MP 1,1-5,1 Psu fan speed CMP

[Ih8reno]

Hi all, on my 2010 Mac Pro quad 3.2 I find that whenever I do anything that slightly taxes the gpu the fans get annoyingly loud in the psu. I am using a flashed gtx 680 with 2gb. Now is there a safe setting I can use on macs fan control that would be a good mix of cooling and noise levels on the psu fans? I am hesitant because I don’t want to kill the psu of the machine.

 

[Flocarino]

Have you ever cleaned the PSU with compressed air to remove the dust......it might help.

 

[tommy chen]

and look at the temperature of the PSU using macsfancontrol

 

[Ih8reno]

Have you ever cleaned the PSU with compressed air to remove the dust......it might help.

Hi yes, cleaned it out

 

[h9826790]

-- How to fix high PCIe / PSU fan speed after GPU upgrade --
- https://forums.macrumors.com/thread...when-using-pc-non-efi-graphics-cards.1946672/
- https://forums.macrumors.com/threads/airout-a-solution-to-the-racing-fan-bug.2143951/

 

[Snow Tiger]

Hi all, on my 2010 Mac Pro quad 3.2 I find that whenever I do anything that slightly taxes the gpu the fans get annoyingly loud in the psu. I am using a flashed gtx 680 with 2gb. Now is there a safe setting I can use on macs fan control that would be a good mix of cooling and noise levels on the psu fans? I am hesitant because I don’t want to kill the psu of the machine.

I'm currently overhauling the following rig :

Mac Pro 4,1 > 5,1 ( factory 2009 )
boot ROM 144.0.0.0.0.
2 x X5690 Six Core 3.46 GHz
96 GB ( 6 x 16 GB ) 1333 MHz DDR3 ECC
AMD RX 480 8GB with Blower type fan
SATA SSD ( temporary )

The System has been undergoing preliminary stability and thermal tests with all her processor cores and memory at continuous load for the last 24 hours . Hardware Monitor is used for the sensor readings .

Everything is cool and pretty quiet with the following MFC settings .

PSU fan = 1200 RPM Constant .
PCIe fan = 1200 RPM Constant .
Intake fan = 1200 RPM Constant .
Exhaust fan = 1200 RPM Constant .
CPU A Booster = 1700 RPM Constant .
CPU B Booster = 1700 RPM Constant .

This is a pretty high end configuration for a Nehalem cMP .

Final configuration will have a Highpoint PCIe miniSAS Rocket RAID Card 27xx ( one internal port and one external port ) . And four 6TB enterprise SATA mechanicals will be installed in reverse facing HDD Bay Sleds . A bootable Highpoint PCIe Flash RAID Drive Card will also be installed with nVME SSDs. The installation of these components later on will likely alter the system fan settings .

This Mac will also have installed later a TransIntl front chassis filter , which will probably require the fan rotationals to be raised by about another 5 to 10 percent from current .

 

[startergo]

I'm currently overhauling the following rig :

Mac Pro 4,1 > 5,1 ( factory 2009 )
boot ROM 144.0.0.0.0.
2 x X5690 Six Core 3.46 GHz
96 GB ( 6 x 16 GB ) 1333 MHz DDR3 ECC
AMD RX 480 8GB with Blower type fan
SATA SSD ( temporary )

The System has been undergoing preliminary stability and thermal tests with all her processor cores and memory at continuous load for the last 24 hours . Hardware Monitor is used for the sensor readings .

Everything is cool and pretty quiet with the following MFC settings .

PSU fan = 1200 RPM Constant .
PCIe fan = 1200 RPM Constant .
Intake fan = 1200 RPM Constant .
Exhaust fan = 1200 RPM Constant .
CPU A Booster = 1700 RPM Constant .
CPU B Booster = 1700 RPM Constant .

This is a pretty high end configuration for a Nehalem cMP .

Final configuration will have a Highpoint PCIe miniSAS Rocket RAID Card 27xx ( one internal port and one external port ) . And four 6TB enterprise SATA mechanicals will be installed in reverse facing HDD Bay Sleds . A bootable Highpoint PCIe Flash RAID Drive Card will also be installed with nVME SSDs. The installation of these components later on will likely alter the system fan settings .

This Mac will also have installed later a TransIntl front chassis filter , which will probably require the fan rotationals to be raised by about another 5 to 10 percent from current .

It is not recommended to use constant speed for the fans. Set some fans based on the sensors (CPU, PSU, Tdiode) and leave the rest in Automatic.

