Jump to content

Voodoo Kernel BETA 2 - Available now

mercurysquad

By mercurysquad,
in New Releases and Updates

 Share
325 posts in this topic

Recommended Posts

aliasa_anderson
aliasa_anderson
Posted
Posted
Just use IntelCPUPMDisabler.kext prior to the 10.5.5 update.

You can get it here: http://www.mediafire.com/?rpjxduvifls

I use it on a T5500 + this kernel and it works great.

 

Hi Badeendje,

 

Thanks for the link.

What does this kext do actually.....what about using it on Vanilla kernel ?

 

Mercurysquad,

 

On google groups link for vodoo kernel you've said that :

"Based on testing done by ritalin, we recommend using vanilla AppleSMBIOS.kext together with Chun-nan's SMBIOSEFI.kext."

 

Did you try

 

http://forum.insanelymac.com/index.php?sho...mp;#entry960278

its improved SMBIOS by Andy.According to him:

 

"I have recently combined all of the advantages of Macdotnub's patches, AppleSMBIOSEFI's changes and Superhai's SMBIOSResolver nicely into my very own blend AppleSMBIOS.

You could say it's a neatly rolled version and it also allows setting of the targeted Mac platform through the Info.plist

If you use this compiled version, set the required memory type (e.g. DDR) you memory speed (for normal SD-RAM your FSB speed, for DDR(2) double your FSB speed), the required Mac platform, your desired serial number, etc...)

I also release the sources for these so have fun."

Link to comment
Share on other sites
mercurysquad
mercurysquad
Posted
  • Author
Posted
Mercurysquad,

 

With latest vodoo kernel will I be able to use stock AppleIntelCPUPowermanagement.kext because in my case I have Intel core duo T2050 @1.6GHZ and as far as I know,it is capable of HPET but with vanilla I(and even others) still get "Cannot enable HPET" message whle booting.

If it does ,then would the IntelEnhancedSpeedStep.kext be useful ?

 

No. Enabling HPET is a function of the BIOS. If the BIOS does not properly assign an IRQ to it (and assign it to the legacy rtc instead), then the kernel can't make use of it. Fixing this requires modifying the DSDT. Work on this is being done by other people independent of the kernel so it is likely this will be fixed.

 

Until then people can use our own speedstep kext which does an okay job and works with a larger variety of processors.

 

Did you try

 

http://forum.insanelymac.com/index.php?sho...mp;#entry960278

its improved SMBIOS by Andy.According to him:

 

"I have recently combined all of the advantages of Macdotnub's patches, AppleSMBIOSEFI's changes and Superhai's SMBIOSResolver nicely into my very own blend AppleSMBIOS.

You could say it's a neatly rolled version and it also allows setting of the targeted Mac platform through the Info.plist

If you use this compiled version, set the required memory type (e.g. DDR) you memory speed (for normal SD-RAM your FSB speed, for DDR(2) double your FSB speed), the required Mac platform, your desired serial number, etc...)

I also release the sources for these so have fun."

 

While this work is cool, a lot of people believe the proper way is to keep patched kexts to a minimum and use enabler/resolver/injector kexts in addition to the stock kexts. If you use a patched SMBIOS kext, come next update and BOOM - it's gone. With a 'resolver' kext you can continue to update while the extra kext will continue to work with the new ones. Case in point: so many people are posting they cannot boot anymore after 10.5.5 update because they simply deleted the AppleIntelCPUPowerManagement.kext instead of using PMdisabler.kext.

Link to comment
Share on other sites
aliasa_anderson
aliasa_anderson
Posted
Posted

Tried the latest beta2c.Woks Perfectly!!!

Thanks Mercurysquad.

 

All tests passed.

 

my config is:

 

Intel Core Duo processor T2050@1.6GHZ

553MHz DDR2 RAM,945M motherboard with ICH7.

 

HP Laptop model dv6114tx.

 

Mercurtysquad,

does this kernel use HPET or RTC ?

Are both necessary ?

and in / directory, what is the porpose of mach_kernel.ctfsys ?

 

Thanks once again ?

Link to comment
Share on other sites
Mad Dogg
Mad Dogg
Posted
Posted

Hi

Here is my Beta 2c testing results:-

 

Sleep/Resume Passed: [X] Yes [_] No [X] With issues (Pressing the power button doesn't work)

Shutdown Passed: [X] Yes [_] No [_] With issues

Audio/Video performance Passed: [X] Yes [_] No [X] With issues (VLC having some video problems after about 9+ hours uptime).

AMD Athlon Dual-core Timing Passed: [X] Yes [_] No [_] With issues

TSC stability Passed: [X] Yes [_] No [_] With issues

Phenom TSC sync Passed: [_] Yes [_] No [_] With issues

Cache information Passed: [X] Yes [_] No [_] With issues

 

Note:

  • Overall performance is at least 50% better (Logic Pro 8 is on fire :( ) even boots up faster.
  • USB Flash Drives and USB Hard drives are not recognized at all.

I hope that was helpful

 

Thank you

Link to comment
Share on other sites
mercurysquad
mercurysquad
Posted
  • Author
Posted
Mercurtysquad,

does this kernel use HPET or RTC ?

Are both necessary ?

and in / directory, what is the porpose of mach_kernel.ctfsys ?

 

1) Kernel uses TSC (timestamp counter) to keep the time. HPET is currently useless unless you want to use AppleIntelCPUPM kext. Kernel supports HPET if possible. RTC is also used for other things in the kernel.

 

2) HPET is not necessary. RTC is.

 

3) It is needed for debugging. Can be safely ignored by most users.

Link to comment
Share on other sites
mitch_de
mitch_de
Posted
Posted

mercurysquad

THANKs for explaining the background things - its very interesting to know why, what how works !

Q:

Do you know when the chameleon team has the (3 weeks;soon) announced new bootloader with some "great new things" available ?

This chameleon team (or netkas group) didnt have so much OPEN SHOWN (updated Blogs or an forum) friedly contact with the uesers like you or also friendly communicating superhai.

Link to comment
Share on other sites
aliasa_anderson
aliasa_anderson
Posted
Posted
1) Kernel uses TSC (timestamp counter) to keep the time. HPET is currently useless unless you want to use AppleIntelCPUPM kext. Kernel supports HPET if possible. RTC is also used for other things in the kernel.

 

2) HPET is not necessary. RTC is.

 

3) It is needed for debugging. Can be safely ignored by most users.

 

Thank You very much for clearing all my doubts!!!

 

I have one more question,

I found that with Leopard i used to get dependecy warnings while booting like:

AppleACPIPlatform.kext has dependencies on both kernel and kpi using only one style.

 

but when i used your voodoo beta2c kernel,i never got those kpi and kernel dependency warnings,even though the Info.plist for this kext still has those dependencies

Is this normal ? Have you used exception handling to catch those exceptions/warnings ?

Link to comment
Share on other sites
mercurysquad
mercurysquad
Posted
  • Author
Posted
mercurysquad

THANKs for explaining the background things - its very interesting to know why, what how works !

Q:

Do you know when the chameleon team has the (3 weeks;soon) announced new bootloader with some "great new things" available ?

This chameleon team (or netkas group) didnt have so much OPEN SHOWN (updated Blogs or an forum) friedly contact with the uesers like you or also friendly communicating superhai.

 

Chameleon is seeing heavy activity too. Zef/Kabyl/Turbo are busy merging and rewriting a lot of the core code to provide new and exciting features not seen in any EFI bootloader before. Hopefully it will be done and released together with Voodoo kernel (which is also almost finished). I can't speak for them but I think the reason they the chameleon team doesn't look so "open" is because: (1) They know what they're doing and don't need a lot of testers or community involvement, so they are able to concentrate on just the actual work, and (2) We're all now more-or-less working together and communicate frequently.

 

I found that with Leopard i used to get dependecy warnings while booting like:

AppleACPIPlatform.kext has dependencies on both kernel and kpi using only one style.

 

but when i used your voodoo beta2c kernel,i never got those kpi and kernel dependency warnings,even though the Info.plist for this kext still has those dependencies

Is this normal ? Have you used exception handling to catch those exceptions/warnings ?

 

This warning was removed in Voodoo kernel (re: Dense), because it was merely annoying and as of now doesn't affect the kext's ability to function in any way. When Apple finally decide to block the ability to load kexts with mixed kpi and kernel components, then we'll revert it. Doesn't seem likely anytime soon though.

Link to comment
Share on other sites
Superhai
Superhai
Posted
Posted
This warning was removed in Voodoo kernel (re: Dense), because it was merely annoying and as of now doesn't affect the kext's ability to function in any way. When Apple finally decide to block the ability to load kexts with mixed kpi and kernel components, then we'll revert it. Doesn't seem likely anytime soon though.

 

Seem like in snow all kernel components are gone. All is moved to kpi.

Link to comment
Share on other sites
entitas
entitas
Posted
Posted

I got a apparent solution for now, I think. As I am using JaS 10.5.4 (updated to 10.5.5) disk and it contained Chameleon 1.0.10. Since it's a older version, it seems that there were some incompatibilities with Voodoo 9.5. Now I've updated to Chameleon 1.0.11 and Voodoo is working fine, feels a lot more snappier. But surprisingly XBench and Geekbench scores are at an all time low! XBench is giving 66% lower score than Voodoo Beta 2c (with Chameleon 1.0.10) and 26% lower score than StageXNU 10.5.4. Interestingly, StageXNU is still giving almost same score when Voodoo Beta 2c scores suck bit time. Why is this discrepancy?

Link to comment
Share on other sites
entitas
entitas
Posted
Posted
XBench is {censored}.
But the same thing happens with Geekbench also. Now after upgrading to Safari 3.2, it has become so crash-prone that I can't browser for 2 minutes with it! Should I post the complete Safari crash report here?

 

Most notably the crash report is saying that

Exception Type: EXC_BAD_ACCESS (SIGBUS)

Exception Codes: KERN_PROTECTION_FAILURE at 0x0000000000000005

Crashed Thread: 0

Link to comment
Share on other sites
mitch_de
mitch_de
Posted
Posted
But the same thing happens with Geekbench also. Now after upgrading to Safari 3.2, it has become so crash-prone that I can't browser for 2 minutes with it! Should I post the complete Safari crash report here?

Exception Type: EXC_BAD_ACCESS (SIGBUS)

Exception Codes: KERN_PROTECTION_FAILURE at 0x0000000000000005

Crashed Thread: 0

Most notably the crash report is saying that

I think its better to add only the last steps before the crash thread 0 info lines also (bottom lines =past, first lines =last "doings")

So look for those thread 0 lines (only they will be interesting) in the crash.log (they ar archived, use Console to get it shown)

Link to comment
Share on other sites
entitas
entitas
Posted
Posted
This is probably because of opcode patcher. Fixed in beta3.

 

Most pleased to hear it's fixed, may I have a link to that beta3 kernel?

 

BTW can you please explain that severe deterioration of Geekbench score?

Link to comment
Share on other sites
mercurysquad
mercurysquad
Posted
  • Author
Posted
Most pleased to hear it's fixed, may I have a link to that beta3 kernel?

 

BTW can you please explain that severe deterioration of Geekbench score?

 

Beta3 is not yet available.

 

As for Geekbench: first you need to run Test_5.command and make sure there are no timing issues. If the timing test passes then attach your Geekbench scores from voodoo and another non-voodoo kernel so we can compare and see which area exactly is taking a performance hit.

Link to comment
Share on other sites
 Share
Go to topic listing
×
×
  • Create New...