Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Status
The first post of this thread is a WikiPost and can be edited by anyone with the appropiate permissions. Your edits will be public.
AMD ATI FirePro - W6170M

Excuse for my dumpness, I've understand you correctly. Inside of a CPU
you cannot exchange an adress, but on a video card there are perhaps more
possibilities to exchange or disconnect an adress... ;).

Since I am a practical person and unfortunately not a theorist it might be
possible to cut two lines of the MXM connector and switch off the I2C Temp
information 😁 then I would prepare these card to cut the lines or insulate
it with tape !!!

I don't need I2C temp information from the GPU, perhaps I look after MXM
infos if the I2C temperature infos have fixed copper lines on the MXM card.

Thanks for your knowledge, very helpful 👍
You cannot change the address of any of those ICs, those addresses are hardcoded, IC-specific, but not unique. Apple's SMC IC uses that address to read sensor (and other) data via (industry standard) I2C serial bus by first addressing (what it thinks is) a unique IC.

And now that you have plugged a W6170M card into the MXM slot you have hardwired two ICs with the same address onto the same bus. (Something that an electronics designer must clearly avoid...) SMC is confused: It reads some data from a specific address (which should be CPU temperature), but it apparently gets data from the GPU temp. sensor IC, which may not even be in the correct format. So you must disconnect one of those ICs from that bus, and clearly not the IC that reads CPU temperature data. (Which is not inside the CPU, by the way. CPU contains a 'temperature-sensitive diode', which is connected to a temp. sensor IC, which then 'offers' its readings to SMC via SMBus.)

And also, yes, MXM connector has two specific contacts for the SMBus, but simply insulating them will lead to other problems, since the temp. sensor IC on the GPU is not the only IC on the GPU connected to SMBus. (GPU die is on those lines, too.)

So, in short (or finally), we have to look for those SMBus lines around the temp. sensor IC on the GPU. I'm very confident that there are two 0402-sized jumpers close to that IC, which will disconnect it from the SMBus.

(Maybe also read my writeup on how we solved those problems for some WX41x0 cards...)
 
Last edited:
AMD ATI FirePro - W6170M

Excuse for my dumpness, I've understand you correctly. Inside of a CPU
you cannot exchange an adress, but on a video card there are perhaps more
possibilities to exchange or disconnect an adress... ;).

Since I am a practical person and unfortunately not a theorist it might be
possible to cut two lines of the MXM connector and switch off the I2C Temp
information 😁 then I would prepare these card to cut the lines or insulate
it with tape !!!

I don't need I2C temp information from the GPU, perhaps I look after MXM
infos if the I2C temperature infos have fixed copper lines on the MXM card.

Thanks for your knowledge, very helpful 👍
If you're feeling adventurous there's a quick and easily reversible experiment you can try on your card: use nail varnish to paint over pin #32 (SMB_DAT) on the card MXM connector. That should isolate SMBUS on the card from SMBUS on the iMAC and it may work on your card.

I won't cut any track, as @dfranetic said, SMBUS is connected to other ICs on card, and we don't know exactly what side effects it may have or if they're the same for all cards, so cutting the track is not a good idea. If the "nail varnish" trick does not work you can easily remove it with a bit of dissolvent. Also I won't use tape as it moves easily and also may get stuck inside the mxm connector.

If you'd like to try, pin #32 is on the top of the card (same side as the GPU chip), and pins are numbered right to left, even numbers only, starting with the first "thin" pin as #2. The way I'd do it is covering with tape the adjacent tracks and then painting over with nail varnish the whole pin #32, wait for it to dry and then remove tape.

1638876442394.png
 
This is good to know. This explains my CPU fan going full pelt after installing a WX4170.
I will try covering this pin and report back on this
 
Hello everyone, I am new to the forum. I have a 2010 iMac 11,2 with an upgraded CPU (i7 860S) and I have the original HD4670 and a K1100M(Samsung memory). The computer works fine with the CPU upgrade and the HD4670, however when I install the K1100M I don't get the third light. The computer chimes and boots into the Linux USB, and the GPU is listed when I use the lspci command, but I can't flash any bios to it because it says I have an id mismatch. I used all the proper commands and the -6 flag. I tried the most recent version of nvflash downloaded from here, but that also didn't work. I have ordered a mini-displayport adapter to see if it's outputting video at all.
 
