Power issues on Mac Pro 4,1?

webhead93 · Sep 8, 2014

Hey Everyone,

I've recently been having issues with my system, and I've narrowed it down to being the graphics card, I currently have 2 graphics cards installed in my rig, an ATI Radeon HD 2600, and a NVIDIA GeForce GTX 570. The issues are now coming from the Nvidia card (Lockups/Freeze's with graphical Intensive games). I have the card setup with the required 2 6-pin cables. So I've been assuming its an issue with the card, but I was doing some tests with Hardware Monitor, and my results make me think that maybe the issue is with one of my PCIe boost lines? I've run these tests while stress testing my machine and 'PCIe Slot 2 12V Boost line' never changes in the amperage and wattage?. Is this an issue?

Thanks!
Steve

MacVidCards · Sep 8, 2014

the app is broken

run fur mark with just 570, you will know

ScottishCaptain · Sep 8, 2014

That app ("Hardware Monitor") is seriously screwed up.

I tried running it and got all kinds of irrelevant and incorrect information from the hardware sensors (which I know to be incorrect). Hell, it seemed to think I had a 5th disk drive and wouldn't even list the 12V booster rails.

I did try another app (iStat Menus) which seemed to work fine. It listed the booster rails at 12.15 and 12.16V (both A and B), drawing approximately 0.29 and 0.34A. This seems about on par with a 680 GTX idling at the desktop. If I enabled iTunes Visualizer, those values jumped up to 0.79 and 0.84A.

So you should be seeing something from those sensors (they certainly work fine over here). Whether or not Hardware Monitor is actually capable of displaying those values, I dunno- I certainly couldn't get it to work.

-SC

webhead93 · Sep 8, 2014

Okay, with Istat Menus I get a normal read, around 1.34A on A & B. So I'm guessing its my GTX 570 failing, seeing as these lock ups don't happen on my ATI. So I've purchased a GTX 760, and I'm unsure on my power setup for this, it says here: http://www.hwcompare.com/14830/geforce-gtx-570-vs-geforce-gtx-760/ that the GTX 570 consumes 50watts more than my new GTX 760, but I hear the GTX 760 needs one 6pin and one 8pin, would it be damaging to go ahead and just use 2 - 6pins like I've been using with my GTX 570? It seems like since it requires 50 less watts it would be fine....

ScottishCaptain · Sep 8, 2014

Pull the ATI. See if the problem persists. It might be an issue of running ATI and NVIDIA in the same system, which I have heard can be problematic in the past.

-SC

webhead93 · Sep 8, 2014

Alright I can try that, but It was working this way for around the last six months, now in the last month or so, it seems to be crashing more and more often every time I'm running a graphical intensive game. (Games that have worked for the past 6+ months.)

MacVidCards · Sep 8, 2014

webhead93 said:
Alright I can try that, but It was working this way for around the last six months, now in the last month or so, it seems to be crashing more and more often every time I'm running a graphical intensive game. (Games that have worked for the past 6+ months.)

have a look at crash logs. Whenever you get the "your computer wasn't shut down properly" dialog you have the option to have a look at them

webhead93 · Sep 8, 2014

MacVidCards said:
have a look at crash logs. Whenever you get the "your computer wasn't shut down properly" dialog you have the option to have a look at them

Alright, here's a crash log, sorry I can't decipher it, its all German to me.

ScottishCaptain · Sep 8, 2014

Kernel panic says the NVIDIA drivers are encountering some sort of "read error". That could mean that the drivers are messing up inside the kernel or the card itself is encountering some sort of hardware issue (potentially when reading from VRAM).

Honestly, it's hard to tell if it's the card or software. I would maybe suggest leaving things as they are hardware wise and installing OS X on a USB key so you have something to clean boot from there, then stress test the hell out of the card. If it's still crapping out and you've eliminated the software, then I'd probably suggest you replace the card.

-SC

webhead93 · Sep 8, 2014

ScottishCaptain said:
Kernel panic says the NVIDIA drivers are encountering some sort of "read error". That could mean that the drivers are messing up inside the kernel or the card itself is encountering some sort of hardware issue (potentially when reading from VRAM).

Honestly, it's hard to tell if it's the card or software. I would maybe suggest leaving things as they are hardware wise and installing OS X on a USB key so you have something to clean boot from there, then stress test the hell out of the card. If it's still crapping out and you've eliminated the software, then I'd probably suggest you replace the card.

