Jump to content

Core 2 Duo with both cores

Shimms

By Shimms,
in OSx86 10.4 (Tiger)

 Share
59 posts in this topic

Recommended Posts

guiyoforward
guiyoforward
Posted
Posted

hey semthex thanks for your interest! I didn't know you had worked on the issue. sorry I don't know what RTC is, but it does look like a bios-acpi issue. besides the ideas to fix it, did you find anything abnormal? it's really weird that single core works flawlessly...

 

thanks again

 

g.

Link to comment
Share on other sites
kevinosx
kevinosx
Posted
Posted

RTC=Real Time Clock? I think,

I'm going to do the dell a06 bios update for december 23rd 2006 and allow both cores to see what happens, then I will post back the results!

 

This is on a dell 9400/E1705 t2500 core duo

 

P.S. if it's a real time clock issue, I've noticed that the bios is always one hour off the mac clock. I don't know if that means anything or not

Link to comment
Share on other sites
equilibriumuk
equilibriumuk
Posted
Posted (edited)

I have just upgraded my 10.4.7 to 10.4.8 with one of the new kernels.

 

I cannot seem to get the new kernel to boot without adding both "cpus=1" and "idehalt=1" but I also cannot get my old kernel to boot now (mach_kernel.bak) without using cpus=1, whereas before I was running on both cores?

 

I also notice now with both kernels even with these boot flags I can only manage to get to the blue screen with the loading swirl in the middle. Nothing else happens :thumbsup_anim:

 

I should've backed up my extensions folder with all my kexts I think :( might have to try a re-install.

 

My Sys: P5W-DH, E6400, X1900XT, SBLive USB

 

EDIT: I notice the kernel I have is - xnu-792.13.8-mifki-3

Edited by equilibriumuk
Link to comment
Share on other sites
kevinosx
kevinosx
Posted
Posted

hmm, that's interesting equilibriumuk, have you tried semthex's newest kernel for 10.4.8? he has been working on those issues for a while. Give it a try first. you can find an auto updating script that downloads the latest kernel from his site and installs it automatically. do a search for script3vcom I believe.

 

UPDATE: I installed the new bios update for the dell 9400/e1705 and it didn't affect the stuttering problem in any way. I'm really thinking about what semthex said in regards to the RTC (if it means real time clock), because in the bios for example it says 1pm then in Windows it says, 1 pm. But if I go into mac it says 2pm.

 

If I make an adjustment to the correct time in mac then it adjusts the bios clock (and windows time) 1 hour prior to whatever I set in mac.

 

I wonder if there is a program that fixes this for mac?

Link to comment
Share on other sites
kevinosx
kevinosx
Posted
Posted
You time zone setting is not related to RTC problems in general. :-)

ah, and here I thought I was getting off on a hunch!! Breakthrough information my friends!! My timezone settings are off!!!

so, I was looking into a way to custom flash the bios with a non-Dell version and it seems that Dell is so proprietry that you can't...according to a website they have custom only dell boards, only dell versions of other stuff etc.

I'm going to look more into it to see if it is even possible!

Link to comment
Share on other sites
equilibriumuk
equilibriumuk
Posted
Posted

I have installed the latest kernel now on a fresh install. The kernel boots ok without the cpus=1 flag, but I had to use the idehalt=0 one for some reason even tho I installed the JMicron patch.

 

I now get an error about the frame pointer.

loading loginwindow.app

panic(cpu1 caller 0x001A2ECF): kernel_trap (0x3a42524) with 64-bit state

no mapping exists for frame pointer
Link to comment
Share on other sites
kevinosx
kevinosx
Posted
Posted

guiyoforward:

Here is the info console gives me when I enable both cores and it stutters:

what can you make of it? I see it messing with the mDNSPlatformRawTime a lot...

Jan 21 23:26:22 localhost kernel[0]: Jettisoning kernel linker.

Jan 21 23:26:22 localhost kernel[0]: Resetting IOCatalogue.

Jan 21 23:26:22 localhost kernel[0]: VID: family specific matching fails

Jan 21 23:26:22 localhost kernel[0]: VID2: family specific matching fails

Jan 21 23:26:22 localhost kernel[0]: Matching service count = 2

Jan 21 23:26:22 localhost kernel[0]: Matching service count = 4

Jan 21 23:26:22 localhost kernel[0]: Matching service count = 4

Jan 21 23:26:22 localhost kernel[0]: Matching service count = 4

Jan 21 23:26:22 localhost kernel[0]: Matching service count = 4

Jan 21 23:26:22 localhost kernel[0]: Matching service count = 5

Jan 21 23:26:22 localhost kernel[0]: VID: family specific matching fails

Jan 21 23:26:30 kevin-porters-computer mDNSResponder: mDNSPlatformRawTime: last_mach_absolute_time 39219C8900000006

Jan 21 23:26:32 kevin-porters-computer mDNSResponder: mDNSPlatformRawTime: this_mach_absolute_time D9312A5900000005

