Is the new Mac Pro a failure?

[mcnallym]

Maybe not, but If Apple would want to update the Mac Pro tower to anything new, it would meet the Rift requirements.

Well if you want to start updating the components in the Tower with new CPU boards then you are now surely able to update the components in the nMP as well?

If you can update the cMP with new CPU's and Chipset Boards etc then why cannot Apple release a new GPU Cards to fit into the nMP format and meet the requirements?

If you want to release new components into either of the form factors then they can be upgraded to meet the requirements of course, can't they?

Or have I misunderstood your post?

 

[koyoot]

The argument is between form factors, not exactly hardware. For most people who relied on upgradeability of the computer, loosing it is painful.


Both computers as you can see have their drawbacks. Both are currently non upgradeble.

 

[ZombiePhysicist]

I am not questioning his knowledge. He is absolutely right about latency with crossfire. But Crossfire will not be working on VR. It will be even possible to have AMD and Nvidia setup working on VR. Like I said: One GPU for one display.

I asked simple question. Why does Nvidia GPU require 57 ns latency for VR?

Dual GPU setups for VR already been discussed by Valve and work for AMD:

Here is first argument what is happening with Nvidia GPUs:

Secondly: You need as direct control by application of the hardware as you can.

Thirdly: Kepler and Maxwell does not have Hardware schedulers. The last architecture from Nvidia that had such thing was Fermi. And was inefficient, unfortunately. Thats why Nvidia went with software scheduling. To save power.

Next thing is that for scheduling Nvidia went with software, for Kepler and Maxwell. For VR, like I said before, you want as direct control by application over the hardware as you can. With Maxwell and Kepler you have software that controls both the application and hardware. That creates lag. What can Nvidia do in this case? If they are able to optimize the data paths - great. But from hardware point - nothing.

AMD on the other hand have Hardware schedulers in their hardware, at least on Tonga and Fiji chips.

First, thank you for these videos. Second, thank you for thinking I have a shot in heck of understanding hours worth of tech explanation. I'm going to try to listen through these later tonight. Thanks for the links.

As for your substantive arguments, again, I'm out of my depth on this (and many other) issues and best I can do is go by what others say here.

Anyway, if I understand what you're saying, it boils down to that you pretty much want the biggest baddest AMD card you can get for this stuff.

 

[996085]

The argument is between form factors, not exactly hardware. For most people who relied on upgradeability of the computer, loosing it is painful.


Both computers as you can see have their drawbacks. Both are currently non upgradeble.

I am surprised at the number of people who don't understand this.

 

[-hh]

depends on if you're simply talking about the various components inside the machine or the whole of the machine itself..

cmp is simple sheet metal construction.. nmp is using manufacturing processes we've never seen in computer construction ...

The caveat of "...in computer construction..." is key to this claim.

Deep draw nonferrous metal parts have been used for the past 100+ years, such as in brass
cased ammunition: the Smith & Wesson Model 1 pistol dates from 1857.

And not just brass: steel and aluminum have also been used, with aluminum having been tested with since at least the 1920s:

'Lewis, L.D., "Calibre .30 Cartridge Cases Made from Aluminum Alloys", Technical Report Report No. R-16, 25 February 1926, Frankford Arsenal, Phila., Pa'

FYI, the conclusion of this report says (words to the effect of): 'we can't say how old the idea is of using Aluminum, but during our study program, we found some old, undated .45-70 cases in Al laying around' ... (!)

...and to tolerances typically reserved for the aerospace or medical device industries.

Just what level of tolerances is this? The video had no such performance data. Besides, the precision capability of a properly set up Waterbury Ferrel presses (a company that's been in business for 160 years) typically used in the above can achieve precision levels to ~0.001" (or better) ... without the need for CNC honing.

just a guess but when you bought a cmp, the housing (minus any components etc) probably cost you less than $500.. nmp is probably costing nearer $1500..

No, the reality is that what they're doing isn't profoundly different than a beer or soda can, and because it probably replaced a lot of CNC machining of Al with a single deep draw process step, the nMP's case will be cheaper to fab than the cMP's case.

 

[koyoot]

First, thank you for these videos. Second, thank you for thinking I have a shot in heck of understanding hours worth of tech explanation. I'm going to try to listen through these later tonight. Thanks for the links.

As for your substantive arguments, again, I'm out of my depth on this (and many other) issues and best I can do is go by what others say here.

Anyway, if I understand what you're saying, it boils down to that you pretty much want the biggest baddest AMD card you can get for this stuff.

If you are pointing the last words directly to me, then you're wrong. I want the biggest and baddest GPU from Nvidia for VR, and I will not be able to get it. For the exact reasons I have shown.

 

[MacVidCards]

Everyone's all over it cause it kinda suits them now.
If we can tap at both GPUs individually in OS X (not XFire or anything), like koyoot says it's one GPU for each eye how the tech works, maybe the D700 will hold it's own, why not? Current res for each eye is low enough, and refresh rate. So why suffer (or beat on it) in anticipation?
By the time anyone has it and it runs properly and smoothly we'll be discussing 2018's nMP.

Unfortunately the nMP has the GPUs hard wired such that all display outputs come from one card. When not using the 2nd one for compute it is literally sitting there taking up space.

As in, the physical connection between 2nd card and display outputs isn't there. So even if AMD comes out with non-crossfire dual GPU VR setup, it won't be running on a nMP.

However, it may very well run on cMP with 2 AMD cards.

The 12 Core nMP does indeed have enough CPU speed for Rift, I tried last night. It was just laughed out due to the GPUs.

If Rift gets written to use multiple CPU cores effectively, the cMP will become the only Mac easily capable of running it.

And as far as Netkas single handedley creating all the cards, Maxwell was first generation where I had no real part in the EFI other than LOTS of testing. For every card, AMD and Nvidia, I personally found the PCIE 2.0 switch.

For Fermi & Kepler I had crucial input on EFI. One of my proudest moments was replying to an email where he had given up on GTX5xx as being impossible. We had identical systems and cards and had spent a few weeks on 570. I replied to his "it's hopeless" email with a screenshot of 570 finally working. I had found last piece myself. He is by far and away vastly more knowledgable but sometimes the puzzles were solved via pattern recognition and brute force trying things that didn't make sense but worked anyway.

 

[ZombiePhysicist]

If you are pointing the last words directly to me, then you're wrong. I want the biggest and baddest GPU from Nvidia for VR, and I will not be able to get it. For the exact reasons I have shown.

Which shows my amazing ability not to get the point. ;p I'll try and watchup on the videos provided. Thanks.
[doublepost=1452554141][/doublepost]

However, it may very well run on cMP with 2 AMD cards.

The 12 Core nMP does indeed have enough CPU speed for Rift, I tried last night. It was just laughed out due to the GPUs.

If Rift gets written to use multiple CPU cores effectively, the cMP will become the only Mac easily capable of running it.

So this is weird. Why would it run on teh 2.7 12core nMP processors and not on the 3.46ghz 12core cMP processors. They are basically even and from what I've seen, at most 5-10% apart.

And as far as Netkas single handedley creating all the cards, Maxwell was first generation where I had no real part in the EFI other than LOTS of testing. For every card, AMD and Nvidia, I personally found the PCIE 2.0 switch.

So my bad, I'm sure there were lots of other folks that contributed to the pool of knowledge and the community, and you certainly have been a huge contributor. Didnt mean to diminish that. So sorry and thanks!

 

[flat five]

This is not going to happen Koyoot. Flat Five (expert in a lot of things) shared with us that it will take 5 - 10 years before this all will be mainstream. Sorry m8. I think you'r wasting your time! (sarcasm)

what i said was this:
#1451
""it's not production ready technology yet.. far from it by the looks of it..
and even further still, none of the industry standard design software makes (any) use of it.

we're looking at at least 5-10 years until this is ready for the big time""
-----

and what i meant by that was VR will not be used in production environments for quite some time.. 'mainstream' is not the same thing as being used in production.. it's possible we'll see mainstream adaption for consumptive purposes prior.. ie- the way others may view one's work..

what we won't see for quite some time is VR tech being used as a serious tool in the design phases.. for one, there is no (standard) software for it yet.. if you think the likes of http://www.skanska.com are going to adopt some non-standard non-proven technology for use in their production environments then you have another thing coming.. and that same mindset trickles down to nearly all the smaller companies as well.. it's going to be multiple years yet until these types of industries begin to use VR tech in their work and even then, it's going to trickle in slowly.. at first, it will be for oohs&aahs&wows for client presentation.

prior to being beneficial or practical as a design tool, we're going to need new input devices or methods for us to interact with a computer in a virtual world.. all cad/design software relies heavily upon mouse & keyboard input.. put a VR headthing on and you can no longer even see the keyboard or mouse.. this is a chasm that needs bridged in order for the tech to become production ready.

