Jump to content

ATI Radeon 6xxx, working QE/CI


Nuphor
 Share

791 posts in this topic

Recommended Posts

Here's a quick log of my own experiences trying to get a Diamond Radeon HD 6850 up and running on an Asus P8P67 Sabertooth w/ Intel 2600K & 16Gb memory. Unfortunately, I don't have any answers to the 6850 issue. Hopefully this will help somebody else, though.

 

First, a big thanks to Andy since it was his P67 boot disk that I used to get my system up and running (here). I followed the instructions to the letter, and the system installed with almost no hiccup whatsoever. The only thing that I had to add to the instructions is that, upon booting the system for the first time, I had to set "GraphicsEnabler=no", and then I had to add it to /Extra/com.apple.Boot.plist to make it permanent. Without "GraphicsEnabler=no", the fans go into high speed, the displays are black, and the system is unresponsive. Only a power cycle fixes the problem.

 

I have 2 monitors, one an Apple 24" Cinema display (about 6 years old), and the other a LG 23" LED LCD. Both are attached to the 6850, and dual-monitor use is no problem at all. I have working QE/CI. However, I can't run either DVD Player or Geekbench. VLC, though, works without any issues at all.

 

After reading through 23 pages of posts in this forum, I decided to try to get the DVD Player/Geekbench issues resolved on my system. It took re-reading a few posts, but I found out that the version of Chameleon on Andy's P67 CD isn't the one that I need to use. Somewhere around page 14 of this post, I found Andy's link to RC 750 with his ATI changes (again, a big thanks to Andy). Good enough - I replaced the /boot file with version and began my experiments. The options that I manually typed in at the boot screen were "GraphicsEnabler=yes AtiConfig=xxx -v" where "xxx" is replaced by the framebuffer of choice for that experiment.

 

Some of the framebuffer attempts halted the system almost immediately when booting, reporting a NULL address (line 750 of ati.c, I believe). Those were the best ones to try because I knew almost immediately that they were not going to work. Examples of these were Duckweed, Pithecia, and Bulrushes.

 

Some of the framebuffers booted, but they didn't resolve the DVD Player/Geekbench issue. Gibba and Cattail are examples. Lotus, on the other hand, brought me to the login screen. Attempting to log in tossed me straight back to the login screen. With Elodea I ended up with a white screen and a mouse cursor. This one and Hydrilla were the worst because they made me power down the system, and the reboot forced a disk repair.

 

In all I went through every one of the 11 framebuffer names, none of which worked. I continued to see "ATI Radeon HD NI40" under "About This Mac", and both DVD Player and Geekbench continually crashed.

 

Currently I'm back to running without any framebuffer specification and with "GraphicsEnabler=no". I'm using the original boot file from Andy's CD because it seems to boot faster than the RC 750 version. DVD Player and Geekbench would have been nice to be able to run, not because I needed them but because there might be another program that I *do* need that won't work without the framebuffer fix. For now, though, VLC works as a nice DVD Player substitute, and Geekbench will just have to get along without my numbers. :rolleyes:

 

Hope this helps somebody else, too.

 

Regards,

John

Link to comment
Share on other sites

Currently I'm back to running without any framebuffer specification and with "GraphicsEnabler=no". I'm using the original boot file from Andy's CD because it seems to boot faster than the RC 750 version. DVD Player and Geekbench would have been nice to be able to run, not because I needed them but because there might be another program that I *do* need that won't work without the framebuffer fix. For now, though, VLC works as a nice DVD Player substitute, and Geekbench will just have to get along without my numbers. :blink:

 

Hope this helps somebody else, too.

 

Regards,

John

 

John,

 

Thanks for the feedback.

Can you do one last test for me please? Can you boot with GraphicsEnabler=y Wait=Yes -v and make a note of any framebuffer or ATI Card messages that are displayed on the page that says 'Press any key to continue' - There should be line such as ATI Barts AMD Radeon HD 6850 with the device id's in brackets etc.

 

You can use either the original boot disk bootloader or the newer one for this.

 

Thanks

 

Andy.

Link to comment
Share on other sites

Tired of booting to get the vendor-id? Here's a nice little app called : System Info

Look for "Display Controller" in the 'Class' column. This works only when you didn't change the vendor-id, and thus should work for John also... since his Diamond Radeon 6850 wasn't recognized at boot time.

Link to comment
Share on other sites

Tired of booting to get the vendor-id? Here's a nice little app called : System Info

Look for "Display Controller" in the 'Class' column. This works only when you didn't change the vendor-id, and thus should work for John also... since his Diamond Radeon 6850 wasn't recognized at boot time.

 

I need him to do it at boot time so I can use the ouput from the bootloader to try and work out why it's either not detecting his card or not loading a framebuffer for it!

Link to comment
Share on other sites

I need him to do it at boot time so I can use the ouput from the bootloader to try and work out why it's either not detecting his card or not loading a framebuffer for it!

I can understand that, but if it's just a matter of yet again a missing vendor-id – hence the error in ati.c – isn't it time to add some logic so that new/unknown vendor-id's can be used without people having to wait for anything?

 

That brings me to the following question; Do you have a repository where I can have a look at your changes?

Link to comment
Share on other sites

John,

 

Thanks for the feedback.

Can you do one last test for me please? Can you boot with GraphicsEnabler=y Wait=Yes -v and make a note of any framebuffer or ATI Card messages that are displayed on the page that says 'Press any key to continue' - There should be line such as ATI Barts AMD Radeon HD 6850 with the device id's in brackets etc.

 

You can use either the original boot disk bootloader or the newer one for this.

 

Thanks

 

Andy.

 

Or use "bdmesg" and post the results !

Link to comment
Share on other sites

I can understand that, but if it's just a matter of yet again a missing vendor-id – hence the error in ati.c – isn't it time to add some logic so that new/unknown vendor-id's can be used without people having to wait for anything?

 

There is no indication that the id's are missing, I'm 99.9% sure that I added the 6850 so it may be something else (it may simply be the id's but I can't tell from here and that's exactly why I wan't to see the output).

 

How do you propose new or unknown id's are automatically added - if they are new and unknown and we don't know the relevant framebuffers to use? Could try hacking the code to add even more kernel flags I guess for chipset, vendor and product id's - may not be an entirely mad idea I guess.

 

You will have to excuse my grumpiness today - just had a hard disk die on me and had to pay pc worlds criminal prices for an emergency replacement (grrr)

Link to comment
Share on other sites

There is no indication that the id's are missing, I'm 99.9% sure that I added the 6850 so it may be something else (it may simply be the id's but I can't tell from here and that's exactly why I wan't to see the output).

This remark: "reporting a NULL address" was what triggered my attention.

 

How do you propose new or unknown id's are automatically added - if they are new and unknown and we don;t know the relevant framebuffers to use lol.

Please. Let me do the thinking. My track record shows that I can come up with solutions that nobody ever considered to be useful or working and these days many developers ask me almost anything, and they value my input... but to get back on topic, I want to see the source code simply because you have to publish them (see forum rules) so please, don't let me waiting. I seriously want to help and at least give it a go :unsure:

 

Thank you.

Link to comment
Share on other sites

That brings me to the following question; Do you have a repository where I can have a look at your changes?

 

The short answer to that is nope I don't have a repository anywhere.

 

The slightly longer answer is that my changes are a mixture of Anval's sandybridge code with the relevant ati changes from kabyl's branch merged in. I have no intention to maintain a seperate branch of chameleon because basically I'm just hacking around with it as a stop gap until somebody performs a proper merge of the branches back into trunk.

 

I have made no specific ati changes other than adding in a couple of extra id's and known framebuffers (some for my own experimental uses and some as requested by others).

 

This remark: "reporting a NULL address" was what triggered my attention.

 

 

Please. Let me do the thinking. My track record shows that I can come up with solutions that nobody ever considered to be useful or working and these days many developers ask me almost anything, and they value my input... but to get back on topic, I want to see the source code simply because you have to publish them (see forum rules) so please, don't let me waiting. I seriously want to help and at least give it a go :unsure:

 

Thank you.

 

 

You replied before I finished editing my post and adding at the end about being grumpy lol.

Link to comment
Share on other sites

Sorry. Dinner time here. Had to wrap it up fast or mom said to pull the plug (like she did it before).

 

No worries though because I will have a go at it no matter what. I may fail, but one cannot fail without even trying :unsure:

Link to comment
Share on other sites

Sorry. Dinner time here. Had to wrap it up fast or mom said to pull the plug (like she did it before).

 

No worries though because I will have a go at it no matter what. I may fail, but one cannot fail without even trying ;)

 

I'm in a much better mood now, I just found out that my deceased 1TB Samsung hard disk was just within warranty still so Samsung have granted an RMA on it (even though I've already bought a new Hitachi 2TB to replace it :P ).

 

:D

 

If anybody is interested in the current source code for the bootloader I have uploaded it to mediafire http://www.mediafire.com/?i11m11nkna75tms

 

There are a few changes since the initial build, mainly a few framebuffer corrections and additions for a couple of AMD mobile cards as requested by users from this forum. Additionally I have fixed the annoying 'DMI (?) table entries list is full! Next entries won't be stored' error messages that sandybridge users and been seeing and defaulted the system definition to a macbookpro 8,1 along with setting default 'reported' dram speed to 1600 and DDR3 for unrecognised or integrated memory controllers so if those settings match what you have there is no need to edit the smbios.

 

If you don't have a Sandybridge processor then you should probably NOT bother with this source code, you would be better to stick with Kabyl's code and add in your own device id's if needed. The whole reason for me putting this version together was that there was no other build containing both AnValv's excellent sandybridge code and kabyl's fantastic ATI code.

 

Regards

 

Andy.

Link to comment
Share on other sites

Hi Andy,

 

Here's the output from running "GraphicsEnabler=y Wait=Yes -v". I didn't bother with the system ID since I don't think that you need that, but everything else has been included. I typed it all in, but I don't think there's a typo. I double checked the framebuff, MMIO, I/O port, ROM address, etc.

 

Loading Darwin 10.6
Loading kernel mach_kernel
ACPI: [hd(0,2)/Extra/DSDT.aml] 35046 bytes.
Using PCI-Root-UID value: 0
Framebuffer @0xD0000000  MMIO @0xFE620000  I/O Port @0x0000E000 ROM Addr @0xFE600000
ATI card POSTed, reading VBIOS from legacy space
ATI Barts AMD Radeon HD 6850 Series 1024MB (Gibba) [1002:6739] (subsys [1787:0b00]):: PciRoot(0x0)/Pci(0x1,0x0)/Pci(0x0,0x0)
DMI table entries list is full! Next entries won't be stored.
DMI table entries list is full! Next entries won't be stored.
DMI table entries list is full! Next entries won't be stored.
DMI table entries list is full! Next entries won't be stored.
Patched DMI Table
Found SMBIOS System Information Table 1
Customing SystemID with : xxxx....
ACPI: [hd(0,2)/Extra/DSDT.aml] 35046 bytes
FADT: Restart Fix applied !
FADT: Restart Fix applied !
Patched ACPI version 2 DSDT
Starting Darwin x86_64
Press any key to continue...

 

I added the DSDT file just last night, after typing my prior post. The DSDT file wasn't in any of the experiments that I ran for the ATI card. The one that's installed was generated by the DSDT Auto-Patcher, though I used the P8P67 Deluxe motherboard selection for my Sabertooth.

 

Hope this helps.

 

Regards,

John

 

 

Or use "bdmesg" and post the results !

 

This would have been simpler, but there's no ATI information :(. For reference, though, here's the output:

