imac 11.3
i7-860
8g DDR3
rx480 4g flash RX480_4GB_EG_adj.rom
no upgrade WiFi
@Ausdauersportler provides CoreEG2.fs and EDIDParser.fs in the eg2 folder modified with uefitool
I wonder if it's because I didn't load amdgop?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
iM 4,1-14,4 2011 iMac Graphics Card Upgrade
- Thread starter MichaelDT
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
@Ausdauersportler provides CoreEG2.fs and EDIDParser.fs in the eg2 folder modified with uefitool
I wonder if it's because I didn't load amdgop?
Really?!
That's not a signature and I don't understand the question.
AMDGOP.efi is required for UEFI Windows installations, but it obviously has no effect on proper operation of the EFI bootpicker. (Which you would have known if you actually read the posts I linked yesterday...)You may be facing a hardware problem. Try replacing the CR2032 cell (BIOS battery) and/or test the EFI picker with WiFi PCIE cable removed. (I'm only adding that last one, since I recently had jerky mouse movements as in your video, simply because I can no longer properly sit that cable in one machine...)
@Ausdauersportler provides IMAC-EFI-BOOT-SCREEN/GOP/AMDGOP.efi and modifies config.plist
Under windows 10 is also the same out of modern code 43
Installing atikmdag-patcher is also the same code 43
Under windows 10 is also the same out of modern code 43
Installing atikmdag-patcher is also the same code 43
Replace vbios RX480_4GB_EG_adj.rom with RX480_GOP.rom
Then Windows will be able to recognize the video card.
All those people wanting dear Windows to run perfectly on their iMacs are driving me mad - I wish it would not be possible to run it at all.
(Being able to run both Windows and macOS without problems in some cases is difficult to obtain with one single VBIOS.)
(Being able to run both Windows and macOS without problems in some cases is difficult to obtain with one single VBIOS.)
Last edited:
Maxwell and Pascal work better in windows 10 with better performance to energy ratio.
Yes, of course those cards are a different thing - somewhat not in the main focus of this thread.
Good evening , soooo.. big update the project was a success and with the new tool released a few days ago i was able to flash my k3100m and now high sierra works just fine ( even if the screen is not very bright) I would like to thank all the people that contributed to this this thread.
The problems started appearing when I tried installing windows 10 , I wanted to install it without opencore at first , but sadly my dvd drive doesnt seem to read data dvd , it would read cd and even dvd movies , but not a single windows dvd setup , not even original ones , so I installed W10 from usb and it worked fine but of cource the audio bug was there , so I sadly had to install opencore to the internal ssd and then boot the w10 partition with opencore at startup to allow the audio drivers to work.(first time using opencore so I do not know if i did things correctly). Now windows 10 audio works but open core always asks me which partition i want to run at boot and it stays there for 5 seconds , not a big deal but itd rather boot directly into the selected partition like it did before installing opencore .The second problem is that once windows 10 installs the nvidia drivers for the k3100m, those will increase w10 boot time by a lot and i mean by a lot , windows 10 without gpu drivers was booting in 10-15 seconds, after installing the nvidia drivers after the w10 logo boot with the spinning dots , the screen goes black and needs 70-80 seconds before i can see the w10 log in screen and desktop , and I have no idea why installing drivers causes this problem .Is there an alternative way to install windows 10 and working audio without first installing win 7 or 8 from dvd or opencore is my only solution . Could you please help me if you experienced these problems before? thank you
The problems started appearing when I tried installing windows 10 , I wanted to install it without opencore at first , but sadly my dvd drive doesnt seem to read data dvd , it would read cd and even dvd movies , but not a single windows dvd setup , not even original ones , so I installed W10 from usb and it worked fine but of cource the audio bug was there , so I sadly had to install opencore to the internal ssd and then boot the w10 partition with opencore at startup to allow the audio drivers to work.(first time using opencore so I do not know if i did things correctly). Now windows 10 audio works but open core always asks me which partition i want to run at boot and it stays there for 5 seconds , not a big deal but itd rather boot directly into the selected partition like it did before installing opencore .The second problem is that once windows 10 installs the nvidia drivers for the k3100m, those will increase w10 boot time by a lot and i mean by a lot , windows 10 without gpu drivers was booting in 10-15 seconds, after installing the nvidia drivers after the w10 logo boot with the spinning dots , the screen goes black and needs 70-80 seconds before i can see the w10 log in screen and desktop , and I have no idea why installing drivers causes this problem .Is there an alternative way to install windows 10 and working audio without first installing win 7 or 8 from dvd or opencore is my only solution . Could you please help me if you experienced these problems before? thank you
The temporary black screen delay when Windows 10 loads Nvidia drivers is a known problem with some Kepler cards (k3100m is one of them), and there is currently no way around it.