 

[Arron Rouse]

That doesnt mean it's not form over function. Spending $1million on a jewel encrusted case and increasing the price while still doing the same thing is form over function.

If it functions in the market segments where the Apple advertises its use, at worst that makes it form and function. The nMP is much faster than the cMP too.

Also, more substantively, new media work like augmented reality. Not happening on the new mac pro.

Not happening on the cMP either. But Apple hasn't advertised any Mac Pro for AR/VR and all the early stuff is Windows only so it's not happening on any other *nix machines either.

Oh, you are a freelancer that does a lot of 4K video production work and you want a fast SSD like this, not happening either: http://barefeats.com/hard210.html

If you've only got a 4K monitor plugged in and switched on, the cMP won't even boot. You have to remember to switch the monitor off before every OS update or restart. And that SSD is overkill even for ProRes 4444 XQ -- 1TB for $1200? You can buy an entire 8 bay Thunderbolt RAID for that which would be much better for 4k video work. (I do think that SM951 RAID is awesome but it's ridiculously niche market.)

You can't just pin all the failure risk on the computer and claim "if my computer with all my data in it goes down, so does my company". Well if you have all your data in one external box and THAT goes down, so does your company.

Ah, you misunderstand. Or more likely I wrote poorly. When I say "my company" I mean it's not just me who'd suffer, the NAS in the office feeds 6 systems. If I had the RAID in the cMP instead of the NAS, the risk would be unacceptable. Even if were just me, a RAID in the cMP would be unacceptable as I use my laptop a lot.

You need to quit thinking of the cMP in its actual implementation and start thinking of it as a cMP form factor with updated components.

There's a reasonably updated cMP sat under my desk in my office. It has maybe another two or three years left because there just aren't any more worthwhile updates. Hell, as mentioned above, if I forget to switch the monitor off when I pack up for the day, the damn thing won't even boot in the morning. I do still love it though.

The argument is between form factors, not exactly hardware. For most people who relied on upgradeability of the computer, loosing it is painful.

A lot of other arguments have been brought out but that seems to be the only valid one left. In my case, the upgradeability has been as much a curse as anything else. The most problematic thing by far has been a Sapphire 7950.

 

[ZombiePhysicist]

If it functions in the market segments where the Apple advertises its use, at worst that makes it form and function. The nMP is much faster than the cMP too.

Disagree, if it has an overpriced case, it's not functioning. Furthermore, 3.46ghz 12 core cMP with titan and 5.9GB/sec SSD decimates the nMP. I love it.

If you've only got a 4K monitor plugged in and switched on, the cMP won't even boot. You have to remember to switch the monitor off before every OS update or restart. And that SSD is overkill even for ProRes 4444 XQ -- 1TB for $1200? You can buy an entire 8 bay Thunderbolt RAID for that which would be much better for 4k video work. (I do think that SM951 RAID is awesome but it's ridiculously niche market.)

I love it, now the cMP's SSD is too fast!

Not happening on the cMP either. But Apple hasn't advertised any Mac Pro for AR/VR and all the early stuff is Windows only it's not happening on any other *nix machines either.

Apple ignoring a market doesnt mean it's not valuable. And apple has to ignore it, because, effectively, it has no hardware to support it. Pixar, in the mean time, isnt ignoring it.

 

[bogg]

Disagree, if it has an overpriced case, it's not functioning. Furthermore, 3.46ghz 12 core cMP with titan and 5.9GB/sec SSD decimates the nMP. I love it.



I love it, now the cMP's SSD is too fast!



Apple ignoring a market doesnt mean it's not valuable. And apple has to ignore it, because, effectively, it has no hardware to support it. Pixar, in the mean time, isnt ignoring it.

5.9GB/s? x16 PCIE connected SSD? Sounds like you could buy a few nMPs for the cost of one of those drives (if you aren't RAID:ing a few x4:s to achieve that speed)

Edit: Aah, it was the amfeltec squid thingy...read about that a while back.
[doublepost=1452555540][/doublepost]

Apple ignoring a market doesnt mean it's not valuable. And apple has to ignore it, because, effectively, it has no hardware to support it. Pixar, in the mean time, isnt ignoring it.

