Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Yeah, you're right about the W5000, didn't have the low-end options below Apples FirePros in mind. The W7000 (Pitcairn XT like D300) does have ECC though.

I still don't see how this makes Apples GPUs (or their MP product cycle in general) less ridiculous though, nor why you justify Apple ripping of their customers.
 
  • Like
Reactions: tuxon86
Yeah, you're right about the W5000, didn't have the low-end options below Apples FirePros in mind. The W7000 (Pitcairn XT like D300) does have ECC though.

I still don't see how this makes Apples GPUs (or their MP product cycle in general) less ridiculous though, nor why you justify Apple ripping of their customers.

AMD's own literature has it listed as W7000 does not support ECC memory for some reason. Unless perhaps they added it later on.

The point I'm making is even some of the PC Firepro's don't use ECC so calling Apples Firepro's fake is not in line with what the PC versions are showing. I believe AMD did not start using ECC until its W series cards.
 
Last edited:
I don't discount that some do indeed burn up. But I'm sure there a lot of reasons for failures other than the GPU's are overheating. Than out of the total nMP users have problems with this actually happening, we don't know and probably never will.

A few user experiences doesn't prove the whole GPU burning up theory. When you start to look into these individual stories in detail they already start to fall apart. Conjecture, no data on temps and for how long. No repair, replacement information. You come into MacRumors with a few unproven stories and soon everyone is passing it around as fact. I'm glad for one this is MacRumors and not MacFacts and we would all have wild stories to tell.

We have three nMP at work, purchased in the past 24 months. They are used with GPU intensive applications and are run all day, 5-6 days a week. All three have had the GPU fail (12c/D700) and were sent in for service. One of them has failed twice.

I have seen similar failure rates for the nMP at other facilities I work at around town.

nMP GPU failure is not an uncommon subject on entertainment industry forums, although less so lately simply because people stopped buying them and went with HP boxes.

This is a very real problem in the creative field. When you see a failure rate that high in the field you don't have to be a rocket scientist to conclude that the product is a lemon.

I have never seen a cMP die of thermal failure. Upgraded cMP are a common sight around facilities in my business and their failure rate is extremely low. Come to think of it I have never seen a dead cMP anywhere in the past 6 years.
 
Last edited:
We have three nMP at work, purchased in the past 24 months. They are used with GPU intensive applications and are run all day, 5-6 days a week. All three have had the GPU fail (12c/D700) and were sent in for service. One of them has failed twice.

I have seen similar failure rates for the nMP at other facilities I work at around town.

nMP GPU failure is not an uncommon subject on entertainment industry forums, although less so lately simply because people stopped buying them and went with HP boxes.

This is a very real problem in the creative field. When you see a failure rate that high in the field you don't have to be a rocket scientist to conclude that the product is a lemon.

I have never seen a cMP die of thermal failure. Upgraded cMP are a common sight around facilities in my business and their failure rate is extremely low. Come to think of it I have never seen a dead cMP anywhere in the past 6 years.

Ok, so how was it determined the GPU's failed other than being used in an intensive GPU applications? What were the GPU temps and did it exceed the TDP of the associated GPU? The GPU's in the nMP do throttle when necessary, so did you see that? Did the technician mention the cause of the failure? Just curious.
 
Ok, so how was it determined the GPU's failed other than being used in an intensive GPU applications? What were the GPU temps and did it exceed the TDP of the associated GPU? The GPU's in the nMP do throttle when necessary, so did you see that? Did the technician mention the cause of the failure? Just curious.

All machines were sent to Apple for service and had the GPU replaced.
 
Ok, so how was it determined the GPU's failed other than being used in an intensive GPU applications? What were the GPU temps and did it exceed the TDP of the associated GPU? The GPU's in the nMP do throttle when necessary, so did you see that? Did the technician mention the cause of the failure? Just curious.
As several of Apple's products have had overheating issues, most likely due to space constraints, it seems reasonable the nMP would not be immune. There is a lot packed into the package and I can see heat related problems cropping up just as they have in their other products.
 
As several of Apple's products have had overheating issues, most likely due to space constraints, it seems reasonable the nMP would not be immune. There is a lot packed into the package and I can see heat related problems cropping up just as they have in their other products.

Our nMP GPU failures all occurred after about 12-24 months of heavy use. Apparently people running DaVinci Resolve were seeing GPU failure in a shorter time. If the box did not fail outright, you first saw corrupted frames being generated during renders, followed by total failure.These machines were run in well cooled bays and offices, so it's not like they were run under extreme conditions.
 
Last edited:
Our nMP GPU failures all occurred after about 12-24 moths of heavy use. Apparently people running DaVinci Resolve were seeing GPU failure in a shorter time. If the box did not fail outright, you first saw corrupted frames being generated during renders, followed by total failure.These machines were run in well cooled bays and offices, so it's not like they were run under extreme conditions.
I've read about the issue and it seems reasonable to me Apple, once again, took "thin and small" too far with the nMP.
 
I have never seen a cMP die of thermal failure. Upgraded cMP are a common sight around facilities in my business and their failure rate is extremely low. Come to think of it I have never seen a dead cMP anywhere in the past 6 years.
Really? My 2008 MP has killed 3 GPU's due to thermal issues. The GT 8800 it came with died after 6 months, due to gaming only.
 
Yes, I'm going to put my foot in it again.

I've been using computers for a long time. I learned to program on a refrigerator sized minicomputer that had to have the boot sequence keyed in from the front panel. I've lived through a huge array of different storage methods. The aforementioned minicomputer used terminals that had paper tape punches for storing user programs on. 8 inch floppies, disk packs, 80 column punch cards, 9 track tape, 5 1/4 inch floppies, Bernoulli drives, mini tape drives, plus all the more modern stuff. I've programmed on several different types of mainframes and minis, plus Atari 800, Apple II, original IBM PC, and so on. I've stuffed 100's of different types of cards in slots, from OEM stuff to custom boards to support bespoke lab equipment.

I say all that so that when I say that those who are holding on to the internal drive bay or internal slots are like those old car buffs who lament that everything is computerized these days, I know what I'm talking about because I have the same experience with computers that those old car buffs have with cars. I don't lament those old days though, because I really like the fact that the mechanic no longer has to spend several days tinkering with my car to figure out what is wrong.

I can't count the number of times we ran out of bays or slots trying to support some university project. The prospect of moving that stuff outside of the box to where you're no longer constrained by space or arbitrary numbers of slots is awesome! Yes, there are teething pains, and yes, stuff is expensive right now. Same thing happened with damn near everything else along that path I laid out.

Drive storage is expensive because of Thunderbolt. It won't always be that way, and the cost right now is pretty much irrelevant to most professional setups, which will eventually drive prices down for everybody else. I love the fact that when software/OS needs to be upgraded, IT can just upgrade a generic "can", and then just unhook the old can and drop the new can in, everything the developer needs that is for that individual developer is on external storage. The savings in developer time from not having to twiddle thumbs while stuff updates pays for the external stuff. Something goes down, IT just swaps it out, without time consuming pulling the computer out to test individual components.

The situation with external "slots" is not good right now. I suspect that Apple skated to where they expected the puck to be, the pass got intercepted, and the external capability they expected to be there hasn't emerged yet. But it won't always be that way. Can't drop 4 GPU's in a bunch of slots? How about being able to drop way more than you could ever drop into slots into an external enclosure?

External hookups don't have to be ugly or sprawling. If you took the case away from a cMP, and just had all the components sitting there hooked up to each other (I have built computers that way for testing, it ain't pretty), it would be the same. So come up with a "case" around all your components that fits what you need, instead of having a one size fits all cheese grater.

As for those who think the MP is going to get EOL'd, what do you think Apple is going to build its own operating systems on if they do that? Any developer who does not at least compile release versions on a machine with ECC memory is playing Russian Roulette. Sure you can do some work on a Macbook/Mac Mini/iMac, but unless you're building toy apps, you've got to grunt it out on a MP for reliability. Unless Apple starts supporting building code on Windows/Linux, there has to be a MacPro.

My one beef with the nMP is that second GPU. It's completely useless to a developer unless they're writing apps specifically for that second GPU. It just costs money and burns electricity. For me personally, the option for a second CPU would be nice (full test suite would finish before I come back from lunch), but my work doesn't need the extra memory that a second CPU can bring to the table. Ironically, some of my hobby work (my personal machine is a 2010 5,1) could sure use some extra memory, but I do have the option of buying a dual CPU tray for it and pushing more memory in it. I'd been hoping to be able to buy a dual CPU 7,1 instead of expensive parts for a computer that has no warranty coverage and will run up against EOL eventually. Maybe they'll release one this fall?

I love my 5,1. It's a great hobby/enthusiast machine. But I don't think that Apple should be holding back from moving forward with better options for professionals just to suit hobbyists/enthusiasts. No more than car makers should be making cars without OBD to suit the old car buffs.

You may now commence the flaming and telling me I don't know what I'm talking about.
 
We have three nMP at work, purchased in the past 24 months. They are used with GPU intensive applications and are run all day, 5-6 days a week. All three have had the GPU fail (12c/D700) and were sent in for service. One of them has failed twice.

I have seen similar failure rates for the nMP at other facilities I work at around town.

nMP GPU failure is not an uncommon subject on entertainment industry forums, although less so lately simply because people stopped buying them and went with HP boxes.

This is a very real problem in the creative field. When you see a failure rate that high in the field you don't have to be a rocket scientist to conclude that the product is a lemon.

I have never seen a cMP die of thermal failure. Upgraded cMP are a common sight around facilities in my business and their failure rate is extremely low. Come to think of it I have never seen a dead cMP anywhere in the past 6 years.

You don't know that those are thermal failures, you just know the GPU failed. I don't do any intensive 3D on my nMP and I had to have the GPUs replaced twice. Yes, there are GPU defects on the nMP. No, I don't think it's conclusively proven they're thermal related.
 
  • Like
Reactions: linuxcooldude
I can't count the number of times we ran out of bays or slots trying to support some university project. The prospect of moving that stuff outside of the box to where you're no longer constrained by space or arbitrary numbers of slots is awesome

(just expanding on that a bit)

cmp hard drive design accommodates only a certain group of people.. those needing exactly 2, 3, or 4 drives..
of which, at least one drive needs to be backup, preferably two..(even though it's probably not too wise to place backups inside the computer meaning to be backed up ;) ) and one generally a scratch..
..which leaves two drives to do what, exactly, with? massive storage? large video projects? probably not.

idk, the arguments for internal storage are just weak.. for a mid-to-high data power user, it's not enough.. (even for someone like me whose files are generally sub-250MB, it's not enough).. for someone using cloud and wireless, it's wasted space..

just sort of makes me laugh when someone is like "i'm a serious computerer so i NeeD multiple internal drives !! :mad:" .. lol, no you're not..
 
If 4 internal bays are not enough, how's the removal of this option helps at anything ? What is the real alternative ?

And, btw, Computing 101: backups are not supposed to be taken in internal storage. It beats the purpose of backup altogether.
 
I love my 5,1. It's a great hobby/enthusiast machine. But I don't think that Apple should be holding back from moving forward with better options for professionals just to suit hobbyists/enthusiasts. No more than car makers should be making cars without OBD to suit the old car buffs.

For some of us, it's a lot more than that.

Personally, I don't do coding or video, but a cMP is what I need professionally to create design and illustration work for my clients. I frequently have around 8 applications running simultaneously (Photoshop, Illustrator, Painter, Lightroom, InDesign, etc.) so I need lots of RAM. I need proper connections for my three monitors (2x Asus 23" + 1 Cintiq 22HD), and one 2GB Nvidia 680 provides all the right connections, no need for a second useless gpu.

The iMac is not an option for several reasons, not the least of which is being stuck with a monitor that I neither want or need. And I prefer the hardiness/longevity/stability of Xeons, even if they aren't necessarily as fast as an i7 for certain tasks, they've never let me down.

What it comes down to for me is choice and long-term viability. I want more choices in my configuration than Apple provides in the nMP, and I want to be able to add/change elements down the road to keep it at peak performance.

I'm not a fan of the nMP and I don't think mine is a particularly Luddite attitude. I think Apple is trying to push us in a direction (more non-upgradeable, disposable computers) that is better for Apple because it decreases our options and encourages full system purchases down the road rather then incremental upgrade purchases. I'm not a fan of that business model, no matter how forward-thinking it may be. So in one sense I suppose I may be a Luddite in that I prefer the old way of having a choice in many of the elements in my system, and a reasonable path for upgrades.

I look at the system options for the nMP and compare them to what is offered by HP or Dell, and I'm blown away by the enormous amount of choices they offer. Sure, Apple simplifies the process a lot, but some of us Luddites enjoy wading into the minutia of processor speeds and gpus and multiple internal PCIe slots. EDIT TO ADD: Also, HP and Dell offer the latest processors and GPUs, not three-year old versions - and they cost less.

As it is, I've got a good system (a 4,1 upgraded to a 5,1 with Dual 6-Core 3.46GHZ and 64GB RAM) and I'm sticking with it for the foreseeable future. The hip kids can go ahead with their TB enclosures and external GPU's, but Old Man River is doing just fine with his wind-up vacuum-tube cheese grater.
 
Last edited:
If 4 internal bays are not enough, how's the removal of this option helps at anything ? What is the real alternative ?

probably something more customizable.. currently, everybody needs at least one drive.. after that, there is no standard.. so have it set up where the user can optimize the storage arrangements to their needs and workflow..
4 drives? that's just an arbitrary number.. odds are, for the vast majority of users, 4 drives is either too little or too much.. for people using high data workflows (video etc), it's certainly too little.
4 drives accommodates low to mid range users.. not 'power' users..

further, it's tying all the data to that one single computer and making it a bit more difficult to access it unless you're at this exact computer.




And, btw, Computing 101: backups are not supposed to be taken in internal storage. It beats the purpose of backup altogether.
i know.
ie- even if your computer has 4 drive bays, you still have to use external drives.. that was what i meant to get across in the earlier post.




[doublepost=1467779750][/doublepost]
[edit]
that said, nmp is designed to house 2 drives..
so 2 @ 2TB+ each should satisfy a significant amount of people desiring multiple internal spinners i would imagine.
 
Last edited:
If 4 internal bays are not enough, how's the removal of this option helps at anything ? What is the real alternative ?

What we used to do was massive SCSI deployment outside of a case. Well, a standard case. Built a little enclosure for the motherboard and all its doodads for protection, and a kind of a semi-enclosure for the drives (cuz we kept having to add more). OMG, WE INVENTED THE nMP AND A THUNDERBOLT ENCLOSURE! The past becomes the future!
 
probably something more customizable.. currently, everybody needs at least one drive.. after that, there is no standard.. so have it set up where the user can optimize the storage arrangements to their needs and workflow..
4 drives? that's just an arbitrary number.. odds are, for the vast majority of users, 4 drives is either too little or too much.. for people using high data workflows (video etc), it's certainly too little.
4 drives accommodates low to mid range users.. not 'power' users..

further, it's tying all the data to that one single computer and making it a bit more difficult to access it unless you're at this exact computer.





i know.
ie- even if your computer has 4 drive bays, you still have to use external drives.. that was what i meant to get across in the earlier post.




[doublepost=1467779750][/doublepost]
[edit]
that said, nmp is designed to house 2 drives..
so 2 @ 2TB+ each should satisfy a significant amount of people desiring multiple internal spinners i would imagine.
Having internal storage does not preclude the use of external storage. The opposite is not true necessarily.
Also, where did you get that the nMP is designed for two drives, are those two sockets identical?
 
IIRC only the one of the nMP's GPUs was able to accept the ssd. I could be wrong, though. It's been a while.

Yes, my point was exactly what H2SO4 said: The option to use internal storage does not prevent the user to also connect external ones. Just like the lack of need of ports (that a few people claiming on rMB forums) for laptops, doesn't mean that actually having ports is a bad thing.

It's all about having the option versus not having it. Apple seriously steers towards the latter even on products that - imho - they shouldn't just yet.
 
Last edited:
Also, where did you get that the nMP is designed for two drives, are those two sockets identical?
it's stated in one of the nmp patents.


/appft1.uspto.gov/

[0073] • It should also be noted that one (as depicted) or both GPU riser boards 412, 414 can include SSD module 420. SSD module 420 can be coupled to a rear surface of one or both GPU riser boards 412, 414.

Screen Shot 2015-12-26 at 1.50.52 PM.png
 
  • Like
Reactions: AidenShaw
Yes, I'm going to put my foot in it again.

I've been using computers for a long time. I learned to program on a refrigerator sized minicomputer that had to have the boot sequence keyed in from the front panel. I've lived through a huge array of different storage methods. The aforementioned minicomputer used terminals that had paper tape punches for storing user programs on. 8 inch floppies, disk packs, 80 column punch cards, 9 track tape, 5 1/4 inch floppies, Bernoulli drives, mini tape drives, plus all the more modern stuff. I've programmed on several different types of mainframes and minis, plus Atari 800, Apple II, original IBM PC, and so on. I've stuffed 100's of different types of cards in slots, from OEM stuff to custom boards to support bespoke lab equipment.

I say all that so that when I say that those who are holding on to the internal drive bay or internal slots are like those old car buffs who lament that everything is computerized these days
, I know what I'm talking about because I have the same experience with computers that those old car buffs have with cars. I don't lament those old days though, because I really like the fact that the mechanic no longer has to spend several days tinkering with my car to figure out what is wrong.
Here we go again...someone telling us they know our needs better than we do. We get it...you're old and no longer want to perform upgrades, rock music is a sign of the devil, and you don't want people on your lawn. Fine, you're welcome to do external upgrades to your hearts content. However us younger guys have our needs and preferences. Please don't tell us they're wrong and yours are right.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.