New kernel 8.8.1 and Xbench for ASRock Conroe945G-DVI

erikk · October 26, 2006

hey ho ...

really nice addition... i get in xbench:

189 points now (without hd testing) (Native)

47 points now (without hd testing) (rosetta)

really amazing..

alicheusz · October 26, 2006

try setting
fsb=<mhz>

boot option with my latest kernel (from http://www.filefactory.com/file/f618b4)

WOW that is amazing, it look like for the first time we have software overcloker on hackintosh

I am interesting did this will also work on my s**t

PS: multiplier would be cool also

erikk · October 26, 2006

WOW that is amazing, it look like for the first time we have software overcloker on hackintosh

I am interesting did this will also work on my s**t

PS: multiplier would be cool also

nice idea... but very strange....

when you use a lower fsb as your cpu is running... the finder animations and clock will be faster... but the overall system performance will be slower...

boo50 · October 26, 2006

there is no perfomance improvement if you'll set fsb=500 or 1000, only with correct fsb you'll get your system working properly.

ReLoad · October 26, 2006

definately this is NOT software overclocking - just put fsb clock equal to BIOS setting

alicheusz · October 26, 2006

definately this is NOT software overclocking - just put fsb clock equal to BIOS setting

Maybe I don't understand something, but if I have 133 MHz FBS * 19 = 2527 MHZ CPU and can change it to np 150 MHz x 19 = 2850 MHz then I definitely have some CPU improvements.

If I only can adjust fsb at boot.

there is no perfomance improvement if you'll set fsb=500 or 1000, only with correct fsb you'll get your system working properly.

According to you If I only could put my fsb to 1000 MHz I will not have any speed improvements,

simple calculation

1000 MHz * 19 = 19 000 MHz this will be faster then core 2 duos

~Neo · October 26, 2006

Maybe I don't understand something, but if I have 133 MHz FBS * 19 = 2527 MHZ CPU and can change it to np 150 MHz x 19 = 2850 MHz then I definitely have some CPU improvements.

If I only can adjust fsb at boot.

According to you If I only could put my fsb to 1000 MHz I will not have any speed improvements,

simple calculation

1000 MHz * 19 = 19 000 MHz this will be faster then core 2 duos

But this option doesnt let you change the fsb!!!

Its only for the kernel, that "he" knows what your fsb is!!

alicheusz · October 26, 2006

But this option doesnt let you change the fsb!!!
Its only for the kernel, that "he" knows what your fsb is!!

Thanks, I wrongly understood this fsb at boot, I will experiment with this later, After we will have some simple installation for new kernel.

PS: Conroe945G-DVI is great mobo I think that this is best you can get for hackintosh for this money. I will buy one when I get some money.

Infamous · October 26, 2006

Well, what you set in BIOS is not always exactly correct.

I mean if we set 200, in reality it's like 200,9 or 199,5 or so... it's not always EXACT.

So it will always break my time? How much? I think I just figured out why the time is always some seconds behind or ahead when I don't sync the time from Timeserver for a month.

Because any OS's kernel reads the FSB and it's not completly exact correct.... ?

cardinal · October 26, 2006

Well, what you set in BIOS is not always exactly correct.
I mean if we set 200, in reality it's like 200,9 or 199,5 or so... it's not always EXACT.

So it will always break my time? How much? I think I just figured out why the time is always some seconds behind or ahead when I don't sync the time from Timeserver for a month.

Because any OS's kernel reads the FSB and it's not completly exact correct.... ?

As far as I understand it, you are close, but it is not because of rounding error of the kind you describe above. Your OS keeps time based on how many ticks passed since computer was turned on, not based on built-in, battery powered, quartz clock (which is what keeps time while computer is off). So, it is not a rounding error, but rather imperfect oscilators that can slightly change frequency with system load or whatever.

kossi · October 26, 2006

System runs nicely now with fsb kernel flag. Thanks mifki!

