An 8-core firestorm design will significantly outmatch any 8 core Tiger Lake or Zen 3 design, especially mobile ones. Looking at the SPEC2017 scores, it would be closer to 12-core desktop Zen 3. If we get a tweaked core with small CPU improvements, 5950X level performance in a thin and light laptop (around 50-60W TDP) is absolutely possible.
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.
M1X worst case scenario
- Thread starter JMacHack
- Start date
- Sort by reaction score
An 8-core firestorm design will significantly outmatch any 8 core Tiger Lake or Zen 3 design, especially mobile ones. Looking at the SPEC2017 scores, it would be closer to 12-core desktop Zen 3. If we get a tweaked core with small CPU improvements, 5950X level performance in a thin and light laptop (around 50-60W TDP) is absolutely possible.
I think that the M1 is enough for the vast majority of people. But running without thermal issues will be a big deal. I like my 2015 MacBook Pro but it can get toasty.
systems vendors have a more steady supply of GPUs than buying from random shop #22 or random dude off the back of a truck on Ebay.
A large number of GPUs have shipped it is just the demand is outstripping supply in the 'spot' markets.
As for the 5950X, we can shave down to the 5900 and still 12 versus Apple's 8 cores. 50% faster is still probably a smaller stretch in many contexts ( especially those with a larger RAM footprint which Apple's laptop SoC has pretty good chance of being capped at a lower amount on. If Apple gets to 64GB it is likely more on the RAM vendor maket than on Apple's tech. ).
There is also some blurriness where "M1X" is being used by some to cover everything from 10 core JadeCut all the way up to 40 core Jade4C. It is very doubtful that Apple is going to have one product name to cover that whole span. In the c6ntext of the MBP 14-16
Apple has been yapping about how they had already covered desktop performance for a long time.
"... Apple famously called their “Cyclone” design a “desktop-class architecture” which in hindsight probably should have an obvious pointer to where the company was heading.... This year’s A14 chip includes the 8th generation in Apple’s 64-bit microarchitecture family that had been started off with the A7 and the Cyclone design. .. "
https://www.anandtech.com/show/16226/apple-silicon-m1-a14-deep-dive/2
This is far from the "first gen" for the cores. The developer transition kit was an A12Z. If want to start counting "macsOS on 'arm' " that would be gen 1. Apple has had this running for a long time versus what got sold to customers.
Minus some relatively minor tweaks to the L2 cache sizes these are the same cores that are in the current iPhone. So casting those cores are "generation 1" cores is a bit of an overreach.
Apple has done a ton of work here. Putting macOS on it later doesn't mean that the silicon foundation somehow completely started from "scratch" for the Mac is a huge oversell. the A14X turning into the M1 isn' turely gen 1.
The "bigger than that" die is bigger. If using A15 cores then probably will have a '2' in the official name somewhere. That would be even less "Gen 1" .
The "uncore" stuff is probably a bigger leap. Sucking in TB controllers (although Intel had already done that by the time Apple did it. ). That was a bigger jump over the A-series than the cores. But there too it is narrow wins. Just four x1 PCI-e v4 lanes would get you whole lot of "enthusiast" market hate mail if Intel or AMD did that. Apple is doing a substantive amount of 'don't look at that , look at this" marketing with the M-series so far.
The "uncore" features what the big gap the A-series had with desktops. M1 closes it a bit, but still some big gaps if talking about the hardcore enthustiant market.
If stick to comparing "apples to apples" then, yeah Apple has more clear winners.
Apple is probably going to pitch to some that they can dump their old MP 2013 , MP 2012, or iMac Pro for a MBP 16" now in many cases. Assisting in moving more people from desktops to laptops.... Apple has a good traction there.
the 16 or 32 GPU core versions? Probably not on the "bottom high end" that is going to be the competition over the next 12-14 months; at least the desktop clocked ones.
Again it will be the "uncore" that will really matter. If Apple solders down all the RAM they aren't going to get welcoming calls from the enthusiants market. If there are no PCI-e slots. Again it will be more " pitchforks and torches" than some stampede inrush of new customers.
It is not necessarily up to the cores themselves to handle the fabric. That interchip fabric is very likely chopped out of the 16 GPU core version though. I highly doubt there is just one die design that folks are covering with "M1X". The bigger one probably could have a large enough area budget to add that interdie linkage fabric. Or there is a third even bigger one that has the interlink and may not be attached to "M1__" sequence.
The high-end 24" iMac now has 2 extra USB-Cs that are USB 3 only, so it seems like Apple are quite prepared to have non-TB USB-C ports where needs must (whether it is because USB-A connectors are too deep, as on the iMac, or that there aren't enough TB controllers, like on the 12" MacBook).