 

[h9826790]

Agree, NEVER use constant fan speed with MFC.

MFC assign a target fan speed to the system, which override the system protection. The fan speed is really locked even something really overheating.

If anyone want to have constant fan speed, better to use SMCFanControl etc, which only alter the minimum fan speed, and leave the overheat protection still serviceable.

The reason why MFC can "suppress" the high fan speed is exactly because it override everything. In this case, that high fan speed is a SMC bug. We can simply clear it by stressing the GPU a bit. This is way better than assign a constant speed for the fans.

If want to use MFC, always make it base on the most critical hardware's temperature. Otherwise, you will have zero protection from hardware overheat.

 

[Snow Tiger]

It is not recommended to use constant speed for the fans. Set some fans based on the sensors (CPU, PSU, Tdiode) and leave the rest in Automatic.

I build systems for clients that need an assurance they won't get any louder if their Mac goes to load during a demanding session . Especially sensitive audio editing guys .

The way I deal with that need is to place the System at load ( all major components ) , check the thermals and adjust system fans with MFC constant speeds as needed to keep everything within spec . If the machine goes to idle , no biggie . It'll stay cool anyways .

Besides , the big problem with setting things to automatic is that the SMC won't always respond properly to upgraded components when they get too hot . Case in point is the X5690s with cMPs and W-3275M upgrades with the MP7,1s . To cool them properly , you need to set the rotationals to constant .

I'm not worried at all about the fans crapping out . These are enterprise grade fans . If they are going to fail , it'll be within the first week of their life .

 

[h9826790]

I build systems for clients that need an assurance they won't get any louder if their Mac goes to load during a demanding session . Especially sensitive audio editing guys .

The way I deal with that need is to place the System at load ( all major components ) , check the thermals and adjust system fans with MFC constant speeds as needed to keep everything within spec . If the machine goes to idle , no biggie . It'll stay cool anyways .

Besides , the big problem with setting things to automatic is that the SMC won't always respond properly to upgraded components when they get too hot . Case in point is the X5690s with cMPs and W-3275M upgrades with the MP7,1s . To cool them properly , you need to set the rotationals to constant .

I'm not worried at all about the fans crapping out . These are enterprise grade fans . If they are going to fail , it'll be within the first week of their life .

Your client has specific need. And I assume you give them warranty etc if the hardware overheat.

But OP isn't your client. IMO, it's not appropriate to suggest a setting for him that will disable the system protection, and can cause hardware overheat.

 

[Snow Tiger]

Agree, NEVER use constant fan speed with MFC.

MFC assign a target fan speed to the system, which override the system protection. The fan speed is really locked even something really overheating.

You are not placing the System at continuous load and checking the thermals of all the critical components like I am . I know exactly how hot everything can get before shutting down or have a durability or performance related issue . Hardware Monitor and other utilities give the sensor readings or I can manually install point sensors with thermocouples and read them with a multimeter .

Once I know how hot everything gets at System load for like a 24 hour period , with a certain system fan profile , I can decide to keep it or adjust it as necessary . If it's OK with factory SMC variable fans , then I'll leave it at that . But most of my rigs are heavily upgraded and require an utility like MFC set at constant RPM .

And if a component is seriously overheating despite receiving proper cooling , it needs to be rebuilt or replaced as it is failing . Factory SMC fans won't save it . Components like CPUs , GPUs have their own safeties and will shut down independently of System .

If anyone want to have constant fan speed, better to use SMCFanControl etc, which only alter the minimum fan speed, and leave the overheat protection still serviceable.

SMCFanControl is a dangerously unreliable program with modern Macs and macOSes . You should not use it . Use MFC instead . In the old days , SMCFanControl was indeed the preferred utility . But not anymore .

The reason why MFC can "suppress" the high fan speed is exactly because it override everything. In this case, that high fan speed is a SMC bug. We can simply clear it by stressing the GPU a bit. This is way better than assign a constant speed for the fans.

Taking a component from idle to load or back again in no ways alters MFC constant fan settings . They always stay constant and I've built several hundred Mac Pros with MFC to prove that statement . If it were otherwise , I would have noticed this by now after 15 years of building editing workstations . I have no idea why someone as experienced as you would make such a silly statement .

If want to use MFC, always make it base on the most critical hardware's temperature. Otherwise, you will have zero protection from hardware overheat.

The only way to deal with the danger of components critically overheating is to maintain the System properly . A wise user will occasionally read the various thermal sensors , maybe once a month , note any trends and take action if needed . To repair , replace or overhaul a System , when needed . Factory SMC variable fans will not help a dying component or System .

 

[h9826790]

You are not placing the System at continuous load and checking the thermals of all the critical components like I am . I know exactly how hot everything can get before shutting down or have a durability or performance related issue . Hardware Monitor and other utilities give the sensor readings or I can manually install point sensors with thermocouples and read them with a multimeter .

This is exactly why you shouldn't suggest the others to use your setting. They do not monitor the components like you do.

SMCFanControl is a dangerously unreliable program with modern Macs and macOSes . You should not use it . Use MFC instead . In the old days , SMCFanControl was indeed the preferred utility . But not anymore .

SMCFanControl is 100% safe, much safer than MFC. I have my own fan control software, I know each single SMC key's function, and how SMCFanControl or MFC work. In fact, I did help fan control software developers to make things better by telling them which key is missing in their software (for cMP).

SMCFanControl NEVER override the system protection. This is why it is 100% safe. MFC, on the other hand, can keep the fans run at too low speed. If you don't agree, please set your fan speed to minimum via MFC, and stress the hardware, and tell me if they can spin up properly.

IMO, this kind of "fail safe" protection is important when we give suggestion to others in this forum. We should not remove the protection without telling the others what the risk is.

Taking a component from idle to load or back again in no ways alters MFC constant fan settings . They always stay constant and I've built several hundred Mac Pros with MFC to prove that statement . If it were otherwise , I would have noticed this by now after 15 years of building editing workstations . I have no idea why someone as experienced as you would make such a silly statement .

I never say stress the component will alter MFC constant fan settings.

I said, MFC constant fan settings CANNOT be altered by anything including system overheat protection.

I said stress the component will clear the SMC bug, and bring the fans back to normal speed WITHOUT any fan control software. Which is a more preferable way to work around this "high fan speed bug".

You are a system builder. So what? I also know this fan control thing inside out. I know exactly which SMC key they are controlling, and how they control it, and what's the consequence.
[automerge]1581419046[/automerge]
If you want to suggest MFC to the others, tell them what the risk is, tell them that there will be no more system overheat protection, tell them that they have to monitor the hardware temperature by themselves. But not just tell them that's the good quiet fan speed, and use it.

 

[Snow Tiger]

This is exactly why you shouldn't suggest the others to use your setting. They do not monitor the components like you do.

They don't need to , as that's why they hired me . I build plug and play systems .

SMCFanControl is 100% safe, much safer than MFC. I have my own fan control software, I know each single SMC key's function, and how SMCFanControl or MFC work. In fact, I did help fan control software developers to make things better by telling them which key is missing in their software (for cMP).

SMCFanControl NEVER override the system protection. This is why it is 100% safe.

Wrong . In the last 5 years I have had at least three Macs built with SMCFanControl suddenly have their fans speed settings go offline and revert to dangerous factory defaults that cannot cool the System . The program is dangerous and should not be used by anyone , let alone ordinary users . That's why I use MFC with constant fan speed settings . Perfectly safe . I have never had MFC fail on me with many hundreds of builds .

MFC, on the other hand, can keep the fans run at too low speed. If you don't agree, please set your fan speed to minimum via MFC, and stress the hardware, and tell me if they can spin up properly.

IMO, this kind of "fail safe" protection is important when we give suggestion to others in this forum. We should not remove the protection without telling the others what the risk is.

When I use MFC , I set the fans settings only at constant . I don't trust the variable "auto" speed settings . Please tell me what scenario you are worried for the System fail safes to kick in ? Assuming that MFC over rides those fail safes ? MFC does not over ride PSU , CPU , GPU or controller chip built in OT protections . If those components get too hot , they will independently shut down regardless of what an utility instructs them . I have seen this with CPU , GPU and controller chips .

You are a system builder. So what? I also know this fan control thing inside out. I know exactly which SMC key they are controlling, and how they control it, and what's the consequence.

If you want to suggest MFC to the others, tell them what the risk is, tell them that there will be no more system overheat protection, tell them that they have to monitor the hardware temperature by themselves. But not just tell them that's the good quiet fan speed, and use it.

This is why my experience is more valuable than yours . I build a lot more Systems with a far greater number of hardware configurations and they all get tested very thoroughly before release to client . I go to lengths to break Systems in my shop first to address any issues , so the Systems when they are used in real life production environments have a greater chance of staying online as expected .

