What are you talking about? Intel released its Ice Lake Xeon workstation line in June of 2021. Apple lied multiple times, but that’s ok we all knew their heart was never in it for Intel.
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 Mac Pro 7.1 owners! What would 8.1 need to be to get you to leave your 7.1 behind?!
- Thread starter maikerukun
- Start date
- Sort by reaction score
What are you talking about? Intel released its Ice Lake Xeon workstation line in June of 2021. Apple lied multiple times, but that’s ok we all knew their heart was never in it for Intel.
THIS THIS THIS all day...I am literally PRAYING that Apple does this. It's what I assumed they were doing when TC said they have a few more Intel Machines in the pipeline, and I'm hoping that's what's going to happen. I'm very excited, and I mean extremely excited for where AS Mac Pro is going to go in 4 to 5 years from now, but until the cards are doubling the power of DGPU's, let the muscle car crowd have their muscle car.
100% agree with this. Hope Apple notices how many of us are wanting them to simply have two paths from here on out.
Hmmmm, I see your point. But Mac Pro has always been TWO things in my opinion. Their top of the one system...but ALSO...Apple showing off that they're still the most innovative company in the computing world. And with the 2019 Mac Pro, I would argue this is the most innovation Apple has demonstrated in the pro market in a VERY long time...and to just let it die a handful of years later would be an extreme disservice to the ingenuity that went into the damn thing IMO.
You two are talking about two different line ups.
Xeon SP Gen 3 (Ice Lake) arrived Q2 '21 ( Xeon Silver/Gold/Platinium x3xx ) [ Older 2017-2019 roadmaps put that closer to Q4 '20 ]
Xeon SP Gen 4 (Sapphire Rapids are sliding. Was suppose to be out by now ( Q2 '22) but now limited around Q4 '22 and more wider rollout Q1 '23 ) . Intel is on its > 10 stepping so there is some 'bridge too far' aspect(s) of the design somewhere.
The workstation products are technically a different set of chip packages ( although the dies are the same ... or have been the same. ).
Xeon W-3300 (Ice Lake) arrived in Q2 '21 ( just on paper.) [ some smaller system vendors like Boxx and others were able to fill long lead time orders in limited volumes by 2H '21. No retail. No motherboards DIY . No Dell/HP/Lenovo or Apple. Has gotten better in '22 but still no top 3 workstation vendors. (Lenovo and others mostly bolting to Threadripper. ) ]
Xeon W-3400 (presumably the number as gen 4 in this line up. also Sapphire Rapids tech ) might make it in Q4 '22 depending upon where the large set of design defects Intel has been fixing are.
[ https://www.techpowerup.com/296005/...-processor-spotted-in-a-16c-32t-configuration
more muddled than that. 'W5-3400' may be the "HEDT" replacement. Not the "Expert Workstation" solution .
https://wccftech.com/intel-fishhawk...ith-8-channel-memory-for-expert-workstations/
Not sure if the "expert workstation" is just plain W-3400 or W9-3400 or whatever kooky name complexity Intel cooks up. Maybe WE-3400; "workstation expert" ]
If the defects that Sapphire Rapids are laboring on are "inter tile networking" , CXL , HBM then pretty good chance that the 3400 ducks several of those. The monolithic (?) lower 'half' of the line up won't have inter tile issues. The HBM mode isn't likely. And CXL is something could flip off on the initial set just to get some stuff out the door.
There is chance that Intel could shuffle these out the door just to start hitting some revenue. Probably prefer to do Xeon SP packages if they can though. But Threadripper W5000 really beats W-3300. High end Zen 4 Ryen 16 core count also for the lower end of the range also.
The Xeon SP Gen 4 is just large a mismatch for macOS. The core count is too high. The TDPs are much higher . That isn't a CPU package that is well suited for the current Mac Pro chassis at all. Most of the upper end of the W-3400 family has similar issues. At the top end the core count is too high for macOS. Similar thermal problems. The original plans for that 4th generation was after 2020 anyway in the pre-2018 Intel roadmaps. Highly doubtful Apple wanted to pick new Intel CPUs after 2020 back when they were planning in 2017-2018 timeframes. If Intel had rapidly fixed the problems that Apple was already seeing then would pick up something that was 2020 and might slide into 1H '21. But Apple picking something that was suppose to be '21 and pretty darn likely to into and possibly out of '22 .... probably not. Post June '22 they would have already hit the 2 year mark of their 2 year transition.... still patiently waiting for an Intel CPU probably would not have been on their wish list.
