Jump to content
12 posts in this topic

Recommended Posts

Here's what I just posted in AGPGart's forum. I just got AGPGart working on my Asus/Intel motherboard, please check the original threat. I thank there to the user Zizou for his compilation.

 

http://x86dev.org/forums/index.php?topic=288.msg2300#msg2300

 

ok, installed, verbose says bridge and device have been detected, and all is good.

 

boots up normally, about this mac says AGP too :poster_oops: Qe/Ci Supported, 128mb apeture too.. that's all good I guess, it's just that there's no speed improvement..

 

btw, compiled version added.

 

Zizou,

 

You rock! Working 100% for me! My hardware is

 

Asus P5PE-VM - Intel northbridge 865G southbridge ICH5

Nvidia 5200FX 256MB AGP working 100% with Natit 1.0 drivers

 

Here's what I saw in verbose boot:

Found Intel 82865 host to AGP bridgeFound Intel 82875P host to AGP bridgeAGP: Starting AGP

AGP: Fond an AGP 3.0 compliant bridge.

AGP: Found an AGP 3.0 compliant device.

AGP: Setting AGP v3 mode 8

AGP: AGP command in bridge = 00000b12

AGP: AGP command in agp device = 1f00e312

 

How I installed it:

 

- I logged in as root.

- Unzipped and copied your AGPGart.kext to ./System/Extensions/

- Opened the terminal

- I typed

chown -R root:wheel /System/Library/Extensions/AGPGart.kext

chmod -R 755 /System/Library/Extensions/AGPGart.kext

(you must type [sudo ] before these lines if you not log in as root, but I didn't try that.)

- I trashed the files

./System/Library/Extensions.kextcache

./System/Library/Extensions.mkext

- I opened Disk Utility and repaired permissions

- I set verbose boot flag in ./Library/Preferences/System Configuration/com.apple.boot.plist just for precaution and eventual crash diagnosis, but this is not needed

- I took a deep breath and rebooted

- I saw the above verbose lines and typed exit (not needed if you don't select verbose boot)

- Everything worked fine.

 

I must confess, I had no performance change, it was everything working (fast) before with Natit drivers, but it now shows as AGP. I wonder if this will make any difference for some graphics applications (would I have the expected 3D acceleration in Parallels Desktop 3 running WinXP?)

AGPGart_INTEL.kext.zip

post-3175-1185970863_thumb.jpg

post-3175-1185970939_thumb.jpg

Link to comment
https://www.insanelymac.com/forum/topic/58678-agpgart-working-on-intel-chipset/
Share on other sites

  • 1 month later...

Let's test... with the classic AGPGart light i get this in the dmesg...

 

AGP: Found an AGP 3.0 compliant bridge.

AGP: Setting AGP v3 mode 8

AGP: AGPCMD in bridge = 00000b12, in agp device = 1f00e312

 

And this is the XBench 1.3 result...

 

Results 55.00

System Info

Xbench Version 1.3

System Version 10.4.8 (8L2127)

Physical RAM 1024 MB

Model ADP2,1

Drive Type Maxtor 6B160P0

Quartz Graphics Test 54.16

Line 56.69 3.77 Klines/sec [50% alpha]

Rectangle 44.86 13.39 Krects/sec [50% alpha]

Circle 44.33 3.61 Kcircles/sec [50% alpha]

Bezier 70.71 1.78 Kbeziers/sec [50% alpha]

Text 63.74 3.99 Kchars/sec

OpenGL Graphics Test 91.71

Spinning Squares 91.71 116.34 frames/sec

User Interface Test 39.71

Elements 39.71 182.27 refresh/sec

 

Pretty slow for a 6600GT 128 Mb... Let's see...

 

These are the results with the modified kext:

 

Found Intel 82865 host to AGP bridge

Found Intel 82875P host to AGP bridge

AGP: Starting AGP

AGP: Found an AGP 3.0 compliant bridge.

AGP: Found an AGP 3.0 compliant device.