Did anyone handle that "about this mac" issue? The gui restarts when you click on it...

Not as bad as a speeding clock, but still an issue

kossi

zam · October 26, 2006

Yes oh yes, respect to mifki/Vitality! The new kernel runs superb and the clock speed is fine with the fsb option. If Seagata drives did decent in Xbench, I would be scoring great, but this is not bad either:

(E6600 on Conroe945G-DVI, running stock 2.4 GHz)

Results 141.31
System Info

Xbench Version 1.3

System Version 10.4.8 (8L2127)

Physical RAM 2048 MB

Model ACPI

Drive Type ST3500630AS

CPU Test 124.82

GCD Loop 282.20 14.88 Mops/sec

Floating Point Basic 136.62 3.25 Gflop/sec

vecLib FFT 100.52 3.32 Gflop/sec

Floating Point Library 89.01 15.50 Mops/sec

Thread Test 240.02

Computation 218.80 4.43 Mops/sec, 4 threads

Lock Contention 265.79 11.43 Mlocks/sec, 4 threads

Memory Test 134.62

System 151.51

Allocate 119.36 438.34 Kalloc/sec

Fill 163.85 7966.80 MB/sec

Copy 187.99 3882.77 MB/sec

Stream 121.12

Copy 115.63 2388.22 MB/sec

Scale 120.98 2499.40 MB/sec

Add 123.92 2639.69 MB/sec

Triad 124.39 2660.90 MB/sec

Quartz Graphics Test 171.91

Line 147.00 9.79 Klines/sec [50% alpha]

Rectangle 180.44 53.87 Krects/sec [50% alpha]

Circle 173.74 14.16 Kcircles/sec [50% alpha]

Bezier 161.22 4.07 Kbeziers/sec [50% alpha]

Text 209.12 13.08 Kchars/sec

OpenGL Graphics Test 264.45

Spinning Squares 264.45 335.47 frames/sec

User Interface Test 537.40

Elements 537.40 2.47 Krefresh/sec

Disk Test 54.14

Sequential 108.55

Uncached Write 72.65 44.61 MB/sec [4K blocks]

Uncached Write 111.06 62.84 MB/sec [256K blocks]

Uncached Read 171.10 50.07 MB/sec [4K blocks]

Uncached Read 121.41 61.02 MB/sec [256K blocks]

Random 36.06

Uncached Write 12.11 1.28 MB/sec [4K blocks]

Uncached Write 98.79 31.63 MB/sec [256K blocks]

Uncached Read 89.91 0.64 MB/sec [4K blocks]

Uncached Read 140.73 26.11 MB/sec [256K blocks]

TopazBar · October 27, 2006

This is the last (hopely) remaining issue. I've looked at it and I know this much:

When you click on about this mac, it crashes loginwindow and then, launchd re-starts it. no biggy.

loginwindow crashes because of EXC_BAD_ACCESS. It is probably accessing unmapped virtual address.

Now, it does not occur on mifki's mobo. So it looks like something particular to our mobo and we may have to solve our selves.

...Did anyone handle that "about this mac" issue? The gui restarts when you click on it...

Not as bad as a speeding clock, but still an issue

kossi

erikk · October 27, 2006

for xbench cheaters its a really nice option with fsb

in normal check with my real fsb i get around 150 points (with hdd)

now i set fsb=350 and im on 190 points (with hdd)

great

icyderguru · October 27, 2006

This is the last (hopely) remaining issue. I've looked at it and I know this much:
When you click on about this mac, it crashes loginwindow and then, launchd re-starts it. no biggy.

loginwindow crashes because of EXC_BAD_ACCESS. It is probably accessing unmapped virtual address.

