Hi i've been looking allot on this forum and know about stressing the card or putting it in slot 2 but does a properly efi flashed 970/980/ti cause the expansion fans to rev? i like the look of the mac pro on the desk and im always going back to the 5770 because it doesn't rev the expansion fans they always stay at 800rpm. with a none flashed card the fans go up to like 1700 while just watching a youtube video...its very very annoying. So want to know if i should shell out some dough for a properly flashed card.
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.
Do UnFlashed/Flashed 970/980/ti Cause Expansion Fan Revs?
- Thread starter Vic Damone Jr
- Start date
- Sort by reaction score
It seems quite a few cards doing this, even though my 7950 Mac Edition card (not the flashed one) sometimes will do this.
And now, I am running two 7950 without Mac EFI in my cMP with the internal power only. The fan behave exactly the same as if the switches are selected to the Mac EFI ROM.
However, I realise that the PCIe fan is not just react to the power draw from the cables or slots, but also the cards' temperature. (Even though the card's temperature is not avail for us to monitor in OSX).
If I install the card in slot 1 and 3, the card in slot 1 will have much better cooling. So, the PCIe fan can stay at 800 when idle. If I install the 2nd card in slot 2, which cause the card in slot 1 run at much higher temperature. Even though both card still at idle, with exactly the same power draw. The PCIe fan idle speed will go up a bit (around 900). Stress the card will only make this worse, because once the card heated up, it's very hard to cool down (e.g. New idle at around 55C rather than 50C after cood boot, PCIe fan idle at above 1100). Select EFI ROM has no effect on this behaviour.
AFAIK, some 9xx card's fan won't spin up until certain temperature (e.g. 60C). May be it's the cause of why the PCIe fan spin up. By watching YouTube, the card running a bit warmer (e.g. 58C), which trigger the PCIe fan to spin up. However, it's just a relatively light loading for the card. The card's fan no need to spin up. And just allow the GPU run a bit warmer.
This is just my hypothesis. You may leave the case door open and run YouTube. If the PCIe fan spin up, but not the card's fan, then this may be the reason.
And now, I am running two 7950 without Mac EFI in my cMP with the internal power only. The fan behave exactly the same as if the switches are selected to the Mac EFI ROM.
However, I realise that the PCIe fan is not just react to the power draw from the cables or slots, but also the cards' temperature. (Even though the card's temperature is not avail for us to monitor in OSX).
If I install the card in slot 1 and 3, the card in slot 1 will have much better cooling. So, the PCIe fan can stay at 800 when idle. If I install the 2nd card in slot 2, which cause the card in slot 1 run at much higher temperature. Even though both card still at idle, with exactly the same power draw. The PCIe fan idle speed will go up a bit (around 900). Stress the card will only make this worse, because once the card heated up, it's very hard to cool down (e.g. New idle at around 55C rather than 50C after cood boot, PCIe fan idle at above 1100). Select EFI ROM has no effect on this behaviour.
AFAIK, some 9xx card's fan won't spin up until certain temperature (e.g. 60C). May be it's the cause of why the PCIe fan spin up. By watching YouTube, the card running a bit warmer (e.g. 58C), which trigger the PCIe fan to spin up. However, it's just a relatively light loading for the card. The card's fan no need to spin up. And just allow the GPU run a bit warmer.
This is just my hypothesis. You may leave the case door open and run YouTube. If the PCIe fan spin up, but not the card's fan, then this may be the reason.
Last edited:
All modern cards rev the fans from time to time. The only way to prevent it is in a driver. Connecting directly to the power supply helps a little.
As far as I know, no one was ever able to determine the cause as to why some people saw their expansion slot fans spin up at start up. I don't think it had anything to do with EFI or non-EFI firmware. The people who suffered from this had to put some load on the GPU by running a benchmark utility such as LuxMark or Unigen Heaven or Valley for a bit, then quitting the apps and removing the load placed on the GPUs would cause the expansion slot fans to return to normal behavior.
I don't recall having read anywhere that power source or drivers had anything to do with the cause.
I don't recall having read anywhere that power source or drivers had anything to do with the cause.
It doesn't happen to PCs like that. I think it has to be the SMC and card not speaking correctly to each other.
Obviously, that's not how it was intended to work... But what made it extremely hard to fix was that not everyone saw this happen. I forgot which thread contained this discussion originally...
i read on here if you add cuda z as a startup program that the fans will start normally, has anyone tried this as i dont have a non apple card right now.
Last edited:
You only need to launch CUDA Z when the fan bug turns up. Then you go to the performance tab, wait for the fans to climb again, then quit the app.
I used to follow the threads about this problem extensively. I never had the problem myself, but I was curious about it.
Anyway, the fan problem has nothing to do with the source of the EFI (or lack of EFI) in the video card. This was evidenced by the fact that it also occurs for some people using official Mac Edition cards from EVGA, and more importantly, even for people with official Apple 5870 video cards. There was another thread where someone claimed to have discovered that it was a bug in the SMC firmware. If true, the problem will probably never be fixed.
I seem to recall that by far most people had the problem with 2009 MP. I cannot remember if anyone experienced it with a 2010. The 2009 and the 2010 do have different SMC firmware.
If the problem occurs, there are two workarounds: (A) putting it in slot 2 instead of 1, and (B) after bootup, run anything that "works" the card by using GPU accelerated effects, such as the aforementioned CUDA Z.
I've always wondered though, with more and more GPU accelerated effects for certain features in the OS X desktop itself, if a simpler (B) solution exists simply by activating one of those OS X features? Can't test this myself as I don't have the problem.
Anyway, the fan problem has nothing to do with the source of the EFI (or lack of EFI) in the video card. This was evidenced by the fact that it also occurs for some people using official Mac Edition cards from EVGA, and more importantly, even for people with official Apple 5870 video cards. There was another thread where someone claimed to have discovered that it was a bug in the SMC firmware. If true, the problem will probably never be fixed.
I seem to recall that by far most people had the problem with 2009 MP. I cannot remember if anyone experienced it with a 2010. The 2009 and the 2010 do have different SMC firmware.
If the problem occurs, there are two workarounds: (A) putting it in slot 2 instead of 1, and (B) after bootup, run anything that "works" the card by using GPU accelerated effects, such as the aforementioned CUDA Z.
I've always wondered though, with more and more GPU accelerated effects for certain features in the OS X desktop itself, if a simpler (B) solution exists simply by activating one of those OS X features? Can't test this myself as I don't have the problem.
what card do you use?
i put a gtx 980 none flashed in a 2009 mac pro and the fans would settle at about 1100 and with any effort go up to about 1700 im speaking of the expansion fan
I currently use an MVC-flashed GTX 980--the fans don't spin at all when the video card is idle.
Previously I had a self-flashed GTX 680 4GB, and before that the stock AMD 5770. I've been happy with all three cards.
Worth mentioning it's practical to start CUDA-Z as Login Item. Only need it for cold starts.
That was me.
I've contacted Apple before about this issue and discussed it with them at great length. They don't feel there's anything wrong with their implementation, even though there clearly is.
On certain cards (mainly, the GT120 and RV770), the SMC will actually read the thermal information from the card directly and use that to base it's fan speed calculations on. They seemed to drop this feature for the 5770/5870, but still assume those cards have their own fans and ignore the power consumption of the slot/aux cables accordingly.
For everything else, the SMC assumes that the card does NOT have a fan, and requires active cooling from the PCI-e expansion slot fan. This is supposed to be based on the power consumption of that slot and the auxiliary cables. Unfortunately the code they're using to do this is completely and utterly bugged up. The way it was explained to me (by an Apple engineer) is not the way it behaves IRL, which is evident by the vast number of people who experience fan related issues that go away after you stress the system a bit.
TLDR; the short version is that Apple's entire thermal management of the Mac Pro is very poorly implemented in software. It's not going to be fixed anytime soon since from what I've heard SMC updates for the 2009/2010/2012 Mac Pros are impossible (there's no way to update the CPU daughter card SMC through the logic backplane, and apparently both need to be updated in tandem so the versions match).
-SC
So do the people who say they don't have a problem actually do and don't have a fan monitor and are oblivious to the issue?
Ive had 6x 4.1/5.1 mac pros in my life and every single one did this
I haven't got a clue.
I've tried to figure out why some machines do it and some don't. The SMC versions are all identical for the 2010/2012 machines, there never were any updates for that.
The only thing I can think of is that it has something to do with the power consumption of the various components located in the upper half of the machine (PCI-e slots & disk drives). The core of the problem seems to be that the SMC bugs out and miscalculates the fan speeds if you're drawing too little or too much power during startup, and the issue will only rectify itself if you load the machine down, forcing the SMC to re-evaluate whatever function or formula it's using to determine the fan speeds. Once that happens the machine will run quietly until you do another cold boot, which seems to wipe away whatever state table the SMC maintains while the machine is operational.
Again, I've talked to Apple about this issue. I sent them a ton of information from my fans, both after a cold boot (running noisily) and after it had been loaded down. Apple claimed the fans were *correct* during a cold boot (which made no sense, because that's when the machine was the most noisy). When I pressed them for details on why the issue resolved itself after I ran an OpenGL intensive application for ~30 seconds, I never got a response. I must have had 15 messages between me and the engineer who was assigned to the problem, they went through great lengths to describe how everything was working properly, and the moment I said "well, why does it do this then?"... that was it, I never heard back from them. That bug is still open too.
-SC
I have iStat Menu monitoring my system and my expansion slot fan is at 800rpm on reboots. This was true when I used my EFI flashed GTX 680 and is still true with my unflashed GTX 980.
My 2009 Mac Pro would rev loudly at startup when the Radeon 4870 was installed. It was irritating, but I assumed that's the way it was supposed to work.
I didn't know it was the GPU/PCIe fan until I removed the card and installed an unflashed PC card.
I was stunned by how quiet the machine was when I powered it on with the PC card in.
It's really nice that the Scottish Captain was able to pinpoint a possible explanation for that revving.
I didn't know it was the GPU/PCIe fan until I removed the card and installed an unflashed PC card.
I was stunned by how quiet the machine was when I powered it on with the PC card in.
It's really nice that the Scottish Captain was able to pinpoint a possible explanation for that revving.
In case it's helpful, you can read through a great deal of information on this topic in my old thread from a while back (link below). It also includes links to other discussions from over the years. Unfortunately, it's just a compilation of observations, tests, and speculation. We never found a solution, and it sounds like the Cap'n didn't have any luck with Apple (thank you though for going through that effort!).
In the end, I gave up and simply changed my computing habits. I don't shut down my Mac Pro anymore. I just use sleep mode overnight, and restart as needed. Correct fan behavior will persist through sleep and restart. On the rare occasion that I do have to do a full shut down and cold boot, I use OpenGL Extensions Viewer to restore normal fan operation. I'd also recommend attaching your Mac Pro to a UPS, both as a general good practice, and also for the peace of mind considering it will be on or sleeping at all times.
https://forums.macrumors.com/thread...-problems-with-non-apple-video-cards.1685629/
I hope this helps!
In the end, I gave up and simply changed my computing habits. I don't shut down my Mac Pro anymore. I just use sleep mode overnight, and restart as needed. Correct fan behavior will persist through sleep and restart. On the rare occasion that I do have to do a full shut down and cold boot, I use OpenGL Extensions Viewer to restore normal fan operation. I'd also recommend attaching your Mac Pro to a UPS, both as a general good practice, and also for the peace of mind considering it will be on or sleeping at all times.
https://forums.macrumors.com/thread...-problems-with-non-apple-video-cards.1685629/
I hope this helps!
Definitely not. First of all my computer is very quiet despite being right there next to me. Second, I do use fan monitoring software and there was no noticeable PCIe fan changes when I upgraded from the stock Apple HD5770 to GTX 680 and again to GTX980.
This is the best theory I've heard so far. I do have a lot of crap installed:
2 optical drives
2 SSDs
2 HDDs
1 GPU card
1 USB 3.0 card
I suppose the theory is easily testable. I could unplug everything except the GPU and a single SSD with OS X on it, then check fan speed. I could then turn off, hook up one more device, turn on, and check again. Repeat until everything is installed. If you think there is some merit to trying that, I'll give it a shot this weekend.
EDIT: It just occurred to me that I have a W3690 installed which in itself probably consumes more power than normal, and there's not much I can do about that. I still think the theory could be checked if enough people with and without the problem listed their installed hardware.
Last edited:
I did this kind of experimentation with my machine back when I was investigating the issue, and was not able to find any variables that affected the fan behavior (except for moving the video card from slot 1 to 2, but that had its own problems). However, more data is always better. If you're willing to donate some time to do this testing as well, it's possible you might turn up something... "actionable."
You'd be coming at things from the other direction (looking for something that triggers the problem) as opposed to my tests (looking for something that solves the problem). For reference, my machine is a 6-core 3.33, 2 SSDs, 1 HDD, and 1 USB3 card. Compared to you, I have one less optical drive and a W3680. My GPU is a GTX670, though it's a quirky FTW model so it's sort of a 670/680 hybrid.
The full context is in the "My Observations" section of the thread I linked above, but here's the relevant part:
"In slot 2, the PCI and PS fans run slightly high, but closer to normal. The OpenGL Extensions Viewer trick has no further effect. One interesting oddity: installing a HDD in sled 3 drops the PS fan speed to normal (600rpm), but the PCI remains the same. Overall, having the video card in slot 2 would be acceptable for me if not for the fact that the fans still ramp up unnaturally high when the video card is under moderate to heavy load in Windows. Others have said that using slot 2 works around the problem, but in my case at least, it's only a partial workaround."
This. I do this quite regularly with my EVGA GTX 980 Ti SC. I usually only have to do it, however, if the computer starts from a complete shut down. Restarting the computer doesn't cause this.
Interesting explanation from Apple. My 4,1 with the stock 4870 never experience the fan issue. So it seems even though it's fan related, it still need some other factor to make the bug actually happen.
On the other hand, when I am using my 7950 / R9 280, regardless it's a Mac Edition / flashed / non-flash card. All of them may suffer from this fan issue (not 100% hit rate even from cold boot).
But all I need to do it just keep use my computer normally. And I usually will open some flash webpage which stress the card a bit, and then the fan will go back to normal by itself (from 1500 idle -> 800 idle).
On the other hand, when I am using my 7950 / R9 280, regardless it's a Mac Edition / flashed / non-flash card. All of them may suffer from this fan issue (not 100% hit rate even from cold boot).
But all I need to do it just keep use my computer normally. And I usually will open some flash webpage which stress the card a bit, and then the fan will go back to normal by itself (from 1500 idle -> 800 idle).
However, more data is always better. If you're willing to donate some time to do this testing as well, it's possible you might turn up something... "actionable."
I pulled or unplugged all the drives and cards I could, even the optical drives. There was no change in fan speeds; they were approximately the same either way.
PCI 1100
PS 700
Exhaust 600
Intake 600
BoostA 850
People are talking about at least two different fans in this thread (PCI and PS), so I actually don't know which one is supposed to be the problem.