don't you think two years is an awful lot of time to decide if you should buy a mac pro or not?
like- pleNty of time
?
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.
Will there be a new Mac Pro in 2015?
- Thread starter Anto38x
- Start date
- Sort by reaction score
- Status
don't you think two years is an awful lot of time to decide if you should buy a mac pro or not?
like- pleNty of time
?
No. Because the time of the need for a new workstation is in a lot of cases unknown.
i.e. You may have a serious failure in your current/old MP and you need a replacement right now to have your work done.
http://ir.amd.com/phoenix.zhtml?c=74093&p=RssLanding&cat=news&id=2083146
OpenCL drivers for OS X, new API for RayTracing in OpenCL for OS X, new drivers for OS X, Metal that is based on Mantle driver(like every other API these days). And now this. Looks like things have changed a bit lately
. Multi GPU is being pushed as hell everywhere it is possible.
http://forums.anandtech.com/showpost.php?p=37669975&postcount=14
Well, I think it is only the beginning
And I don't know why but I have a feeling that Nano, will become a FirePro GPU in new Mac Pro offering with all of these capabilities.
OpenCL drivers for OS X, new API for RayTracing in OpenCL for OS X, new drivers for OS X, Metal that is based on Mantle driver(like every other API these days). And now this. Looks like things have changed a bit lately
. Multi GPU is being pushed as hell everywhere it is possible.http://forums.anandtech.com/showpost.php?p=37669975&postcount=14
Well, I think it is only the beginning
And I don't know why but I have a feeling that Nano, will become a FirePro GPU in new Mac Pro offering with all of these capabilities
.
Last edited:
i don't think i follow what you're saying.
if your computer explodes and you need a new one right now, seems you need to make a decision within hours.. not years.
?
I'm sorry but you 're the one talking about years. In the text you have quoted from tomvos there is not anything about a decision taking years.
If I have missed something please inform me.
If I have missed something please inform me.
tomvos post stems from:
https://forums.macrumors.com/thread...mac-pro-in-2015.1840458/page-78#post-21759873
it's been years since the nmp has been out there.. and years of people posting thousands of gripes with the nmp.
do you think they're making buying decisions? or something else?
I still have my 5.1 cMP, so unless there is a very expensive component failure, I have some time to make up my mind.
I think we all agree one of the major questions everybody asked about the nMP was: How to I extend and customize this system? Apples answer was: Use Thunderbolt. Many folks said: Well, thats expensive and comes with some technical limitations.
What I and many other in this thread like to know is if Apple decides to target the nMP to more people than just the film and video crowd. Apple does not need to abandon the nMP concept. Personally, I'd like to see the following changes:
- SSD connected with standard M.2 connectors and perhaps have two of these internal connectors
- Offer a mixed GPU combination: one D300 and one D700.
This would not compromise the original idea of the nMP too much. And it would show that Apple acknowledges the requirements of the non-video-film crowd out there.
But unless I see the 2nd generation nMP my opinion about the nMP is … still forming.
Last edited:
I don't think that Apple will ever support m.2 connectors. It's not their intention/policy to support standard/non proprietary connectors. They like to have the ultimate control of the hardware they sell.
What are the benefits of this mixed GPU environment you are suggesting?
Well, they did sell PowerMac G3 with standard ATA, PowerMac G4 with standard ATA, PowerMac G5 with standard SATA and MacPro with standard SATA. Why not switch to standard M.2 connectors? Up to now nobody could explain to me what's the benefit of the propiertory connector (except that M.2 was perhaps not ready when the nMP was designed - so even more reason to switch to an open standard).
About the GPUs: Primarily it's about costs. See, at the moment Apple sells you a Ferrari. But only as long as you buy two of them. For the price of two. Even if you can only drive one at a time.
On the nMP there is no crossfire. Most applications and games use only one of the GPUs. There are a few applications which use one GPU for OpenGL and the other GPU for OpenCL. Mostly I figure these are some dedicated video/film applications - hence I said the nMP is targeted to the film and video crowd.