Now, it does not occur on mifki's mobo. So it looks like something particular to our mobo and we may have to solve our selves.

it also doesnt occur on my system (975 xbx/6600 conroe)

Bamboo · October 28, 2006

I managed to install the nex kernel on my little homebrewn PC (E6600@2.82GHz, FSB@312MHz/PCIE Sync.@117MHz, dual screen with the onboard GMA950) which was using the 'old school' 10.4.4/10.4.8 JaS upgrade (I didn't really use any method to install it, just launched the official combo update, added mifki's files, rebooted and it worked). The overal speed is amazing (and yes, I did set the FSB settings right).

com.apple.Boot.plist						

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple Computer//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
	<key>Kernel</key>
	<string>mach_kernel</string>
	<key>Kernel Flags</key>
	<string>fsb=312 -v</string>
	<key>Boot Graphics</key>
	<string>Yes</string>
</dict>
</plist>

(Posted it full as a reference so that others won't have to get to the trial and error process I went through.)

Everything (including the onboard network thanks to DaemonES's Realtek r1000 driver) works. And it now actually 'feels' like a Mac.

And finally, a little Xbench vanity benchmark (HD disabled since they mean nothing (4k random block issue)) :

Results	223.34   
System Info	   
	Xbench Version		1.3
	System Version		10.4.8 (8L2127)
	Physical RAM		4096 MB
	Model		ACPI
	Drive Type		SAMSUNG SP2004C
CPU Test	145.83   
	GCD Loop	331.73	17.49 Mops/sec
	Floating Point Basic	159.64	3.79 Gflop/sec
	vecLib FFT	116.70	3.85 Gflop/sec
	Floating Point Library	104.37	18.17 Mops/sec
Thread Test	281.49   
	Computation	258.73	5.24 Mops/sec, 4 threads
	Lock Contention	308.63	13.28 Mlocks/sec, 4 threads
Memory Test	159.24   
	System	188.84   
		Allocate	166.36	610.92 Kalloc/sec
		Fill	189.84	9230.25 MB/sec
		Copy	217.02	4482.46 MB/sec
	Stream	137.66   
		Copy	129.72	2679.37 MB/sec
		Scale	134.84	2785.76 MB/sec
		Add	143.01	3046.43 MB/sec
		Triad	144.11	3082.82 MB/sec
Quartz Graphics Test	211.30   
	Line	172.79	11.50 Klines/sec [50% alpha]
	Rectangle	218.45	65.22 Krects/sec [50% alpha]
	Circle	209.35	17.06 Kcircles/sec [50% alpha]
	Bezier	190.83	4.81 Kbeziers/sec [50% alpha]
	Text	304.81	19.07 Kchars/sec
OpenGL Graphics Test	310.11   
	Spinning Squares	310.11	393.39 frames/sec
User Interface Test	450.94   
	Elements	450.94	2.07 Krefresh/sec

Take care,

B.

Infamous · October 28, 2006

Thanks for the info Bamboo...

a little bit offtopic, but I want to ask you, why are you running 4x 1GB only at 333mhz? Is it the board/chipset limitation? So it can not utilize that much of memory at decent speeds?

I'm interested because i'm planning to get 4GB too... Thanks.

Bamboo · October 28, 2006

a little bit offtopic, but I want to ask you, why are you running 4x 1GB only at 333mhz? Is it the board/chipset limitation? So it can not utilize that much of memory at decent speeds?

I'm not sure if it's a limitation of the chipset (but 945's are getting old so it's very possible) and the mobo itself has 4 settings : Auto, 200MHz (DDRII400), 266MHz (DDRII533) and 333MHz (DDRII667).

I bought DDRII800 capable memory chips because they were hardly more expensive than DDRII667 chips and a bit more future-proof. The day Apple moves to 965's with GMA3000's, I'll (hopefully) just have to buy a new mobo to have a sensible speed increase. In the meantime, you've seen my Xbench. This hackintosh is quick !

In fact, even Firefox (2.0), which has the reputation of being dog slow on Macs and quick on PC's seems incredibly quick on my hackintosh since I've upgraded it to this real 10.4.8. I'm used to placebo-effect updates, but I doubt I'm its victim this time. It was just faster than my Firefoxes I use on my (less powerful) PC's at work.