Jan 21 23:26:32 kevin-porters-computer mDNSResponder: mDNSPlatformRawTime: last_mach_absolute_time 456172E700000006

Jan 21 23:26:32 kevin-porters-computer mDNSResponder: mDNSPlatformRawTime: this_mach_absolute_time D9433FCC00000005

Jan 21 23:26:30 kevin-porters-computer mDNSResponder: mDNSPlatformRawTime went backwards by 1814 ticks; setting correction factor to -2118906982

Jan 21 23:26:37 kevin-porters-computer mDNSResponder: mDNSPlatformRawTime: last_mach_absolute_time 612F7FC400000007

Jan 21 23:26:37 kevin-porters-computer mDNSResponder: mDNSPlatformRawTime: this_mach_absolute_time F529DF4200000006

Jan 21 23:26:40 kevin-porters-computer mDNSResponder: mDNSPlatformRawTime: last_mach_absolute_time 356D655F00000008

Jan 21 23:26:40 kevin-porters-computer mDNSResponder: mDNSPlatformRawTime: this_mach_absolute_time C94FA08C00000007

Jan 21 23:26:39 kevin-porters-computer mDNSResponder: mDNSPlatformRawTime went backwards by 1813 ticks; setting correction factor to -2118905169

 

So in Semthex's words, it looks like an RTC error maybe?

Link to comment
Share on other sites
Andrmgic
Andrmgic
Posted
Posted (edited)

About this Mac reports 1 core, but Activity Monitor has 2 boxes that have different results, so it would appear that both cores are working

 

However, I think my xbench scores look a little low for it to be working.

 

Any ideas?

 

Results 106.73

System Info

Xbench Version 1.3

System Version 10.4.8 (8L2127)

Physical RAM 2048 MB

Model ADP2,1

Drive Type MAXTOR STM3200820A

CPU Test 77.47

GCD Loop 128.88 6.79 Mops/sec

Floating Point Basic 93.55 2.22 Gflop/sec

vecLib FFT 46.59 1.54 Gflop/sec

Floating Point Library 85.30 14.85 Mops/sec

Thread Test 167.14

Computation 111.80 2.26 Mops/sec, 4 threads

Lock Contention 330.89 14.23 Mlocks/sec, 4 threads

Memory Test 117.59

System 112.26

Allocate 115.67 424.76 Kalloc/sec

Fill 117.60 5718.09 MB/sec

Copy 104.44 2157.26 MB/sec

Stream 123.45

Copy 111.72 2307.61 MB/sec

Scale 115.44 2385.03 MB/sec

Add 136.40 2905.56 MB/sec

Triad 134.08 2868.23 MB/sec

Quartz Graphics Test 141.92

Line 112.29 7.48 Klines/sec [50% alpha]

Rectangle 151.46 45.22 Krects/sec [50% alpha]

Circle 140.98 11.49 Kcircles/sec [50% alpha]

Bezier 128.25 3.23 Kbeziers/sec [50% alpha]

Text 206.91 12.94 Kchars/sec

OpenGL Graphics Test 125.20

Spinning Squares 125.20 158.83 frames/sec

User Interface Test 231.83

Elements 231.83 1.06 Krefresh/sec

Disk Test 53.08

Sequential 109.32

Uncached Write 76.84 47.18 MB/sec [4K blocks]

Uncached Write 122.71 69.43 MB/sec [256K blocks]

Uncached Read 116.00 33.95 MB/sec [4K blocks]

Uncached Read 146.96 73.86 MB/sec [256K blocks]

Random 35.05

Uncached Write 12.26 1.30 MB/sec [4K blocks]

Uncached Write 66.30 21.23 MB/sec [256K blocks]

Uncached Read 93.77 0.66 MB/sec [4K blocks]

Uncached Read 146.29 27.15 MB/sec [256K blocks]

 

System is as follows:

 

(I realize that this thread is about Core2 chips, but I seemed to be having a similar issue, so I thought someone here might know :) )

 

Kernel date from using uname -a

 

Darwin drew 8.8.1 Darwin Kernel Version 8.8.1: Sat Dec 9 22:18:27 AZOT 2006; semthex:/nebukadnezar/BUILD/obj/RELEASE_I386 i386 i386

 

Amd Athlon 64 X2 4400+ (@2.2ghz stock)

2048mb pc3200 ddr

Asus A8N-SLI Premium

Ati Radeon X1900XT 512mb

HDDs:

Maxtor 7200rpm ATA100 200gb (OS X)

Samsung Spinpoint SATA 300Gb/s 200gb (Windows Xp)

 

what would be considered a good xbench score for a setup like this? (Also, I don't have any of the natit or boris ati drivers installed at the moment because I'm missing a VGA cable (will pick one up tomorrow)

 

Thanks!

Edited by Andrmgic
Link to comment
Share on other sites
kevinosx
kevinosx
Posted
Posted (edited)

Seems the problem is in fact with the mDNSPlatformRawTime

 

Ok, did a little more digging, looks like there was an open source project for working on a VxWorks plugin that was having RTC issues aswell, their solution was to implement this string in the driver:

//================================================================================

========================================== // mDNSPlatformRawTime //================================================================================

==========================================

mDNSs32 mDNSPlatformRawTime( void )

{

return( (mDNSs32) tickGet() );

}

 

taken from http://www.opensource.apple.com/darwinsour...WorksIPv4Only.c

 

and I found another post where this problem was mentioned and Daemon gave his advice. he said:

"it seems to timer problem. Try to enable all ACPI, RTC stuff in your BIOS settings."

http://forums.osx86project.org/lofiversion...t32381-150.html

 

Here is a description of the mDNSResponder:

http://developer.apple.com/documentation/D...esponder.8.html

 

EDIT: I thought it was interesting that they made an RTC Compensation Patch in linux to help with cpu timer issues.

http://www.captain.at/howto-linux-real-time-patch.php

 

If I run Cinebench 9.5 with both cores enabled, it runs the program perfectly with no stuttering whatsoever, but everything around it stutters... so there has to be a solution.

And by the way, my results were right up there with the macbooks and macbook pro's!!! :D without both cores, it rates around a powerbook g4. Dell 9400/E1705 core duo 2.0 t2500 1 gig ram GMA 945 video

 

LET'S UNTAP THE POTENTIAL OF OUR CORE DUO'S!!!!!!!!!!!!!

Edited by kevinosx
Link to comment
Share on other sites
caracols
caracols
Posted
Posted

Both cores are working perfectly on my system. I'm using latest semthex kernel and audio and ethernet patches,and not a single kernel panic since I first installed. Has to be a specific combination of mobo+dual core cpu what is causing most users to have problems, not the kernel itsef.

Link to comment
Share on other sites
equilibriumuk
equilibriumuk
Posted
Posted (edited)
equilibriumuk: what did you enter in the plist? was it something like this: platform=X86PC cpus=2 -lagacy ??

All I'm using in my com.apple.Boot.plist are:

<key>Kernel Flags</key>
	<string>-legacy fn=3</string>

I didn't need the cpus= or the platform.

On system profiler it picks up both cores etc perfectly :guitar:

Edited by equilibriumuk
Link to comment
Share on other sites
semthex
semthex
Posted
Posted

Let me try to explain (AGAIN) what I think it is up to:

 

Dell as well as some other manufactures use messed up BIOS with a bad Windows trimmed ACPI. This ACPI lacks features which are required for the timing and will not let it work proper. You can't fix this since there is nothign you can do. The darwin kernel was designed for Apple computers using dual core themself, but they have proper hardware for this and not a {censored} Dell blackhole Dimension BIOS controlled one.

You could solve this, if you would be able to enable things in your bios, but Dell is not that nice, they keep you short as well as other manufactur do. Get a read on ACPI and how {censored} this table can be on serveral machines, than you might have a clue what is going on, but hell you won't be able to fix this.

 

:thumbsup_anim:

Link to comment
Share on other sites
guiyoforward
guiyoforward
Posted
Posted

thanks for your intervention, semthex. I will try Ubuntu or some decent linux distro on my laptop. if dual core works in there, with no stuttering or problems, it would mean that somehow these guys have overcome the ACPI problems, and thus perhaps we can use their source for inspiration? my 2c/1000 :)

Link to comment
Share on other sites
kevinosx
kevinosx
Posted
Posted (edited)

yeah they got it working, be cool to see what they did, I'll look forward to your research guiyoforward!

 

check this out for linux

http://linux-on-laptops.com/forum/archive/....php/t-218.html

 

and I think I might play around with this a bit!

ACPI TOOLS

http://en.ecomstation.ru/projects/acpitools/

 

According to my research linux Ubantu and Debian have a tool I believe called pmtools that will let you edit the ACPI tables in the BIOS from Linux

Edited by kevinosx
Link to comment
Share on other sites
consolation
consolation
Posted
Posted
You time zone setting is not related to RTC problems in general. :-)

 

just to be anal retentive and expand semthex's answer:

 

OsX like most systems with a unix heritage assumes that the clock is displaying UTC (kind of like the computer world's Greenwich Mean Time). It then uses your localisation settings to adjust it to display the correct LOCAL time. Windows assumes that the clock runs in LOCAL time and whenever it needs to use UTC applies an adjustment. Hence the weird resetting of the clock.

Link to comment
Share on other sites
myzar
myzar
Posted
Posted
Duel cores work in all OSs, except OS X :thumbsdown_anim:

 

If Linux kernel can overcome this, why cant Mac kernel?

 

Because i bet they are using Hpet for timings and not the rtc. Try using the hacked 10.4.9 kernel to see if it helps if it doesn't you are pretty much sol

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