AGP: Setting AGP v3 mode 8

AGP: AGP command in bridge = 00000b12

AGP: AGP command in agp device = 1f00e312

 

And the Xbench results:

 

Results 80.65

System Info

Xbench Version 1.3

System Version 10.4.8 (8L2127)

Physical RAM 1024 MB

Model ADP2,1

Drive Type Maxtor 6B160P0

Quartz Graphics Test 76.00

Line 71.91 4.79 Klines/sec [50% alpha]

Rectangle 57.93 17.30 Krects/sec [50% alpha]

Circle 56.38 4.60 Kcircles/sec [50% alpha]

Bezier 84.05 2.12 Kbeziers/sec [50% alpha]

Text 200.44 12.54 Kchars/sec

OpenGL Graphics Test 118.62

Spinning Squares 118.62 150.47 frames/sec

User Interface Test 64.07

Elements 64.07 294.04 refresh/sec

 

There is a nice improvement... let's see if UT2004 runs faster...

UT2004 a little faster... Google Earth stutters much less then before... I think this is the right way to move to... but I can't figure out how u can obtain such an high xbench result with a GeForce FX 5200 which is WAY FAR slower than my 6600GT... maybe a kernel problem, or do u use a particular natit????

UT2004 a little faster... Google Earth stutters much less then before... I think this is the right way to move to... but I can't figure out how u can obtain such an high xbench result with a GeForce FX 5200 which is WAY FAR slower than my 6600GT... maybe a kernel problem, or do u use a particular natit????

 

I'm using Natit 1.0 and the AGPGart I got from Zizou. I and Emersonfx checked extensively the IO registry of my computer and there is no reason to tell that it's not working for me.

 

I posted this to the AGPGart forum:

 

Well I did have the same result in Xbench, as I told before. In my opinion the system was just misreading the bus. The fact is that it was very fast for me already with Natit. It looked faster than with the native drivers on Windows I'd dare to say. I'm not a hacker nor developer, but emersonfx is - and he I exchanged a lot of emails about this matter, I sent him dozens of screenshots og IORegistry Explorer showing details about my system and we both concluded that for all matters it seemed that using the AGPGart "for Intel" everything was working indeed as it should.

 

-it shows as AGP

-IORegistry Explorer shows shows in every aspect what it was expected to show with a supported system, I and emersonfx reviewed dozens of reports

-both outputs (VGA and DVI) are recognized

-my monitor brand and model is recognized

-all supported resolutions are recognized

-it's fast, it's rendering everything it should be supposed to render, including DVD playback, QE, CI, OpenGL, etc. It's actually more than twice as fast as my Mac mini which have a built-in ATI Radeon 9200 (Xbench).

-it has no screen artifacts whatsoever

-I can calibrate the monitor with the ColorSync calibrator

-it shows messages recognizing the AGP bus appearantly correctly in verbose mode

 

I see no reason to say it's not working, at least so far.

 

Can u disable one core and do the same benchmarks? I suspect that the second core is doing the trick for u, accelerating the graphics performance...

 

There's no way this is happening. I 've seen this computer running without acceleration and the difference is brutal. And I've double-checked the IORegistry. Anyway I did the xbench test four times (two with two cores, two with one core) and there was virtually no difference. You know that these tests never give the same results twice, maybe due to the OS's background processes. One of the two-core result is actually slower than the one-core results, another is faster. So you take the average. I'm posting the results to the forum.

 

One Core test 1

 

Results 95.44

System Info

Xbench Version 1.3

System Version 10.4.10 (8R2232)

Physical RAM 1536 MB

Model ACPI

Drive Type SAMSUNG SP0842N

Quartz Graphics Test 79.75

Line 64.86 4.32 Klines/sec [50% alpha]

Rectangle 64.96 19.39 Krects/sec [50% alpha]

Circle 62.92 5.13 Kcircles/sec [50% alpha]

Bezier 86.15 2.17 Kbeziers/sec [50% alpha]

