Do I even need a Mac Pro anymore?

[deconstruct60]

.... It's too bad the dual workstation graphics cards are the price of a single socket Xeon.

On which alternative universe?

E5 1620 --> $294
E5 1650 --> $583
E5 1660 --> $1080

The v2 versions of those are going to be about the same price.

Apple is charging $250 for a 1GB AMD 5770. A workstation card with 4-6GB of VRAM would likely to be in the $400+ zone. (AMD FirePro 7000 w/ 4GB VRAM goes for ~$699. If Apple wrangles a $200 discount out of that the price that would be $499) So two would be minimally be $800+ (and more likely around $1000+). While $800 is lower than the 1660 it is definitely lower the CPUs that are probably the "good" and "better" standard new Mac Pro configurations.

But it also a bit myopic to cast this as purely only being graphics cards. If all ever push them on a Mac Pro is simply graphics then the two card set up doesn't have much value. There little to no reason why LR can't leverage the cards as computations devices though. OpenCL is portable and fits in with Adobe's need for a cross-platform infrastructure to layer features on top of.

The iMac and MBP 15" will get a roughly similar percentage performance bump since they too can leverage OpenCL.

An operative question with the 2013 Mac Pro is whether you are buying it for by primarily looking into the past and minimizing the number of applications and peripherals evolving. Or are you looking at where software and needs are going to be 2-3 years from now.

 

[badlydrawnboy]

Anyone know whether the Mini is due for, or likely to get, a refresh anytime soon?

 

[deconstruct60]

Anyone know whether the Mini is due for, or likely to get, a refresh anytime soon?

No. Probably alongside the iMac again in a Sept-Oct rollout.

For last couple of years, typically the Mac Mini is largely a refactored MBP 13". Now if the MBP 13" stalled perhaps that shifts to the rMBP 13", but still a variant of a laptop that hasn't been released yet. The clock on the mini release probably starts after the MBP 13" variant ships that is shares a high number of parts with.

 

[badlydrawnboy]

Given that Fall/late Fall is also probably when the MP will be released, seems to make sense to wait until then. I'm not in a huge hurry. Thanks.

 

[ValSalva]

On which alternative universe?

E5 1620 --> $294
E5 1650 --> $583
E5 1660 --> $1080

The v2 versions of those are going to be about the same price.

Apple is charging $250 for a 1GB AMD 5770. A workstation card with 4-6GB of VRAM would likely to be in the $400+ zone. (AMD FirePro 7000 w/ 4GB VRAM goes for ~$699. If Apple wrangles a $200 discount out of that the price that would be $499) So two would be minimally be $800+ (and more likely around $1000+). While $800 is lower than the 1660 it is definitely lower the CPUs that are probably the "good" and "better" standard new Mac Pro configurations.

With the new Mac Pro one Xeon is paired with two workstation graphics cards. That's my only point. That's overkill for many users (not all).

But it also a bit myopic to cast this as purely only being graphics cards. If all ever push them on a Mac Pro is simply graphics then the two card set up doesn't have much value. There little to no reason why LR can't leverage the cards as computations devices though. OpenCL is portable and fits in with Adobe's need for a cross-platform infrastructure to layer features on top of.

The iMac and MBP 15" will get a roughly similar percentage performance bump since they too can leverage OpenCL.

An operative question with the 2013 Mac Pro is whether you are buying it for by primarily looking into the past and minimizing the number of applications and peripherals evolving. Or are you looking at where software and needs are going to be 2-3 years from now.

What about the present? And in 2-3 years those non-upgradeable graphics cards will be yesterday's news. It's an expensive gamble for the future. There are no guarantees about where software is headed.

 

[derbothaus]

Wait, just found this: http://macperformanceguide.com/macmini2012-speed-diglloydSpeed1.html

Seems the 2.6 Mini is faster than the 3.3 6-core Mac Pro with 24 GB RAM. http://macperformanceguide.com/macmini2012-speed-diglloydSpeed1.html

And adding a PCI-e Acceslior using something like this (http://macperformanceguide.com/Reviews-OWC-Mercury-Helios.html) could make it even faster.

You might be on to something!

Did you also see how the performance takes a nosedive when the 'medium' test is used? The lite fluffy test favors the mobile chips extra turbo bins. That is it really. Read through the lines a bit. Also those are 'estimated'. FYI.

 

[deconstruct60]

With the new Mac Pro one Xeon is paired with two workstation graphics cards. That's my only point. That's overkill for many users (not all).

Probably not over the service lifetime of the machine. If buying for a 4-5 year service lifetime then there is headroom for many initially but that will shrink somewhat over time. The longer the surface lifetime the more the shrinkage. I'm not a "future proof" zealot, but you're missing a major point with the Mac Pro design that if skipping over that there is "head room" built into this. It is aimed at what going to do not where are now.

Sure for users whose workload has absolutely plateaued they probably do want to start looking "down market" from the Mac Pro. The current Mac Pro or the 2013 Mac Pro neither one is a good fit so the specific parameters of the 2013 Mac Pro are moot.

What about the present? And in 2-3 years those non-upgradeable graphics cards will be yesterday's news.

Either the hardware is overkill and the customer is already riding the "yesterday's bleeding edge front" news or it isn't overkill for longer term needs of the customer. If customer takes 2-3 years to grow into a box going to keep for 4-6 years that is a pretty good fit.

It's an expensive gamble for the future. There are no guarantees about where software is headed.

There are not guarantees that new Macs , OS X , or Apple inc will exist in 2-3 years either.

 

[fa8362]

Did you also see how the performance takes a nosedive when the 'medium' test is used? The lite fluffy test favors the mobile chips extra turbo bins. That is it really. Read through the lines a bit. Also those are 'estimated'. FYI.

The tests are irrelevant. badlydrawnboy doesn't do anything that requires a Mac Pro.

 

[mBox]

Maybe ask that question to yourself once you get to try one at the Apple Store months from now.
I know that most of my pro photographer friends and film production peeps are waiting for this to drop.
The promise of multiple 4K res is just too enticing for most of us.
I work with 4K and up (RED) and this might help.