I have a return rate of around 3 or 4 Systems out of the last thousand if you want to talk about warranties . That's an excellent build quality track record . My Systems work at continuous load , they rarely fail in real life environments and they last much longer than average at load because I often rebuild the cooling subsystems with better than factory materials and better system fan settings .

There isn't any risk to using MFC once it has been set properly . The machines run for years properly once set at my shop . I work on the same units over a long period of time , so I'll get my own gear back at my shop for servicing . It's then I'll get the opportunity to observe them . In as far as fully informed users , I strongly encourage this and will tell them what utilities to use to observe system sensors . But most just want plug and play for production and won't even alter the configurations once they leave my shop .

 

[h9826790]

They don't need to , as that's why they hired me . I build plug and play systems .

Did OP hired you?

Wrong . In the last 5 years I have had at least three Macs built with SMCFanControl suddenly have their fans speed settings go offline and revert to dangerous factory defaults that cannot cool the System . The program is dangerous and should not be used by anyone , let alone ordinary users . That's why I used MFC with constant fan speed settings . Perfectly safe . I have never had MFC fail on me with many hundreds of builds .

Wrong for what? factory fan speed is danger? There is no prove of that, it's just your opinion. But MFC disable the system hardware overheat protection is fact. And this is the real danger.

This is why my experience is more valuable than yours . I build a lot more Systems with a far greater number of hardware configurations and they all get tested very thoroughly before release to client . I go to lengths to break Systems in my shop first to address any issues , so the Systems when they are used in real life production environments have a greater chance of staying online as expected .

I have a return rate of around 3 or 4 Systems out of the last thousand if you want to talk about warranties . That's an excellent build quality track record . My Systems work at continuous load , they rarely fail in real life environments and they last much longer than average at load because I often rebuild the cooling subsystems with better than factory materials and better system fan settings .

There isn't any risk to using MFC once it has been set properly . The machines run for years properly once set at my shop . I work on the same units over a long period of time , so I'll get my own gear back at my shop for servicing . It's then I'll get the opportunity to observe them . In as far as fully informed users , I strongly encourage this and will tell them what utilities to use to observe system sensors . But most just want plug and play for production and won't even alter the configurations once they leave my shop .

All I can see is just you provide a danger solution to others. Who is NOT your client. And you didn't tell him what the potential risk is.

 

[Snow Tiger]

Did OP hired you?

Nope . Nor did he hire you . So , I guess he should ignore your advice as well .

Wrong for what? factory fan speed is danger? There is no prove of that, it's just your opinion. But MFC disable the system hardware overheat protection is fact. And this is the real danger.

Not my opinion . It's my observation . There's a difference . It is you who is giving opinions . Where is your proof that MFC ever prevented a System from shutting down when a critical component has OT ? How many Systems have you worked with ? One , two , a dozen ? I've worked with hundreds with MFC .

QUOTE="h9826790, post: 28209194, member: 884762"]
All I can see is just you provide a danger solution to others. Who is NOT your client. And you didn't tell him what the potential risk is.
[/QUOTE]

B.S. My advice , under the limited amount of information that was made to me , is decent enough to help get his System stable for further observation . If required . It may actually get his system up and running properly . MFC is a safe program . If it weren't , there would be hundreds of ppl here saying don't use it .

 

[h9826790]

Not my opinion . It's my observation . There's a difference . It is you who is giving opinions . Where is your proof that MFC ever prevented a System from shutting down when a critical component has OT ? How many Systems have you worked with ? One , two , a dozen ? I've worked with hundreds with MFC .

MFC is not danger, it's your suggested setting disabled the system overheating protection.

I never say MFC disable shutdown protection. But why rely on shutdown protection if we can prevent overheat by allowing the system to spin up the fan? What logic is it?

Anyway, all I want is just to tell OP what the associated risk is. You can say whatever you want as long as don't break the forum's rule. You can share your point of view. I can share my point of view. OP will pick the solution he want. This is how forum works. I don't have to agree with you, you don't have to agree with me.

 

[Ih8reno]

-- How to fix high PCIe / PSU fan speed after GPU upgrade --
- https://forums.macrumors.com/thread...when-using-pc-non-efi-graphics-cards.1946672/
- https://forums.macrumors.com/threads/airout-a-solution-to-the-racing-fan-bug.2143951/

Thanks for this. I used the second solution which stresses the system and left the fans as they were. It seems to have worked. For the other posters I appreciate your suggestions as well!