I don't really see what Pixar has to do with a discussion regarding the _Apple_ nMP?

 

[flat five]

The caveat of "...in computer construction..." is key to this claim.

Deep draw nonferrous metal parts have been used for the past 100+ years, such as in brass
cased ammunition: the Smith & Wesson Model 1 pistol dates from 1857.

And not just brass: steel and aluminum have also been used, with aluminum having been tested with since at least the 1920s:

'Lewis, L.D., "Calibre .30 Cartridge Cases Made from Aluminum Alloys", Technical Report Report No. R-16, 25 February 1926, Frankford Arsenal, Phila., Pa'

FYI, the conclusion of this report says (words to the effect of): 'we can't say how old the idea is of using Aluminum, but during our study program, we found some old, undated .45-70 cases in Al laying around' ... (!)

just don't forget.. the comparison is being made to cmp.. sheet metal production.

i could, quite literally, model a cmp case tonight.. send it to my guy in the morning.. and have a non-apple made cmp case in my hands by tomorrow afternoon.. like- i'm not even exaggerating..
sheet metal work is really easy and really fast.

how many sheet metal guns do you know about?

Just what level of tolerances is this? The video had no such performance data. Besides, the precision capability of a properly set up Waterbury Ferrel presses (a company that's been in business for 160 years) typically used in the above can achieve precision levels to ~0.001" (or better) ... without the need for CNC honing.

would really like to see them press a chunk of aluminum to a thousandth that's the size of nmp instead of the size of a bullet casing.

No, the reality is that what they're doing isn't profoundly different than a beer or soda can, and because it probably replaced a lot of CNC machining of Al with a single deep draw process step, the nMP's case will be cheaper to fab than the cMP's case.

we could argue about this all night and never conclude.. i definitely believe, without components, the nmp costs much more than cmp.. but you'll just say 'no, you're wrong.. the cmp cost more'
well. ok. whatever you'd like to believe then.. because i highly doubt you or i or anyone here is going to gain access to the actual costs.

also, it's not only about the shell..
it's the core and fan and custom boards which are also adding to the cost.

 

[996085]

There's a reasonably updated cMP sat under my desk in my office. It has maybe another two or three years left because there just aren't any more worthwhile updates. Hell, as mentioned above, if I forget to switch the monitor off when I pack up for the day, the damn thing won't even boot in the morning. I do still love it though.

No, there is not. The cMP uses older processors, older disk interfaces, older USB, and lacks Thunderbolt...just to name a few. Again: You're comparing technology instead of form factor. There's nothing which would prevent the current nMP technology from having been used in the cMP form factor.

 

[MacVidCards]

If you've only got a 4K monitor plugged in and switched on, the cMP won't even boot. You have to remember to switch the monitor off before every OS update or restart. ...In my case, the upgradeability has been as much a curse as anything else. The most problematic thing by far has been a Sapphire 7950.

This is an error.

In point of fact, our Maxwell EFI cards offer BETTER 4K boot support than the nMP with it's 4 year old GPUs. The 7950 has the issue becuase it is using a very old EFI that was Pre-4K. The nMP doesn't like SST 4K and last time I checked wouldn't show 4K bootscreen there. The fact that a 2 man shop can do a better EFI than Apple has bothered to put on nMP is a testament to just how little they care.

(following not addressed to Arron)

And I have to say that there is no way in high holy heck that nMP case costs $1,500.

It is using same manufacturing tech as a Soda can. No more, no less. They send it through the buffer one more time than Coors does for their Silver Bullet and call it a day. Guzzling Kool Aid will stain your teeth badly and cause you to mistake a soda can for $1,500 in Aluminum. (At $ .66/lb for Al, it's closer to $3)

 

[flat five]

There's nothing which would prevent the current nMP technology from having been used in the cMP form factor.

well, except that the designers/manufacturers didn't want to build another cmp.
because, that, is pretty damn preventative.

 

[ZombiePhysicist]

5.9GB/s? x16 PCIE connected SSD? Sounds like you could buy a few nMPs for the cost of one of those drives (if you aren't RAID:ing a few x4:s to achieve that speed)

Edit: Aah, it was the amfeltec squid thingy...read about that a while back.
[doublepost=1452555540][/doublepost]

Right, lets move the goal posts. It's fast. It's probably not as expensive as you theorize to make you point, much like the statement asserting cMPs being non bootable in 4k mode (not yours) that was argued to be wrong by MVC in mere minutes.