Take care.

B.

Infamous · October 28, 2006

Oh.. sure.. I didn't realize you mean 333mhz as 667 option...

alright then, cool

I'll try the kernel in a while and post my success (if any).

aberracus · October 29, 2006

i have a 10.4.8 kernel version running right now,

And having some ISSUES, everything works perfectly, The Vide drivers are WOW!!

DVI output everything is recognized

look:

Radeon X1600 Series:

Chipset Model: Radeon X1600 Series

Type: Display

Bus: PCIe

VRAM (Total): 256 MB

Vendor: ATI (0x1002)

Device ID: 0x71c0

Revision ID: 0x0000

Displays:

SyncMaster:

Resolution: 1680 x 1050 @ 60 Hz

Depth: 32-bit Color

Core Image: Supported

Main Display: Yes

Mirror: Off

Online: Yes

Quartz Extreme: Supported

Rotation: Supported

i will see if it got two screens support as soon as i can it really looks awesome.

The neopheus installer for Video and Lan worked wonderfully

BUT

i have the works Xbench results since i got a conroe

look:

Previous xbench

Results 159.83

System Info

Xbench Version 1.3

System Version 10.4.7 (8J2135a)

Physical RAM 2048 MB

Model ADP2,1

Drive Type WDC WD3200KS-00PFB0

CPU Test 127.73

GCD Loop 308.36 16.25 Mops/sec

Floating Point Basic 149.10 3.54 Gflop/sec

vecLib FFT 90.02 2.97 Gflop/sec

Floating Point Library 97.48 16.97 Mops/sec

Thread Test 259.77

Computation 238.50 4.83 Mops/sec, 4 threads

Lock Contention 285.20 12.27 Mlocks/sec, 4 threads

Memory Test 159.90

System 151.53

Allocate 116.37 427.34 Kalloc/sec

Fill 194.01 9433.21 MB/sec

Copy 165.29 3414.07 MB/sec

Stream 169.25

Copy 158.30 3269.70 MB/sec

Scale 161.23 3330.88 MB/sec

Add 180.49 3844.90 MB/sec

Triad 179.40 3837.86 MB/sec

Quartz Graphics Test 191.04

Line 157.22 10.47 Klines/sec [50% alpha]

Rectangle 200.55 59.88 Krects/sec [50% alpha]

Circle 192.45 15.69 Kcircles/sec [50% alpha]

Bezier 173.60 4.38 Kbeziers/sec [50% alpha]

Text 258.44 16.17 Kchars/sec

OpenGL Graphics Test 205.63

Spinning Squares 205.63 260.85 frames/sec

User Interface Test 479.96

Elements 479.96 2.20 Krefresh/sec

Disk Test 73.09

Sequential 109.83

Uncached Write 124.17 76.24 MB/sec [4K blocks]

Uncached Write 111.04 62.82 MB/sec [256K blocks]

Uncached Read 87.74 25.68 MB/sec [4K blocks]

Uncached Read 125.57 63.11 MB/sec [256K blocks]

Random 54.77

Uncached Write 20.70 2.19 MB/sec [4K blocks]

Uncached Write 173.11 55.42 MB/sec [256K blocks]

Uncached Read 90.18 0.64 MB/sec [4K blocks]

Uncached Read 127.45 23.65 MB/sec [256K blocks]

With new kenrel

Results 86.31

System Info

Xbench Version 1.3

System Version 10.4.8 (8L2127)

Physical RAM 2048 MB

Model ACPI

Drive Type SAMSUNG SP2004C

CPU Test 82.64

GCD Loop 188.85 9.95 Mops/sec

Floating Point Basic 90.55 2.15 Gflop/sec

vecLib FFT 65.63 2.17 Gflop/sec

Floating Point Library 59.44 10.35 Mops/sec

Thread Test 160.54

Computation 146.94 2.98 Mops/sec, 4 threads

Lock Contention 176.92 7.61 Mlocks/sec, 4 threads

Memory Test 105.06

System 105.35

Allocate 88.97 326.73 Kalloc/sec

Fill 119.86 5827.93 MB/sec

Copy 112.42 2321.97 MB/sec

Stream 104.78

Copy 98.39 2032.12 MB/sec

Scale 101.21 2090.98 MB/sec

Add 110.75 2359.19 MB/sec

Triad 109.89 2350.74 MB/sec

Quartz Graphics Test 117.01

Line 96.91 6.45 Klines/sec [50% alpha]

Rectangle 118.47 35.37 Krects/sec [50% alpha]

Circle 113.03 9.21 Kcircles/sec [50% alpha]

Bezier 106.15 2.68 Kbeziers/sec [50% alpha]

Text 175.31 10.97 Kchars/sec

OpenGL Graphics Test 130.47

Spinning Squares 130.47 165.51 frames/sec

User Interface Test 285.03

Elements 285.03 1.31 Krefresh/sec

Disk Test 29.82

Sequential 60.79

Uncached Write 78.01 47.90 MB/sec [4K blocks]

Uncached Write 68.08 38.52 MB/sec [256K blocks]

Uncached Read 42.41 12.41 MB/sec [4K blocks]

Uncached Read 67.95 34.15 MB/sec [256K blocks]

Random 19.75

Uncached Write 6.60 0.70 MB/sec [4K blocks]

Uncached Write 54.32 17.39 MB/sec [256K blocks]

Uncached Read 51.11 0.36 MB/sec [4K blocks]

Uncached Read 76.29 14.16 MB/sec [256K blocks]

As you can see WORST results in every section....

How do i installed it:

Jas 10.4.6 + Mac official 10.4.8 update + Vitaly kernel

This is my boot psllist not sure if its ok...

<?xml version="1.0" encoding="UTF-8"?>

<!DOCTYPE plist PUBLIC "-//Apple Computer//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">

<dict>

<key>Kernel</key>

<string>mach_kernel</string>

<key>Kernel Flags</key>

<key>Boot Graphics</key>

</dict>

</plist>

Anyway everything "looks" perfect

but i think the clock is a liittle fast

gona test it now against real watch clock

anyone got an idea, help appreciated...

thanks

aberracus · October 29, 2006

my bootplist was wrong now is ok

<?xml version="1.0" encoding="UTF-8"?>

<!DOCTYPE plist PUBLIC "-//Apple Computer//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">

<dict>

<key>Kernel</key>

<string>mach_kernel</string>

<key>Kernel Flags</key>

<string></string> <<<<<<<<<< Line removed!!!!

<key>Boot Graphics</key>

</dict>

</plist>

now my xbench is better but my disk are slower maybe thats happening cause this disk is SATA I and teh other is SATA II ?

new xbench after correction boot.plist

Results 136.68

System Info

Xbench Version 1.3

System Version 10.4.8 (8L2127)

Physical RAM 2048 MB

Model ACPI

Drive Type SAMSUNG SP2004C

CPU Test 135.56

GCD Loop 307.42 16.20 Mops/sec

Floating Point Basic 148.07 3.52 Gflop/sec

vecLib FFT 108.90 3.59 Gflop/sec

Floating Point Library 96.91 16.88 Mops/sec

Thread Test 261.89

Computation 239.81 4.86 Mops/sec, 4 threads

Lock Contention 288.44 12.41 Mlocks/sec, 4 threads

Memory Test 174.87

System 180.15

Allocate 150.73 553.53 Kalloc/sec

Fill 194.66 9465.02 MB/sec

Copy 204.87 4231.59 MB/sec

Stream 169.89

Copy 158.38 3271.29 MB/sec

Scale 162.91 3365.62 MB/sec

Add 180.84 3852.33 MB/sec

Triad 179.79 3846.15 MB/sec

Quartz Graphics Test 197.40

Line 159.34 10.61 Klines/sec [50% alpha]