From my perpective, the dual GPU setup is mainly useless. But I would like to have one fast GPU. This would be sufficient for my needs. I would like the option to minimize the cost of the second GPU.
Last edited:
For me I would be happy if there would be a Mac Pro without GPU's. I just need to drive displays I have nothing which can be faster with GPU's. (Well a small GPU to drive the displays since the xeons dont have on chip graphics)
I would really like if the Mac Pro was just a CPU RAM and a SSD in a box. Max out the Thunderbolt interfaces with all the pcie lanes available and make eGPU a thing with Thunderbolt 3. Now everyone can put his system together how he wants and I dont care if this stuff is external and sits under the desk. Storage is networked who really cares about 10 TB storage in the computer itself. Fast scratch disks okay but the bulk storage can live elsewhere and there may be the better place for it (redundancy)
Edit: And give some pcie lanes to 10Gbit ethernet. With 10gbit in the box. We are waiting so long that 10gbe is getting down in price and with this "pro" machine can we please start to put this stuff in that people think its useful. 10gbe Cards got cheaper the last years, there are 1000 bucks 10gbe switches we are getting there. Then we will have really cool networked storage.
I would really like if the Mac Pro was just a CPU RAM and a SSD in a box. Max out the Thunderbolt interfaces with all the pcie lanes available and make eGPU a thing with Thunderbolt 3. Now everyone can put his system together how he wants and I dont care if this stuff is external and sits under the desk. Storage is networked who really cares about 10 TB storage in the computer itself. Fast scratch disks okay but the bulk storage can live elsewhere and there may be the better place for it (redundancy)
Edit: And give some pcie lanes to 10Gbit ethernet. With 10gbit in the box. We are waiting so long that 10gbe is getting down in price and with this "pro" machine can we please start to put this stuff in that people think its useful. 10gbe Cards got cheaper the last years, there are 1000 bucks 10gbe switches we are getting there. Then we will have really cool networked storage.
At first I have to inform you that I agree completely with your thoughts. They 're reasonable.
The real problem is that they always wanted/liked to have proprietary connectors if they had the option.
remember ADB, ADC, mini dvi, micro dvi, even mini displayport etc.
About the PowerMac G3 (the beige ones, I believe) you mentioned... every single computer of its time was capable of having two internal IDE HDs connected with the same ATA cable. The Powermac not ... only one HD, it was freezing if you tried to connect a second one.
The last years they have moved to a even closed ecosystem with the glued batteries in the MacBooks, the glued screens, the connectors of the SSDs, in both portable and desktop systems, the non upgradeable GPUs, the soldered RAM, etc...
Now about the proprietary SSD connector.
(edited/removed the: Imho the only benefit is for Apple, you depend from Apple to upgrade)
Imho this is a Apple, they 're pioneers and innovation leaders and they needed this connector because they had to accommodate the faster SSDs, the downside of course is that you depend from Apple to upgrade (if possible/available), and now that this connector is shared between their systems, I don't think that they 're going to change it for a standard one.
I agree that a nMP with a single GPU should exist, for the less cost and /or better flexibility , i.e. instead of the cost of the second GPU you could have more RAM, larger SSD, or better CPU, things that many people would prefer.
With the GPUs we have always wanted something more.Remember the lack of officially supported upgrade GPUs for all of their systems in time, from PowerMacs to cMPs.
So, after all, even if the closed system case is the reason that we have systems with better stability, we have a downside, at least in present time, because of the less customize/upgrade options.
Last edited:
I think that you 're describing a super mac mini.
Imho I don't think that there are enough PCIe lanes to give 10 of them to networking.
I think that you 're describing a super mac mini.
Imho I don't think that there are enough PCIe lanes to give 10 of them to networking.
Well kinda yes. It should still be able to power 3 4k displays (which to be honest isnt too much to ask for for a graphic chip today) and come with the better processors. I need CPU cycles and cores more than GPU's
And more than a mini is currently capable (Well more cores than the 2012 mini lets say)I dont know if the new skylake Xeons (and broadwell) will come with more lanes than before but the i7s come with a couple more I think.