I see
thank you for letting me know , thats really bad , if i knew before I would have bought a different gpu.
Mojave installs from scratch do not work with Kepler cards. You have to upgrade in place from High Sierra. It’s a bug that Apple never fixed.
Oh, how irritating. Well it's too late now, not worth going back. I should say, I was already starting from High Sierra, unless you mean that the Mojave installer has to be run from within High Sierra? Rather than from the installer USB drive, that is.
EDIT: The GTX870M does not seem to be on this list of "Kepler" cards from Nvidia (which may be just a mistake on their part, this is not my expertise at all):
NVIDIA Support
nvidia.custhelp.com
-= NVIDIA Geforce GTX880M + HDD FAN AIRFLOW Redirect =-
In my last post I briefly covered reassembling the iMac with the GTX880 and flashing the vbios using XANDERSON’s USB (@xanderon) method with the hacked ROM provided by @nikey22. Now that I’ve successfully upgraded to macOS 12.6.1 using OCLP0.5.1 and bootcamped the second internal drive with Windows 10, it’s time to perform some benchmarking and reveal what kind of performance improvements we can expect with a GTX880M + HDD FAN airflow redirect (using a COOLMAC addon).
1. Test Configuration
Figure 1.0: iMac 2011 (12,2) System/GPU information and test configuration.
 |  |  |
- iMac12,2 (27-inch Mid 2011) (Tested machine)
- Intel i7-2600 3.4GHz
- Mem 16334MB @ 1332MHz
- Windows10 (Test OS) - Bootcamped on 2nd internal drive
- DELL Nvidia GTX 880M N15E-GX-A2, MXM-B (3.0) (@nikey22 vbios)
- ODD sensor relocated to GPU Heatsink
- COOLMAC (HDD Fan airflow pipe) - eBay Link
- MacsFanControl (v1.5.11)
2. GPU Benchmarking
First a disclaimer, I note that not all Nvidia GTX880M GPU cards are the same so the results reported herein are in relation to the Dell model using the @nikey22 vbios patch.2.1 Unigen (OpenGL/DirectX)
 |  |  |  |
Valley (2020) | Valley (2022) | Superposition OpenGL (2022) | Superposition DirectX (2022) |
To start, I decided to benchmark this build initially with Unigen Valley @ 1080p (1920x1080) to get a baseline comparison with what was performed earlier by @nikey22 back in 2020. Much to my surprise, direct-to-card air-cooling provided by the HDD FAN (or similar COOLMAC addon) can aid in maintaining some relatively minor clock boost achieving ~993MHz (100% load) at temperatures <65C . Under these conditions the GPU is capable of operating at frame rates FPS(average)>75 without thermal throttling which amounts to a performance improvement of ~23%.
A further comparison of the OpenGL scores reported with the AMD GCN v4 GPU RX480M (2800) and WX7100M (2949) cards clearly demonstrate that the Nvidia GTX880M (with direct-to-card air-cooling) can maintain a performance edge - albeit limited to the tested resolution. However, it is acknowledged that the boosted clock on the GTX880M is only comparable to the base clock of the RX480M.
Table 1.0: Unigen Valley @ 1080p (1920x1080) BenchMark Results
| Parameter | nikey22 | Test#1 | Test#2 | Test#3 | % Improvement |
|---|---|---|---|---|---|
| HDD FAN (RPM) | - | 1250 | 2500 | 4500 | - |
| FPS | 62.2 | 75.6 | 76.3 | 76.7 | 22.8 |
| SCORE | 2604 | 3162 | 3192 | 3210 | 22.8 |
| Min FPS | 8.4 | 29 | 29 | 14.9 | 233 |
| Max FPS | 98 | 141.7 | 143.1 | 145.6 | 46.7 |
| GPU Clock Freq (MHz) | - | 993 | 993 | 993 | - |
| GPU Temp (min/max) | ?,70-75 | 31,64 | 31,63 | 31,62 | - |
Next a similar framework of GPU tests were applied using Unigen's latest Superposition benchmark.
Table 1.1: Unigen Superposition @ 1080p (1920x1080) BenchMark Results
| Parameter | Test#1 | Test#2 | Test#3 |
|---|---|---|---|
| HDD FAN (RPM) | 1250 | 2500 | 4500 |
| FPS | 31.74 | 31.81 | 34.87 |
| SCORE | 4243 | 4252 | 4641 |
| Min FPS | 25.94 | 26.15 | 28.41 |
| Max FPS | 39.82 | 39.79 | 44.66 |
| GPU Clock Freq (MHz) | 990 | 993 | 993 |
| GPU Temp (min/max) | 32,65 | 31,63 | 31,62 |
Figure 2.1: Unigen Superposition benchmarked systems and equivalent ranking at 1920x1080 and medium quality.
Under the examined settings we can see the performance benchmarks for the GTX880M (GK104) under boost can be comparable to newer desktop variants of the Geforce GTX1080 (GP104) and even the GTX980 Ti (GM200) graphics cards.
2.2 3DMark (DirectX)
Now that we’ve established in the benefit of using the HDD FAN for direct-to-card cooling (using this COOLMAC addon), it’s time to delve a little deeper into the performance metrics. Below are screenshots of 3DMarks Fire Strike suite showcasing the DirectX 11 benchmark together with the online comparison. |  |
As shown above, 3DMark benchmarks for 1920x1080 using the DirectX11 graphics API demonstrate a similar spread of FPS metrics: (min) 23, (avg) 36 and (max) 42 at GPU temperatures, (min) 36C and (max) 59C. Unfortunately, at time of writing there isn't any online data available on the 3DMark benchmark database to directly compare a similar iMac build.
Note: The 3DMark benchmark data base provides a good online capability to publish iMac GPU upgrade results for future comparison.
3. Discussion
Benchmarking the Nvidia GTX880M (GK104) with direct-to-car air-cooling reveal on average improvements for the FPS metrics: (min):~200%, (avg) 23% and a (max) 47%. These results demonstrate a capability at least on a fundamental level to achieve baseline performance metrics comparable with lower end Nvidia Maxwell GM204 architectures such as the Quadro M4000M and GTX980M.3. Conclusion
Overall, the integration of the iMac12,2 and Nvidia GTX880M with some additional direct-to-card air-cooling has shown the ability to maintain a boosted clock frequency of ~993MHz at temperatures < 65C and overall increased performance metrics. The GTX880M may be the last in the line of compatible Nvidia Kepler cards but this low-cost workhorse is still able to compete with some of the more modern Nvidia and AMD GPUs (at base clock levels) that are being pursued in the upgrade market."This is where it started for me. This is where it ends." - Bourne
Next step - iMac 2011 Maxwell and Pascal Upgrades
Last edited:
Hello ! Is there any difference between installing windows 7 using the dvd drive then upgrade to 10 and via usb with open core to get the audio drivers working in terms of use/performance? also would a gtx 670 be compatible with these macs ? thank you
You should not install Windows on an old iMac.
If you want Windows that bad, build a Hackintosh PC. It would be much more powerful than the iMac 2011.
So, I am back. I found out what the problem was. Unfortunately it seems that on my many attempts to position the problematic clip correctly, I scrapped off a tiny resistor right next to the Bios chip.
This was the reason it was not booting even if I was flashing the modded bioses correctly.
I realised this when I compared two photos of my mobo and I saw an empty space where this tiny resistor should have been.
Ofcourse I could not find it anywhere.
Then, @dfranetic contacted me, I explained the issue and he gave me the exact particulars of the transistor. I could not find the exact same small one, so I got a bigger one (but with same specs) and I soldered it as best as I could.
