"Industry Standard Architecture (ISA) is a retronym term for the 16-bit internal bus of IBM PC/AT and similar computers based on the Intel 80286 and its immediate successors during the 1980s."
Will Apple really use a parallel 16-bit bus for the mMP?
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.
The new Mac Pro - What are your biggest fears?
- Thread starter iBrooker
- Start date
- Sort by reaction score
"Industry Standard Architecture (ISA) is a retronym term for the 16-bit internal bus of IBM PC/AT and similar computers based on the Intel 80286 and its immediate successors during the 1980s."
Will Apple really use a parallel 16-bit bus for the mMP?
hopefully it means apple isn't going to make a hacintosh
[doublepost=1492368865][/doublepost]
i was linked to this guy elsewhere on these forums.. entertaining enough i suppose.
but here:
(at 4:50)
he's claiming the same person who designed the nMP also did the mbp touchbar.. and that this person is no longer working at apple.. which points to c.stringer who curiously enough, announced departure the day before the schiller/Federighi interview..
https://www.macrumors.com/2017/04/03/designer-christopher-stringer-leaving-apple/
so either Stringer is being thrown under the bus or there's truth behind this -or- it's just coincidence and/or me being dramatic ; )
----
but fwiw.. i really like the nMP design and decently like the touchbar idea.. both are nice and highly skilled work.. just maybe not well suited for mass consumption.
Last edited:
"Industry Standard Architecture (ISA) is a retronym term for the 16-bit internal bus of IBM PC/AT and similar computers based on the Intel 80286 and its immediate successors during the 1980s."
Will Apple really use a parallel 16-bit bus for the mMP?
I name ISA PCIe, not just ISA (the pre PCI interface), while you're right about the name, it's AWG-PCIe (the actual name for the PCIe slot) how many time 'll need to explain AWG means Arapaho Work Group (HP/IBM/DELL/Intel partnership behind the industry standard, not ISO neither IEEE std)
[doublepost=1492376313][/doublepost]
hopefully it means apple isn't going to make a hacintosh
[doublepost=1492368865][/doublepost]...really like the nMP design and decently like the touchbar idea.. both are nice and highly skilled work.. just maybe not well suited for mass consumption.
the nMP (or tcMP) wrongs where about TDP and thermal core flexibility, from there its an great achievement as very few PCIe cards are actually needed in mac ecosystem (storage/gpu dont account, since are supposed to be provided by the system configuration), and Apple was looking to promote Thunderbolt, cant compete with thunderbolt allowing PCIe slots, same situation now, but they are aware the lack of flexibility the thermal core imposed, and need to offer single/dual GPU options in different power categories, Apple releasing a PCIe slot in the mMP means give up about thunderbolt and its not a good message for its partners developing TB3 peripherals, unless TB3 becomes a real bottleneck I don't believe Apple would include AWG-PCIe slots in the mMP, its too naive to consider that 4 years before the last cMP was sold.
Last edited:
maybe they'll do a breakout box of some sort and get macOS to a plug&play level via tb3?.. also, maybe the new display will have a GPU in there?
i personally don't think apple is going to do a big all-encompassing box which could house 4+ gpus..
i think, in a very small amount of space, they could house 4+ SSDs, 1 or 2 CPUs, 4-8 rams slots, and a single GPU (or even 2).. and create within itself a powerful little computer with more thermal headroom.. maybe ditch the custom gpu board/cooling and instead, utilize the heatsink&fan(s) from the factory.. then they only need to cool the CPU/PSU/ssds/etc.. nMP might of actually of handled this type of situation very well had it not needed to also cool the GPUs ?
then manufacture/sell Mac GPU boxes.. which will not only work with mMP, but all the other macs too.. because it's not like the nMP is the only mac which super-power-users complain about lack of GPU power/expandability.. it's also the mbp and imac.. so a solution that more-or-less solves the dilemma across the board might be more viable than dealing with each model individually.
?
(just guesses- in case that's not obvious)
ISA its shorter, and only very old dogs recognize the actual name of the very first PC-IBM used exansion slot (also shared with other platforms)
ISA: Instruction Set Architecture
ISA: old PC slot
I don't know what else it would mean in computing.
I fear that Apple has lost the will to come up with the "fastest computer in the world".
Steve did it, every time. Whether it was an apple vs oranges comparison (IBM PowerPC vs Intel Pentium) or with the Intel cMP. The aim was to deliver power.
Please, don't start the new design from the looks! We want SUBSTANCE... and some good one. I fear that we may go again with a mid range Pro machine, with watered down CPUs and GPUs. No way.
Steve did it, every time. Whether it was an apple vs oranges comparison (IBM PowerPC vs Intel Pentium) or with the Intel cMP. The aim was to deliver power.
Please, don't start the new design from the looks! We want SUBSTANCE... and some good one. I fear that we may go again with a mid range Pro machine, with watered down CPUs and GPUs. No way.
well, the blurb we've been told so far (from p.Schiller):
"we want to architect it so that we can keep it fresh with regular improvements, and we’re committed to making it our highest-end, high-throughput desktop system, designed for our demanding pro customers."
..doesn't necessarily instill confidence that they're aiming for the title of 'world's fastest computer'.. however, if they keep the CPU/GPUs up to date ('keep it fresh with regular improvements'?), then it should certainly be competitive in the arena you speak of.
i guess one problem is that 'world's fastest computer' isn't so clear today as it was 10+ years ago.. for my workload, i'd argue a 48core computer isn't going to allow me to work any faster than a 6-core computer.. and that in most instances, the 6-core will probably be faster.
but at the same time, it does seem apple has recognized the iMac form can accommodate the type of speed which my workflow benefits from and that they're going to do 'even more' with the imac to reach this realization.
so, maybe they will actually build a mac pro that's capable of outputting the highest geekBench score via lots of cpu cores? which to a lot of people (and some workflows) will be the type of measurement determinant of how fast the computer is.
i imagine the iMac will fill this slot. especially if apple begins marketing an 'iMac Pro'.. we'll probably see an updated imac before the macPro so i guess when that happens, it should be an okay indicator for how much more powerful the base mac pro will need to be in order to not directly compete against the imac. (because, i think when that happens, most people end up going with the imac instead)
Some of the above has been hashed out already ... but to toss in 2c as well as to focus on the second half of this.
Point the first ...
In reading (& learning) about the film/video industry and their high degree of reliance on networked data storage, I think a primary point here is high performance network interfaces which no longer is Gigabit Ethernet. As such, the Trash Can was woefully underequipped for this on its launch date, even if pedantically one could sell a young child to get a TB-to-FibreChannel (8 or higher) adapter.
Lesson learned here for the next Mac Pro is that there needs to be a clear solution for how it is going to have a high performance network interface ... in no small part because when that connection is slow, then it drives up the needs for high performance local storage as the alternative solution.
Point the second ...
Apple has never really played well in the large Corporate Environment ... and there's not any clear signs that Apple now intends to. As such, use cases such as the above are more likely to remain "cobbled together" (such as through a TB-FC adapter) rather than baked into the basic design. Maybe we'll see 10GB Ethernet as a step up, but all of these circle back to how much including this level of networking capability is going to cost (and drive up the minimum product MSRP).
In any event, I think that there's something slightly larger being overlooked here, which is that Pro's who run their own Small Business has been a quiet staple for Apple, and for better & worse, as "One Man Shops" (either literally or figuratively), there's not the compelling business collaboration environment need to drop $10K or more for a high performance networked file server when a couple of TB's of internal SSD's functionally accomplishes the same workflow needs at a fraction of the cost & complexity.
To a certain degree, a daisy-chain of external Thunderbolt SSD drives does suffice for this customer base needs, but the cost which was incurred by making that external TB box the **only** solution was a huge hit to the Apple = Value paradigm. I've personally priced out some stand-alone (off-network) workstations for this use case and found that what originally cost ~$5500/seat for a cheese grater stayed the same going to Windows, but jumped to $7500/seat for a Trash Can (with no real growth in data management capability), which translates to a +35% "Trash Can Tax".
Understood, but there's always going to be something better around the corner, and since the core issue with the Mac Pro has been that it hasn't been maintained, there's really not a strong justification for Apple to delay a new MP for months while they "wait for" AMD Vega: what is needed is a system whose architectural design will be able to integrate that enhancement once it does come to market ... and which WILL BE supported by Apple, too.
Since the topic asks for fears, I'd fear that Apple would drop in a cheap 10GbE card - rather than a converged adapter with offloads for TCP, iSCSI and FCoE.
A cheap card that is a CPU hog isn't going to meet requirements for many users - these puppies can pump 1.25 GigaBytes per second in and out simultaneously. The CPU needs help for that.
10 GbE cards new run the range from about $60 to $1600. The cheap ones are, well, cheap. Once you hit about $300 you see decent offload engines for TCP and iSCSI.
Edit: Add missing decimal point.
Last edited:
Agree, but I convert from my faith to whatever is the FSM if apple comes with something better than intel X550.
I am not convinced Apple is talking about reverting back to a "cMP-style user-accessible everything" when they say modular design.
From the transcript, it seems to me they are talking about modular design in the sense that they (Apple) can easily update or replace its internals. Not necessarily something that end-users can do themselves.
But I could be wrong, of course.
From the transcript, it seems to me they are talking about modular design in the sense that they (Apple) can easily update or replace its internals. Not necessarily something that end-users can do themselves.
But I could be wrong, of course.
they did say something to the effect of-- since the computer will be modular by definition, we will also be making a new display to go with it.
?
idk, while i welcome apple giving somewhat of a roadmap, i still think they were being more vague than necessary.. especially with the use of the word 'modular'
The X550 is a pretty decent chip - I could live with that (and do have some for the 3rd and 4th 10 GbE ports in some systems).
My fear would be something like the Marvell-Alaska-X-88X201x-PHY .
Last edited:
Literarily one man shop, maybe, but even a few people their is a need for collaborative and shared media. For small file types, AI, PSD, Word etc.. storage by and large has been taken over by the CLOUD and other online storage options, BUT a lot of studios do not allow CLOUD storage or CLOUD rendering.. Any studio with the legal means will not allow CLOUD anything.. You will get sued if caught. All media and assets have to be on a local network and all rendering local as well. Your security has to be verified as well, most of the time... But for most intensive media tasks, TV/FILM/VFX the CLOUD is still too slow and un-usable, I know some studios that take their legal chances and render via AWS, and if your rendering and copying locally at the same time, it can work. BUT by and large lots of local data, LOTS, shared and rendered via a shared network.
If Apple goes VEGA we are all screwed, first they must have INTEL for Thunderbolt... Second if they decide to VEGA and want thunderbolt we will need extra chips and controllers, wasting time, money and space. Apple is INTEL only for at least the next 10/15 years.
But we also need to take into consideration of the rest of the "chat", where they went into specifics of how they see the nMP as a failure. One would assume, if Apple weren't as dumb as it's usually implied here, they ought to take the most effective changes to accommodate the "lost Pros" who were well served by the Cheesegrater design.they did say something to the effect of-- since the computer will be modular by definition, we will also be making a new display to go with it.
?
idk, while i welcome apple giving somewhat of a roadmap, i still think they were being more vague than necessary.. especially with the use of the word 'modular'
As a result, Apple needs a very strong reason to not go back to a box design with a few PCI card slots. If they try something clever, which I think they may, like an externally modular stack of Minis for example, these components will need to excel as standalone devices working with other Apple products in the ecosystem. A display with a hub is one of these which they had done in the past, and other modules can be like TB3 equipped hard drive with WiFi / Time Capsule which can be used by Mac with USB-C and iDevices concurrently. The difficulty is to make it a very Apple design that fits well into their current lifestyle roadmap, while at the same time deliver performance that is needed int their targeted "Pros", whoever they are.
I thought I was done with this. But, I have some witty comments to add!
So, you're saying that this hypothetical film/video industrial company has spent money on high speed fiber-optic network and are thinking of getting 2013 nMP's. But, the extra cost of Thunderbolt-FiberChannel Adapters are cost prohibitive? Figuratively, it would be like the price of a young child?
So, you're saying that Apple should have learned that Dual Gigabit Ethernet, Thunderbolt 2, etc. will be slower in 2017 than it would have bee in 2013?
So, you're going back to that whole adding high speed network to 2013 nMP is going to cost like the price of a young child, figuratively-speaking, again.
Okay. A small business is not, technically, a One Man Shop. A business relies on like, more than one person. So, a $2000 trashcan tax as you say--if, this business really wants to use the trashcan--should be like adding a couple of fresh, young potatoes to the dish, figuratively-speaking.
The waiting thing you replied to on the above-quote is not like you or me waiting for Nvidia Volta or AMD Vega, which you make it sound like by saying "there's always going to be something better around the corner." Do you think Apple, a freaking multibillion dollar company with smart people working for them doesn't know this?
So, you're saying that this hypothetical film/video industrial company has spent money on high speed fiber-optic network and are thinking of getting 2013 nMP's. But, the extra cost of Thunderbolt-FiberChannel Adapters are cost prohibitive? Figuratively, it would be like the price of a young child?
So, you're saying that Apple should have learned that Dual Gigabit Ethernet, Thunderbolt 2, etc. will be slower in 2017 than it would have bee in 2013?
So, you're going back to that whole adding high speed network to 2013 nMP is going to cost like the price of a young child, figuratively-speaking, again.
Okay. A small business is not, technically, a One Man Shop. A business relies on like, more than one person. So, a $2000 trashcan tax as you say--if, this business really wants to use the trashcan--should be like adding a couple of fresh, young potatoes to the dish, figuratively-speaking.
The waiting thing you replied to on the above-quote is not like you or me waiting for Nvidia Volta or AMD Vega, which you make it sound like by saying "there's always going to be something better around the corner." Do you think Apple, a freaking multibillion dollar company with smart people working for them doesn't know this?
It is the risk of "One More Straw" on the Camel's back (or nail in coffin ... your choice of analogies).
When I first priced FC8+higher interfaces back in 2013, they were adding IIRC roughly another +15% to the cost of each seat that was to be so equipped. This isn't trivial, so it functionally was an additional limit on the market for adoption.
Incorrect interpretation.
1GB Ethernet (even as a dual) was already a trailing technology in 2013 - - and this was a point of hardware discussions back in 2013. When the design concept is to minimize/obviate local data storage capability because its workflow paradigm is for it to be networked for accessing its data, then its networking connection shouldn't suck (or, as per the above, be a cost barrier to adoption).
Integral 10GB Ethernet was a technology option in 2013. Not super-cheap, true, but Apple's economies of scale would have made it reasonable, and offering this may have been reasonably forward looking enough for this data on the network paradigm. But the history as we now know it is that Apple instead chose to stick with the 1GB Ethernet (which had dropped to a cheap commodity component by this time too).
No, this is looking at a different use case, and how the nMP wasn't a good fit for him either.
Not because of the lack of a high performance network port, but because it wasn't able to as readily have its data storage localized (lack of drive bays).
Not really: the question is to what degree there is a workflow capability requirement for collaboration on the same dataset where high performance is concurrently required.
For example, when a small shop only has one primary editor, he doesn't "share" his dataset with anyone while he's working on it - - that's why they don't materially benefit from having a high performance data server with wicked fast/fat connects. Their workflow is going to be more individualistic-isolated, where one person works on their thing, then "throws it over the wall" to the next specialist type who takes their finished product as their input and does some different workflow on it (eg, burn it to DVD, or whatever).
And for such a SMB, the TB Tax was a motivator to switch to Windows to solve his workflow needs.
What's the evidence suggest?
The fact of the matter is that because of how the Trash Can went 3+ years with nothing, they're NOT doing a good job thinking (& designing) strategically.
And quite frankly, I don't need to care if the reason is because of lack of talent ... or because corporate culture ignored the advice of talented people: either way, they very clearly failed.
And FWIW, this is something that terrifies me as an Apple Stockholder: are they falling into the US Corporate paradigm trap where their leadership's vision is blindered to only "Next Quarter's Profits"? Given the rumors about how much attention the 'Space Ship' (and Car) is dominating their attention, Apple is not skating to where the puck is going to be ... except for two guys working on the iPhone 8, they've taken their hockey skates off and are down at the pub having a latte.
I'm only gonna, ummm... reply to this part because it's the most interesting, objectively-speaking.
I don't have evidence, perse. But, I have guesses....
Guess #1) AMD
Guess #2) Could
Guess #3) Not
Guess #4) Deliver
Guess #5) A GPU
Guess #6) Appropriate
Guess #7) For the trashcan Mac Pro
Guess #8) And, Nvidia
Guess #9) Apple is not interested in
Guess #10) For whatever reason
Guess #11) Maybe, something to do with CUDA
Guess #12) Maybe, something to do with Greedvidia
Guess #13) But, Novidia will be going in the trashcan Mac Pro
Guesspinion #14) Probably, the right reason
Guesspinion #15) Don't buckle under their greedy grip
Guesspinion #16) And, so on and so forth
Last edited:
Guesspinion #17) Apple ignores what Apple's customers are asking for, and continues to fight to right Steve Job's imaginary slights.
"Industry Standard Architecture (ISA) is a retronym term for the 16-bit internal bus of IBM PC/AT and similar computers based on the Intel 80286 and its immediate successors during the 1980s."
Will Apple really use a parallel 16-bit bus for the mMP?
it'd at least have a huge amount more cards available than the trash can!
[doublepost=1492489210][/doublepost]
Apple is definitely not above allowing 3rd party standard PC video cards and calling it a "revolutionary UNHEARD OF idea." Maybe they can have bootscreens with standard EFI and that'll be the big "achievement." Can we think of a reality distortion field name for it ? Fusion cards? iUbuquity technology?
I do think by "modular" they do mean easily upgradeable, probably by the user. Maybe flat five will get his wish and they'll make proprietary upgrade cards OR a daughter board with the CPU slotted and the GPU soldered on there -- sharing a heat sink maybe? That idea's just bad enough to be the ticket!