Rectangle 204.51 61.06 Krects/sec [50% alpha]

Circle 196.61 16.03 Kcircles/sec [50% alpha]

Bezier 178.49 4.50 Kbeziers/sec [50% alpha]

Text 287.77 18.00 Kchars/sec

OpenGL Graphics Test 214.99

Spinning Squares 214.99 272.72 frames/sec

User Interface Test 450.52

Elements 450.52 2.07 Krefresh/sec

Disk Test 44.72

Sequential 73.39

Uncached Write 44.08 27.06 MB/sec [4K blocks]

Uncached Write 113.07 63.97 MB/sec [256K blocks]

Uncached Read 70.16 20.53 MB/sec [4K blocks]

Uncached Read 114.72 57.66 MB/sec [256K blocks]

Random 32.15

Uncached Write 10.64 1.13 MB/sec [4K blocks]

Uncached Write 97.38 31.18 MB/sec [256K blocks]

Uncached Read 80.75 0.57 MB/sec [4K blocks]

Uncached Read 128.92 23.92 MB/sec [256K blocks]

is SATa I te culprit?

~Neo · October 29, 2006

i have a 10.4.8 kernel version running right now,

And having some ISSUES, everything works perfectly, The Vide drivers are WOW!!

DVI output everything is recognized

look:

Radeon X1600 Series:

Chipset Model: Radeon X1600 Series

Type: Display

Bus: PCIe

VRAM (Total): 256 MB

Vendor: ATI (0x1002)

Device ID: 0x71c0

Revision ID: 0x0000

Displays:

SyncMaster:

Resolution: 1680 x 1050 @ 60 Hz

Depth: 32-bit Color

Core Image: Supported

Main Display: Yes

Mirror: Off

Online: Yes

Quartz Extreme: Supported

Rotation: Supported

i will see if it got two screens support as soon as i can it really looks awesome.

The neopheus installer for Video and Lan worked wonderfully

Good to hear that it worked with my installer :2cents:

Could you post a "review" in this thread?

Would be perfect!

cYa NeoPheus

denisvj · October 29, 2006

Hi NeoPheus .

Thanks for the new driver ! , it is working fine but i am having an issue with the 3.4 drivers ( testing the new kernel ) and don't know what is the problem, im still trying to fix it but i put a picture here if you have an idea of what the problem can be.

By the way my card is Ati x1600 Pro.

http://img216.imageshack.us/img216/6162/errorlc4.png

Thanks !

Bamboo · October 29, 2006

now my xbench is better but my disk are slower maybe thats happening cause this disk is SATA I and teh other is SATA II ?

We have seen in another thread (can't remember which) that Xbench HD benchmarks weren't accurate and did a horrible job, especially when testing random 4k blocks. Nobody knows for sure why, but it happens (albeit slighly less extreme) on real Macs too. All it would take is somebody with guts to ask Xbench's author to fix it for us, hackintosh users, or at least to give us some insight.

Nobody has seen real life bad disk access performance with the ConRoe 945G-DVI, so there's a high probability that the problem is in Xbench, not on our machines. Anyway, if you want a fair number for your hackintosh, you have to disable HD benches for now.

Take care,

B.

TopazBar · November 1, 2006

Little off topic, but this issue fixed: see http://forum.insanelymac.com/index.php?showtopic=31778

This is the last (hopely) remaining issue. I've looked at it and I know this much:
When you click on about this mac, it crashes loginwindow and then, launchd re-starts it. no biggy.

loginwindow crashes because of EXC_BAD_ACCESS. It is probably accessing unmapped virtual address.

Now, it does not occur on mifki's mobo. So it looks like something particular to our mobo and we may have to solve our selves.

New kernel 8.8.1 and Xbench for ASRock Conroe945G-DVI

Recommended Posts

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites