Was it a promise? A statement?
All of what you written is correct.
And the more it shows brilliance of this design.
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.
MP 7,1 Waiting for Mac Pro 7,1 A1991 (no more)
- Thread starter Mago
- Start date
-
- Tags
- macpro 2016
- Sort by reaction score
And the more it shows brilliance of this design.
I said, that I was not supposed to come to this forum, ergo: "I did not wanted to post here". But I came.
Deal with it
.
Any MCM or multi-socket solution adds latencies for the comm links between the components - especially if the comm fabric is cache-coherent. The first two letters of NUMA mean "non-uniform" - some memory is faster, some is slower. In other words, steps backwards.
Of course - comm links between chips have always underperformed communications on a die.You judge it before seeing it?
Always.
Have you seen it?What I mean: you judge performance of the CPU, before you saw it?
That idea, that Apple has "acknowledged" what was "wrong" with the 2013, is exactly the sort of memory-holing rewrite of history, exactly the same gaslighting of customer concerns and feature requests, that got us here in the first place.
The problem with the 2013 Mac Pro was that its graphic cards were not standard, off the shelf, pci graphics cards. That is the problem, and Apple has never acknowledged it. Everything else Apple blamed, every goal that allegedly "necessitated" that solution, is a symptom of that decision, and of that problem.
- Overheating GPUs? Standard cards would have mandated an enclosure in which cards had room, and power, to cool themselves.
- Unbalanced cooling? As above.
- Under-powered, and out of date GPUs? As above.
- AMD being incapable of producing a good (Apple Branded) card? Customers would have voted with their wallets for better, reliable cards, either different brands of AMD, or Nvidia solutions.
This can't be cured by slapping a poultice on it and taking a couple of aspirin. The minds that dreamed this up, this corrupt lie sold to customers (corrupt being the very word Jobs used to describe the profit-maximisers who ran Apple after he left), is a necrotic infection that requires brutal amputations. It requires a wholesale s**tcanning, Papermaster-style, of everyone involved in the decision to design it the way they did, so their ideas are removed from Apple's corporate mind. It needs their reputations hoisted on pikes in the middle of that ridiculous, and insufficient-for-needs temple to self referential, and self reverential onanism they've built, as a warning that corporate ego has no place in the creation of tools.
But of course, that won't happen. The compliant Mac press, will repeat the thermal corner lie, they'll hail in lockstep, that "Apple admitted they were wrong", that the "new Mac Pro" is the "future of computing" - a future whose details, and perpendicularity to everyone else in the Pro workstation space, will be ignored. People with no skin in the "getting good tools" game, but their very livlihoods in the "talking about Apple game", will repeat the lie that iMacs and Macbook Pros are "good enough" for "professional" users, because you can run a glorified text editor on them.
*drink*
...and another thing...
No. And I am in the same situation as you, and I do not say anything about the performance.
Besides, We don't know the physical design of the CPUs. And the physical design can mitigate the latency.
P.S. This design of EPYC means - there is no more NUMA architecture. NUMA is between sockets.
Last edited:
Well said !
Not sure if Apple hasn't gotten too big for any kind of decisive desicion making though, let alone self reflection .
Just imagine the army of middle management drones they must have by now .
Considering themselves machiavellian masterminds as those types do, while pretending to be devout followers of the cult battling the headphone jack and setting up creative hotspots around the world - stores with all the charm of a public toilet , but without the excitment of a possible casual encounter ...
Apple do have to decide on whether or not they want to make tools, or just keep on selling toys and call them tools at their gatherings .
Not just for the nextMP , but across the Mac lineup .
And for OSX, which they've been developing and carelessly changing like a mobile OS for years now .
All this fake new dawn about modularity and workflow analysis seems only to suit Apple itself rather than its customers.
Aside from speculations on what future IEEE standards might bring, external modules based on current technology are in general more expensive than internal similarly specced components because price markups from vendors, the cost of an external enclosure, the cost of supporting electronics, additional power supplies and cables. Just look at the Blackmagic eGPUS... We should also mention the 15%-20% performance degradation that any eGPU suffers because of the T3 bottleneck. It is a very real tax on performance.
Also, when it comes down to upgrades and/or expansions, many mention that some/most pros never tinker with their machines. Those same people however fail to recognize that for serious medium to large size organizations their "pro" community is actually made of system engineers, system administrators and then pro users. If the latter don't even know how to do a hard reset it doesn't mean that the other two "pro" categories don't need maximum freedom in configuring, managing and maintaining hardware on a large scale securely, efficiently and economically.
Open architectures have consistently proven their advantage over closed ones since since the advent of the PC in 1981. I love Apple but it's not down to luck that HP and Dell are the workstation market leaders in an established industry that does not need new "workflow" solutions in search of a problem that isn’t there. Their margins are thin yes but it's to the advantage of keeping open standards and lower prices.
Apple strategy and propaganda arrogantly wanting to teach pros how to do pro stuff seems more dictated by the restrictions they imposed on themselves to keep their architecture closed. Ok, Mac OS still makes the difference but for how long? News of the T2 chips limiting OS choices are a very sad story proving Apple own fears of open systems.
It is not even clear (in revenue terms) if this direction is good for Apple but it is certainly not good for the industry they still claim to cater for: the slow death of the Mac Pro is here for anyone to see.
A workstation is not a wheel that needs disruptive reinventing unless you are a company that needs to fit the square peg of propping up a closed architecture in the round hole of an established industry based on open standards.
Aside from speculations on what future IEEE standards might bring, external modules based on current technology are in general more expensive than internal similarly specced components because price markups from vendors, the cost of an external enclosure, the cost of supporting electronics, additional power supplies and cables. Just look at the Blackmagic eGPUS... We should also mention the 15%-20% performance degradation that any eGPU suffers because of the T3 bottleneck. It is a very real tax on performance.
Also, when it comes down to upgrades and/or expansions, many mention that some/most pros never tinker with their machines. Those same people however fail to recognize that for serious medium to large size organizations their "pro" community is actually made of system engineers, system administrators and then pro users. If the latter don't even know how to do a hard reset it doesn't mean that the other two "pro" categories don't need maximum freedom in configuring, managing and maintaining hardware on a large scale securely, efficiently and economically.
Open architectures have consistently proven their advantage over closed ones since since the advent of the PC in 1981. I love Apple but it's not down to luck that HP and Dell are the workstation market leaders in an established industry that does not need new "workflow" solutions in search of a problem that isn’t there. Their margins are thin yes but it's to the advantage of keeping open standards and lower prices.
Apple strategy and propaganda arrogantly wanting to teach pros how to do pro stuff seems more dictated by the restrictions they imposed on themselves to keep their architecture closed. Ok, Mac OS still makes the difference but for how long? News of the T2 chips limiting OS choices are a very sad story proving Apple own fears of open systems.
It is not even clear (in revenue terms) if this direction is good for Apple but it is certainly not good for the industry they still claim to cater for: the slow death of the Mac Pro is here for anyone to see.
A workstation is not a wheel that needs disruptive reinventing unless you are a company that needs to fit the square peg of propping up a closed architecture in the round hole of an established industry based on open standards.
Last edited:
I feel most people have taken Apple's mention of a 'Modular' and 'Upgradeable' Mac pro in the wrong sense.
Sure, the new Mac Pro may very well be upgradeable, but its most probably all proprietary components - ie, if you want to replace your GPU it may be in some unique housing and the only upgrade options are available from Apple - again at a premium price point. This would be good for an end user who simply needs to take out the existing and slide in the new, with nothing in the way of drivers etc to concern themselves over, but it still means the 'Pro' user wont have the option to really customise their system, like the original 2010 Mac pro for example.
So yes, it'll be upgradeable, but expect components to be at a premium price point and users to still be locked into the Apple pricing structure.....
Anyway, considering a year or so has passed since originally announced that they are working on a new system, we've had no 'leaks' or no other rumours at all???
Sure, the new Mac Pro may very well be upgradeable, but its most probably all proprietary components - ie, if you want to replace your GPU it may be in some unique housing and the only upgrade options are available from Apple - again at a premium price point. This would be good for an end user who simply needs to take out the existing and slide in the new, with nothing in the way of drivers etc to concern themselves over, but it still means the 'Pro' user wont have the option to really customise their system, like the original 2010 Mac pro for example.
So yes, it'll be upgradeable, but expect components to be at a premium price point and users to still be locked into the Apple pricing structure.....
Anyway, considering a year or so has passed since originally announced that they are working on a new system, we've had no 'leaks' or no other rumours at all???
Tangent subthread on a "how to fix the trashcan" (thermal management)
This is the start to a good observation.
To distill it to core essentials, there's two basic things to keep in mind, namely (1) heat spreaders and (2) heat rejection.
For (1), there are heat spreader (and heat mover) technologies that are better than what Apple used for the core of the Trash Can, which includes simple stuff like using copper for better Thermal conductivity, as well as more exotic solutions, such as liquid cooling, heat pipes, or even TPG (Thermal Pyrolytic Graphite). However, all these do is to move the heat from its (increasingly high density) source to some other location, where it is more readily rejected overboard, which is step (2).
For heat rejection (step 2), the only venue that PCs have is effectively into the atmosphere (air), so a source-to-air heat exchanger has to exist in some form, and because air has a low heat capacity, as the thermal load goes up, it means that you need to move more & more air over/through your heat exchanger.
Even for a fanless laptop, which tries to use natural convection on a slab exterior surface, that's still a source-to-air heat exchanger.
Again, while there are some interesting heat exchanger technologies ($$) which can offer higher average efficiency values, these are not also without their design trade-offs. For example, metallized open-cell foam is more efficient than traditional "fin" designs, so you can have a smaller heat exchanger ... but the same design attributes which let you shrink the heat exchanger makes it more vulnerable to clogging, so you end up with more stringent air filtration requirements to pay for...ie, a bigger air filter and bigger fan (and bigger power to run it).
The resulting trade space may or may not be worth it, depending on how far you have to move the waste heat before you can finally get it overboard, along with what your space budget limitations are for each stage.
FWIW, I was closing this by saying "...but of course what always screws over the ME who's been given the thermal management design task is a project manager who insists that Jony Ive's aesthetic wishes can't be compromised...", but Apple's engineering group's problems run much deeper than merely thermal management: these are largely been very much self-created problems and span multiple engineering disciplines, such as has been recently illustrated in a wireless mouse that can't be used while its charging, and the multiple tries to fix the MBP keyboard...
...and it looks like that within Apple Leadership, this is the responsibility of Dan Riccio.
This is the start to a good observation.
To distill it to core essentials, there's two basic things to keep in mind, namely (1) heat spreaders and (2) heat rejection.
For (1), there are heat spreader (and heat mover) technologies that are better than what Apple used for the core of the Trash Can, which includes simple stuff like using copper for better Thermal conductivity, as well as more exotic solutions, such as liquid cooling, heat pipes, or even TPG (Thermal Pyrolytic Graphite). However, all these do is to move the heat from its (increasingly high density) source to some other location, where it is more readily rejected overboard, which is step (2).
For heat rejection (step 2), the only venue that PCs have is effectively into the atmosphere (air), so a source-to-air heat exchanger has to exist in some form, and because air has a low heat capacity, as the thermal load goes up, it means that you need to move more & more air over/through your heat exchanger.
Even for a fanless laptop, which tries to use natural convection on a slab exterior surface, that's still a source-to-air heat exchanger.
Again, while there are some interesting heat exchanger technologies ($$) which can offer higher average efficiency values, these are not also without their design trade-offs. For example, metallized open-cell foam is more efficient than traditional "fin" designs, so you can have a smaller heat exchanger ... but the same design attributes which let you shrink the heat exchanger makes it more vulnerable to clogging, so you end up with more stringent air filtration requirements to pay for...ie, a bigger air filter and bigger fan (and bigger power to run it).
The resulting trade space may or may not be worth it, depending on how far you have to move the waste heat before you can finally get it overboard, along with what your space budget limitations are for each stage.
FWIW, I was closing this by saying "...but of course what always screws over the ME who's been given the thermal management design task is a project manager who insists that Jony Ive's aesthetic wishes can't be compromised...", but Apple's engineering group's problems run much deeper than merely thermal management: these are largely been very much self-created problems and span multiple engineering disciplines, such as has been recently illustrated in a wireless mouse that can't be used while its charging, and the multiple tries to fix the MBP keyboard...
...and it looks like that within Apple Leadership, this is the responsibility of Dan Riccio.
Yesterday AMD introduced Vega 20 (the one compute focused) as the Radeon Instinct MI60 and MI50 300W TDP, 32gb ECC hbm2 dual infinite in fabric{enable 4gpu interconnect), faster FP64 than Tesla and Volta, pricing unknown but likely closesy to nVidia Titan than V100
FP64 is 10x faster than RX Vega 64 on iMac pro.
It could be the Mac pro/next iMac pro top GPU option.
FP64 is 10x faster than RX Vega 64 on iMac pro.
It could be the Mac pro/next iMac pro top GPU option.
I think too much is being read into the Mac mini clustering. The MacStadium feature at the product event was simply to demonstrate what you _can_ do with a Mac mini because of its size, not that Apple is shifting to some massive parallel computing paradigm or that Mac Pro will be a "multi-mini" of some kind. MacStadium has been doing Mac mini colocation for a long time now.
Yeah, especially when you wait until four years after you've hit the iceberg to feed in any steering input...
Sure, they don't OWE us anything. But even given the insignificant fraction of income (next to its beloved iOS lineup) now coming from the Mac > Desktop Mac > Pro Desktop Mac customer base, it boggles the mind that they don't put any more effort and intelligence into what they're designing and selling. Even if you make mistakes, you can do things to mitigate and partially correct. But do they? No. They give up for YEARS!