I don't really see what Pixar has to do with a discussion regarding the _Apple_ nMP?

They have been held here as pretty much the only thing that counts as professional use, so now those arguments get hoisted by their own petard. And if you missed it, pixar is developing movies in VR, see here:
http://www.thedailybeast.com/articl...-oculus-rift-virtual-reality-movie-henry.html

Since the nMP and basically all macs are suboptimal or outright cant do VR development, it's another professional market going to windows, in substantial part because apple isn't make a real professional machine to serve the professional markets.

 

[flat five]

And I have to say that there is no way in high holy heck that nMP case costs $1,500.

i definitely agree with you Mac.

for clarity, i mean -> take out the intel/amd/motorola/etc chips.. take out the ram modules.. and whatnot..
what you have left is what i've been talking about.. not only the shiny shell.. rather the whole infrastructure

---
edit
according to this search:

apple part 923-0498

we're looking at somewhere between $3-400 as a retail cost on a nmp outer shell.

 

[996085]

well, except that the designers/manufacturers didn't want to build another cmp.
because, that, is pretty damn preventative.

Didn't want is not the same as prevent.

 

[wallysb01]

Ah, you misunderstand. Or more likely I wrote poorly. When I say "my company" I mean it's not just me who'd suffer, the NAS in the office feeds 6 systems. If I had the RAID in the cMP instead of the NAS, the risk would be unacceptable. Even if were just me, a RAID in the cMP would be unacceptable as I use my laptop a lot.

Why do you view a RAID in a workstation more risky than a RAID in a NAS?

 

[flat five]

Didn't want is not the same as prevent.

sorry but yes, it is..
it's the entire reason the cmp has been discontinued.

just accept the chain of events for what they are.. not sure why you insist on fighting these things as if they didn't happen.

 

[MacVidCards]

And here we are, the nMP is excluded from participating in a new Tech, ENTIRELY due to lack of GPU upgrades.

There is literally no other reason.

Form (shiny soda can) has won out over function (running or even being involved in VR).

No amount of Kool-Aid induced apologies and and praising the lovely finish is going to fix that. Ditto on posting pix of the amazing facility, the smiling designers, etc. They compromised the function, and it is now, officially, a laughingstock.

Someone posted above that it costs "at least $1K" to get a VR capable PC. Think about that.

The fact that tossing $10K at Apple won't get you VR but $1K from a PC builder will is really sad.

At least it's shiny.

 

[flat five]

And here we are, the nMP is excluded from participating in a new Tech, ENTIRELY due to lack of GPU upgrades.

[...]
The fact that tossing $10K at Apple won't get you VR but $1K from a PC builder will is really sad.
.

here's what's sad --

we're quite literally on the verge of the most important, or- most life changing technology ever brought to us by fellow humans..

..and here, it's devolved into a freaking mac vs pc vs amd vs nvidia debate.

 

[996085]

sorry but yes, it is..
it's the entire reason the cmp has been discontinued.

just accept the chain of events for what they are.. not sure why you insist on fighting these things as if they didn't happen.

Uh no, no it's not.
[doublepost=1452559591][/doublepost]

here's what's sad --

we're quite literally on the verge of the most important, or- most life changing technology ever brought to us by fellow humans..

..and here, it's devolved into a freaking mac vs pc vs amd vs nvidia debate.

Ah, no...no it hasn't. It's the Mac is incapable due to Apple choosing form over function. That's it. It has nothing to do with the PC.

 

[flat five]

Uh no, no it's not.

look pl..

you're incredibly insistent on people getting and understanding and agreeing with your point:
There's nothing which would prevent the current nMP technology from having been used in the cMP form factor.


..i bet you've said this over a hundred times by now and i get it, you're frustrated because nobody is hearing you.

but guess what.. e v e r y b o d y .. gets it.
you're arguing n o b o d y.
nobody is saying you're wrong.
every single person here knew exactly what you're saying on the very first day the nmp was previewed to us.

ok?

what's happening though is that you're so set on making your point, the one that is already completely obvious to even your nemesis, that you're not listening to what others are saying in return.. others aren't saying "you're wrong".. they're saying other things.