EPDM Posted January 26, 2007 Share Posted January 26, 2007 Hi ppl, Since a few days I've been reporting problems with OpenGL on my rig. The strange part is that I build 2 identical systems though one has 1GB instead of 2GB ram and other brand of Radeon X1600pro (but with same specs) The 2 systems are as follows ECS915P-A -CMI9880 audi -Realtek 8110S (gigabit LAN chip onboard) -ICH6 The differences are: System 1 ----------- P4 630 (2MB cache) 2GB DDR1 (dual channel) Saphire X1600pro-256 PCI-e System 2 ----------- P4 521 (1MB cache) 1GB DDR1 (dual channel) Peak X1600pro-256 PCI-e Both VGA cards have the same vendor and device ID's so basicly these are the same systems. The socket 775-P4 cpu's are appart from the 2nd-level cache identical running at 3GHz with Hyperthreading, NX bit, SSE3 etc... These systems are as close to Apple 915-development systems as possible using the bits (VGA cards) that gotten upto now the highest levels of compatibility with the least hassle. I discovered the following odd behaviour. I've tested on system 1 the following OSX versions: 10.4.1 -LAN and audio-out work with no editing -system profiler work - using ATI Radeon X600 PCI-E QE and CI work with mouse tearing 10.4.3 -LAN and audio-out work with no editing -system profiler work -using ATI Radeon X600 PCI-E QE and CI work with mouse tearing 10.4.5 -LAN and audio-out work with no editing -system profiler work -using ATI Radeon X600 PCI-E QE and CI work with mouse tearing --->full screen GL-gaming cause broken pointer ------> also used Callisto drivers with same results -using ATI Radeon X1600PCI-e QE and CI work, no mouse tearing with kext-editing --->full screen GL-gaming cause no broken pointer ----> also used Boris 3.2 installer with same (perfect results) 10.4.6 -LAN and audio-out work with no editing -system profiler work -using ATI Radeon X1600pro PCI-e QE and CI work ---> using kext editing with Boris3.2=OK & ATIX1x00_3.3=OK ----> using ATIX1x00_3.5=no QE, no CI ----> using Natit_Uni_1.0=no QE, no CI 10.4.8 updated from 10.4.6 using info on InsanelyMac -LAN and audio-out work with no editing -system profiler work -using ATI Radeon X1600pro PCI-e ---> using Boris3.2=NO BOOT & ATIX1x00_3.3=NO BOOT ----> using ATIX1x00_3.5=QE & CI work, Full game freeze on System 1; OK on system 2 -----> using ATIX1x00-3.6=installer aborts <--- driver set is incomplete for X1600!!!! ----> using Natit_Uni_1.0=QE & CI work, Full game freeze on System 1; OK on system 2 10.4.8 clean Install from JAS DVD Intel SSE3 (on system 1 ONLY) -NO LAN and Audio-out??? -System profiler gives error (after changing loginapp, with original loginapp "about this mac"-exits) -using Radeon X1600pro PCI-e ----> using ATIX1x00_3.5=QE & CI work, Full game freeze on System 1 ----> using Natit_Uni_1.0=QE & CI work, Full game freeze on System 1 ----> manual editing of ONLY ATIX1000.kext (changing device ID only)=QE & CI work, Full screen OGL-game work, mouse tearing and broken pointer after returning to desktop. To sum up: System 2 is an upgraded 10.4.6 to 10.4.8 -system profiler works as expected -with natit-driver QE and CI work -resolution changing works (no mouse tearing) -full screen GL-gaming don't freeze System 1 with upgraded 10.4.6 to 10.4.8 -system profiler works as expected -with natit-driver QE qnd CI works -resolution changing works (no mouse tearing) -full screen GL-gaming freezes System 1 with clean install from 10.4.8 JAS DVD -system profiler gives error -manual editing of ATIRadeonX1000.kext (only) enables QE and CI -resolution changing works with mouse tearing -full screen GL-gaming work with broken pointer aftwerwards. ----> I haven't yet edited more ATIRadeon1000xxxx-kext to try to fix this. I haven't on the native installed 10.4.8 system 1 used th ATI-drivers from 10.4.5 (or earlier) nor the included Nucleus drivers only the included and installed standard kext which I only tested by applying Device ID in ATIX1000.kext. The weird thing is that this system used to be fully compatible out o/t box with every OSX clean install until 10.4.8 where suddenly audio-out and LAN-card doesn't work anymore. I patched LAN-card with RTL1xxx_1.02.pkg (I dunno who made this but this does work also on version 10.4.8) These are just my own experiences I hope they can be of use for the ppl involved. If you want more information about this rig (device ID's etc...) don't hesitate to ask. Regards, EPDM Link to comment https://www.insanelymac.com/forum/topic/39975-weird-osx-1048-woes/ Share on other sites More sharing options...