  • Like
Reactions: Shawn09345
Hello everyone, I am new to the forum. I have a 2010 iMac 11,2 with an upgraded CPU (i7 860S) and I have the original HD4670 and a K1100M(Samsung memory). The computer works fine with the CPU upgrade and the HD4670, however when I install the K1100M I don't get the third light. The computer chimes and boots into the Linux USB, and the GPU is listed when I use the lspci command, but I can't flash any bios to it because it says I have an id mismatch. I used all the proper commands and the -6 flag. I tried the most recent version of nvflash downloaded from here, but that also didn't work. I have ordered a mini-displayport adapter to see if it's outputting video at all.
Take a look here....
 
Take a look here....
I saw that post and I have one in my Amazon cart, I am wondering if it will work though since in the FAQ about nvidia cards it says that most of them can't be flashed that way. Also I took a screenshot of nvflash --verify if maybe that helps understand what could be wrong with the card.
 

Attachments

  • Screenshot from 2021-12-07 11-15-27.png
    Screenshot from 2021-12-07 11-15-27.png
    49.3 KB · Views: 76
I saw that post and I have one in my Amazon cart, I am wondering if it will work though since in the FAQ about nvidia cards it says that most of them can't be flashed that way. Also I took a screenshot of nvflash --verify if maybe that helps understand what could be wrong with the card.
Did you use the protectoff option, too?
 
Last edited:

MXM 3.0 Specification Sheet


MXM 3.0.png



 
Last edited:
5B0836F8-BA0B-4590-9BB2-C8412C70CB4A.jpeg
Hi. I bought the W6170M vCard but after the vbios change it’s booting without boot screen.. I try also with open core and still nothing.. is ther something that I missed? Can someone help me? Many thanks!
 
  • Like
Reactions: CooperBox
Hi. I bought the W6170M vCard but after the vbios change it’s booting without boot screen.. I try also with open core and still nothing.. is ther something that I missed? Can someone help me? Many thanks!
You need OpenCore to see the OC boot picker! Try OLCP 0.3.1 - should work.
There is no EFI Boot screen available.

BTW: Disable sleep from energy settings!

Edit:
I assume you did a cold boot and experienced the NVRAM corruption. This effectively prevents the screen to turn on and you have a black screen boot, sometimes even only remote access since the screen does not come up on macOS startup. Have no solution for this problem, currently I do not recommend to run these cards in the iMac12,x systems.
 
Last edited:
Hi. In the picture is with 0.3.1 booted like you see at the cpu is xeon but still no successful..
Hi!

I will not and I cannot repeat the contents of the posts I made about this card.

1. flash the vBIOS from the first post
2. run OCLP again and write new OC to internal disk in case you have not changed the config after changing the card. you need to write a new OC after changing the hardware.
3. restart with PRAM reset
 
Nvidia Quadro K2100M - 2Gb Elpida VRAM


The following is a K2100M vbios for those of you that have the Elpida based GPU. The K2100M came with 3 different memory vendors: Hynix, Elpida and Samsung. By far the least common type was Elipda.

View attachment 1699002 View attachment 1699014


  • The original K2100M_EG2.rom did not have entries specific for initializing -Elpida vram based cards and therefore defaulted to a incorrect vram size of 1024Mb (1Gb). Many posts by members have identified this issue. @kuoshen <post>
  • MacOS applications addressing anything above 1Gb (1024Mb) could cause a hang in the GPU controller code or <artifacts>. If you were lucky, it just produced lower performance. Many of you returned the cards to the original Ebay sellers suggesting a fault in the hardware. We know this not to be true now.
  • The memory clock currently run at 2000Mhz. While this is good from a performance perspective and likely okay for Samsung & Hynix based cards, I have noticed some individuals who have reported some blown cards which run very hot and have short life expectancies. For the -Elpida based cards, I have chosen to keep the mem boost to 1700Mhz and the base clock to 705.5Mhz. This still represents about a 200Mhz overclock. However, it should be more stable and produce less heat and less fan noise. Possibly for those of you with the -7A (7.0Gbps) varient cards, you may indeed tolerate higher frequencies as those ram chips would be top end Micron/Elpida.
  • You still will need to use OpenCore for your boot screen and brightness parameters with this vbios.

Thank you to @BadBiscuit for doing the testing & screenshots:

2000Mhz mem VRAM:
View attachment 1699027

1700Mhz mem VRAM addressing full range 2048Mb:
View attachment 1699028


BigSur 11.0.1
View attachment 1699036

insanely great!
- Steve Jobs



I see this is a needed correction for my card. I've already been using the computer for a year with the unfortunate hobbled VRAM. At the moment, I'm not sure what exactly needs to be done to implement this after downloading. Thanks.
 
Hi!

I will not and I cannot repeat the contents of the posts I made about this card.

1. flash the vBIOS from the first post
2. run OCLP again and write new OC to internal disk in case you have not changed the config after changing the card. you need to write a new OC after changing the hardware.
3. restart with PRAM reset
Well I think I followed all the step, installed oclp on internal HDD modified config with that code line save & restart pram reset .. by the way with oclp I have random shut down. I look at the temp most are at 40/45 Celsius.. without oclp I have no issues with random shut down..
 
W
Well I think I followed all the step, installed oclp on internal HDD modified config with that code line save & restart pram reset .. by the way with oclp I have random shut down. I look at the temp most are at 40/45 Celsius.. without oclp I have no issues with random shut down..
Well,

how can I be of any help here? User claims to have followed the docs to the letter, nevertheless something is broken! Have five of those identical cards, the procedure has been tested by me several times with several cards in all 27 inch iMacs.

- so I ask you to repeat the whole procedure of installing OpenCore using OLCP to your internal HDD
- it will not be necessary to alter the plain new config to get the boot screen working. The settings are just necessary to get the external display working (pikera) and change the name to R9 M280X (rebuild device tree).

BTW:
- we asked all users to take pictures of the card and post them in case of trouble
- we asked users to save the original vBIOS of the new card in case of trouble
- the screen shot you were posting had clearly not applied the OpenCore settings as described, you would have seen a R9 M280X GPU....
- disable system sleep, do not even send the iMac to sleep using the Apple menu item
- OCLP tries to disable a non existing iGPU in your XEON based system

Please give it at least a try and report back.
 
I see this is a needed correction for my card. I've already been using the computer for a year with the unfortunate hobbled VRAM. At the moment, I'm not sure what exactly needs to be done to implement this after downloading. Thanks.
How did you apply the currently VBIOS ROM on your card? The process could be the same, except using the K2100M_EG2.rom. If you previously used the clip method to flash it, then I would suggest you explore the linux method (well documented on page 1) to avoid the need to disassemble everything.

See the 5th "Spoiler":
Questions and Answers. > Q3: Flash VBIOS? How do I do that?

Note: You shouldn't need to SSH from another system (for the screen already works), though it could be a good learning experience if you have not done that before.
 
Last edited:
Hello everyone.

Present system/hardware:
iMac12,2 (27-inch, Mid 2011)
3,4 GHz Quad-Core Intel Core i7
32 GB 1333 MHz DDR3
AMD Radeon HD 6770M 512 MB
BigSur 11.6.1
OCLP: 0.3.1

Expecting from China the following GPU:
FirePro M5100 (Red Dell PCB) flashed with "M5100-DELL-RED-GOP-BLUE_OBJI.rom.zip" from Ausdauersportler's post #16552.

The routine to install M5100 is nicely described in post 16695, however I'm a bit concerned due to the fact that I'm already having BigSur installed and OpenCore (including the postinstal legacy GPU patch) are copied to EFI.
Do I need just follow the guide (post 16695) and write OpenCore in safe mode to EFI once again, or do I need some extra steps, or contrary, do I have some shortcuts and some steps are not necessary?


And as usually hail to the thread maintainers and contributors!
Heartily thanks!
 
Hello everyone.

Present system/hardware:
iMac12,2 (27-inch, Mid 2011)
3,4 GHz Quad-Core Intel Core i7
32 GB 1333 MHz DDR3
AMD Radeon HD 6770M 512 MB
BigSur 11.6.1
OCLP: 0.3.1

Expecting from China the following GPU:
FirePro M5100 (Red Dell PCB) flashed with "M5100-DELL-RED-GOP-BLUE_OBJI.rom.zip" from Ausdauersportler's post #16552.

