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.
Actually that's a good point...
I've noticed that the Fans were all running at their lower speed even though the card was reaching 96 degrees C.

Mac fan control was detecting the temperature as well but the ODD fan never kicked in.
Is it supposed to do that?
Can I fix this under Opencore somehow?
No, there is no software solution available and this happens to all replacement cards! You need either configure Macs Fan Control or move the ODD sensor to the head sink, too. This sensor is able to control the fans, too.
I did install the card with some good thermal past and pads on each RAM module so I should be good with that I guess.

EDIT: Unigine Valley has started crashing already so I don't know what's up with that.
Just use Macs Fan Control to ramp up the ODD fan before the GPU reaches 80C+ and the system crashes!!
 
Well I don't think it was caused by the iMac as I was able to get it running well with a GTX765M in the end and sold it.
It was most likely caused by my setup at the time somehow.

I am now getting 1500 score under Unigine Valley in the mentioned settings.
No crash during the benchmark. I was able to see the memory speed mostly stuck at 1500Mhz and the Core got up to 1200Mhz.

Brightness control with this card is working just fine and I have to see if deep sleep works also.
But it looks like the card is working in the end, right?!
The card is working as expected.

From the first post, last section:

7. GPU Temperature Monitoring is available on cards marked with a + or ++when using both custom VBIOS and OpenCore. This shows the temps, but does not control the fans!

Use Macs Fan Control to control the ODD fan. Check the PDF attached to this post for details.

The best solution as of today is to relocate the ODD sensor as described in this solution! The ODD sensor will drive the fan up using the Apple SMC software.

If you sell modded systems move the ODD sensor!
 
  • Like
Reactions: Rendell001
You are just missing the debugging session ahead!

Since this has not been a problem in the past we have two possibilities:

1. you found an OCLP bug
2. you found an Apple bug

To debug the first one please just try older releases of OCLP starting with 0.2.4 and go back until 0.1.7 and check if you find fully working version.

If you find a working older OCLP release please open a new issue on the OCLP Github page and describe you observations.

Thanks in advance!
Alright, 0.2.4 seems to fix the issue on my iMac 11,3 running Big Sur Version 11.6. Having BacklightInjector.kext enabled still broke backlight control but disabling it and adding AppleBacklightFixup.kext instead did the trick. Backlight control now works fine with full brightness.

This did not work on 0.2.5 and 0.3.0.

I'll open a new issue on Github.
 
In the past month, @edwardgeo, @internetzel, @m0bil and me had some very fruitful conversations. Result: We now have (a) workable solution(s) to make various makes of AMD WX4130/WX4150/WX4170 cards function properly in all iMacs from the iMac11,1 - iMac12,2 range! ('Big' WX4170 cards only fit 'big' iMacs, of course!) I have updated my original posts, so if you just want to know what to mod, simply visit the WX4130/WX4150 mod post or the WX4170 mod post. Curious types - read on below.
As I have reported in my HP WX4150 mod post, and @m0bil reported later for a Dell WX4130, some of those cards caused 'SMC confusion' in iMac11,2. I also found that the (originally) modded HP WX4150 will not POST in an iMac12,1. We found the culprit: All problematic cards have a temperature sensor IC, TMP431B (with top marking DUFC) present. This IC is connected directly to the iMac's SMC "0" SMBus. The problem: I2C address of this IC is 0x9A/0x9B, which is also the address of the TMP423B on iMac11,2 logic board (on the same bus). This causes a so-called 'address collision' and consequently 'SMC confusion'. My proposed solution: Disconnect TMP431B (on the dGPU) from SMBus to remove the offending I2C address. Luckily, there are two jumpers on the WX4130/WX4150 we can remove to achieve that. And it works!