EPDM Posted January 26, 2007 Author Share Posted January 26, 2007 Just to confirm that the problem IS related to the OpenGL-framework. When using the standard (the one that gets installed from the Apple update 10.4.8 or from the JAS 10.4.8 Intel SSE3-DVD without selecting ATIX1x00-drivers) and editing ATIRadeonX1000.kext does enable Quartz extreme and Core image. And OpenGL games DON'T crash. But Natit-1.0 crashes the machine completely (so does ATIX1x00_3.5.mpg). The problem is that manually editing the ATINDRV.kext from 10.4.8 doesn't do thing since that Info.plist is completeley differeint from the one that is described in Boris Badenov's manual install Can someone point out the PROPER procedure to manual adjust kext for Radeon X1600pro with standard supplied kexts (as these work with OpenGL-programs) Or if someone can tell me How to install photoshop beta CS3 on NON-10.4.8 systems. This is important for me otherwise I'll just have to ditch this system and put Windows on it (at least THAT works without this {censored}). Thanks in advance, EPDM Link to comment https://www.insanelymac.com/forum/topic/39975-weird-osx-1048-woes/#findComment-285630 Share on other sites More sharing options...
EPDM Posted July 29, 2007 Author Share Posted July 29, 2007 To follow up on this thread. I have investigated this a bit further especially the GFX-cards since I assumed that thats where the problems are. It appears that both cards are functionally and electronically the same: ATI Radeon X1600pro (RV530) Use 256MB GDR2 memory Both have 1x 15-pins VGA, 1x DVI and mini-din-TV out. So basicly both cards are identical appart from the VGA-bios on the cards. This is ROM-readout of the Saphire card: Adapter 0 (BN=01, DN=00, PCIID=71C21002, SSID=0840174B) Asic Family : RV53x/Mx6 Flash Type : M25P05/C (64 KB) Product Name : DDR2 RV530 BIOS 500e/405m Bios Config File : 2C8405SA.003 Bios P/N : 11x-2C8405SA-00 Bios Version : 009.012.012.002.020511 Bios Date : 07/13/06 21:44 And here is the ROM-readout of the cheaper Peak (probably standard reference design): Adapter 0 (BN=01, DN=00, PCIID=71C21002, SSID=2173148C) Asic Family : RV53x/Mx6 Flash Type : M25P05/C (64 KB) Product Name : RV530PRO DDR II 256MB VBIOS 148C-52619LBB.IEA-9.12.8.2 Bios Config File : 52619LBB.IEA Bios P/N : P/N Not Available Bios Version : 009.012.008.002.000000 Bios Date : 06/19/06 08:56 It seems that the Peak card (though bought more recently) has an older BIOS. I have swapped the Saphire card on my rig for the Peak and gues what. NO crashes anymore with OpenGL games!!!! I flashed the Saphire card with the Peak VGA-Bios. The system booted but the image of the startup BIOS screen was very weird. I did not go further and re-flashed the original Saphire VGA-Rom back. I used the ATIFlash v3.41 program to perform my actions. So it seems I have to ditch the higher quality VGA-card for the cheap Peak variant. I went for the ATI cards because real iMAcs (still) use the ATI X1600pro chip in their configurations. I presumed by going for similar components, I would have the least compatibility problems. I also tried 10.4.9 Upchuck DVD and one OpenGL game and that too worked flawless with the Peak card. My conclusing in this endavour is that Apple uses cheap probably reference design ATI cards with the older VGA-BIOS for their ATI X1600-iMac. I hope this information is usefull for potential users. Cheers Link to comment https://www.insanelymac.com/forum/topic/39975-weird-osx-1048-woes/#findComment-417313 Share on other sites More sharing options...
Recommended Posts