-SC

Thanks for looking over the log, I've also tested my card running BF3 on windows bootcamp, and had a Lockup/freeze after about 30sec, this leads me to believe that its not drivers but the card itself?

ScottishCaptain · Sep 9, 2014

Probably, unless it's a PSU issue. You'd have to either replace your Mac Pro's PSU to test that, or chuck the card in another computer (probably a PC).

-SC

666sheep · Sep 9, 2014

Hardware Monitor in mine shows amperage and wattage on PCIe lines +/- accurately. Verified it with HWSensors and readings are about match.

OP, did you try to swap GPU power cables? I'm curious if it would change anything in readings... Worth a shot.

webhead93 · Sep 9, 2014

ScottishCaptain said:
Probably, unless it's a PSU issue. You'd have to either replace your Mac Pro's PSU to test that, or chuck the card in another computer (probably a PC).

-SC

Sounds like that would be my best option, but unfortunately I don't have another desktop/PSU to test with, the strange thing is, I never seem to get a lockup when doing a stress test with FurMark/LuxMark, but lockups have been happening more and more frequently with any graphic intensive game I play? I've also ran the AHT, didn't seem to find anything.

h9826790 · Sep 9, 2014

666sheep said:
Hardware Monitor in mine shows amperage and wattage on PCIe lines +/- accurately. Verified it with HWSensors and readings are about match.

OP, did you try to swap GPU power cables? I'm curious if it would change anything in readings... Worth a shot.

In his case, the reading of PCIe boost line 1 is actually the reading of the PCIe slot 1. And the slot 1 and 2's reading should belongs to the aux cable 1 and 2. That's why he was confused.

It seems that he has the 570 in slot 1, the 2600 in slot 3, and no other card in slot 2 and 4. HW Monitor able to get all the reading correctly indeed, but the naming may be wrong.

No matter is there any loading or not, the 12V supply is always there. That is correct as well. If the current is zero, the overall power consumption is still zero. It means, there is no loading in slot 2, which HW Monitor call it the boost line 2.

webhead93 · Sep 9, 2014

Alright, I just had another crash (Running Minecraft for about 20sec), and I think this one might provide a little more information.

webhead93 · Sep 9, 2014

Anyone out there wanna take a look at my crash log and see if they can decipher anything? Would be much appreciated!

ScottishCaptain · Sep 9, 2014

One of your CPUs appears to be defective.

Does that make anything more clear?

Your machine is very sick. If you do not have the hardware to test things, then you need to take it into someone who does. Posting more kernel panics here isn't going to solve anything. Something is very, very wrong with your hardware and it is impossible to diagnose anything like this over the internet.

-SC

webhead93 · Sep 9, 2014

That can't be it, I have a 2 monitor setup, one connected to my ATI, and one to my GTX 570, running the same game on both of them, at identical graphics, my ATI will play fine, and my GTX will Freeze/Lockup after 30sec. How could this possibly be a CPU issue?

ScottishCaptain · Sep 9, 2014

Look, you wanted to know what the kernel panic said. The kernel panic says that a machine check exception (a hardware error) has occurred, and the only kernel extension in the backtrace has nothing to do with your graphics cards (AppleIntelCPUPowerManagement deals with your CPUs, not the graphics cards).

I really don't know what else to tell you. Your machine is very sick. The only way you're going to get to the bottom of this is to start troubleshooting the hardware. Posting more kernel panics here isn't going to solve a single thing.

Pull the 570 and see if your machine becomes stable. If it doesn't, then you might have more then one problem (in the RAM or CPU). If it does, then it's probably the card. You're going to have to start eliminating components sooner or later, so you might as well get to it and start with the GPU.

-SC

webhead93 · Sep 10, 2014

Alright just to update everyone, I purchased a GTX 760 and installed it, and voila! no more lockups! thanks everyone for your help.

Power issues on Mac Pro 4,1?

macrumors newbie

Suspended

macrumors 6502a

macrumors newbie

macrumors 6502a

macrumors newbie

Suspended

macrumors newbie

Attachments

macrumors 6502a

macrumors newbie

macrumors 6502a

macrumors 68040

macrumors newbie

macrumors P6

macrumors newbie

Attachments

macrumors newbie

macrumors 6502a

macrumors newbie

macrumors 6502a

macrumors newbie

Our Staff