Comment 2 by Bugs B, Mar 29, 2012
The Park chip (used by Apple themselves) is part of the Manhattan chip family, see http://www.tomshardware.com/forum/53336-35-apple-update-imac-macbook- lineup-autumn, and also http://semiaccurate.com/2009/07/26/ati-evergreen-mobile-parts-outed Thanks for looking into this, Cosmosis Jones, very much appreciated :) !!
Comment 3 by Bugs B, Apr 2, 2012
OMG - the entire 'MANHATTAN' chips family (mobile GPUs Park,
Broadway, Madison) is currently NOT AT ALL LISTED in ati.c and
therefore NOT supported in Chameleon .. :((
However, my HD 5450 is NOT a mobile but a desktop grfx card.
Therefore the Park GPU on it can technically be grouped into the
CEDAR family, as is the 5430M with the same DID 68e1 (but different
SubSys), and as are two other 5450's (but w. different DIDs).
So, may I for the time being (until 'MANHATTAN' has officially been
integrated as Family), kindly request the following addition to
ati.c:
{ 0x68E1,␉0x174b3000, CHIP_FAMILY_CEDAR,␉␉"ATI Radeon HD
5450",␉␉␉␉kEulemur␉},␊
Feel free to let me know if more information (different ioreg than
the one in the zip, etc.) is needed.
Comment 4 by Bugs B, Apr 6, 2012
Question: does the GPU have a different DID than the grfx card? Or is there only ONE DID (GPU DID = gfrx card DID)?
Comment 5 by Bugs B, Apr 7, 2012
Jesus, the two blocks in the Subsystem ID need to be reversed, why
ever .... -->
{ 0x68e1,␉0x3000174b, CHIP_FAMILY_CEDAR,␉␉"ATI Radeon HD
5450",␉␉␉␉kEulemur␉},␊
(thanks to Vlad for pointing this out)
Comment 6 by machinist, Apr 9, 2012
The "Baboon" FB works better with this "Park" chipped card then "Eulemur." "Baboon" tests significantly better with openglviewer, cinebench, novabench, and other assessment tools. I find it "feels" faster and more stable in real world applications, such as writing this. And ioreg seems to report "Baboon" has the same output config as "Eulemur." Have not finished reediting the "monkey" fb personalities so each can be tried with this 5450 "Park," but I can say "Eulemur" performs the worst of all so far. I suspect it's a framebuffer designed for mobile platforms, so that's to be expected? AMD's software ID's the Park chip as a Mobility series card (in Win7 and XP), so whether it's in a desktop or not is irrelevant; it's a 5450M "Park," not "Cedar." So, my two cents as an owner of an identical card, please don't encode the "wrong" default fb. (Not that it really matters to me, I'll use whatever works best thanks to the override "AtiConfig" option.)
Comment 7 by Bugs B, Apr 12, 2012
so yes convinced, all three of us now agree w. machinist (after a
long try-and-error testing phase):
{ 0x68e1,␉0x3000174b, CHIP_FAMILY_CEDAR,␉␉"ATI Saphire Radeon
HD 5450",␉␉␉␉kBaboon␉},␊
If you want to look at the result of the testing phase:
http://www.insanelymac.com/forum/index.php?showtopic=266182&view=
findpost&p=1812826
The trick (thanks to machinist) is to find the correct PCIRootUID
value via
ioreg -l | grep -15 "AppleACPIPCI" | grep "_UID"
which will output a # (known values betw. 0 and 4, but higher values
might appear)
then enter that value into the org.chameleon plist as follows:
<key>PCIRootUID</key>
<string>4</string>
remove any EFI strings from device properties
delete extension (if present) AND kernel caches in (Lion)
/System/Library/Caches/Startup
reboot
Comment 8 by Bugs B, Apr 15, 2012
N O T Baboon!! THIS CARD's FB is EULEMUR - Period !!
reason: Baboon might deliver a couple of fps more on the DVI port,
BUT disables HDMI on this card!!
THEREFORE:
{ 0x68e1,␉0x3000174b, CHIP_FAMILY_CEDAR,␉␉"ATI Saphire Radeon
HD 5450",␉␉␉␉kEulemur␉},␊
this is the very final result!!
Comment 9 by Bugs B, May 4, 2012
Who is able to include
{ 0x68e1,␉0x3000174b, CHIP_FAMILY_CEDAR,␉␉"ATI Saphire Radeon
HD 5450",␉␉␉␉kEulemur␉},␊
into the the trunk ati.c
Cosmosis?
Sign in to reply to this comment.
Reported by Bugs B, Mar 23, 2012