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.
Looks like everything went well, the iMac misses an operating system to boot.
Try to create an High Sierra USB installer and start up the installation.
EnableGop seems to work since you have a working display, in addition the correct eDP support was already included in your base vBIOS.
You may upload the modded and original vBIOS? (both zipped)
Yeah, i still don't have installed an OS. (new SSD)

Sure thing, here are the vbios files.
 

Attachments

  • m1000m vbios.zip
    289.8 KB · Views: 60
  • Like
Reactions: Ausdauersportler
Success story: I got an iMac12,1 for free with a broken GPU and broken ethernet. I bought a cheap (25€) Dell NVIDIA Quadro M1000M (Maxwell, MXM-A). The first issue was that the DELL backplate didn't fit into the iMac cooling and was glued, so i removed it with a hairdryer and replaced with the stock one from the iMac.

View attachment 2386581

After the GPU was installed in the MXM slot, i disconnected the HDD and SSD (so that it automaticly boots from the USB drive) and tried to dump the vbios with GRML. This was not easy due to no networking. I used an usb ethernet adapter and tried to boot GRML from usb. For whatever reason, the newest GRML froze at starting (this one) so i used an older one. (2011_imac_usb.zip) This one worked perfectly.

By default, the second USB network card is not enabled, so i typed commands blind on my keyboard to enable it. Maybe it can help someone in the future:


After this i dumped, patched the bios on my mac mini (with OpenCore's EnableGop tool) and verified the firmware.


Result:
View attachment 2386580

I also tried OpenCore Legacy with Nvidia GOP injection before patching, but for whatever reason it didn't worked.
amazing! I don’t have the original rom (I got the m2000m from ali express, and it was shipped with the fw from this thread already 🤦‍♂️, at least with the same md5 sumcheck).

After fiddling a lot with this, I wasn’t able to show any menu picker with opencore. I had both macOS and linux previously (before installing the card) loaded with nvidia drivers.
Everything works eventually, with or without opencore, I just don’t have the picker, and I have to blindly select at boot.
Without opencore, booting linux, for a second I can see the mac boot option menu. sonstrange. Basically opencore wasn’t very useful to me.

I’ll try to add the gop on a original m2000m if I find one. if I can have the screen from the beginning of the booting would be great gor me!!

My only issue is that the card is stuck at 405Mhz, max. so the performance is not great. however being supported by the latest nvidia proprietary drivers (vs kepler that are not supported anymore) makes it more compatible with linux wayland and enable vulkan 1.3, making many steam games still playable.
So all in all it’s still appealing vs the old radeon

I’m curious to know about your results if you are planning to do some benchmarks (maybe on linux)

EDIT:

I tried to by adding the gop in the vrom, I tried both versions of the m2000m as well 2 from techpowerup.com for Dell quadro m2000m.
All the version gives the same outcome, black screen until a nvidia driver (web driver in macOS or the proprietary Nvidia driver linux) is loaded.

I removed the glass to have a better look at the screen, and with a bright light I can actually see the apple boot menu, and I can see basically everything. This happens with all the vbios tested.

I was never able to handle the backlight via opencore, I have tried different approaches, there might be some special combination, but to be honest I would like to have backlight from the start, needing OC to be able to see something (which is not the case for me anyway) doesn't sound good to me (eg changing hdd would require thinking about backing up OC etc). So I'm going to try the hardware approach. it looks easy to remove if I'm going to fed up with the brightness of the screen

EDIT2: hotwire method worked. I just had to remember to cut the 3.3v wire otherwise the system was going crazy (high vents and all).

Curiously it was shutting down to safety when the backlight were supposed to turn on but never did; so the psu was sending some current to the backlight, but apparently not enough, I think (I’m not an expert in electronics, I’m just guessing)
 
Last edited:
  • Like
Reactions: derplayer
I'fiddled with the vbios to boost performance in lnux, my primary OS. So the common issue of the card being stuck at P5, in linux it's exacerbated by being stuck at the lowest freq of P5. So the GPU is stuck at 405 MHz.
I managed to bump the P5 memory frequency, and bump the GPU P5 base freq to 540 MHz, but nothing I can do makes it go higher.

Linux also have a open source driver (which is in a pitiful state right now for older cards), for which Maxwell support is broken but still functional. Notably, there is the option to manually decide which performance profile to use.
Well, I tried and boy, it worked! Despite the fact that nouveau drivers are known to perform 10x worse than the binary drivers from NVidia, I managed to outscore the benchmark by 1.5x time.
The valley benchmark was even better than the score I got in windows getting 1600 in linux (no nvdia drivers, but nouveau drivers manually boosting to "P0") vs 1200 in windows (cards is able to scale up to 1.3 GHz in P5)

But the nouveau is terribly broken for every day usage (eg needing vulkan), it's getting there though. So I'll stick to the linux 550x drivers. Also because at 540MHz, I get some performance that are slightly worse than in windows, however, without opening a flame, the windows system is so sluggish and slow, I guess the lack in performance in windows is also because of the heaviness of the OS (fresh windows 11 install).

In retrospective, for linux user, the best option out there would be to spend a tiny bit more and get directly a HP Quadro T1000, for which recent opensource effort makes it possible to use the card as-is with vulkan and all, without the nvidia binary (they are not using the nouveau either, so the experience might be different to what I described above).

This also makes it more long lived support since the driver will be carried over the linux kernel, instead of being dropped unsupported in few years by nvidia.

I'll get one of those card in the future hopefully.
 
Last edited:
I'fiddled with the vbios to boost performance in lnux, my primary OS. So the common issue of the card being stuck at P5, in linux it's exacerbated by being stuck at the lowest freq of P5. So the GPU is stuck at 405 MHz.
I managed to bump the P5 memory frequency, and bump the GPU P5 base freq to 540 MHz, but nothing I can do makes it go higher.

Linux also have a open source driver (which is in a pitiful state right now for older cards), for which Maxwell support is broken but still functional. Notably, there is the option to manually decide which performance profile to use.
Well, I tried and boy, it worked! Despite the fact that nouveau drivers are known to perform 10x worse than the binary drivers from NVidia, I managed to outscore the benchmark by 1.5x time.
The valley benchmark was even better than the score I got in windows getting 1600 in linux (no nvdia drivers, but nouveau drivers manually boosting to "P0") vs 1200 in windows (cards is able to scale up to 1.3 GHz in P5)

But the nouveau is terribly broken for every day usage (eg needing vulkan), it's getting there though. So I'll stick to the linux 550x drivers. Also because at 540MHz, I get some performance that are slightly worse than in windows, however, without opening a flame, the windows system is so sluggish and slow, I guess the lack in performance in windows is also because of the heaviness of the OS (fresh windows 11 install).

In retrospective, for linux user, the best option out there would be to spend a tiny bit more and get directly a HP Quadro T1000, for which recent opensource effort makes it possible to use the card as-is with vulkan and all, without the nvidia binary (they are not using the nouveau either, so the experience might be different to what I described above).

This also makes it more long lived support since the driver will be carried over the linux kernel, instead of being dropped unsupported in few years by nvidia.

I'll get one of those card in the future hopefully.

EDIT:

Thanks to @galeon34 that sent me a vrom edited by @m0bil for the m2000m, I was able to get super performances out of my card, it was able to get at P0 performance, so it's amazing.

It's not my rom to share, but I think anyone with a Quadro M2000M card would want this ROM instead of the. one linked in OP
 
Yeah, i still don't have installed an OS. (new SSD)

Sure thing, here are the vbios files.
nice! When you can test the card with an OS, check that the card can boost clocks. Some Maxwell cards need a "power patch" in the vbios to enable high power states, otherwise they stay at the lowest power state.
 
  • Like
Reactions: derplayer
nice! When you can test the card with an OS, check that the card can boost clocks. Some Maxwell cards need a "power patch" in the vbios to enable high power states, otherwise they stay at the lowest power state.
I installed now High Sierra and Windows 10 on SSD. High Sierra booted with the basic GPU driver. But after installing the NVIDIA webdriver it stopped to work (black screen after boot). To make it work, i needed OpenCore. I couldn't change the boot parameter in safe mode, because partition mounting is locked there?
But i found out that when the Mac goes into sleep mode and wakes up, the GPU starts to work properly and the blackscreen disappears temporary until reboot. From here on, i could edit the OpenCore's bootargs and added "agdpmod=vit9696" to it. This fixed all blackscreen issues and the webdriver now runs fine.


2024-06-13_22-45-40.jpg


I did some first benchmarks, but noticed then that my CPU cooler is dead and doesn't spin anymore. For the other running fans i use Macs Fan Control. Now im waiting for a replacement to be shipped that i bought.

Geekbench benchmarks on High Sierra: Seems to look fine?

Geekbench CPU
Geekbench Compute

On Windows i had issues with the Intel iGPU. I tried to flash the iMac firmware with your patches to disable it, but my programmer threw errors because of overvoltage protection with a clip at reading. With the iMac connected to power, the erroes were gone, but the readed-out binary blob was garbage :(. I bought a cheap CH341A because i saw on YouTube that it works better with the iMac 2011. Maybe this one will work (i also want the turbo OC patches haha). But for now, i just blacklisted the iGPU driver in safe mode.

I also ran some Witcher 3 to test the GPU driver on Windows for a short time without a CPU cooler 😅.

2024-06-13_22-26-09.jpg
 
Last edited:
I have a question about web driver : If I run a 980m on high sierra, would there be any downside using Macvidcards.com drivers or is it a flawless experience ? Thanks in advance.
 
After having read the entire thread, I answer my own post : web drivers are official nvidia drivers that support modern cards, can be installed on macos up to High Sierra. Links provided at macvidcards.com are officials drivers from nvidia.com (I was under the impression they were hosting modified drivers).

At some point in the thread, nvidia drivers were reported to not work anymore because that had been un-certified by nvidia. I guess it's ok now because they are still recommended in the first page.

---

A Kepler card recently died in my 27' 2011 and I'm looking to buy a Maxwell card. I use High Sierra and Win10, and do some light gaming. I would rather avoid using opencore.

Good options seems to be gtx980m and m4000m. Price is 140e and 100e. m4000m is cheaper but requires opencore for brightness control.

What would be another good option is the m3000m, it's cheap (about 50e), has a TDP of 75W, fast enough, sadly no vbios yet.
 
Last edited:

iMac GTX 980M rom:


- Enables the internal display.
- Enables native boot screen using EnableGop.
- Brightness control and sleep/wake work
- External display should work on the mini DP closest to USB ports.

Many thanks to @dorninko for donating the card and @Ausdauersportler and @darioNV for testing this card and vbios.

View attachment 2224073


3 pipe heatsink and install with good thermal paste is a must.

Rom can be programmed from windows or linux using patched nvflash, or using clip + hardware programmer at 3.3V.

Similar to GTX 1070, there are gsync and non-gsync versions of this card. This rom is specific for the non G-Sync GTX 980M with Vendor/Device ID 10DE:13D7. Cards with Device ID 10DE:1617 are G-Sync and will not work with this rom.

The rom has included EnableGop and will provide native Apple bootscreen without the need of OpenCore.


Display tables changes based on @Santa's Little Helper work for the M4000M/M5000M cards.
Hi m0bil,
Thanks a lot for this vBios.
I was able to run the GTX 980M :)

But for me the Brightness control is not working under Windows. (macOS too but you wrote there is work to do)
Do i need to do something to make it work under windows?

best regards :)
 
Good options seems to be gtx980m and m4000m. Price is 140e and 100e. m4000m is cheaper but requires opencore for brightness control.

What would be another good option is the m3000m, it's cheap (about 50e), has a TDP of 75W, fast enough, sadly no vbios yet.
All Maxwell/Pascal cards in the thread behave the same regarding brightness control (opencore is needed for brightness control in macOS, but not for windows or linux).

I got a donated M3000M pending to be tested, sadly don't have much spare time, but will try to make it work over the summer.
 
My GTX 980M is now running without issues.
But i have termal problems
The card is installed on the grinded 3 pipe cooler.

If i run now a GPU stresstest like Furmark, the temperature is after 10 minutes so high, that GPU-Z is showing "Trm" in "PerfCap". Its resulting in a lower Clk for GPU and vRam.

Sounds this familiar or is my cooler somehow bad?
 
My GTX 980M is now running without issues.
But i have termal problems
The card is installed on the grinded 3 pipe cooler.

If i run now a GPU stresstest like Furmark, the temperature is after 10 minutes so high, that GPU-Z is showing "Trm" in "PerfCap". Its resulting in a lower Clk for GPU and vRam.

Sounds this familiar or is my cooler somehow bad?
what temperatures are you seeing on the card ? Are you using MacsFanControl or glued the odd temp sensor to the gpu heatsink ? What rpm do you see on the ODD fan ?

The grinded 3 pipes cooler is capable of cooling the card, but you need to make sure of proper installation and configuration, otherwise with high temps it is possible to break the card.
 
what temperatures are you seeing on the card ? Are you using MacsFanControl or glued the odd temp sensor to the gpu heatsink ? What rpm do you see on the ODD fan ?

The grinded 3 pipes cooler is capable of cooling the card, but you need to make sure of proper installation and configuration, otherwise with high temps it is possible to break the card.
let me show you a picture :)

Unbenannt.PNG


the pink bar in GPU-Z starts at 87° (reported by MacsFanControl) which is 101° in GPU-Z for Hot-Spot sensor.

the ODD fan is at max from 60° in MacsFanControl for the card sensor.

i used propper amount of thermal paste. also i used thermal pads for the vRam and voltage converter on the top of the card.

My idle temp is round about 50° in MacsFanControl for the card sensor.

thanks a lot for your input.
 
let me show you a picture :)

View attachment 2396333

the pink bar in GPU-Z starts at 87° (reported by MacsFanControl) which is 101° in GPU-Z for Hot-Spot sensor.

the ODD fan is at max from 60° in MacsFanControl for the card sensor.

i used propper amount of thermal paste. also i used thermal pads for the vRam and voltage converter on the top of the card.