Before soldering the resistor I flashed the modded EG2 bios, because I could not place the clip with the resistor in place.
Then used the GMRL usb and flashed the modded EG2 bios to my m5100, rebooted, held the “option” key and…
SUCCESS !!
Now the card is being recognized as R9 M370X and I have the OEM boot screen working.
HUGE thanks to @Ausdauersportler , @dfranetic , @jay508 , @duvelhedz
and to anyone else who helped.
I am writing this post using the iMac!
P.S.: For anyone caring, it was the R6100 resistor which is in fact a 3.3kOhm resistor in 0402 package. The iMac cannot read the ROM, so it does not POST. (credit to @dfranetic ). So be careful when placing the clip on the chip.
This was the reason it was not booting even if I was flashing the modded bioses correctly.
I realised this when I compared two photos of my mobo and I saw an empty space where this tiny resistor should have been.
Ofcourse I could not find it anywhere.
Then, @dfranetic contacted me, I explained the issue and he gave me the exact particulars of the transistor. I could not find the exact same small one, so I got a bigger one (but with same specs) and I soldered it as best as I could.
Before soldering the resistor I flashed the modded EG2 bios, because I could not place the clip with the resistor in place.
Then used the GMRL usb and flashed the modded EG2 bios to my m5100, rebooted, held the “option” key and…
SUCCESS !!
Now the card is being recognized as R9 M370X and I have the OEM boot screen working.
HUGE thanks to @Ausdauersportler , @dfranetic , @jay508 , @duvelhedz
and to anyone else who helped.
I am writing this post using the iMac!
P.S.: For anyone caring, it was the R6100 resistor which is in fact a 3.3kOhm resistor in 0402 package. The iMac cannot read the ROM, so it does not POST. (credit to @dfranetic ). So be careful when placing the clip on the chip.
Attachments
Last edited:
I had the same iMac 12,1 2011 with the same m5100 as you a few days ago, and I used CH134A to read the bios directly and back up the changes. I didn't add any resistors.So, I am back. I found out what the problem was. Unfortunately it seems that on my many attempts to position the problematic clip correctly, I scrapped off a tiny resistor right next to the Bios chip.
This was the reason it was not booting even if I was flashing the modded bioses correctly.
I realised this when I compared two photos of my mobo and I saw an empty space where this tiny capacitor should have been.
Ofcourse I could not find it anywhere.
Then, @dfranetic contacted me, I explained the issue and he gave me the exact particulars of the transistor. I could not find the exact same small one, so I got a bigger one (but with same specs) and I soldered it as best as I could.
Before soldering the resistor I flashed the modded EG2 bios, because I could not place the clip with the resistor in place.
Then used the GMRL usb and flashed the modded EG2 bios to my m5100, rebooted, held the “option” key and…
SUCCESS !!
Now the card is being recognized as R9 M370X and I have the OEM boot screen working.
HUGE thanks to @Ausdauersportler , @dfranetic , @jay508 , @duvelhedz
and to anyone else who helped.
I am writing this post using the iMac!
P.S.: For anyone caring, it was the R6100 resistor which is in fact a 3.3kOhm resistor in 0402 package. The iMac cannot read the ROM, so it does not POST. (credit to @dfranetic ). So be careful when placing the clip on the chip.
The cpu was also upgraded from i5 2400s to i7 2600 using everything normally without any problems.
Attachments
Hi. I had to add the new resistor because (by mistake) I unfortunately and carelessly scrapped off the OEM one when trying to position a problematic clip on the bios chip...
So I had to put a replacement otherwise the iMac was not booting.
Success upgrading to the EG2 mod on my own daily driver machine. Had image for the linux flasher through OCLP originally. Used a Raspberry Pi with flashrom over SSH to flash the BootROM chip. I buggered this up last week as thought I could get away with flashing the card and then the BootROM a few days later. If you do this, you WILL end up with a black screen. They both must be flashed at the same time.
Attachments
In the published repo there is currently only a license file and the usual markup readme file to be found. Could it be that a commit is missing? According to the readme there ought to be at least a disk image present somewhere.
I have to admit, that I would not have found that and I use git daily. Thanks a lot!
Last edited: