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
At the WWDC 2019 .... 1:27:51 the it talks about RED and AVID "announcing support today". Doesn't explain what that support refers to but it can't be alimony. It has to refer to something in regards to the AfterBurner.
It is support for new features in Metal API and support for Infinity Fabric Link for new MPX GPUs
Maybe 19.11? Maybe they're playing with numbers, since the mMP is A 1991, would be almost a match 
It would be 11.19 in the US, right. Still, close enough.
It would be 11.19 in the US, right. Still, close enough.
Well I'm really hoping for something today. I've been running my Mac Pro 3,1 for nearly ten years! It's been upgraded to the max and has worked nearly flawlessly all this time, but it's really starting to show it's age and I'm desperate for the new MP already.
Nope, it's todayMaybe 19.11? Maybe they're playing with numbers, since the mMP is A 1991, would be almost a match
It would be 11.19 in the US, right. Still, close enough.
[automerge]1573566616[/automerge]
Maybe an updated iMac Pro besides the all New MBP 16 and the almighty new cheesegrater is announced today past 1pm NYT
Last edited:
Sleepy Hollow, NY sounds like a good place to intro the new "Rip van Winkle" Mac Pro.
Only intrigued about afterburner/Vega ii final upgrade cost, the rumours about an iMac Pro update totally unexpected to me, as I consider it a single generation product
Like the smoothly rolled out $999 monitor stand?
Or the multiple years of "dog ate my homework" excuse of why didn't have Mac Pro yet.
Yes there is a "hot mess" PR problem to clean up with the MBP but the Mac Pro isn't in that much better shape. And this new 16" solution doesn't appear to be be an affordable solution either. If Apple has more "Cadillac pricing" surprises on Mac Pro BTO options then it would be better done where they can explain it a bit better.
It will be interesting if the new MBP 16" has a Navi and the Mac Pro rolls out with none. I can see Apple really not wanting to put those to in the same picture.
[automerge]1573568066[/automerge]
Apple doesn't. It would be basically just a CPU update that fits in the same socket. It would be curious if Apple didn't do an update if actually were just limping along with a single iteration design product.
Apple has left a large price hole that the iMac Pro will probably be used over several generations to fill. ( not going to make everyone happy but the discounts on the CPU if passed through will generate a bigger positioning gap between iMac Pro and new Mac Pro )
Last edited:
Hopefully the new iMac Pro besides predictable upgrades includes a DIY RAM upgrade door this time, sadly don't include the also just announced AMD threadripper 3 with a all pcie 4 internals it would be a home run with filled bases
I stopped and thought about it and it was that interview John Gruber did with Craig at WWDC.
The Talk Show Live From WWDC 2019, With Craig Federighi and Greg Joswiak
Recorded in front of a live audience at The California Theater in San Jose on Tuesday 4 June 2019, John Gruber is joined by Craig Federighi and Greg Joswiak to discuss the news from WWDC 2019.daringfireball.net
There's fun stuff throughout that interview but there's a short bit of around the fourteen minute mark where Craig acknowledges that Afterburner is an FPGA and can be reprogrammed in seconds for new purposes. He states that "absolutely" it can reprogrammed to do new things. He doesn't go so far as to say it will be used for other codecs but he says "more to come, I don't have anything to announce today."
I can't remember if they circled back to that later and I'm not planning to rewatch the whole thing now ;-) but that's what had struck me as this could be extended to other codecs.
The other major thing there is just how "it just get used if use standard foundation library calls" there. Apps which use Apple's library to AV work just get the Afterburner boost if it is there and get some "best effort" if it is not.
IHMO, it will be more likely that Afterburner will be expanded in the future to make more Apple Audio, Video, and/or Image handling calls just get better in a very similar way. ( works "fast enough" on many "pro" models and works much faster on some models with an extra Afterburner card attached. )
For example, Apple stuffs stuffs image processing improvements into each A-series update. Afterburner means they could track those updates over time with a card in a Mac Pro. Tether a camera to a Mac Pro and it could just do better over time.
There is a wide variety of things Apple's AV/Image library does now and Afterburner will only cover a very narrow subset at the beginning. That could be expanded as the API calls grow or more horsepower ( "Afterburner" ) is applied to an incrementally wider set of existing APIs.
I think "Afterburner" is more of a "preBurner" ( or "IngestBurner"). But they could add fixed function export to the mix too.
TBH, the Mac Pro would likely be a better candidate for PCIe 4 at the moment because of the accessible slots. With everything built in, and GPUs getting practically nothing out of PCIe 4 so far, it’d be only for the SSD, assuming Apple remembers to update their T2 chip to support it.
Threadripper would be useful for other reasons, but if Apple hasn’t reached the end of their rope with Intel yet, it doesn’t seem likely. Especially with Intel willing to produce what amounts to custom SKUs for devices like the Mac Mini.
That would be a case redesign. Early 2019 Apple did a memory capacity "speed bump" on iMac Pro. This CPU update would be essentially the same thing. Quick to market because avoiding going through industrial design chokepoint on resources and simply just "keep up". ( with the Ryzen 9 and Thunderipper onslaught on price ; not PCI-e v4. That saving pass along is generally going to to the Intel workstations too. The Mac Pro is going to out of position at the lower end. It makes about zero sense to also misposition the iMac Pro too. ).
If Apple as doing a case redesign I'd wouldn't expect that until 2020. Apple has demonstrated almost no ability to walk and chew gum at the same time. If there is a new MBP chassis and new MP chassis in the same year they are most likely "spent" on doing any new industrial design in the Mac space. ( the other speed/fix bumps on a subset of the reset of the line up makes it even more so. )
A deep upgrade of the iMac Pro would need a new GPU... and there really isn't one to move to in the AMD line up at the moment that is equally space efficient. Much better chance in 2020. And Apple probably won't do much with the Mac Pro in 2020 (all the same chokepoint industrial design reasons) except perhaps some MPX card update(s). (the 580X being most prime candidate).
Sadly 90% of it's potential won't be available as macOS don't have any official fpga toolchain, and an SDK from apple for it's fpga it's like asking too much, to someone used not to walk while chewing gum at the same time.
If Apple is sticking to Intel, then it isn't. The next gen Sunny Cove or Willow cove core designs with better IPC are coupled to the PCI-e v4 updates. What the iMac Pro needs more is better coverage of the 8-12 core range with better performance at lower prices ( then the W 3xxx series ).
The Mac Pro is in far more pressing need to just ship and get the inevitable bugs out of the design.
Since there is little evidence that folks are putting PCIe v4 and low power into the same packages then it isn't coming to the T-series any time soon. GPUs don't get much out of PCIe v4 on gaming workloads, but there are a number of computational workloads that do. The iMac Pro would need a bitter GPU target to work with though.
the iMac Pro also doesn't maximize the space between GPU and CPU which is more of an issue with PCI-e v4. Yes redrivers are a workaround but there is already alot going on with the Mac Pro logic board and that is something that should be throughly worked out before jump into those trade-offs.
Technically putting something into a BGA (ball grid) instead of a LGA ( pins) package is a SKU but that is probably not outside of AMD's skill set range ( since it would probably be outsourced anyway). It isn't really customer "Just for Apple". It is more the case that Intel has so many customers that finding a few others that want what Apple wants isn't a problem.
The Mac Pro is shipping with 580X .... AMD's track record with Apple probably isn't spotless either. Is AMD going to hold it together with Zen3 or have a hiccup on perf/W or something. AMD has a much better window than they have had over the last several years, but they have issues too on crossing t's and dotting i's.
[automerge]1573573112[/automerge]
The point is they don't need a general access FPGA toolchain. They just need Afterburner to be useful for a substantial number of folks. If it is in the sub $1K range then it may be useful in a few TBv3 contexts too. (just not 8K work).
It isn't Apple's FPGA to have a toolkit for. If Apple made the FPGA , but they highly likely are not. And since the FPGA is underlying core foundational libraries, it doubtful Apple is going to want random folks poking at that then at the binaries in the kernel. Apple is probably using someone else's FPGA SDK and can't give it away for free. ( nor do they probably particularly want to subsidize that ).
Apple is using a FPGA probably because it is cheaper (full lifecycle wise) than using an ASIC (application specific IC) . But they are primarily trying to do something "application specific" (i.e., add horsepower to their application APIs. ).
If some 3rd party wants to sell a general usage FPGA card for the Mac Pro, then they have the same work to do as as any other PCI-e add-in-card vendor. Write the drivers for macOS , be a trustworthy partner ( follow Apple's guidelines and strategic directions) , and ship the product. Afterburner doesn't have to solve everything for everybody. Neither do the Apple MPX modules have to solve everything for everybody.
Last edited:
I can actually see that happening, because while Apple doesn't love user-accessible upgrades, I think the motivation behind the iMac Pro was less about stopping said upgrades than it was retooling an existing product to new demands with the least time and effort possible. But if that happens, it would be a chassis redesign, and while on an infinite timescale it's coming, we've heard nothing to suggest it's gonna' be soon. An iMac with thinner bezels and chin would be great, though. The problem with the 27" is it's damned hard to pair with another monitor on a regular-sized desk when it's so huge, and an inch or two extra room would definitely help.
On the other hand: the Mac Pro will exist for the people who really want access, and RAM access is always going to be lower on Apple's priority list. So who knows. That said, if the iMac chassis redesign is only in the early stages now, it's possible that it'll be a product with minimal Ive input.
Tea leaf readings and speculation. Some are "citing" this article too. Until the release BTO pricing and take orders for actual deliveries, it really means nothing.
16-inch MacBook Pro reveal likely this week as Apple appears to be holding private press briefings - 9to5Mac
Where is the rumored 16-inch MacBook Pro? Apple hasn’t officially announced the larger entry in the notebook lineup, but there’s...
9to5mac.com
I’m pretty sure we are going to have weekly circle j*rks until December 17, since the 21st is considered winter...
c’mon Tim make me wrong!
Cause the usual suspects (YouTubers) flew yesterday for an not so secret meeting at NYC by apple's sponsorship likely
Preacher, meet choir. Don't mistake my brevity for a lack of engineering knowledge, or my willingness to entertain hypothetical ideas and discuss them as endorsement.
That said, at least what what little CUDA data I've been able to find (based on the GTX 1070), even dropping to 8x 3.0 lanes isn't a huge loss. Unless you are also doing SLI for certain simulations like nbody which apparently winds up leveraging PCIe lanes to coordinate work between the two cards.
Better eGPU compute, and SLI setups tight on lanes are really the biggest win here when it comes to GPUs and PCIe 4 in the near term. And in the eGPU case, I wouldn't be surprised if we are stuck waiting for "USB 4.1 gen 2" or some nonsense to get PCIe 4 support. I'm not sure how much more effort I can expect Intel to put into Thunderbolt 3 controllers going forward. Maybe we'll see OCuLink-2 connectors on Macs (I doubt it).
Logistics and support for a new SKU aren't free. But my argument is more that Intel has a good working relationship with Apple where Apple can say "we need this to make product X" and Intel is generally able and willing to oblige. There's no pressure right now for Apple to abandon that working relationship.
AMD is also willing to build custom silicon, but that would require Apple expand their current working relationship with AMD to the CPU division, and without something souring the relationship with Intel, why do it? Apple would need more custom silicon from AMD than they currently do from Intel, with how much Apple has bought into on-die GPUs in their designs now outside the 15" MBP and iMac lines.
The breaking point may be if Intel can't recover from their mis-steps in the face of what AMD is bringing to market this year. But by the time Apple is forced to move, who knows. Maybe they will have a workstation-class ARM chip ready (but I honestly suspect that's going to be a lot longer of a wait, and that the MacBook Air will be the first to switch). But at least Apple has AMD as an option these days along with both AMD and Intel producing parts Apple can use, unlike when IBM and Motorola both shrugged their shoulders at Apple over a decade ago.
AMD's track record is spotty in general, IMO. I'm honestly a little surprised how favorable the Ryzen 3000 launch coverage was in the face of the microcode/driver bugs that took months to sort out and get fixed for customers (and not just the RDRAND issue). That launch reminded me why I don't build PCs often.