My idle temp is round about 50° in MacsFanControl for the card sensor.

thanks a lot for your input.
At first sight, a 26°C difference between card (87°C) and GPU Heatsink (61°C) seems way too high, indicating maybe some installation problem (thermal paste, heatsink not fully touching the gpu chip...). That said I remember my gtx1070 reaching 84°C under stress with fan at 2500 rpm, I'll check my GPU heatsink temps...
But be careful, above 90°C it's possible to damage the card.
 
I wouldn't worry too much about the 980M hitting that temp under furmark, that's a serious worst case test that you'll likely not see those type of temps in a game scenario.

I've ordered a 1070M and 2700k which should arrive soon. Will put my WX7100 flashed up for sale.

Does the 1070M work under OCLP with accerlation or is it only under High Sierra?
 
At first sight, a 26°C difference between card (87°C) and GPU Heatsink (61°C) seems way too high, indicating maybe some installation problem (thermal paste, heatsink not fully touching the gpu chip...). That said I remember my gtx1070 reaching 84°C under stress with fan at 2500 rpm, I'll check my GPU heatsink temps...
But be careful, above 90°C it's possible to damage the card.
ok, thanks for help. I just wanted to get more insides, before i disassemple everything.
I did it now. I figured out that one thermal pad was to thick. So your sugesstion was correct. the chip did not have enough touch to the Heatsink.


difference between card and GPU Heatsink is now round abpout 10° and thermal throteling did not happen anymore.
i also changed the voltages in the biops for each clk step by -25mV. Still stable!

Thanks a lot for your help.
 
Hi everyone!

I've tried the P3000 in mid-2010 27" with windows 11 and everything works fine until the driver gets installed. I can hear windows welcome sound, backlight is on but there is no picture. RDP-ing into the machine shows that the driver is installed and reports that no display is connected.
I guess this could be due to the LVDS display connector in the older mid-2010? Card works like a charm in mid-2011 A1311 and A1312 but they both have eDP connectors.

Did anyone get the P3000 working on older machines?
 
Hi everyone!

I've tried the P3000 in mid-2010 27" with windows 11 and everything works fine until the driver gets installed. I can hear windows welcome sound, backlight is on but there is no picture. RDP-ing into the machine shows that the driver is installed and reports that no display is connected.
I guess this could be due to the LVDS display connector in the older mid-2010? Card works like a charm in mid-2011 A1311 and A1312 but they both have eDP connectors.

Did anyone get the P3000 working on older machines?
What do you mean by "everything works fine" ? You mean using EnableGOP the screen works at boot and until the windows Nvidia driver loads ?

Yes, there are no LVDS display/connector defined in the iMac P3000 vbios.
 
Hi everyone!

I've tried the P3000 in mid-2010 27" with windows 11 and everything works fine until the driver gets installed. I can hear windows welcome sound, backlight is on but there is no picture. RDP-ing into the machine shows that the driver is installed and reports that no display is connected.
I guess this could be due to the LVDS display connector in the older mid-2010? Card works like a charm in mid-2011 A1311 and A1312 but they both have eDP connectors.

Did anyone get the P3000 working on older machines?

iMac 2010 27" also uses eDP protocol for its internal display.
 
What do you mean by "everything works fine" ? You mean using EnableGOP the screen works at boot and until the windows Nvidia driver loads ?

Yes, there are no LVDS display/connector defined in the iMac P3000 vbios.
Can LVDS be added to the BIOS by any chance?

Edit: Sorry, I didn't answer your question - yes using enableGOP works OK and I have a screen in MacOS installer and also in Windows installer. When windows installs and picks up the driver from the update the screen goes blank. Remote desktoping into the machine and opening Nvidia control panel shows that no panel is detected. This led me to believe that the BIOS has only eDP ports configured (which you confirmed :)).
 
Last edited:
iMac 2010 27" also uses eDP protocol for its internal display.
Well, the connector on the motherboard is definitely LVDS as can be seen on motherboard type 820-2901-A. This is mid-2010 A1312 27"
Mid-2010 A1311 21.5" with motherboard type 820-2784-A has an eDP connector.
Maybe you mixed up those two?
 
Well, the connector on the motherboard is definitely LVDS as can be seen on motherboard type 820-2901-A. This is mid-2010 A1312 27"
Mid-2010 A1311 21.5" with motherboard type 820-2784-A has an eDP connector.
Maybe you mixed up those two?

I don't know. Perhaps you are correct.

There is a post that affirms that iMac 2009, 2010 use eDP protocol in the link below. Perhaps you can crosscheck with the poster about it.


There is also a Technical Datasheet issued by LG as well. On page 6, it clearly stated that "It has been designed to apply the 8Bit 4Lane Display port interface." Perhaps you can crosscheck with LG about their LCD panels.

 
Last edited:
  • Like
Reactions: m0bil
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.