iMac11.2_SMC0.png
Surprisingly, this also works in iMac12,1 - and it POSTs and runs with that (additional) mod. I did't find 'an offending I2C address' in the iMac12,1 schematic so I do not have an explanation why it works in this machine - possibly some SMC code 'remnant' from previous generation tries to read from that offending address, preventing the machine from POSTing.
Would, alternatively, simply removing the IC with that offending I2C address from the dGPU also work? I don't think so. (And it did not work in one iMac12,1 case.) There are other (safety) signals from that IC directly connected to the MXM connector, which can prevent the computer from POSTing when absent. While there are WX4130/WX4150 cards without that IC, they have different placing of some other components to route appropriate signals to the MXM connector.
While I was trying to upgrade my iMac11,1 with a modded (originally EEPROM-less) WX4170 I encountered a whole lot of issues - the machine would not POST, but it did allow a PRAM reset and did POST after that. While running macOS it would randomly shut down and wouldn't POST again until another PRAM reset. I (again) traced the problem to the TMP431B temperature sensor IC present on that card. Again, we have that IC on the iMac's SMC "0" SMBus, along with EMC1402-2 on the iMac logic board on the same I2C address (0x9A/0x9B), ie. another address collision. Proposed solution: Disconnect TMP431B (on the dGPU) from the SMBus (again). Luckily, WX4170 also has appropriate jumpers to do that!

iMac11.1_3_SMC0.png
I still don't have an explanation why the (modded) HP WX4150 did not cause such problems in iMac11,1 and iMac11,3. Maybe it's a combination of other additional factors...
Possibly. Maybe replacing the temperature sensor IC (TMP431B on those cards) with a supported one would work. I tried to do this with an ADM1032, but it did not report the temperature to SMC. (Even though ADM1032 is used on Apple's HD 6970M.) Another possibility would be EMC1402-1 (or maybe even TMP431A). (Currently both unobtainable at my usual suppliers.) If any of these ICs (in place of the original temperature sensor IC on those AMD cards) actually works, it should even report the appropriate temperature to SMC, which would make SMC properly control iMac's fans wrt. GPU temperature. Anyway, @Ausdauersportler 's ODD temp. sensor solution (probably) keeps the GPU even coller, so I won't be investing more time into such experiments.
I'm also happy to report that no GPU was damaged/broken during all the experiments and tests @edwardgeo, @m0bil and I did. One came close, though.
 
Last edited:
Thanks for your promp reply.
I use it for normal 2D Adobe illustrator and AutoCad work only.

May i get your advice is K2100M can support it?

I have that card in a iMac 11,3 in big sur 11.6.

Yes, as I can test, illustrator and autoCAD are working fine.

BUT, if I was you, I'll go with a newer AMD card if I could afford it. It's more future proof, as Monterey dropped support for older nvidia cards. So, I'm planning to get a AMD card and a bluetooth/wifi card to upgrade to Monterey in the next few months.
 
The card is working as expected.

From the first post, last section:

7. GPU Temperature Monitoring is available on cards marked with a + or ++when using both custom VBIOS and OpenCore. This shows the temps, but does not control the fans!

Use Macs Fan Control to control the ODD fan. Check the PDF attached to this post for details.

The best solution as of today is to relocate the ODD sensor as described in this solution! The ODD sensor will drive the fan up using the Apple SMC software.

If you sell modded systems move the ODD sensor!
Well I am not so sure it's working as expected cause the image/iMac is freezing and apps are crashing even when the temps of the card are cool.
I've taken your advice and moved the ODD sensor on the GPU itself however there is already a sensor in the heatsink which is why it's not making sense to me to do that.

Anyhow, I have had a bad experience with this card so far but it could just be because of a bad card.

In the end, I did reflash the customer GTX765M with the latest bios and I am now getting brightness control back + sleep is working so that's the most important.

I'll keep that WX4170 aside until the mod becomes more stable. Hopefully the card is fine.
If not, oh well, that would be just another bad eBay experience.
 
Well I am not so sure it's working as expected cause the image/iMac is freezing and apps are crashing even when the temps of the card are cool.
I've taken your advice and moved the ODD sensor on the GPU itself however there is already a sensor in the heatsink which is why it's not making sense to me to do that.
The original sensor on the sink does not control (for unknown reasons) the ODD fans with every replacement card listed on this thread. That is why we need to use Macs Fan Control to control the fan - otherwise your new card will burn itself under load to death. I cannot believe you had not recognised this problem so far?
Anyhow, I have had a bad experience with this card so far but it could just be because of a bad card.

In the end, I did reflash the customer GTX765M with the latest bios and I am now getting brightness control back + sleep is working so that's the most important.
Only the new vBIOS posted by @Santa's Little Helper offers Brightness Control using OpenCore. So I doubt now that your customer had a recent vBIOS and that he had Brightness Control at all.
I'll keep that WX4170 aside until the mod becomes more stable. Hopefully the card is fine.
If not, oh well, that would be just another bad eBay experience.
No evaluation possible based on the few data you posted...
 
Alright, 0.2.4 seems to fix the issue on my iMac 11,3 running Big Sur Version 11.6. Having BacklightInjector.kext enabled still broke backlight control but disabling it and adding AppleBacklightFixup.kext instead did the trick. Backlight control now works fine with full brightness.

This did not work on 0.2.5 and 0.3.0.

I'll open a new issue on Github.
I can confirm that I've having this issue as well. iMac 11,3 + K3100M latest firmware v4 + macOS 11.6
I tried to revert to OCLP 0.2.4 and disable BacklightInjector.kext and re-add AppleBacklightFixup.kext and modify config.plist accordingly after it was built and installed on EFI.

Still no brightness control though. I must have done something wrong, but at least I can confirm that the issue is there.
 
I know this may be a long shot, but I have a favor to ask. As some of you have already guessed, I am trying to find a hardware mod for the VER:1.0 WX7100 and/or V1.0 RX480, which would make these cards POST and run in 27' 2011 iMacs. This time I only have a single (non-POSTing) RX480 and images of VER:1.1 WX7100 (but no actual POSTing cards) to compare. Before making changes (and possibly destroying the RX480 in my hands), I'd prefer to know the values of certain 0201-sized resistors (which V1.0 cards lack).

So does anybody have a VER:1.1 WX7100 'lying around' (ie. possibly not mounted into an iMac)? If so, would you be so kind and use a multimeter to measure the resistances of 6 0201-sized SMD resistors - circled in blue rectangles in attached pictures (3 on top layer, 3 on bottom layer)? I (and perhaps others) will be eternally grateful. :)

Getting close to the upgrade, df; I will try to remember to test the R's before I install the 1.1 card.

Regards, splifingate
 
  • Like
Reactions: dfranetic
I can confirm that I've having this issue as well. iMac 11,3 + K3100M latest firmware v4 + macOS 11.6
I tried to revert to OCLP 0.2.4 and disable BacklightInjector.kext and re-add AppleBacklightFixup.kext and modify config.plist accordingly after it was built and installed on EFI.

Still no brightness control though. I must have done something wrong, but at least I can confirm that the issue is there.
Take a look at the OCLP Github. There is a new BacklightInjector.kext that includes the display panel ID for the iMac 11,3. I don't know if Apple used more than one type of panel in the iMac 11,3 but mine works fine now after providing the ID of my display panel. AppleBacklightFixup.kext is not necessary anymore after replacing BacklightInjector.kext.
 
I can confirm that I've having this issue as well. iMac 11,3 + K3100M latest firmware v4 + macOS 11.6
I tried to revert to OCLP 0.2.4 and disable BacklightInjector.kext and re-add AppleBacklightFixup.kext and modify config.plist accordingly after it was built and installed on EFI.

Still no brightness control though. I must have done something wrong, but at least I can confirm that the issue is there.
Could you please provide your panel type using this procedure:

Your panel ID can be found by going to System Preferences > Displays > Color > Open Profile > mmod

Do this without OpenCore in use booting into High Sierra!

