Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

glawrie

macrumors member
Original poster
Oct 31, 2005
56
5
Maidenhead, UK
I have a mid 2010 5,1 2x2.4 Quad Xeon, 12 GB with one spare harddisk bay and one spare optical bay.

I've got a flashed MSI R9 280X Gaming 3G, currently powered via PCI slot plus the two six-pin PCI boost cables, one with a 6-8 converter on it.

In operation it appears that one of the two PCI boost connectors provides about half of the power draw required - when running FurMark in a window at 1024x680 with no anti-aliasing the card is drawing about 2.5amps from PCI slot 1, 3amps from PCI Boost B, and about 6amps from PCI Boost A.

However when subject to high graphics processing loads (e.g. FurMark at significantly higher resolutions and / or anti-aliasing) the card draws too much current / overloads something and trips the Mac Pro.

One possible solution would be to combine the current two PCI Boost cables into a single connection (using a 2-1 cable) and use this to feed the high-current socket, and to get a supply to feed the other socket from elsewhere in the 5,1.

So my question is would the power connections to either the spare HardDisk or Optical bay be strong enough to provide the second PCI boost power requirement? I'm the lower-current power PCI boost input to the card will need up to about 5amps.

Does anyone know whether either connection can provide this?
 
Last edited:
So my question is would the power connections to either the spare HardDisk or Optical bay be strong enough to provide the second PCI boost power requirement? I'm the lower-current power PCI boost input to the card will need up to about 5amps.
According to this link, the max current for each sata power is 4.5 amps, so adding two more sata power output should provide enough juice as mentioned in this post. But I have to question your obsession with FurMark, which isn't a normal application and some people would call it a power virus. If you do a search on that topic, you'll find out how people think about it.
 
Thanks for the info and links.

I quite agree about the futility of benchmark apps. However I also am a bit wary of configuring my Mac Pro in a way that I know is liable to fail if overloaded: I would be happier knowing that the machine is set up such that it is able to operate safely without a risk of crashing induced simply by asking it to do a 'too difficult' task whatever that might be. So in this case the benchmark is simply a way of validating the configuration of the Mac Pro - I'm not really interested in the benchmark results.

Hope that makes some sense.
 
Is it typical to see the two PCI boosts drawing such disparate power? Does the system wait to tap the 2nd PCI boost until after the PCI bus and first PCI boost have been saturated, or does the system draw from all 3 sources at once all the time and at lopsided rates? (Keeping in mind the PCIE bus is 150 and the boosters are only 75 each of course.)

I've never watched the power draws on my 2008 because my 6870 is a pretty efficient GPU - never had to worry about it.
 
Some time ago I did a bunch of testing that basically showed that an approved GTX285 would and could pull more than 100 watts from one of its connectors. This was same power port that was 8 pins on a GTX280.

Very few people have ever killed logic board by drawing too much power, and those who have typically had something plugged into every possible power plug.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.