as said before, the kepler does also display with the apple drivers and just need the web driver for full support, while the maxwell chip just send pixel with nvidia's driver...
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.
Frequently Asked Questions About NVIDIA PC (non-EFI) Graphics Cards
- Thread starter Asgorath
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
- Status
Can anyone confirm if FileVault is now working "properly" in High Sierra 10.13.X with non-EFI cards on MacPro5,1? Meaning you can turn on the machine, clearly select user and enter password, then boot macOS. Have read conflicting reports and feedback. Considering an upgrade from GTX 680 (official Mac version) to GTX 1080 (unflashed) before end of this year.
I apologize in advance if this has been answered elsewhere, I didn't see it and in all honesty was skimming through this thread w 170 pages of content. 
Feel free to link to a post that I might have missed!
I'm a grad student working with ML/RL and would like to *add* a single GTX 1080/1070 Ti to my cMP 5,1 (Mid-2010) while retaining my stock ATI 5770 for display (it has been perfectly fine for what I use my system for). I just want to use the GTX for compute with ML/RL code, so will need CUDA to be available but will not be using the card for any form of display. Kind of eyeballing the Zotac 1070 Ti Mini since it has a relatively-low power draw for the performance (rated at 180W max) but should significantly speed up my ML/RL schoolwork.
I'm good on RAM and CPUs (2 x 3.46 Xeons, 26GB), no other slots taken but #1 for the 5770 card.
Is this possible, and if so does anyone know if this is just a "drop in" upgrade or is there a documented build for getting this config to work? Trying to avoid building a dedicated ML box, but will if it's going to require to much hackery on my cMP...
Thanks!
Feel free to link to a post that I might have missed!I'm a grad student working with ML/RL and would like to *add* a single GTX 1080/1070 Ti to my cMP 5,1 (Mid-2010) while retaining my stock ATI 5770 for display (it has been perfectly fine for what I use my system for). I just want to use the GTX for compute with ML/RL code, so will need CUDA to be available but will not be using the card for any form of display. Kind of eyeballing the Zotac 1070 Ti Mini since it has a relatively-low power draw for the performance (rated at 180W max) but should significantly speed up my ML/RL schoolwork.
I'm good on RAM and CPUs (2 x 3.46 Xeons, 26GB), no other slots taken but #1 for the 5770 card.
Is this possible, and if so does anyone know if this is just a "drop in" upgrade or is there a documented build for getting this config to work? Trying to avoid building a dedicated ML box, but will if it's going to require to much hackery on my cMP...
Thanks!
should be possible, if you get the power plug from somewhere...the pci-e gives you 75w of power and you need at least another 2x6pin for the card, you should get the power from the second superdrive connector ( afaik... )
good luck!
edit : the zoatc 1070 just need that 1x8pin ?
good luck!
edit : the zoatc 1070 just need that 1x8pin ?
Thanks for the quick reply! So it sounds like power shouldn't be an issue, what about drivers? Are they needed if I'm just going to use the card for CUDA ops, or should CUDA-enabled libs just work out of the gate?
edit: Re the 1x8 pin, looks like it does! According to NewEgg specs, it comes with a 2x6pin => 1x8pin adapter
you need at least the webdriver and the cuda drivers. both is available with a little time shift after apple release a new security update from nvidia direct.
to be true, i know about issues some years ago while im still on usual pc...there it was not possible ( without some nerdy things... ) to run the red and green together...would you mind to google that before you buy ?
to be true, i know about issues some years ago while im still on usual pc...there it was not possible ( without some nerdy things... ) to run the red and green together...would you mind to google that before you buy ?
I apologize in advance if this has been answered elsewhere, I didn't see it and in all honesty was skimming through this thread w 170 pages of content.
I'm a grad student working with ML/RL and would like to *add* a single GTX 1080/1070 Ti to my cMP 5,1 (Mid-2010) while retaining my stock ATI 5770 for display (it has been perfectly fine for what I use my system for). I just want to use the GTX for compute with ML/RL code, so will need CUDA to be available but will not be using the card for any form of display. Kind of eyeballing the Zotac 1070 Ti Mini since it has a relatively-low power draw for the performance (rated at 180W max) but should significantly speed up my ML/RL schoolwork.
I'm good on RAM and CPUs (2 x 3.46 Xeons, 26GB), no other slots taken but #1 for the 5770 card.
Is this possible, and if so does anyone know if this is just a "drop in" upgrade or is there a documented build for getting this config to work? Trying to avoid building a dedicated ML box, but will if it's going to require to much hackery on my cMP...
Thanks!
Possible, but...
1) Which OS you are running?
2) Why pick this non-reference mini card if there are plenty of room in your PCIe compartment? I can't see the benefit but just potential trouble.
3) How you power the card? The 5770 already occupy one of the mini 6pin. And the 1080 / 1070Ti need two more, but only one extra available on the logic board.
power could be grab from the sata power connectors of the superdrive...
https://www.amazon.de/Adaptare-35109-Grafikkarten-Stromkabel-2-mal-SATA-Strom/dp/B004KMJQHW
https://www.amazon.de/Adaptare-35109-Grafikkarten-Stromkabel-2-mal-SATA-Strom/dp/B004KMJQHW
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Sorry if this is a really noobish question, but does the webdriver only work on a specific version of macOS, or only on a specific version or older.
As in can you update to a newer (bugfixed) webdriver, while staying on an older OS if you need to keep using older apps?
As in can you update to a newer (bugfixed) webdriver, while staying on an older OS if you need to keep using older apps?
One Web Driver for one macOS release.
Sadly, if you want to get new driver, you have to upgrade os too.
I don't have a second computer to flash my GTX 680.
I'm going to boot in to FreeDOS and flash the card, but before I do, I wanted to know if I can have my stock video card (Nvidia GT120) installed when I flash the card. I've read that you definitely can not, but I've seen it done, albeit with an AMD GPU.
So my question is: can I boot using my Nvidia GT120 to run nvflash against the device index # of my GTX 680? Or am I going to have to do this "blind" (script it and cross my fingers)?
Thanks a ton.
I'm going to boot in to FreeDOS and flash the card, but before I do, I wanted to know if I can have my stock video card (Nvidia GT120) installed when I flash the card. I've read that you definitely can not, but I've seen it done, albeit with an AMD GPU.
So my question is: can I boot using my Nvidia GT120 to run nvflash against the device index # of my GTX 680? Or am I going to have to do this "blind" (script it and cross my fingers)?
Thanks a ton.
Hi all, just a quick Q about this GFX, because I'm bidding ATM!
EFI Rom available/flashable that works in a Mac Pro 3.1 - OS X 10.10.5?
Thanks!
EFI Rom available/flashable that works in a Mac Pro 3.1 - OS X 10.10.5?
Thanks!
Last edited:
Probably flashable but no guarantee. We can't see what manufacturer made this but with those heat pipes it doesn't look like the Nvidia reference design. Of preference I would always choose an EVGA GTX680 as they manufactured the Mac version & use the reference design & are guaranteed flashable.Hi all, just a quick Q about this GFX, because I'm bidding ATM!
EFI Rom available/flashable that works in a Mac Pro 3.1 - OS X 10.10.5?
Thanks!
Forgot to mention that the GFX is from ASUS. Thanks for your answer.
I already own an EVGA GTX-680-4GB that I also have flashed with the help of this great MR community and is still working like a (race)horse in my cMP 5.1 (mid-2012).
Have a splendid 2018 ya all!
Lechaim!
Sucessfully flashed an Asus I picked up for 100$ including shipping off of eBay.
I'm now able to use my 4k monitor, but there is no audio through HDMI. I read the instructions for enabling it, but I don't think I have it in me. I'm going to buy some bookshelf speakers and call it a day.
I'm now able to use my 4k monitor, but there is no audio through HDMI. I read the instructions for enabling it, but I don't think I have it in me. I'm going to buy some bookshelf speakers and call it a day.
Hey all, sorry if this has been answered upthread but search didn't turn up anything so:
I am running 10.12 Sierra on a classic Mac Pro 5,1. Installed the webdrivers and a stock PNY Nvidia P600 (Pascal, half-length single-slot quad mDP card) around August last year, and all was working perfectly well. Left the machine running for weeks, had no wake from sleep issues, etc.
AFAIK I had not updated the operating system beyond minor security updates, as I was waiting on compatibility reports with apps I use before moving to High Sierra.
Updating some software the other weekend, lengthy install process ran overnight.... and then bam. Cannot get any signal out of the video card anymore. Rebooted, reset NVRAM, took it all apart and put it back together, cycled power on monitors, swapped ports, nothing. No sign of any physical damage to the card. Computer *seems* to boot up normally, from what I can hear, no disk issues. Everything is plugged into a Furman rackmount power filter / surge suppressor.
Talked to the software company and they confirmed that their install does not touch anything that would interfere with drivers (no .kexts, etc)
Next steps: I'm going to try a Safe Mode boot (don't have much optimism about that), then rolling back the software install via Time Machine once I can get into the computer via Firewire Target Disk Mode, and if that doesn't work, I will double-check if the card went bad by installing it in a PC (don't have one at home to test with).
I have an AMD WX3100 (Polaris) card coming which in theory is now supported natively, so I'll give that a shot.
Has anyone else experienced a similar issue? I'm just wondering if something might have happened to zap the drivers entirely, or if this is something else to do with my Mac's hardware.
I am running 10.12 Sierra on a classic Mac Pro 5,1. Installed the webdrivers and a stock PNY Nvidia P600 (Pascal, half-length single-slot quad mDP card) around August last year, and all was working perfectly well. Left the machine running for weeks, had no wake from sleep issues, etc.
AFAIK I had not updated the operating system beyond minor security updates, as I was waiting on compatibility reports with apps I use before moving to High Sierra.
Updating some software the other weekend, lengthy install process ran overnight.... and then bam. Cannot get any signal out of the video card anymore. Rebooted, reset NVRAM, took it all apart and put it back together, cycled power on monitors, swapped ports, nothing. No sign of any physical damage to the card. Computer *seems* to boot up normally, from what I can hear, no disk issues. Everything is plugged into a Furman rackmount power filter / surge suppressor.
Talked to the software company and they confirmed that their install does not touch anything that would interfere with drivers (no .kexts, etc)
Next steps: I'm going to try a Safe Mode boot (don't have much optimism about that), then rolling back the software install via Time Machine once I can get into the computer via Firewire Target Disk Mode, and if that doesn't work, I will double-check if the card went bad by installing it in a PC (don't have one at home to test with).
I have an AMD WX3100 (Polaris) card coming which in theory is now supported natively, so I'll give that a shot.
Has anyone else experienced a similar issue? I'm just wondering if something might have happened to zap the drivers entirely, or if this is something else to do with my Mac's hardware.
I'm not familiar with that card but any system update means that I have to wait for a web-driver-update before I apply the system update. Else the screen stays black after the next restart. (I keep an older, but OS X compatible card for those moments because sometimes I forget…)
What did you update? It will be helpful to know. I have a p4000 and have not had that happen. I think this has something to do with something you updated.
Where can anyone get actual Nvidia-manufactured cards? I thought they didn’t do any of their own manufacturing anymore. I see ebay listings for cards that use stock Nvidia card pics with the logo, and many manufacturers (PNY, ASUS, etc) use this for their box art as well, so it’s hard to even tell if it’s a reference design or not. Is there a comparison table somewhere out there?
[doublepost=1515166394][/doublepost]
Weird. I wonder if that installer did an automatic restart? It usually doesn’t, because it wasn’t installing anything at the system level, just application data. In any case, I hadn’t updated the system, because that would have caused a restart... and it was working fine before. I guess I will Target Disk into it and install the latest web drivers and see if that fixes it. Gotta update to High Sierra anyway for this Spectre CPU issue...
Of course the Nvidia logo will be on their box, because there is a Nvidia GPU inside. However, PCB is another matter. I don't think any PCB manufacture can randomly put the Nvidia logo on their custom PCB. From memory, the Nvidia logo only exist on the reference PCB (most likely are the Founder Edition card).
Nvidia brand graphic card may be still available on the 2nd hard market, but they are generally expensive, so I didn't really play attention on those cards. I aim at the reference PCB card with after market cooler, IMO, that's the best balance between compatibility and cooling ability (which eventually will affect the GPU's performance). But this is very personal.
- Status