I'm still slack-jawed at the trash can MP. For FOUR YEARS they didn't change anything, not even pricing. In a computer product!!!!! Over time, if there couldn't be any increases in specs, there should have been gradual decreases in price and/or increases in capacity. But nope. They just let it SIT THERE. Even w/ the original $2999 quad-core gone and the $3999 6-core taking its place at $2999, it is a huge insult to "Pro users". Now that another year and a half has gone by w/ no reassurance, product road-map, or further price drops and capability increases, it's even more so.
Yes, way more if we hopped in our Deloreans and went back to 2014. But it's not 2014, it's 2018. Instead of a 1TB HD that might cost us $40, we could buy a 512GB Samsung 970 EVO for under $100. And 8GB of RAM is quite reasonable now too. Low-end consumer CPUs for cheap? Yep. So what does Apple give us for $300/60% more? One fourth the storage, 8GB RAM, and an inexpensive consumer CPU. Nice.
Yep. That it has taken this much longer w/ zero information or reassurance given makes that a near certainty.
Just once, it would be nice if Apple really, truly listened to its customers, rather than taking the hyper-arrogant stance that it knows what its customers want better than they do. What percentage of Mac Pro customers would be thrilled if they could buy a minimally refreshed cheese grater? Or an HP Z workstation that ran macOS? Or just any decent looking tower w/ sockets, slots, and bays that Apple blessed? They could have probably the MOST popular option w/ the LEAST amount of thought...just as long as they GET IT OUT THERE! Sigh...
No. Just showing off a customer that buys and deploys a crapload of minis.
They are there yet. Compressor allows you to distribute encoding jobs across a network of Macs. It's had this capability for probably a dozen years or more.
Could just be posturing, but the attitude they seemed to convey at the MacBook Air and Mac mini event spoke volumes to me. They seemed so triumphant to provide— after 4 years of Mac mini stagnation, and 8 years since any design change on the MB Air— mostly what everybody had been wanting and expecting for years. I'm glad they did, but the reaction for me (and I suspect most) was more like "Okay, thanks...it's about time."
Apple's response to hitting an iceberg and sinking the ship "the problem was, that we didn't hit it hard enough".
The crowd cheering (screaming) in that video had a genuinely unsettling quality to it - got a very strong experience of what non-apple users used to say they felt made Apple too culty. Tim talks like he's a kindergarten teacher, speaking to children... it's getting creepier with each event, especially now that all the other presenters are starting to use the same cadence and voice coaches.
Agreed. Tim's got the charisma of a garden slug during these things. So forced...
If I want to get pedantic, I'd point out that had Titanic just rammed the iceberg, it probably would have survived (most damage would be in the fore and it'd have had enough watertight compartments to maintain bouyancy for much longer.)
If you don't like his style, that's just taste. But this is nonsense. Spend some time in an actual kindergarten.
Ha! Jinx! While I wasn't specifically referencing the Titanic in my initial mention, that exact same thought went through my head about the Titanic being better off ramming the iceberg instead of turning and letting the 'berg eviscerate it down the side...
hah, yes, but technically correct is the most satisfying form of correct, so point granted
In a proper collision, the titanic would have simply decelerated instantly and killed or injured a huge portion of those on board, who would have kept on travelling at the speed the ship was doing, until they collided with something that wasn't.I've seen people reading stories to children, Tim seems a lot closer to that, than to someone speaking to adults.
That or a hamster in a spinning wheel, same difference .
The way Apple has positioned the Macs in the past few years, the MP and MBP in particular, has made the underlying technology a secondary concern , if that .
Actually, I believe Apple can no longer afford to switch to a different CPU architecture; it's not 2006 anymore and they discouraged a lot of crucial 3rd party software and hardware support since then .
A dual AMD Radeon instinct mi60 Mac pro will be 20x faster than the iMac pro at FP64 and 9x and about 4x faster at FP64/ fp32 over dual d700 tcMacPro
CPU performance gains still very modest at about 1.4x for 18 core Xeon W-2195 vs 12 core E5-2697 is just 42% faster at multicore and 23% faster at single core (passmark}
So expect Tim cook or Craig announcement of that amazing trashcan 20x faster than the iMac pro..
CPU performance gains still very modest at about 1.4x for 18 core Xeon W-2195 vs 12 core E5-2697 is just 42% faster at multicore and 23% faster at single core (passmark}
So expect Tim cook or Craig announcement of that amazing trashcan 20x faster than the iMac pro..