Also anyone who thinks Apple would have watched the high number of erratas flying off Gen 4 dies and not walked away in disgust is delusional. Apple was already complaining about errors in Intel implementations. If the count jumped even higher still then Apple would have quit faster ; not waiting until after the 2 year transition announcement mark to jump onto an Intel product.
Apple had an "Ice Lake" generation laptop ( MBP 13" four port 2020 ). It would have made some sense to do a MP early or mid-2021 (with a W-6300) if it has been shipping in volume , at more reasonable TDPs, and the target date for Apple Silicon MP was late Fall '22 .
Didn't happen. AMD 'sat' on Threadripper WX5000 and Intel slow rolled W-3300. Both threw as many dies as possible into the Epyc and Xeon SP products (respectively) to fight a 'death match' market share war there ( at higher profit margins for both). Apple tossed some MPX GPU updates into the MP 2019 and just limped along with the same platform.
While there may later be a follow on to what Intel has been calling the W-1x00 series, that is not the very high core count Xeons. Those the desktop die with ECC turned on. That will be something like W-1400 or W3-1400 ( used to be Xeon E3 xxxx was the desktop feature bumped die and the E5-xxxx was the workstation die in the previous naming prefix scheme. )
Alder Lake is the desktop/laptop family of dies; not the server ones. Different code name. All "P" cores. no 'E' cores.
Suspect there will may not be a "Alder lake" xeon at this point. Desktop Gen 13 ( Raptor Lake) seems to be on track. [ that doesn't fit with the everything is doom and gloom at Intel narrative... so it doesn't get highlighted much. ]. Decent chance that would be a better fit for them as a W-1400 product. E-core count is higher and would be more competitive with Ryzen Zen4 offerings (which can do ECC without gratuitous market segmentation , but required logic board / firmware changes. ) . Usually Intel rolls out the Xeon variant about 3-6 months after the desktop version rolls out. It hasn't really happened. And AMD is not hitting the snooze button.
Intel has a slew of leaking boat problems that they are chasing and Raptor Lake is exactly the same socket (and chipset for workstation). So why bother? Sliding an "Alder Lake" product into Q1 '23 for volume customers is kind of goofy when rolling out Gen 13 at the same time that is better and Zen 4 likely better still. ( You'd think Intel would want to use the wafers to ramp Gen 13 as opposed to fill year old stuff. )
Only reason would be bunch of board vendors had already queue'd up production and have to sell something. The expectation was that W-1400 would be rolling out by June 2022 and it was a "no show". ( was it even at Computex in May-June ??)
Sapphire Rapids with same P ( Golden Cove) cores with the AVX-512 turned on is the one that will go out super limited distribution late '22 and full product line in '23. That has been sliding.
Apple would never use Xeon W=1400 in a Mac Pro update. Gen 12 or 13 . Doesn't have the I/O bandwidth to pull the wait. it is the W-3000 variant that would be a even a remote plausible candidate.
That makes sense, I was confused so thank you for clarifying. There is always going to be another cpu update in the future but Intel has already released new Ice Lake Xeon W’s that Apple could use in a refreshed Mac Pro. These processors support PCIe 4.0 so it’s kind of a big deal. Although as deconstruct60 noted it doesn’t seem like Intel is shipping these processors in volume for OEMs yet so Apple may have no choice here but to continue to ship Cascade Lake. No wonder Apple is working so hard on Apple Silicon with Intel dragging their feet!
"may be", "subject to" give Apple a way out.Apple's products are usually considered by Apple to be systems as opposed to pure hardware and pure hardware. There is some large wiggle room though for macOS .
They do state clearly for hardware service support
"..
Service and parts may be obtained for longer, as required by law or for up to 7 years, subject to parts availability. Additionally, Mac notebooks may be eligible for an extended battery-only repair period for up to 10 years from when the product was last distributed for sale, subject to parts availability. .."
Obtaining service for your Apple product after an expired warranty - Apple Support
Learn about your options for getting service and parts for Apple devices that are past their warranty period.support.apple.com
Errr some of those spots you are cherry picking.
" parts may be obtained for longer, as required by law or for up to 7 years ..."
The 'may' largely isn't Apple. That is being driven by local law policy. Apple has to write this document so it around 100 different sets of laws. For folks in a specific country with specific laws there is little ambiguity implied there in the context of the overall document. Picking a specific location largely removes it as the Apple or local constraint takes precedent.
Even if Apple hadn't written anything down, the local laws with specific spans would apply.
I wasn't trying to post the whole vintage policy. The "lower bound" limit that Apple specifically puts on themselves is 5 years. So of course 7 years is a "maybe" since there is a range. If Apple gives themselves a 5-7 year boundary then 7 is not required . 5 is more of a hard constraint.
The reason why I picked that paragraph was folks are arm waving about 10 years. (often because " Microsoft goes 10 years" ) Even with the "may be" stuff there Apple's commitments aren't close to 10. Someone in the USA claiming that Apple owed them 10+ years support because paid a lot they doesn't have a legal leg to stand on. If elsewhere there is a quirky local law that pushes Apple to limit they rather not do, then maybe yes.
I also wasn't addressing the other extreme of folks who state Apple will throw all the Intel systems completely under the bus in 1-2 years.
Apple generally doesn't like carrying long term inventories. This whole "may be if inventories don't run out
is a somewhat new thing they didn't do back in he Jobs era. I wouldn't 'bet the farm' on how fast the parts run out at the end. However, as Apple has grown bigger and has to comply with a more complex set of local laws (and gotten burnt on sanctioned recalls ) there has been some drift toward tolerating some inventory. It is a high compliance cost try grudgingly do (and crank prices to limit impact on margins.)
The issue that seems to echo around on the Mac Pro forums is that Mac Pro users are somewhat privileged. And that "of course" that 5 year can't possible apply to them and that Apple has to give them highest amount possible because they paid more. There is nothing about "paid more is more privileged" in Apple's written policies.
The other reoccurring issue is that because Apple can't name an exact specific day, April 1st 2026 at exactly 12:00 pm GMT, that it is impossible to do any advance system lifecycle planning because Apple told them nothing.
Vintage/Obsolete dates are not exact in part because when Apple replaces a product is generally not exact. ( leading edge iPhones are a quirky corner case. Most of Apple's revenue but no where near most of Apple products models produced. ) .
Plain and simple those are wiggle words which could be used by Apple to avoid support.
Like when Jobs got on stage and pulled the MBA out of an envelope and said that it was the key to the future of Macintosh. Or the can't innovate my a** commentary when rolled out the MP 2013?
To position the the bulk of Apple's Mac technology efforts are solely grounded in the Mac Pro updates is to largely miss the mark. Apple has plowed lots of technological moves into the other system in the Mac line up. The Mac Pro 2019 didn't get T2 first. Didn't get AMD 580 GPUs first, Or the first Vega GPU. It lost the optical drive just like the whole rest of the Mac line up. Throughly integrated Thunderbolt just like the rest of the Mac line up had done years before. First PC tower to have GPUs hooked up in a NVLInk/InfinityFabric class link? Nope. An internal USB socket? Hardly revolutionary.
Mac with RAM mounted on bottom of logicboard... done on Mac laptops before.
Yes there is some 'new' things on the Mac Pro 2019 but the scope isn't that broad. The CPU and GPUs all have references designs for Apple to work with. The CPU and GPUs themselves are not upfront directly funded by Apple. The Plex PCI-e switch was a product pulled from server implementations and admittedly insightfully 'repurposed'.
The MPX socket is a naturally evolutionary path from the full custom socket on MP 2013 cards that sent DisplayPort to the distant TB controllers and got rid of discrete wires for power. If move some TB controllers back to the card while still needing to send DP off , that really wasn't a huge leap. Drop the proprietary aspects of the PCI-e feed for the GPU chip and it is substantive engineering work. But huge innovation is a stretch.
There was lots of engineering to do, but Apple product constraints paint almost every Mac product into a corner on some aspect that some clever engineering tapdancing has to try to mitigate.
The VERY long time is actually quite indicative that large amount of Apple technological time and effort has been spent elsewhere. To a very large extent the MP 2019 was just playing "catch up" as a jump from a 2010 MP to a 2019 one. If measure the 2019 against the other slot boxes Apple released from 2014-2018, then it isn't hard to win the race of "Most innovative". Apple's Mac sales had grown quite substantially in that nine year gap without an "expensive as possible" slot box to 'lead' the line up.
Afterburner 'solved' a problem that M-series has now taken off the table.
T2 ... same thing.
MPX connector ... very similar issue. ( and Apple sank probably an order of more magnitude more money into developing UltraFusion. So if it a lament sunk costs contest ... that is even bigger thing. )
If Apple provisioned out two x16 PCI-e v4 bundles from some M2 SoCs for the Mac Pro Apple could probably reuse :
1. The Plex Switch baseline purpose. Not as many slots but could provision out several slots to keep some legacy PCI-e v3 and v2 cards happy.
2. The case and space frame are all likely highly reusable even if chop the height in half. Same general 'look' and pull off handle. Could even keep around the 'innovative' $100/wheel rolling option.
Most of the fans and dual sided logic board could be kept also. ( One less big , 'front' fan ).
Could probably tone down the 1400W power supply but the format where the backside fan has side effect of cooling power supply doesn't have to go away.
3. SSD modules are still in the Studio ... ( like the iMac Pro and Mac Pro). Same thing likely coming.
4. They'd need to toss a discrete SATA controller on the logic board but the USB+SATA+SATA port thing could stay.
If there is still a "void" inside and distance to SATA sockets is the same , then folks could transfer their Promise J2i brackets.
5. The I/O card thing would work with adjustments. ( just retimers now instead of TB controllers. ).
The notion that the only innovative thing Apple did on the MP 2019 was the GPU MPX modules is rather myopic.
if Apple swapped two M.2 SSD sockets on the 'backside' for MPX connectors on the front there would be same fans of that "no innovation" change. [ get past the one ,and only one, drive support out of the box. ]
It isn't just Intel "dragging their feet'. The dynamics of what SoC is competive where is shifting.
The W-3300 really wasn't a Threadripper 3000 'killer'.
Intel Xeon W-3300 Processor Review Roundup
Intel's new Xeon W-3330 series of workstation CPUs are here, ranging from 12 to 38 cores, and touting up to 18% IPC improvements. But are these features worth the higher cost of the Xeon platform, and how do they fare against AMD's Threadripper Pro line?
www.pugetsystems.com
Isn't any better with the Threadripper 5000 out
DaVinci Resolve Studio: AMD Threadripper PRO 5000 WX-Series vs Intel Xeon W-3300
AMD's new Threadripper PRO 5000 WX-Series CPUs have arrived, promising faster performance with the same high core count and platform features found in the previous generation. Workstation-class CPUs like Threadripper Pro are often used for high-end workflows in DaVinci Resolve Studio, both for...
www.pugetsystems.com
Adobe Photoshop: AMD Threadripper PRO 5000 WX-Series vs Intel Xeon W-3300
AMD's new Threadripper PRO 5000 WX-Series CPUs have arrived, promising faster performance with the same high core count and platform features found in the previous generation. Workstation-class CPUs like Threadripper Pro are usually extreme overkill for lightly threaded applications like...
www.pugetsystems.com
Dell/HP/Lenovo have more resources allocated toward shipping out AMD Epyc (and now Threadripper) systems than before. So, there is also a "demand" problem that W-3300 had also time got closer to Q4 '21 and into 2022. It was also easier when GPU were "un-obtainium" for the bigger OEMs that had more regular GPU supply contracts to just sell new box with trailing CPU as a path to a new GPU. Throw in on top Ryzen 9 and i-9 Gen 12/13 starting to eat away at the old HEDT space from 3-4 years ago. HP's Z series website says the Z4 is their best selling workstation. (similarly Apple said the 16 core MP 2019 was most popular CPU configuration).
Yes there are still vocal folks saying they gotten have 50-60 cores to get anything done but that's increasingly being filled by vendors like Boxx , PudgetSystems ,etc. The market isn't all that large.
The W-3300 for Apple would more so be doing a inexpensive way of bumping the Mac Pro for on last gasp. Limit changes to just socket/chipset. Route just two x16 PCI-e v4 to slots 1 and 3. They rest into the Plex Switch as a better backhaul network for mostly v3 (or less) cards). Keep TBv3, T2 , DP1.2 routing. Apple is killing off the i9 so the creep from below (where need high RAM capacity) isn't as substantive change ( why the intel Mini is still around. two years into the transition). However, a major slide on timing subsstainally ruins the last gasp aspect. The longer into that closing window for Intel using the MP 2019, the easier it is to just cost to end.
A GPU upgrade would let them cost longer for less money if carefully manage the costs on that.
What was the point of this reply? lol. The only part that made any sense was the part where you acknowledged the innovation in engineering that the Mac Pro 2019 is. Period. It is their most innovative machine and all you did was list bits and pieces from other things that voltroned together to create this monster of a machine and it is the only one like it. No other machine compares or even comes close...Next response you should probably just condense it to focus on that part lol.