Config: [hd(0,2)/Extra/com.apple.Boot.plist] 339 bytes.
CPU: Mobile Intel® Core(tm) i7-2600K CPU @ 3.40GHz
CPU: Flex-Ratio = 34 
Sticking with [bCLK: 103Mhz, Bus-Ratio: ??]
CPU: Tjmax ~ 98°C 	           Temperature= ~ 40°C
CPU: IDA:                          Enabled!
CPU: Vendor/Family/ExtFamily:      0x756e6547/0x6/0x0
CPU: Model/ExtModel/Stepping:      0x2a/0x2/0x7
CPU: Multipliers x10:              max=340, min=160
Turbo Ratio:                       59/59/59/59
CPU: MaxDiv/CurrDiv:               0x0/0x0
CPU: TSCFreq:                      3502MHz
CPU: CPUFreq:                      3502MHz
CPU: FSBFreq:                      103MHz
CPU: Enhanced SpeedStep:           1
CPU: NoCores/NoThreads:            4/8
CPU: Features:                     0x36d803a3
Theme: [hd(0,2)]/Extra/Themes/Default
Theme: [hd(0,2)]/Extra/Themes/Default
Found mode 1024x768 in VESA Table
Config: [hd(0,2)/Extra/com.apple.Boot.plist] 339 bytes.
Setting boot-uuid to: 25E497F2-D8EE-3217-B096-4335A8220BD4
Loading Darwin 10.6
Loading kernel mach_kernel
LoadDrivers: /Extra/Extensions.mkext
LoadDrivers: /System/Library/Caches/com.apple.kext.caches/Startup/Extensions.mkext
ACPI: [hd(0,2)/Extra/DSDT.aml] 35046 bytes.
Using PCI-Root-UID value: 0
DMI table entries list is full! Next entries won't be stored.
DMI table entries list is full! Next entries won't be stored.
DMI table entries list is full! Next entries won't be stored.
DMI table entries list is full! Next entries won't be stored.
Patched DMI Table
Found SMBIOS System Information Table 1
Customizing SystemID with : a0eb001e-8c00-003f-80f8-bcaec57607f6
efi_inject_get_devprop_string NULL trying stringdata
ACPI: [hd(0,2)/Extra/DSDT.aml] 35046 bytes.
FADT: Restart Fix applied !
FADT: Restart Fix applied !
Patched ACPI version 2 DSDT
Starting Darwin x86_64

 

 

Tired of booting to get the vendor-id? Here's a nice little app called : System Info

Look for "Display Controller" in the 'Class' column. This works only when you didn't change the vendor-id, and thus should work for John also... since his Diamond Radeon 6850 wasn't recognized at boot time.

 

I ran the utility, just in case there was a mismatch between the output from the boot loader and this utility. It looks like they're consistent with one another. Vendor ID = 6739, Device ID = 1002 (unknown vendor and unknown device). It recognizes it as a "Display controller", subclass "VGA compatible controller" (no big surprise there).

 

Regards,

John

Link to comment
Share on other sites

I have recently installed the new 10.6.7 kexts from Andy, [url="http://www.insanelymac.com/forum/topic/279450-why-insanelymac-does-not-support-tonymacx86/"]#####[/url] ati drivers, put in GraphicsEnabled=Yes, and tried every framebuffer available with AtiConfig. It would turn off one screen and not the other and sometimes no screens would work. I can't seem to get display port to work properly. Duckweed seems to work best since it shows a picture on both monitors but I get scrambled colors and it is just really fuzzy. When I plug the displayport out and plug it back in. It seems to work fine. I have a club3d Ati Radeon 6870. Any ideas? Works fine with 2x dvi out. But its just displayport giving me issues.

Link to comment
Share on other sites

I'm in a much better mood now, I just found out that my deceased 1TB Samsung hard disk was just within warranty still so Samsung have granted an RMA on it (even though I've already bought a new Hitachi 2TB to replace it :( ).

 

:(

 

If anybody is interested in the current source code for the bootloader I have uploaded it to mediafire http://www.mediafire.com/?i11m11nkna75tms

 

 

Hey Andy,

 

I can't build this, it seems to be missing the boot1.asm as well as the ati.h. Have you posted the obj from this build anywhere? I don't feel the need to monkey with it but I would like to fix the annoying boot errors.

 

thanks

Link to comment
Share on other sites

Following up on my comments a few pages back about DisplayPort not working on boot, in case anyone else was having the issue and not speaking up, it looks like I was blaming the Hackintosh (an easy target) when in fact it looks like it's a common fault. Widening my Google search I found that users of genuine Macs etc were having this problem, often with Dell monitors like mine.

 

The real problem? Not Dell. Not ATI. Not Apple. Improperly manufactured DisplayPort cables! They're rife at the moment!

See http://forums.amd.com/game/messageview.cfm...threadid=127563

 

It boils down to manufacturers not consulting the spec and instead assuming every pin should be wired through. Pin 20 should NOT. I just put a multimeter to Pin 20 (the one closest to the angled corner on the connector) on both ends and sure enough it's connected through. I'll have to try and find a known good cable (this is the £6 one from scan.co.uk so don't buy that). Then will confirm if it works.

Link to comment
Share on other sites

I'm open to DP pin-crippling suggestions - can't fit a pair of snips down these little connectors! :D

 

E: managed to nudge the pin out of the way with.... the other type of pin (sharp and pointy). No difference - but I can't say I'm hugely confident in my connector modding abilities.

