In my opinion, the much-hyped "thermal core" design is inadequate for the power it must dissipate. Not that this is necessarily the problem, but when was the last time you saw a premium fan/heat-sink assembly that did not have heat pipes and a heck of a lot more fin surface area? And the "thermal core" has to support the CPU plus two GPUs, not just one chip.
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.
MP 7,1 Waiting for Mac Pro 7,1 A1991 (no more)
- Thread starter Mago
- Start date
-
- Tags
- macpro 2016
- Sort by reaction score
AidenShaw
macrumors P6
Sorry to bring up a tangent - but I haven't figured out how to set the forum to show a threaded view.
I only have a viewport in linear time (for DS9 fans), with no indication as to the "replied to" structure. (And as a result, I'm sure that many of my replies have gone to the wrong ancestor.)
Where's the knob for a "thread view"?
I am not aware if there is or is not a way to view posts by "thread view" as I've never really had the need for one. Maybe it would be a good feature and help prevent the mistake several people made over the weekend where they attributed someone else's comment to me.
AidenShaw
macrumors P6
But, without a thread view - how do you even know to what message they are replying?
The only chance I see for context is quoting the ancestor (like I'm doing here).
I can't provide you an answer. In the post which cause the confusion (that would be post #1527 which I did write) I thought I was quite clear where the quoted material was coming from. Right above the quote I clearly wrote: From post #1522. Here is the relevant part of that message:
From post #1522:
The way I am reading this is:
As I've seen it described by people I have good reason to believe know very specifically what the problem is, the addition of the extra 3GB of vram to the Radeon design has given rise to a localised thermal problem that the system cannot cool, or protect against.
The way I am reading this is:
The text indented and italicized above is the exact way the text appears in post #1527. The part I quoted was even indented from what I wrote. The words following the quote, "The way I am reading this is" should also have been an indicator that the material which appeared above it was someone else's words. There is no reason for me to say "The way I am reading" something wouldn't be applicable to something I wrote.
I thought I was being about as clear as I could be that the indented material (of post #1527) was quoted from someone else out of post #1522.
AidenShaw
macrumors P6
Again, you're trying to get the context from the text.
Many (maybe most) message boards give you the option to view posts in a tree-shaped threaded context. You can traverse the new messages in linear time, but as you move through the posts the left "nav pane" for each message gives you a tree-structured view of the post in the context of the posts above and below it.
The context is obvious even if the post doesn't quote it's ancestor.
I'm trying to find how to turn on that view in MR.
Only thing fair to said here is your don't like the tcMP.
FACTS: the Airflow inside the Thermal Core is bigger than the airflow you can have from std cpu/gpu coolers (case fans dont account since only drive air inside-outside the TC is more like an open-rig).
tcMP issues affects a limited number of users, some users on extreme climate or environment plus heavy loads (as a friend from Portugal which used to do heavy renders all weekend on 40 deg Celsius or about 100deg F, c'mon there even a mac mini playing videos melts).
Mine works 10h-14h every day, even I use to carry it to my home (not easy task on a cMP), even some time I run a heavy process for a week using full CPU (no GPU, but also I use to run few cpu heavy apps specially when I run some theoretical experiments), an nowadays its is the most reliable Mac I've ever own, previously I ve kernel panics, issues with PCIe slots, a spinner HDD failure and lots of jammed DVDs, that's history I love absolutely the tcMP is über powerful and insanely compact even portable.
But i concede you something, Apple should uprate the tcMP TDP to allow more thermal flexibility and less throttling, from 450W to 600W is perfectly possible w/o complications, just a new speedy fan and new fin arrangements inside the thermal core, even a heat-pipe based thermal core could handle 1200W easy w/o size increase (something i demonstrated on another thread).
[doublepost=1473211297][/doublepost]
here are two engineers with large experience on thermodynamics, one of them it's me, I can say you're wrong, the TC its efficient and effective, the problem is on the fan which is de-rated and the PSU overloaded, other issues as cracking solders and thermal paste aren't TC issues the same problem would have been surfaced on a classic design.
PD Thermal pipes aren't used on Clusters, just simple aluminum fins, and those behemoths run full 24x7xMonths solving a single challenge. Look at SuperMicro and HPE's Moonshoot, only exception is when they decide to use liquid cooling on some rigs to provide heating to some near buildings but this is unusual since increases the cost and taxes reliability.
Last edited:
Well, that may all be true and while your personal machine may run fine it can't be denied that there is a reoccurring problem with the nMP dying from thermal issues, bad solder or whatever. That's also a fact.
I've had two of them die on me at one facility and seen other dead ones around town. I have friends who also work in post production who have had the same experience. I travel and work outside of my home base to facilities in other cities and see the same problem. The only machine I have seen with more failures was the SGI 320 and that thing truly was a piece of junk (and helped sink SGI)
So, in theory and on paper the design may appear solid, but that has not proven to be the case in practice.
Like I said, it's time to put this one on the shelf, learn some lessons and move on.
Last edited:
developer13245
macrumors 6502a
The advertised nMP operating temperature range is 50-95 F (10-35 C), so your Portuguese friend is running the machine outside supported parameters. But you claim to be the thermodynamic expert... So pack a fire extinguisher if you visit... A real friend with experience in thermodynamics would call and tell them to turn up the air conditioning.
BTW, 40C = 104F (Not 100F as you claim), but again, you claim to be the thermodynamic expert, hmmmm... Wait, maybe you shouldn't call your friend.
And, claiming the Thermal core is a success while claiming the need for a "speedy" fan is just doing "a dance" because increasing airflow will "blow" the acoustical (noise) performance. So let's just rephrase the criticism: The cooling system just plain sucks, because you cannot ignore the the overall system performance claims.
In order to keep the same acoustical performance (whisper as you concede "non-speedy" fan) with higher cooling capacity, the overall size of the core would have to increase, and so would the case dimensions . But you being the thermodynamic expert obviously know this.
And of course, retooling the USA MP factory for producing new tube lengths and widths for every mod is simple. There are two knobs in the break room next to the microwave that are currently set to 9.9, and 6.6 for height and width - just that simple. So.. at 993 days and counting since the last MP update, just WTF is taking them so long? Maybe you should take your thermal expertise and mosey over to the factory to change those knobs....
Last edited:
As thermodynamics expert I noted (take your time to read) the thermal core to increase to 600W TDP only needs internal fin re-arrange (to increase the laminar contact) with this and a faster fan (not noisier there are fans with higher CFM on the same noise dB).
PD when I scaled 40degC to Fahrenheit I did checking an classic mercury thermometer where both scales are very close, of course this is an simple conversation not a mathematical analysis on why the tcMP fails on certain conditions.
And the mod I propose don't require re-tooling it's the same core with an different fin pattern and the fan just an uprated replacement.
Who's fault is that? A nMP can only be configured with a limited number of configurations all of which come from Apple. If the system is operating outside of its design specification then Apple is to blame.
The fact is isn't unusual operating conditions like the ones on the Portuguese Mac, most workstation and servers expand to 50degC or about 110F their thermal envelope, however seems Apple thinks everybody lives at California"s north, not everybody loves to work at 70degF having 90 outside and at 90deg it's easy for an system to scale to 100, sorry 104...
This is a general fault on Apple designs (from the tcMP to the iMac and the MacBooks everybody living at warm places (not so far as Miami) has something to tell about Apple heating.
AidenShaw
macrumors P6
HPE ProLiant DL380 nodes use heat pipes for any CPU that 120 watts or more - and that's a simple, popular 2U two socket server.
Not a cluster server, look at the Moonshot you find either copper or aluminum coolers
AidenShaw
macrumors P6
There are lots of ProLiant clusters, did you mean "not a blade server"?
And, by the way, the Moonshot server cartridge has a 45 watt quad core Xeon E3-1585L v5 CPU - not really in the class of high end workstation CPUs.