...as for the "backslide" - the current 16" and high-end 13" MBPs have 4 TB ports but each pair "shares" a controller, and the old Intel 2-port modules only had a single controller. With the M1 machines they've gone for each port having a dedicated controller, so they have the same TB bandwidth as the old 4-port models - and the availability of USB4 multiport hubs means you can get those ports back. 3 ports on the M1X models would be 50% more bandwidth than the old 4 port models,
Except the stolen schematics also suggested that there would be a SD card slot. Now, I might buy Apple trading a TB port for a HDMI port (esp. since losing the TB potentially frees up a display stream), but while SD would certainly be on my "nice to have" list I don't think it would make the cut if it really came down to "SD vs TB port #4". So I don't think the missing USB-C port is necessarily just a space thing - it will be down to only having 3 TB controllers. Or, assuming your point is right about needing simultaneous video on all ports for TB4 certification, just not being able to support 3 external displays...
I actually said "HDMI 1.4 or 2.0" - yes, 2.0 would be good enough for most things, but it isn't the "new shiny". 1.4 would be dirt cheap to implement and satisfy the #1 reasons for wanting dongle-free HDMI.
I was going for the "most pessimistic" and "most optimistic" options, not the most likely.
Maybe they are genuine Apple docs (or at least the "originals" were - most of the news stories seem to be rather garbled second-hand accounts backed up by bogus 3D renders...) but that doesn't guarantee that they are up-to-date schematics of actual machines in the pipeline for production. They could be old, obsolete, prototypes, contingency plans for M1-based 14/16" MBPs if the M1X fails...
I expect it is a mix of the iMacs having USB-A connectors so swapping them for USB-C was fine. And, as noted, M1 only supports two TB channels. The laptops have always been USB+TB so dropping TB would be a regression and "M1X" will have at least three (and probably four, even if only three are connected to ports) TB channels.
He confirmed that he believes they were delayed in a recent report:
“M1X MacBook Pro: 14-inch and 16-inch MacBook Pros with high-end M1 chips, MagSafe magnetic charging, miniLED screens, and no Touch Bars are in the offing. After delays, look out for these to hit shelves in the next several weeks.”
Based off of the rumors that the M1X macs were delayed, I’m almost wondering if the 4-port iMac was supposed to come with the higher-end chip the rumored 14” and 16” MacBooks are going to have (and they’d all be announced at the same time during WWDC). It would make some sense as this variant has two fans as well. My guess is this higher-end chip will support 4 TB ports and the 2 USB-only were a plan B option. Just my theory.
The 24" has been rumored since early 2020 so I presume it was always meant to be on M1 and that it ended up being more complicated then Apple expected, which pushed back it's release date later than the other M1 Macs. It has been claimed that the more-powerful iMac 5K replacement has been delayed till (likely) 2022 because Apple needed the engineering talent to get the 24" out the door.
I say MagSafe will be useful. With kids or dogs or sometimes our own clumsiness
One thing for sure they were never be presented at WWDC even if they were on track. Why?
Kuo said M1x macs will launch Q3 2021. Apple never launched a macbook pro in july, aug, or sep.
Kuo probably meant mass production and guess what mass production started in August.
Mark gurman is saying that because he predicted a summer launch.
Apple designed the M1X to mobile first. People on this forum should stop comparing the 8-core M1X to desktop solutions. It's clear Apple markets the M1X to be mobile first, it's competing with AMD mobile H series and Intel 11th gen/12th gen Mobile series.
Again the power draw will be significantly different between M1X and the 5950x or the 5900.
Please compare M1X with mobile AMD processors.
Apple did not launch Macbook Pro in July , Aug , or Sep of 2021. However, they never , ever did a launch in the summer?
https://buyersguide.macrumors.com/#MacBook_Pro_13
July 2018 , June 2017 , July 2014 , June 2012
( and a whole bunch of May's . )
Before Intel's fab process went off the rails , they typically arranged things so that there was new CPU products to showcase at CompuTek conference in Taipei every year. That typically happens last week in May or first week in June.
Basically it occurs in the 1-2 two prelude to the normal WWDC event time. Hence, not only Apple, but most major system vendors relatively regularly introduced new laptops during the May-July time frame as that was the time when Intel's "tick/tock" was delivering new items in most year.
Each Spring there is a lot of "Apple better introduce XXX Mac product at WWDC " to signal that Apple deemed that product important ( i.e., shine a spotlight at one of the biggest dog and pony shows that Apple runs; WWDC keynote). But that is lost in the context that Intel was driving several system vendors here ... not just caving into Apple fanboy requests .
Good points. I do agree it would be very unlikely to release the M2X before M2. I was just posing the scenario and musing over it. Plus it was more about how long in the tooth M1 now is!
M1X is gonna be mighty powerful and plenty good for most of us.