If you are not sure what I am asking for take a look here
 
Could you please provide your panel type using this procedure:

Your panel ID can be found by going to System Preferences > Displays > Color > Open Profile > mmod

Do this without OpenCore in use booting into High Sierra!

If you are not sure what I am asking for take a look here
Also, from terminal:

Code:
ioreg -n AppleBacklightDisplay -arxw0 | grep AppleBacklightDisplay


<string>AppleBacklightDisplay</string>
        <string>IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/P0P2@1/IOPP/GFX0@0/NVDA,Display-A@0/NVDA/display0/AppleBacklightDisplay-610-a007</string>
        <string>AppleBacklightDisplay</string>
        <string>AppleBacklightDisplay</string>

I think last four characters of long string, in this case a007 is the important part
 
I ordered a WX7100 graphics card. Version 1.1 The video card will arrive soon and I will check. It's new. View attachment 1857227
Spending $700 for a WX7100 is an overkill for a 10+yrs old iMac.
The M1 MacMini is sold at a similar price, and a newer version might be a/v in this month from Apple.

At 1/5 to 1/4 the price you can get a WX4130 or WX4150, which should match a 10+yrs old iMac.

But to each his own.
 
  • Like
Reactions: Ausdauersportler
Could you please provide your panel type using this procedure:

Your panel ID can be found by going to System Preferences > Displays > Color > Open Profile > mmod

Do this without OpenCore in use booting into High Sierra!

If you are not sure what I am asking for take a look here

Also, from terminal:

Code:
ioreg -n AppleBacklightDisplay -arxw0 | grep AppleBacklightDisplay


<string>AppleBacklightDisplay</string>
        <string>IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/P0P2@1/IOPP/GFX0@0/NVDA,Display-A@0/NVDA/display0/AppleBacklightDisplay-610-a007</string>
        <string>AppleBacklightDisplay</string>
        <string>AppleBacklightDisplay</string>

I think last four characters of long string, in this case a007 is the important part
It's 9CD6. I have that immediately in my head because I've done AppleBacklight.kext modification as per nikey22's instruction several times over the year lol. (I double checked with ioreg just now to be doubly sure)

Take a look at the OCLP Github. There is a new BacklightInjector.kext that includes the display panel ID for the iMac 11,3. I don't know if Apple used more than one type of panel in the iMac 11,3 but mine works fine now after providing the ID of my display panel. AppleBacklightFixup.kext is not necessary anymore after replacing BacklightInjector.kext.
How awesome it is to wake up and see that 9CD6 was specifically added to Github during the sleep. Everything is working now. Thanks to all contributors.

I've been using dosdude's patched Catalina for a long time, but then recently it would not wake from deep sleep, showing kernel panic before it automatically rebooted. I'm certain it's a recent issue because initially it can sleep/wake just fine. Anyhow so I came here again to see that everyone's encouraged to migrate to OCLP. I also see that Nvidia sleep/wake cycle is also addressed. So I performed clean install + Migration Assistant and voila!

Thanks again. I respect you all.
 
Last edited:
Can you please post a link where I can get a WX4130 or WX4150 for $140-175?
Cheers.
Just search in AliExpress or eBay.
A few are selling at the high end of the price.
Every now and then, they will pop up in varying price.
I bought them last year at roughly half the price, though it has climbed up significantly.
 
Spending $700 for a WX7100 is an overkill for a 10+yrs old iMac.
The M1 MacMini is sold at a similar price, and a newer version might be a/v in this month from Apple.

At 1/5 to 1/4 the price you can get a WX4130 or WX4150, which should match a 10+yrs old iMac.

But to each his own.
I don't want a MAC MINI M1.
I bought for $700 once and for a long time. Let the iMac be old. But useful!
 
Question on the OCLP install process

I've had my 12,2 working (almost always, save sleep issues) on High Sierra with a K3100M for a couple of months now and want to give Big Sur a go. Can I upgrade my High Sierra after I install OCLP with the Big Sur distro on a flash drive or do I have to reformat the whole thing and start from scratch and then migrate from the backup?

thanks
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.