Investigating Titanium CPU Throttling

[Doq]

There are two states: 667 with no L3 and 1000 with L3; these correspond to the "Slower Processor Speed" and "Faster Processor Speed" options respectively.

With a good battery, you can use either mode, regardless of whether you are plugged in or not.
With a not-so-good battery, you are locked to the slower 667 with no L3, regardless of whether you are plugged in or not.

I have not tested this (because my 50W charger is in use by the Twelve), but I'm also under the assumption that you'll be locked to the slower speed if you use a charger that provides less than 65W of power. Not the case, see below post.

The 867 also does this, as does the previous generation 800 MHz model.

 

[Sebasboy]

There are two states: 667 with no L3 and 1000 with L3; these correspond to the "Slower Processor Speed" and "Faster Processor Speed" options respectively.

With a good battery, you can use either mode, regardless of whether you are plugged in or not.
With a not-so-good battery, you are locked to the slower 667 with no L3, regardless of whether you are plugged in or not.

I have not tested this (because my 50W charger is in use by the Twelve), but I'm also under the assumption that you'll be locked to the slower speed if you use a charger that provides less than 65W of power.

The 867 also does this, as does the previous generation 800 MHz model.

cpu speed is not limited with a slower charger

got a bunch of chargers here and functions fine on my 45watt one aswel

 

[Doq]

Thanks for the confirmation!

 

[Macbookprodude]

What chud tools version can I use with Leo on the Titanium 1ghz ?

 

[ervus]

With a not-so-good battery, you are locked to the slower 667 with no L3, regardless of whether you are plugged in or not.

Thanks for investigating this. So on a weak battery, is the option greyed out or missing? Or does it let you choose "Highest" performance but it doesn't actually do it?

 

[Doq]

You can click the option, but it won't select.

 

[Sebasboy]

i just found old developer documentation on the cpu throttling., heres the official reasons for throttle

 

[Macbookprodude]

i just found old developer documentation on the cpu throttling., heres the official reasons for throttle
View attachment 1840847

You want to know what i think ? I think Apple was stupid to do this. stupidity to allow the battery to control your cache ? But what am I saying.. Steve was so great but he failed to see this HYPOCRISY !! Such a stupid move. Also, Machspeed control by Daystar also damages and destroys the L3 Cache, well not destroys it, but disables it. AVOID THIS APP IF YOU CAN ! I had to reinstall Leo to get it to recognize L3 again. What a stupid move by Apple !

 

[B S Magnet]

You want to know what i think ?

N… Not really?

 

[Macbookprodude]

N… Not really?

How come PC laptops then didn’t do this ?

 

[Sebasboy]

How come PC laptops then didn’t do this ?

plenty of newer PC's throttle if you remove their battery

msi ge66. razer blade stealth. ect.

some are more aggressive than others

 

[Macbookprodude]

plenty of newer PC's throttle if you remove their battery

msi ge66. razer blade stealth. ect.

some are more aggressive than others

That is now, not 2000-2002.

 

[galgot]

Titanium 1Ghz (...?), what seems a working battery, benchmark results from Geekbench 2.2.7 on a fresh 10.5 install :

Benchmark Summary
Integer Score 777 ||||||||||||||||||||||||||||
Floating Point Score 664 ||||||||||||||||||||||||
Memory Score 439 ||||||||||||||||
Stream Score 301 |||||||||||

Geekbench Score 622 ||||||||||||||||||||||

System Information
Operating System Mac OS X 10.5.8 (Build 9L31a)
Model PowerBook G4 (1 GHz/867 MHz)
Motherboard PowerBook3,5
Processor PowerPC G4 (7445/7455) @ 1.00 GHz
1 Processor
Processor ID 8001h
L1 Instruction Cache 32.0 KB
L1 Data Cache 32.0 KB
L2 Cache 256 KB
L3 Cache 1.00 MB
Memory 512 MB PC133U-333 SDRAM
BIOS

Integer Performance
Blowfish
single-threaded scalar 825 ||||||||||||||||||||||||||||||
multi-threaded scalar 883 ||||||||||||||||||||||||||||||||
Text Compress
single-threaded scalar 694 |||||||||||||||||||||||||
multi-threaded scalar 627 |||||||||||||||||||||||
Text Decompress
single-threaded scalar 813 |||||||||||||||||||||||||||||
multi-threaded scalar 811 |||||||||||||||||||||||||||||
Image Compress
single-threaded scalar 901 |||||||||||||||||||||||||||||||||
multi-threaded scalar 888 ||||||||||||||||||||||||||||||||
Image Decompress
single-threaded scalar 669 ||||||||||||||||||||||||
multi-threaded scalar 678 ||||||||||||||||||||||||
Lua
single-threaded scalar 771 ||||||||||||||||||||||||||||
multi-threaded scalar 769 ||||||||||||||||||||||||||||

Floating Point Performance
Mandelbrot
single-threaded scalar 501 ||||||||||||||||||
multi-threaded scalar 509 ||||||||||||||||||
Dot Product
single-threaded scalar 800 |||||||||||||||||||||||||||||
multi-threaded scalar 848 |||||||||||||||||||||||||||||||
single-threaded vector 948 ||||||||||||||||||||||||||||||||||
multi-threaded vector 1088 ||||||||||||||||||||||||||||||||||||||||
LU Decomposition
single-threaded scalar 149 |||||
multi-threaded scalar 133 ||||
Primality Test
single-threaded scalar 962 |||||||||||||||||||||||||||||||||||
multi-threaded scalar 724 ||||||||||||||||||||||||||
Sharpen Image
single-threaded scalar 602 ||||||||||||||||||||||
multi-threaded scalar 611 ||||||||||||||||||||||
Blur Image
single-threaded scalar 714 ||||||||||||||||||||||||||
multi-threaded scalar 717 ||||||||||||||||||||||||||

Memory Performance
Read Sequential
single-threaded scalar 294 ||||||||||
Write Sequential
single-threaded scalar 808 |||||||||||||||||||||||||||||
Stdlib Allocate
single-threaded scalar 544 ||||||||||||||||||||
Stdlib Write
single-threaded scalar 271 |||||||||
Stdlib Copy
single-threaded scalar 282 ||||||||||

Stream Performance
Stream Copy
single-threaded scalar 284 ||||||||||
single-threaded vector 307 |||||||||||
Stream Scale
single-threaded scalar 293 ||||||||||
single-threaded vector 304 |||||||||||
Stream Add
single-threaded scalar 291 ||||||||||
single-threaded vector 324 |||||||||||
Stream Triad
single-threaded scalar 316 |||||||||||
single-threaded vector 292 ||||||||||

 

[B S Magnet]

Titanium 1Ghz (...?), what seems a working battery, benchmark results from Geekbench 2.2.7 on a fresh 10.5 install :

Yah, 622 sounds about right for 1.0GHz with an L3 cache. I’m guessing your battery is OK enough for the system not to downclock.

On my A1138 and A1139, I find the 1.67GHz (without L3 cache) flirts between 984 and 996 in Geekbench 2.2.7. If an L3 cache existed in the 7447, I would expect that figure to flirt closer to 1040.

 

[Doq]

Can confirm-- that's actually a fair amount higher than what I get, but my Tiger was nowhere near fresh.

As for why it doesn't track in OpenBSD, #32 might have your answer:

The system software uses a reduced processor speed
- during system startup

So it may be possible that OpenBSD doesn't ever switch to the full speed.

 

[DearthnVader]

i just found old developer documentation on the cpu throttling., heres the official reasons for throttle
View attachment 1840847

That's interesting stuff, I wonder if there is anyway for us to control it from Open Firmware?

With the 7447a/b iBooks and PowerBooks there are some DFS words on the CPU node of the device tree in OF that allow us to control DFS from OF.

This was useful for Mac OS 9 on these "unsupported" models, because Mac OS 9's power management stack can't control DFS as no 7447a/b ever shipped in a Mac supporting OS 9.

The default is dfs-low so these 'Books operate at reduced speeds during boot, thus they only operate at reduced speeds running OS 9, unless you set-dfs-high in OF before you boot OS 9.

Then they operate at only full speed in OS 9.