The routine to install M5100 is nicely described in post 16695, however I'm a bit concerned due to the fact that I'm already having BigSur installed and OpenCore (including the postinstal legacy GPU patch) are copied to EFI.
Do I need just follow the guide (post 16695) and write OpenCore in safe mode to EFI once again, or do I need some extra steps, or contrary, do I have some shortcuts and some steps are not necessary?


And as usually hail to the thread maintainers and contributors!
Heartily thanks!
Well, well! You guys are all very brave :)

There is in fact a short cut.

1. First of all you need to reinstall Big Sur after - or better before changing the GPU. There is a good chance the patcher has deleted the AMD7000 drivers and you will not get to the desktop using the old patched installation and the new card.
2. You should load the upcoming OCLP 0.3.2 TUI version and rewrite OpenCore to your internal EFI in the next step, you will be able to pre select the AMD GCN GPU from the patcher settings deep down and get a fitting OpenCore without any manual "safe mode" intervention. You will find the entry...
3. Turn system off and replace GPU, disconnect the internal drives
4. Turn system on and hold all fingers crossed and boot the Linux flash tool
5. flash the GPU
6. reconnect boot drives and internal display
7. boot, if all went well you will find yourself with an OC boot screen
8. turn system sleep off!

Have fun!
 
Last edited:
Hi, I've just flashed an WX4130 on my iMac12,2 with @internetzel latest GOP and now I don't know how to properly activate the main display. I've unplugged everything and put in a USB with OCLP for iMac12,2 and nothing shows up.
 
Hi, I've just flashed an WX4130 on my iMac12,2 with @internetzel latest GOP and now I don't know how to properly activate the main display. I've unplugged everything and put in a USB with OCLP for iMac12,2 and nothing shows up.
You should have created an OC rescue CD in advance and force booting it. This way you would have a OC boot screen.
Nevertheless the card will switch on the display after booting into any macOS version, if you cannot have OC booted you need to use High Sierra.

Note: You cannot use the NVIDIA OpenCore on the USB. Whenever you change the hardware OpenCore has to be rewritten. To make this easy there is a patcher setting to select the GPU within the system (Nvidia, AMD, AMD GCN).
 
Well, well! You guys are all very brave :)

There is in fact a short cut.

1. First of all you need to reinstall Big Sur after - or better before changing the GPU. There is a good chance the patcher has deleted the AMD7000 drivers and you will not get to the desktop using the old patched installation and the new card.
2. You should load the upcoming OCLP 0.3.2 TUI version and rewrite OpenCore to your internal EFI in the next step, you will be able to pre select the AMD GCN GPU from the patcher settings deep down and get a fitting OpenCore without any manual "safe mode" intervention. You will find the entry...
3. Turn system off and replace GPU, disconnect the internal drives
4. Turn system on and hold all fingers crossed and boot the Linux flash tool
5. flash the GPU
6. reconnect boot drives and internal display
7. boot, if all went well you will find yourself with an OC boot screen
8. turn system sleep off!

Have fun!
Thank you!

Shall I wait for OCPL 0.3.2 first, prepare new MacOS installer based on 0.3.2 and then reinstall BigSur using that media?
Or may I just use my "old" usb stick prepared with 0.3.1, reinstall BigSur without any postinstall patches and when available copy new OpenCore 0.3.2 to EFI?
 
Thank you!

Shall I wait for OCPL 0.3.2 first, prepare new MacOS installer based on 0.3.2 and then reinstall BigSur using that media?
Or may I just use my "old" usb stick prepared with 0.3.1, reinstall BigSur without any postinstall patches and when available copy new OpenCore 0.3.2 to EFI?
I believe that your "old" version of OpenCore will function, but you will need to manually tweak several settings (as described in the instructions for your specific AMD GPU model) for it to work. The advantage of the new version of OpenCore is that it will make it easier for you to change these settings by using a checkbox. It might be possible to get a pre-release version 0.3.2 if you are impatient, or you can just manually change the required settings for your model of GPU in 0.3.1.

It is important to reinstall a fresh version of macOS Big Sur (you can do this on top of your existing install to preserve the user-data, but replace the OS), as it sounds like your currently installed version has had some post-install patches applied to it. I don't believe the same kinds of post-install patches will be necessary for your new GPU, and trying to use the preexisting version of Big Sur with it will probably just not work properly.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.