Ok enough MP thermal core debate, today is keynote day, so a small recap about today speculation;
A boring iPhone 7 it's über confirmed, how boring it's the surprise:
DarkNetGuy talked about the iPhone 7 plus to support the Apple Pen ala Galaxy Note 7, this is the most crazy speculation about from him or her or it.
My personal speculation is the iPhone 7 plus will have some litro-like photo capabilities being you capable to re focus before a shot is recorded.
Another speculation is that Apple could introduce the Macbook Pro by assault at keynote end as the ideal iPhone sidekick.
Also thee is last minute information about updated timecapsule and airport devices being introduced off-keynote today.
A boring iPhone 7 it's über confirmed, how boring it's the surprise:
DarkNetGuy talked about the iPhone 7 plus to support the Apple Pen ala Galaxy Note 7, this is the most crazy speculation about from him or her or it.
My personal speculation is the iPhone 7 plus will have some litro-like photo capabilities being you capable to re focus before a shot is recorded.
Another speculation is that Apple could introduce the Macbook Pro by assault at keynote end as the ideal iPhone sidekick.
Also thee is last minute information about updated timecapsule and airport devices being introduced off-keynote today.
Put 64 of them together and you get some serious compute power.
There are plenty 1U and 2U Xeon coolers rated for 160W w/o thermal pipes.
The point is that a thermal pipe isn't a must to cool down a high performance system, actually most the time it's just a marketing tool than an actual need.
I do justify thermal pipes when you need to move the heat far away the contact area as on tower like coolers or on large passive coolers as those on home theater cases.
Also there is the issue with pure copper coolers as the copper cost sky rocket, most integrators are using pipes to enable aluminum fins instead pure copper fins (better but more expensive)
AidenShaw
macrumors P6
If your application is OK with quad core CPUs with a max of 64 GiB RAM.
If you'd rather have 44 cores and up to 3 TiB of RAM per system, then go with 1U ProLiants. And, of course, you could put 64 of those together! 😉
You only need a single Moonshot chassis to load 45 cartridges each with upto 4 cpu, this is in 4.3U if you wanna build a Peta flop cluster with moonshot you may need only 1 or 2 racks with DL300 you need a full floor...
Man, there are really some nasty people here.
Sometimes one goes, but soon after another one pops in. Is this a plague?
On a serious note, it's not that uncommon to have here 40+ ºC, just this week it happened. Mago, does your friend live in the south? It's where it gets worse.
Great weather for some beach though 🙂 Go there whenever I can, a couple of minutes away from home 🙂
Great to know that the nMP can handle the "heat".
We surely can't ignore the problems reported, but if it was a poor design, in these more extreme conditions it would show even harder, right?
It's something else.
[doublepost=1473259299][/doublepost]The rMBP refresh might come only later, I hope so at least. Kaby Lake would be awesome in there, unless Apple could secure CPUs from Intel to release it this soon.
Maybe the closing thoughts will give us a hint at the new nMP, with the support ending and all.
Sometimes one goes, but soon after another one pops in. Is this a plague?
On a serious note, it's not that uncommon to have here 40+ ºC, just this week it happened. Mago, does your friend live in the south? It's where it gets worse.
Great weather for some beach though 🙂 Go there whenever I can, a couple of minutes away from home 🙂
Great to know that the nMP can handle the "heat".
We surely can't ignore the problems reported, but if it was a poor design, in these more extreme conditions it would show even harder, right?
It's something else.
[doublepost=1473259299][/doublepost]The rMBP refresh might come only later, I hope so at least. Kaby Lake would be awesome in there, unless Apple could secure CPUs from Intel to release it this soon.
Maybe the closing thoughts will give us a hint at the new nMP, with the support ending and all.
It's indeed very tiresome to have to read a great amount of posts containing non stoping attacks, most of the time without a serious reason, while trying to find a useful post. So, yes, it's a plague, a lot of patience needed.
About the thermals now, I can't be sure if this is only an issue with the D700 GPUs, but our D500 are running just fine at 30º to 40º C temperatures, when the AC happens to be off, at least till now.
Last edited: