Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
[G5]Hydra;18335329 said:
Very true but have you ever perused Apple OEM replacement parts lists? The prices will make you dizzy. Also if Apple goes the two different PSU route they will likely always have a couple of lower power cards and a higher power card, so there will always be options. The higher powered cards will likely cost you a couple grand minimum to get a hold of. At that point doesn't it just make sense to sell you MP for decent money and buy a new one instead of plunking another few grand into a machine that is already old? Also to keep in mind these cards and connectors are all proprietary as you mentioned before. What is to stop Apple form changing them all around every revision. PCIe cards had to be held to standards but Apple's own internal working are theirs to play with as they please. They make no assurances that any card they build in the future will even fit physically into an older machine.


Holy cow I said the exact same thing here 2 weeks ago.... Are we like... the same person?

----------

What happens when there isn't "extra thermal capacity " left?

Either they clock down (like what we've been saying here), that "18dBA under load, with unicorn hair" figure gets annihilated, or the dBA goes to Zero because it powers off. Perhaps the next Mac Pro wont include a power supply at all, MOST SILENT COMPUTER EVAR!!
 
In my opinion this thread is already comical. Too many people tend to claim too much without any basis in the facts. I'll wait until december/january until I'll start making any decisions.
 
This was 2nd conjecture, either heavy throttling or different PSU for different GPUs. If PSU needs to be upgraded to move up in GPUs then they have effectively WELDED the GPUs in. So much for that front page repost from French site claiming easy upgrades.
I think the most sensible possibility is that the PSU is upgraded when you select higher parts, or that they use a sufficient PSU for the highest option to begin with.

If they did the former, they wouldn't have to worry about people upgrading themselves, which I'm sure they don't want.

It's probably not easy to replace the PSU.
 
This is why I plan one waiting this first generation out. It is a major redesign and the only true test will be when these units are out and in the hands of users. Once you get a year of people using them for day to day work any design kinks and issues will be corrected on the next gen.

I expect issues to crop up.
 
OK, I'll drop shoe #1 again:

" It works by conducting heat away from the CPU and GPUs and distributing that heat uniformly across the core. That way, if one processor isn’t working as hard as the others, the extra thermal capacity can be shared efficiently among them."

What is shoe #2?

What happens when there isn't "extra thermal capacity " left?

Also, note that the "thermal core" has about twice the cooling for the CPU compared to each GPU....
 
How do you figure that?

apple-mac-pro-2013-cooling-system.jpg

The CPU has twice as many fins as either of the GPUs.
 
Surely there must be some apps that tax the GPU and the CPU to the fullest.

Oh there are, I use one every day. Bunkspeed Shot via Bootcamp. It can be set to utilize all available CPUs & GPUs simultaneously for rendering. I work on my scenes during the day and stack them in a queue. Then at bedtime I start the queue. It pegs all 24 CPU threads and GPUs at 100% for hours on end. It's a beautiful thing to see. :D


http://DG-Digital.com
 
Can't you just add another PSU and fan via thunderbolt?

It's the answer to any problem.
 
Apple has posted the terraflop performance for the D700 cards. Very doubtful they'd be lying and gimping the cards due to the power supply.
 
They'll offer another "external" device, the iPSU which comes in a surplus cMP case.
 
They'll offer another "external" device, the iPSU which comes in a surplus cMP case.

But it will be epoxied shut, so that you can't expand it.

----------

Apple has posted the terraflop performance for the D700 cards. Very doubtful they'd be lying and gimping the cards due to the power supply.

But, they could be running one card and doubling the numbers - even though you couldn't actually run both cards at the same time....
 
My theory is that the OS will throttle whichever components it has to intelligently. Very few situations use 100% of all resources, they do exist, but it's not a common usage scenario.

If you put a 1000w PSU in the Mac Pro I highly doubt it would work thermally. You're not going to dissipate 1000w of heat with something this small.

So basically Apple will balance everything so that 1.5 (out of 2) + 1.5 (out of 2) =3
 
My theory is that the OS will throttle whichever components it has to intelligently. Very few situations use 100% of all resources, they do exist, but it's not a common usage scenario.

If you put a 1000w PSU in the Mac Pro I highly doubt it would work thermally. You're not going to dissipate 1000w of heat with something this small.

So basically Apple will balance everything so that 1.5 (out of 2) + 1.5 (out of 2) =3

Is this part of them "blowing away limitation after limitation"?

Or something else?
 
My theory is that the OS will throttle whichever components it has to intelligently. Very few situations use 100% of all resources, they do exist, but it's not a common usage scenario.

If you put a 1000w PSU in the Mac Pro I highly doubt it would work thermally. You're not going to dissipate 1000w of heat with something this small.

So basically Apple will balance everything so that 1.5 (out of 2) + 1.5 (out of 2) =3
This is what I expect them to be doing; I mean, every Thunderbolt port can potentially provide 10W of power to attach devices, but the majority of Thunderbolt devices seem to rely on their own power supply (not much you can power from 10W that needs Thunderbolt speeds I suppose). In theory you could end up having to give 60W to Thunderbolt, but the vast majority are never going to need that; I mean, I'd be surprised to see setups with even a single bus-powered device to be honest.

My money's on the PSU being enough to handle both GPUs plus the CPU in one of its lower powered states. Plus other combinations, as 130W max for the CPU still leaves plenty of room for a single maxed out GPU or both running at 50% etc.
 
Apple has posted the terraflop performance for the D700 cards. Very doubtful they'd be lying and gimping the cards due to the power supply.

As pointed out earlier, If they're binning and running efficiently as the Reference 7990, they could easily be under the 450 mark even at full load (325W for GPUs)... you just can't run the CPU at much higher than idle at the same time.

Therefore for GPU benchmarks, they can get maximum performance (7TFLOPS, say), they just can't do anything real-world that requires the GPU and CPU at the same time.
 
Speaking of power requirements - this is what can happen if you let the horses

Run Wild:

http://www.legitreviews.com/nvidia-geforce-gtx-780-ti-video-card-review_128012/11

Over 300 watts for one unbridled GPU. But this kind leads the herd in CUDA performance. Against the three fan, clock tweaked, versions, even others of their own kind will savor the taste of dust [ http://videocardz.com/47831/custom-geforce-gtx-780-ti-gigabyte-inno3d-tested ]. The nMP's horses will have to be heavily bridled and Broadway choreographed - great for a musical, but comical for a true horse race. Truth in advertising would require caveats like the 30 second drug commercials where they spend less than 5 seconds telling what the drug is for and the remaining time telling you what the drug can adversely do to you. By the way, what is a "fatal event" and how many of those can a drug's user experience before things get serious? Is it like a computer crash where an application besieges the GPUs and CPU simultaneously with work orders when carrying out that job with the 5 minute away deadline?
 
Last edited:
...
If you put a 1000w PSU in the Mac Pro I highly doubt it would work thermally. You're not going to dissipate 1000w of heat with something this small.
...

Dissipating 1000W, using air with a 20*C temperature rise, through a 15cm diameter circle, could be done with an air velocity of 2.4m/s. I suspect that that's a bit too fast for this application.
 
Last edited:
Dissipating 1000W, using air with a 20*C temperature rise, through a 15cm diameter circle, could be done with an air velocity of 2.4m/s. I suspect that that's a bit too fast for this application.

Its be fun to make your own vertical wind tunnels out of the nMP for playing with paper helocopters though!
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.