 

E2: Found scripts for a) triggering "Detect Displays" and ;) Resetting screen resolution and triggered them both from startup and known/tested spotlight keystrokes. No luck there.

 

I know I can just give up and use DVI, but it's annoying having an unsolved problem hehe

Link to comment
Share on other sites

Hi Andy sorry again

Please help me.

I've a vaporx Ati 5870 of my daughter, the package of ati upgrade 10.6.7, it works also with the 5870? If it works….which framebuffer I will have need?

Thank's Andy...best regard... :)

 

I have a Sapphire Vapor X HD 5870, it works perfectly in 10.6.7. I use the Kabyl booter (v 748) and it automatically selects Uakari (GE=Y). All ports work and I get audio on HDMI and DisplayPort without editing. I tried the ATI package and got the DP black screen on boot (unplug DP to fix) and lost audio. Uninstalled the ATI package.

Link to comment
Share on other sites

I have a Sapphire Vapor X HD 5870, it works perfectly in 10.6.7. I use the Kabyl booter (v 748) and it automatically selects Uakari (GE=Y). All ports work and I get audio on HDMI and DisplayPort without editing. I tried the ATI package and got the DP black screen on boot (unplug DP to fix) and lost audio. Uninstalled the ATI package.

 

Where is boot Kabyl...please...i've a chamaleon r700 , with this dont work?

I need only Kabyl boot?

Thank's

Link to comment
Share on other sites

Mentioned a jittery mouse earlier in the thread - after various things improved it (different port, going through a hub etc) I seem to have finally cured it completely! Woohoo! (I bet that's tempted fate... still it's been perfect for the longest I've known)

 

As it happens I accidentally stumbled across this post from Tony ("The" Tony) with exactly the same issue:

http://www.tonymacx86.com/viewtopic.php?f=79&t=11592

Sure enough, once I looked up my processor's native bus ratio (multiplier) from here:

http://en.wikipedia.org/wiki/List_of_Intel...microprocessors

and appended busratio=22 to my Kernel Flags under com.apple.boot.plist (other processors have different ratios)

and plugged the mouse into my "worst" port that always had the problem within a minute or two, for an easy test

the problem went away! :)

 

I've only seen the busratio command mentioned as a requirement for Sandy Bridge processors, but I have an "old" Arrandale, i7-870. Interesting.

 

It's as if the same kind of thing is happening as the ATI framebuffer issue (and could well be related as it happened ever since then, but it could've been the 10.6.7 update) - it can't detect the correct one, so it just falls back to an incorrect default with crappy results. Maybe I ended up installing a Sandy Bridge kernel with all the tweaks for getting an ATI 6xxx working properly and that one isn't quite right for an Arrandale. I don't know. But manually setting my bus ratio to what it's supposed to be seems to have fixed it nicely.

 

Didn't fix my non-booting Displayport, but can't have everything :P

Link to comment
Share on other sites

You are right. I tracked it down and made a post about it in this very forum before I made my comment here. I did this because I know that your -x boot ups are producing the same white screen so what I was trying to do, without saying anything, is to make you aware of the fact that they may be related.

 

And again. Welcome back Andy :(

 

Oh and my AMD Radeon 6950 should arrive today... confident that it will work soon.

 

Have you managed your 6950 to work properly under 10.6.x / 10.7 ?

 

If anybody will be successful with 6950 card I will also try with my ASUS 6950 DCII... WIth UBUNTU studio 11.04 it was straightforward to get full support with double monitor setup and OpenGL acceleration. There is an official linux AMD Catalyst 11.x suit which works fine. It’s too bad that AMD is not releasing OSX driver package for HD cards...

Link to comment
Share on other sites

Hey Andy,

 

I can't build this, it seems to be missing the boot1.asm as well as the ati.h. Have you posted the obj from this build anywhere? I don't feel the need to monkey with it but I would like to fix the annoying boot errors.

 

thanks

 

Did you do a make clean first - the ati.h isnt used in that build (it uses ati_reg.h instead) so unless I uploaded a dud build it should compile - be pretty surprised if I did manage to upload a dud because I compile from that exact same folder :D

Link to comment
Share on other sites

 Share

×
×
  • Create New...