A dual 10gbe Card comes normally in a x8 lane form factor that means it would need 8 lanes. If you drop the GPU's you have more lanes available, you could add them to more Thunderbolt ports and still use 8 of them for 10gbe. USB 3.1 will still need a couple of lanes because I dont think its going to be in the workstation chipset but then there isnt much stuff in the Mac Pro which would need lanes except the SSD
Remember, Apple doesn't adopt "proprietary" connectors for the sake of them being proprietary, they enhance the functionality of the product. Those display adapters exist because nothing else would serve the functionality Apple wanted at the time. In the case of the SSD connector, there was no M2 connector when Apple started using SSDs in its machine. Rather than leave out the SSD, Apple chose to use its own connector.
This is a unique capability that Apple can offer because they offer a full product (hardware + software). Other PC manufacturers have decided its not worth the added cost of adding this functionality and wait around for the whole industry to implement the standard. When Phil Schiller claims that no other company can deliver something like the Macbook Pro, iMac 5k or Mac Pro, he is right. No one developed a computer with a retina screen because they couldn't force microsoft to implement good resolution scaling. No one introduced an iMac 5k because they didn't want to develop their own internal display connector. No one else can produce a machine like the Mac Pro because of Apple's ecosystem of thunderbolt peripherals (among other things).
Love it or hate it, this is what Apple prides itself in. Designing systems that other companies are not capable of and leveraging their unique capabilities. Of course, with any design decision are trade offs, and Apple has decided that using non-standard components makes their products better, despite limiting configurability and upgradeability.
Stacc, I fully understand these facts, we (the old mac users) have got used to Apple's continuous inventions, it became a habit and something common to us, to think that Apple is the natural innovation leader.
The display adapters were necessary because they wanted the ports to take less space of the continuously shrinking laptop shell (and mac mini's afterwards). By the way I have a drawer full of them.
The proprietary stuff is something needed if you have to move forward. Someone has to make the first steps, to lead the way. This is Apple, and I certainly don't hate it for this. In fact I am grateful for the existence of their systems because they are more helpful/convenient to do my intended works.
Yes their products are first class and they 're expensive, they 're also innovative and have the needed proprietary stuff but as you wrote there is always and the other side, another point of view, which is good to know about.
Now, I would like to have your opinion,
, do you think that now that there is availability of the m.2, Apple is going to use it?Will they switch to a standard connector?
Last edited:
Apple will eventually use in the Pro whatever they're using in the laptop product line. Doesn't look like any of Apple's 2015 models are using m.2 (although some reviews claimed they were but I think they're mistaken, the parts show in the teardowns don't look like standard m.2 connectors), but given the pace of laptop refreshes, if Apple decides to go to m.2, we'll see it there first. I really don't think this is a religious issue for Apple, it's whatever gets them to the form factor they want first, with supply chain cost concerns afterwards. I could be wrong but I don't think the m.2 modules have any size or speed differences anymore.
Now, I would like to have your opinion,
Will they switch to a standard connector?
Looking at available M2 drives, it seems that the form factor is shorter than what is used in Apple products. This limits their capacity to 512 GB, less than the 1 TB options Apple offers. Until the capacities and cost change significantly, I think we will see the Apple connector. In the volumes that Apple is ordering for their SSDs, the added cost of the proprietary connector probably doesn't matter. One thing that may help is the Skylake/Z170 platform has native M2 support, which is probably some encouragement for Apple to switch.
Thank you very much for taking the time.
I appreciate your analysis and it's really nice to discuss with people like you.
I wouldn't count on M.2 in Apple machines, they don't follow others or standards, they make standards.
They expect other to follow them instead, and move on to other proprietary stuff.
And note that I'm not saying this with any kind of resent, much on the contrary - this is what I love about Apple.
10GbE will be available with TB3, as well as USB3.1 so we're not that far off.
Still, you need the lanes for the controllers of course.
Different GPUs seems far fetched, will probably never happen. There will be a time when they'll both be used to the max and we'll thanks Apple for having them both there
They expect other to follow them instead, and move on to other proprietary stuff.
And note that I'm not saying this with any kind of resent, much on the contrary - this is what I love about Apple.
10GbE will be available with TB3, as well as USB3.1 so we're not that far off.
Still, you need the lanes for the controllers of course.
Different GPUs seems far fetched, will probably never happen. There will be a time when they'll both be used to the max and we'll thanks Apple for having them both there
GDDR5 still has potential to grow it seems. Micron is readying 8Gb modules and preparing a new standard GDDR5X for next year, faster of course. Will it survive HBM?
Nvidia is preparing Pascal lineup. Only the top end GPUs will have HBM2 while the rest of will use GDDR5.
AMD readies top-to-bottom HBM2 setup.
P.S. I would not count on improved Asynchronous Compute capabilities in Nvidia Pascal GPUs.
DirectX 12.1 did not came to life for no reason This is all I can tell right now.
AMD readies top-to-bottom HBM2 setup.
P.S. I would not count on improved Asynchronous Compute capabilities in Nvidia Pascal GPUs.
DirectX 12.1 did not came to life for no reason
This is all I can tell right now.Different GPUs seems far fetched, will probably never happen. There will be a time when they'll both be used to the max and we'll thanks Apple for having them both there
The related to GPUs news are coming the one after the other, also metal and the upcoming driver releases will make us really thankful, or so I hope, for our dual GPUs, current and upcoming ones.
I guess that someone with better knowledge, koyoot?, will have a really useful opinion about this.
If you're looking at the R9 Nano and thinking "wow, that seems almost tailor-made for the nMP" you've basically got the causality here almost completely backwards. The nMP was designed with the physically-compact, probably on-package next-generation memory technologies like HBM and HMC in mind; these have been on the roadmap since before the last cMP shipped, and their implications for the physical design of a higher-end computing system were -- and are -- easily foreseeable.
These types of components finally starting to ship for GPUs, and will start happening for CPUs soon; ironically a future nMP may wind up the first ARM-based Mac, if Apple wants to accelerate that transition. Mainstream GPUs will have their RAM on-package, mainstream CPUs will have their RAM on-package, high-end SSDs will take over any remaining DIMM-style interfaces. Tradeoffs will be involved -- your RAM will be non-upgradable and of capacity like today, but substantially faster -- and overall the nMP will be tuned for responsiveness in interactive use, not for pure compute/throughput (go call HP/Dell/Boxx/Supermicro/etc if you just need racks of cores / racks of Teslas / etc).
The nMP is thus a very sensible physical design from, let's say, the year 2020, that was inexplicably released several years too early; the types of components that motivated the redesign are rather late even after accounting for the typical productization delays, and at this point may not be widely available until 2018-2019 or so. Until all those components are available it's a machine of unfortunate compromises, and upgrading it isn't a priority.
Expect the next release by the end of the year, but just a spec-bump with minor upgrades across the board; no form-factor changes (including no TB3), and quite possibly a quiet spec bump, e.g. with little-to-no time during any of the remaining stage-shows this year.
These types of components finally starting to ship for GPUs, and will start happening for CPUs soon; ironically a future nMP may wind up the first ARM-based Mac, if Apple wants to accelerate that transition. Mainstream GPUs will have their RAM on-package, mainstream CPUs will have their RAM on-package, high-end SSDs will take over any remaining DIMM-style interfaces. Tradeoffs will be involved -- your RAM will be non-upgradable and of capacity like today, but substantially faster -- and overall the nMP will be tuned for responsiveness in interactive use, not for pure compute/throughput (go call HP/Dell/Boxx/Supermicro/etc if you just need racks of cores / racks of Teslas / etc).
The nMP is thus a very sensible physical design from, let's say, the year 2020, that was inexplicably released several years too early; the types of components that motivated the redesign are rather late even after accounting for the typical productization delays, and at this point may not be widely available until 2018-2019 or so. Until all those components are available it's a machine of unfortunate compromises, and upgrading it isn't a priority.
Expect the next release by the end of the year, but just a spec-bump with minor upgrades across the board; no form-factor changes (including no TB3), and quite possibly a quiet spec bump, e.g. with little-to-no time during any of the remaining stage-shows this year.
- Status