Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Next mac processor update?

D

dred67212

macrumors newbie
Original poster
Jun 1, 2014
21
1
I am looking into a mac pro and was looking at the 3.7 quad core and then upgrading the graphics cards to d500s. My question is I know the 3.7 has been out for a while and and I was wondering if they are close to updating them with something better soon because i know they are a year old. Should i wait or just buy it?

thanks
 
They were replaced by Intel at the end of August, whether Apple will even update to the latest platform is up for discussion. There aren't huge performance increases from the lower end of the latest platform. You can hold on in order to get the latest machine, but if you need this for productivity then pick up a refurbished unit and get on with it :). You'd be able to sell such a system for very little loss anyway most likely.
 
"Something better" is going to be defined differently by different people. The processors that are replacing the ones currently used are either exactly the same speed for the "lower" core CPUS, or at the higher end, have a slight uptick in speed (quad-core will remain at 3.7, 6-core at 3.5, same turbo boost, etc).

However, the architecture is moving up to the newer Haswell. Haswell brings improvements in different areas, which may or may not apply to the type of work being done. AVX2 is going to be supported, as well as Intel's Quick Sync which can dramatically reduce the time it takes to complete certain rendering tasks (which is what caused some people to believe that the Haswell iMacs were faster than the Ivy Bridge Mac Pros).

All in all, Haswell is a decent update from Ivy Bridge, but it's not dramatic. The biggest changes, I feel, will be the new GPUs as well as DDR4 RAM. Also, there's a new PCI-e SSD by Samsung floating around which is about 30-40% faster than the current ones used by Apple that are based on the XP941 chipset. This will likely be used in the Mac Pro refresh (as well as the other systems when they get refreshed this year).
 
dred67212 said:
I am looking into a mac pro and was looking at the 3.7 quad core and then upgrading the graphics cards to d500s. My question is I know the 3.7 has been out for a while and and I was wondering if they are close to updating them with something better soon because i know they are a year old. Should i wait or just buy it?

thanks
Click to expand...

If you must have a Mac Pro 6,1 (2013) in your life , please keep the following in mind :

The GPUs have a proprietary form factor and can only be upgraded at time of system purchase from Apple . There are no third party alternatives . And the GPU technology is already three years old. You might be able to borg some used upgrade cards off of eBay , but I imagine they are in high demand . That said , the AMD D500 and D700 cards run Apple's own apps nicely due to optimization tricks (they make both the hardware and software .) It's other software publisher's apps you really need to worry about , especially for Cuda based apps like Octane .

The Mac Pro 6,1 (2013) CPUs are Ivy Bridge EP and use a single first generation LGA 2011 socket . The 12 Core 2.7 GHz version is already the highest performing chip possible in this model . There will be no upgrades possible with succeeding versions of Xeon processors since both the Haswell EP and Broadwell EP Xeons will require a LGA 2011 v3 socket . So, no CPU upgrades are possible if you already have the 12 Core 2.7 GHz Ivy installed .

If Apple were to release a new "Black Tube" with Haswell EP Xeons , a logic board redesign would be required . I suspect that will not happen as it's expensive . I think the nMP form factor will be retired as it's just a modern version of the PMG4 Cube (a concept computer that went into production.)

My local Creative clients do not want to own an appliance , no matter how cool looking . Only a handful need the portability of the Black Tube (it can fit in a back pack.)
They want rigs that have two processors and greater future-proofness . They want upgradeable systems since they can't afford a fully decked out rig as it is . They slowly upgrade their gear as funds become available, not Apple's take it now or leave it approach with the Mac Pro 6,1 (2013) .

I'm a major league cheerleader of the Mac Pro , so it hurts me to say these words .

My best estimate of what the next Mac Pro will be and when she's released :

A dual socket Mid Tower with Broadwell Xeons , released in Spring , 2016 . She'll have two or three PCIe Rev 3.0 nVidia Maxwell GPUs installed , since GPGPU is here to stay . You won't like the price tag of the top model , though . A 2 X 18 Core (36 Cores total) Mac Pro might very well cost in excess of 12,000 USD based on current chip cost estimates (4,000 USD per chip for a Broadwell EP 18 Core E5-2600 v4 series CPU.) And she'll run hot, too, despite the die shrink from Haswell . Apple might simply make a much larger cylinder with two or three thermal cooling cores with one large fan each core to keep things quiet and a larger PSU to feed everything . Base model to have just one CPU installed .
 
Machines said:
My best estimate of what the next Mac Pro will be and when she's released :

A dual socket Mid Tower with Broadwell Xeons , released in Spring , 2016 . She'll have two or three PCIe Rev 3.0 nVidia Maxwell GPUs installed , since GPGPU is here to stay . You won't like the price tag of the top model , though . A 2 X 18 Core (36 Cores total) Mac Pro might very well cost in excess of 12,000 USD based on current chip cost estimates (4,000 USD per chip for a Broadwell EP 18 Core E5-2600 v4 series CPU.) And she'll run hot, too, despite the die shrink from Haswell . Apple might simply make a much larger cylinder with two or three thermal cooling cores with one large fan each core to keep things quiet and a larger PSU to feed everything . Base model to have just one CPU installed .
Click to expand...