Text 228.18 14.27 Kchars/sec

OpenGL Graphics Test 86.37

Spinning Squares 86.37 109.57 frames/sec

User Interface Test 136.68

Elements 136.68 627.30 refresh/sec

 

Two cores test 1

 

Results 91.53

System Info

Xbench Version 1.3

System Version 10.4.10 (8R2232)

Physical RAM 1536 MB

Model ACPI

Drive Type SAMSUNG SP0842N

Quartz Graphics Test 75.47

Line 79.42 5.29 Klines/sec [50% alpha]

Rectangle 66.66 19.90 Krects/sec [50% alpha]

Circle 64.89 5.29 Kcircles/sec [50% alpha]

Bezier 88.57 2.23 Kbeziers/sec [50% alpha]

Text 83.61 5.23 Kchars/sec

OpenGL Graphics Test 95.82

Spinning Squares 95.82 121.55 frames/sec

User Interface Test 110.05

Elements 110.05 505.06 refresh/sec

 

One core test 2

 

Results 93.49

System Info

Xbench Version 1.3

System Version 10.4.10 (8R2232)

Physical RAM 1536 MB

Model ACPI

Drive Type SAMSUNG SP0842N

Quartz Graphics Test 79.25

Line 63.90 4.25 Klines/sec [50% alpha]

Rectangle 64.72 19.32 Krects/sec [50% alpha]

Circle 62.91 5.13 Kcircles/sec [50% alpha]

Bezier 85.69 2.16 Kbeziers/sec [50% alpha]

Text 225.88 14.13 Kchars/sec

OpenGL Graphics Test 85.81

Spinning Squares 85.81 108.85 frames/sec

User Interface Test 127.96

Elements 127.96 587.26 refresh/sec

 

Two cores test 2

 

Results 98.97

System Info

Xbench Version 1.3

System Version 10.4.10 (8R2232)

Physical RAM 1536 MB

Model ACPI

Drive Type SAMSUNG SP0842N

Quartz Graphics Test 84.98

Line 79.23 5.27 Klines/sec [50% alpha]

Rectangle 66.74 19.93 Krects/sec [50% alpha]

Circle 64.70 5.27 Kcircles/sec [50% alpha]

Bezier 88.19 2.22 Kbeziers/sec [50% alpha]

Text 225.49 14.11 Kchars/sec

OpenGL Graphics Test 95.62

Spinning Squares 95.62 121.31 frames/sec

User Interface Test 123.63

Elements 123.63 567.39 refresh/sec

voxxdigital_s_ioregistry.txt

AGPGart___INTEL.kext.zip

Natit_1.0.pkg___USE_THIS_ONE_FOR_NVIDIA.zip

No way, even with Natit 1.0 you posted... something still brakes graphics performance, which, at least, gained a nice boost... AGP Aperture set to 128 mb... Having the same chipset as yours I should tear your score apart with a 6600GT... can we compare our ioregister dumps? Which is the command to obtain the log you've just posted?

 

Could it be the kernel? Do you remember having the same graphics performance with 10.4.8???

That's

 

ioreg -l

 

(the "-l" is a small "L", not a capital "i")

 

And I got the same performance with JaS 10.4.8, JaS 10.4.9 and now with Koolkal 10.4.10. Remember, it was fast already with Natit only, I believe it was just misreading the bus before I installed this AGPGart. When I installed the standard AGPGart nothing changed, not even the bus name.

  • 3 weeks later...

New Apple LOGIC 8 needs to see QE\CI to install (also needs 10.4.9).

I just want L8 to install, do not care if graphics are actually any better or worse.

Going to try AGPGart on my ATI 9800 Pro 128 w/ uphuck 1.0.9 1.4i r3.

Thanks for your directions and info guys, esp. voxxdigital

  • 1 month later...
  • 1 month later...

This worked on an Intel D875PBZ Mobo with an Nvidia FX5200 128meg Sweex in it... using NVinject 0.1.7.

 

No performance gain at all though!

 

All other drivers lead to a black screen

 

Running EFI 8.0 Leopard 10.5.1 from Flat Install

×
×
  • Create New...