to run at a very high speed, really loud.
So I gave the CoolBook app a try. (I got an email with the registration information soon after I bought it using PayPal. Then you just enter it into the app. ) (I thought it was awfully quick to get a registered user reply back the same day it was posted.-Mike)
The setting I used was 1837MHz 1.0000V. Then I played the game again, the difference was huge. I also tried some tests with Cinebench, to compare the performance. The performance was equal, but the difference in heat and sound was stunning.
I still have to verify the long term stability with this setting, but so far it's looking good.
(I asked him what the default voltage was for that speed)
The stock voltage for 1,8 GHz is 1.2625V.
-Mikael A. "

Mikael later sent screenshots showing temperatures, CPU speeds and fan speeds while running CineBench 9.5 using his CoolBook settings (lower voltage) vs the stock system control:

" Here are some screenshots of Cinebench. (Below) I put SMCfancontrol and Coolbook controller on top to show the values.
The "original" files uses Apples driver. (standard/default system control/thermal mgmt) I also took a shot at the end (you can see that it throttles down to 1GHz without CoolBook settings). I haven't crashed it yet so now I'm trying 0.9875V.
(I'd be leery of lowering it too far - as you're already more than .2V below the standard setting. (I assume the default control uses Intel recommended settings) Might be worth cloning/backing up the drive when testing like this, just in case.-Mike)

I've had one kernel panic trying 0.95V at 1.83GHz.

And you can see in the pictures what I mean by "stunning". Both the temperature and fan-RPM is much lower (when using CoolBook modified voltage setting). I also included a shot of the available voltages to show that there are many stages between my current setting and the default (maximum) voltage, and I think even the middle ones will save a lot of power.

I'm going away for the weekend so I have to do the "torture test" later.

One more thing; the original tests were done before the coolbook ones, as you can see on the clock. This was done to prevent the computer from being "preheated", making the test unfair to the Apple driver.

(I asked him some questions about these tests-Mike)

Were all tests run on AC power? (not battery)
All test used AC power.

Had you applied the MacBook 1.1 firmware update?
(Note: this was before apple released the MacBook EFI Firmware Update 1.0)
This was with SMC firmware update 1.1. I have not seen any effects of the new firmware. (I thought he had sent this before I first saw the SMC firmware update - unless it appeared in Software Update before it did at Apple's support/downloads page.-Mike)

Another thing, Xbench scores are higher with CoolBook activated. I did many Xbench runs to confirm this. All scores are about the same. But the Quartz graphics score is up from 100 to 115. User Interface score went up from 250 to 305.
-Mikael A. "

If there's as much difference in CPU throttling as the CineBench 9.5/CoolBook screenshots he sent showed (from 1.8GHz down to 1GHz), the scores should have been much better I'd think. Note: I removed the Cinebench screenshots he sent (comparing start/end CPU speeds, temperatures, fan speeds) - when looking at the original full size screenshots, the 'end of test' screenshot showed literally the same CineBench scores even though (per Coolbook panel) the CPU speed had throttled down to 1GHz (from 1.83GHz) without the coolbook voltage tweak. That doesn't make sense to me (nor did the lower fan speed even though the temperature shown as higher).
If any other readers with an Intel Mac notebook try this utility, let me know what you think of it and if you can confirm/replicate the results above. (And call me cautious but I'd suggest having a backup of the boot drive when experimenting like this, just in case the system would ever crash or result in a corrupted drive.)