From your lips to God's ears. I fear Apple has not yet finished to shoot oneself in the foot. Also regarding TRIM refusal for common SSD's.
 
Machines said:
A dual socket Mid Tower with Broadwell Xeons , released in Spring , 2016 . She'll have two or three PCIe Rev 3.0 nVidia Maxwell GPUs installed , since GPGPU is here to stay . You won't like the price tag of the top model , though . A 2 X 18 Core (36 Cores total) Mac Pro might very well cost in excess of 12,000 USD based on current chip cost estimates (4,000 USD per chip for a Broadwell EP 18 Core E5-2600 v4 series CPU.) And she'll run hot, too, despite the die shrink from Haswell . Apple might simply make a much larger cylinder with two or three thermal cooling cores with one large fan each core to keep things quiet and a larger PSU to feed everything . Base model to have just one CPU installed .
Click to expand...

They don't do case redesigns very often, and a change in logic board applies either way. They knew that when they released ivy. I think it's more likely that as long as they opt to continue the line, you'll see another tube with updated internals. If this was to be a concept computer, it wouldn't have displaced the old one.
 
Machines said:
If Apple were to release a new "Black Tube" with Haswell EP Xeons , a logic board redesign would be required . I suspect that will not happen as it's expensive .
Click to expand...

Why do people think that a logic board design is expensive?

If tweaking for a new socket is "too expensive" for a money machine like Apple, how can companies like Asus and Gigabyte and MSI and ... sell dozens of different motherboards. (Asus is currently selling at least three different 2011v3 motherboards. SuperMicro has over 45 different 2011v3 Xeon motherboards for sale.)

Circuit boards are designed by software programs - think of them as "board compilers" that take the input program (parameters) and compile the source to manufacturing instructions for the assembly line robots.

Apple adjusts the "socket" parameter set to "2011v3" and clicks "compile", and the manufacturing directions are sent to the robots that build the boards. All the hard work is done quickly and cheaply by CAD/CAM software.

(Note also that the 2011v3 isn't a radically different chipset - a big reason for changing the pins was to make sure that E5-v2 CPUs aren't plugged into a DDR4 mobo.)
 
Last edited:
Machines said:
I stand corrected. some recovery services claim it is possible to recover data from failed non self encrypted SSDs. None of my clients have ever tried these services , though .
Click to expand...

One of the more common failure modes of an SSD is that it loses the ability to update the block re-mapping tables (the tables that map the logical sector numbers that the OS uses to the actual physical locations in the flash array). This effectively makes the entire drive read-only.

When this happens, the drive often appears dead - especially if it's a boot drive. Often, however, the drive has become read-only (the existing mapping tables are OK, but they can't be updated). Attempts to mount the drive read/write will fail.

If the drive is mounted "read-only" from another system or bootable utility, it can be possible to copy the data off. (Of course, the recovery services won't tell you that it is that easy ;) .)

http://superuser.com/questions/345997/what-happens-when-an-ssd-wears-out
 
Last edited:
AidenShaw said:
One of the more common failure modes of an SSD is that it loses the ability to update the block re-mapping tables (the tables that map the logical sector numbers that the OS uses to the actual physical locations in the flash array). This effectively makes the entire drive read-only.

When this happens, the drive often appears dead - especially if it's a boot drive. Often, however, the drive has become read-only (the existing mapping tables are OK, but they can't be updated). Attempts to mount the drive read/write will fail.

If the drive is mounted "read-only" from another system or bootable utility, it can be possible to copy the data off. (Of course, the recovery services won't tell you that it is that easy ;) .)
Click to expand...

Interesting tidbit of info. Rarely do I have a client with a failed SSD (one out of hundreds I have sold, if I remember), but nice to know to check if it can be read in another machine.
 
reco2011 said:
Data "recovery" from a failed SSD is restoring from backup.
Click to expand...

Exactly. :) It doesn't matter if its an SSD or not, backups are your true data recovery - why would anyone want to count on anything else?

(and come on, there are *lots* of us in here doing things like running on RAID0 SSD boot drives - I certainly hope everyone has good constant backups)
 
dmylrea said:
Interesting tidbit of info. Rarely do I have a client with a failed SSD (one out of hundreds I have sold, if I remember), but nice to know to check if it can be read in another machine.
Click to expand...

I have not had a client report any SSD failures of drives installed by me in systems , but I went to great lengths to avoid OCZ . I use Toshiba , Samsung PROs (great for RAID 0) , OWC , some Sandisk and on occasion Kingston .

I've had OWC drives fail on me , used in house for diagnostics . But they failed because the connectors became loose and unreliable due to heavy use with adapter cables . Chips were probably still good . I now use USB thumb drives , but I'm experiencing the occasional lost file . Might try USB thumb drives with those fancy high speed SSD controllers next for reliability .

Big thing you need to worry about is 4K editing with SSDs . A client will eat through the TBW part of the warranty in a month if you're not careful , even if the drive is known to have survived PBs of writes in a lab somewhere .
 
AidenShaw said:
Why do people think that a logic board design is expensive?

If tweaking for a new socket is "too expensive" for a money machine like Apple, how can companies like Asus and Gigabyte and MSI and ... sell dozens of different motherboards. (Asus is currently selling at least three different 2011v3 motherboards. SuperMicro has over 45 different 2011v3 Xeon motherboards for sale.)

Circuit boards are designed by software programs - think of them as "board compilers" that take the input program (parameters) and compile the source to manufacturing instructions for the assembly line robots.

Apple adjusts the "socket" parameter set to "2011v3" and clicks "compile", and the manufacturing directions are sent to the robots that build the boards. All the hard work is done quickly and cheaply by CAD/CAM software.

(Note also that the 2011v3 isn't a radically different chipset - a big reason for changing the pins was to make sure that E5-v2 CPUs aren't plugged into a DDR4 mobo.)
Click to expand...

Interesting . But if money or complexity is not an issue in a logic board update, why did not Apple, for instance, use PCIe REV 3 slots in the Mac Pro 2012 "refresh" ? The standards were already released in 2010 . Was Apple just being conservative ?
 
flowrider said:
You may be interested in this thread:

https://forums.macrumors.com/threads/1459847/

I wouldn't trust an OWC SSD as far as I could throw it. I had one in my MBA that failed shortly after installation. My MBA is now happily running a Transcend internal SSD. My Mac Pro has 4 Samsung SSDs mounted on 2 Velocity x2 Duos.

Lou
Click to expand...

Did you RAID 0 all four drives on the Velocities , Lou ?

A client of mine is looking into just one Velocity Duo to Stripe , since he needs both decent OS X and Windows bootability . Windows is very finicky when it comes to booting from RAID 0 . I believe it requires a hardware solution .
 
AidenShaw said:
Why do people think that a logic board design is expensive?
Click to expand...

I wonder this myself. If it was cost prohibitive relative to whatever the line takes in, I assume they would kill the line in its entirety.
 
Machines said:
Interesting . But if money or complexity is not an issue in a logic board update, why did not Apple, for instance, use PCIe REV 3 slots in the Mac Pro 2012 "refresh" ? The standards were already released in 2010 . Was Apple just being conservative ?
Click to expand...

I think that you answered your question yourself when you put the quotation marks around refresh.

The 4,1 and 5,1 are the same mobo.

Apple was being cheap, and just milking the cow.
 
thekev said:
I wonder this myself. If it was cost prohibitive relative to whatever the line takes in, I assume they would kill the line in its entirety.
Click to expand...

In the scheme of things, the motherboard redesign itself isn't expensive, but QA'ing the design takes resources and processes that Apple has been well documented as of late to be challenged in. It's highly likely there will be a new motherboard this year, but it's likely that they're not doing a yearly rev of the motherboard. They don't even do that on their higher volume Macs, let alone a line that's sold only reportedly 1.2M units since it's inception. So, my prediction is they're waiting as long as possible so that they can get everything into this hypothetical 2015 motherboard so that they then don't have to do a subsequent rev till 2017. Now, of course, if you're working on the hypothesis that they'll be returning back to the tower by that time, then maybe they wouldn't rev this model's motherboard at all. Of course, all of this is speculation which is why everyone has their own pet theories and there are now two threads on this topic in the forum.
 
AidenShaw said:
I think that you answered your question yourself when you put the quotation marks around refresh.

The 4,1 and 5,1 are the same mobo.

Apple was being cheap, and just milking the cow.
Click to expand...

the 2009 , 2010-11 were the same motherboard for all of the workstation vendors. They all do it. For the tick-tock cycle of the Xeon E5 class ( and previous equivalents) Intel only does one chipset and the same socket. There really isn't much to radically change.

Apple didn't do anything in 2012 more so because of two reasons.

a. probably weren't doing anything in late 2010 - 2011 on upgrades at all motherboard or not. ( note that the 2012 Mac Pro were boat anchored with the same GPUs from 2010. ).

b. If they had "fixed" the legacy Mac Pro design to be up to code for a new Xeon E5 v1/v2 tick tock cycle there wouldn't be room to bring in a "new vision" Mac Pro later until the next tick-tock cycle ( v3/v4 Haswell/Broadwell which is now. ). Apple may have gobs of money in their hedge fund but they only have a limited set of engineers. They just don't have redundant teams on single products. In some cases don't even have some specialites assigned to a single product at all ( they are stretched over multiple products producing even more resource constraint choke points. Ive's design team is a huge choke point. )


Apple didn't do anything in 2012 because "bet the farm" on 2014 era parts ( Thunderbolt v2 and E5 v2 ). It probably also took time to get the Mac Pro R&D restarted and moving again.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back