Intel created the certification standards for TBv4 . You really think they wouldn't have product that implements it?
the Gen 10 ( Ice Lake ) mobile chips integrated the TB controllers back in 2019 .
https://fuse.wikichip.org/news/2628/a-look-at-the-ice-lake-thunderbolt-3-integration/
The Gen 11 ( Tiger Lake ) H models have an even better implementation. ( Intel is on 2nd generation of doing this and about to ship 3rd Generation in Alder Lake Gen 12 ).
All the new laptops announced with Intel’s latest 11th Gen H-series processors
Including models from Alienware, Razer, Asus, and MSI.
www.theverge.com
There have by Gen 11 H models all summer. ..... And Apple hasn't shipped 8 P cores yet.
The MBP 13" four port model has Gen10 chips inside of them. So the M1 was only really keeping up.
The MBP 16" is just old at this point ( Gen 9). If Apple was like Dell/HP and have two or three laptop product lines there would already be a 16" model before the "M1X" shipped also.
There is a quite high probabliity that Intel showed Apple everything on that slide back in 2017 ( or before) in a NDA session. Ice Lake had a long gestation period.
Finally that there is some "defect" that the port pair share a controller, that isn't necessarily a defect. Thunderbolt allows for point to point transmissiong of data between two TB devices. All the data on the network isn't 100% required to be loaded back into the Host system. If Peripherhal A wants to DMA data to Peripheral B that is suppose to be allowed in a full implementation. It would actually more power inefficient to decode the data , send it to the main PCI switch in the CPU package , find out it was suppose to go out the other adjancent port, re-encode it as TB data, and ship it back out.
In the diagram above the pairs share a FIA/CIO Router precise so traffic can move from "Daisy Chain 1" to Daisy Chaine 2" if that is were it being directed to.
There are multiple PCI-e and Display Feeds.... but the ports sharing a switch is actually more efficient for point to point routing.
Edit: MBP 13" early 2020 has Gen10 ( not Gen11 ).
Last edited:
If they did that after including the GPU, SSD controller and all other essential I/O in an SoC for slim laptops and thin clients, very few would care.
What I want from my future MacMini (be it a grey M1x next month or a redesigned consumer grade M2 next year) is:
- run 2 4k monitors with no issue
- have 1 proper USB4/TB port so I can run a dock with an SSD inside at full speed
- 16GB (32GB would be nice but is no deal breaker)
The "M1X" (I hate how you folks make me use this mnemonic
) will power both Mac laptops and desktops, just like M1 did, so should absolutely fair to compare it to the desktop offerings in the same price range. Personally, I'd expect the SoC to be more scalable (somewhere between 30W and 80W) than M1 and thus more adaptable to various usage scenarios. Extrapolating from M1 benchmarks, I'd expect the version in the 16" to be at least on par with the 12-core AMD 5900X.Thanks. Sorry about that. Typo. Had just written "Gen11" in previous sentence and it came out again.
Who thinks these M1X models will get wi-fi 6e (6ghz)? The M1 models did not (only 6/5ghz), but rumors are the iPhone 13 will get it.
I believe all of the wi-fi and bluetooth is offered via a separate module, so it shouldn't be limited to the M core development if the M1X is based on the M1.
In a teardown, the M1 macbooks used this:
Apple USI 339S00758 – Wi-Fi 6/Bluetooth 5.0 module
Broadcom/USI released a 6e SIP more than a year ago but Apple didn't use it then:
en.prnasia.com
WM-BAX-BM-62
I believe all of the wi-fi and bluetooth is offered via a separate module, so it shouldn't be limited to the M core development if the M1X is based on the M1.
In a teardown, the M1 macbooks used this:
Apple USI 339S00758 – Wi-Fi 6/Bluetooth 5.0 module
Broadcom/USI released a 6e SIP more than a year ago but Apple didn't use it then:
USI and Broadcom Launch Wireless SiP Module for Wi-Fi 6E Market - PR Newswire APAC
SHANGHAI, June 17, 2020 /PRNewswire/ -- USI and Broadcom Launch Wireless SiP Module for Wi-Fi 6E Market
WM-BAX-BM-62
USB-C is much smaller. It was design more so as an alternative to Lightining than a replacement for Type-A.
USB Type-C Connector Specifications Finalized
Type A isn't the legacy connector that plays in the small device space. There is also mini and micro.
(from article link above. )
".. First, the mini connectors were introduced, which, at approximately 3 mm x 7 mm, were significantly smaller than the original connector, .... The smaller size did allow USB to be used on a much different class of device than the original connector, with widespread adoption of the mini connectors on everything from digital cameras to Harmony remotes to PDAs of the day. .."
Pragmatically Apple has turned the iMac 24" into a 24" iPad on a pedestal. Type A is out for the same reason it is out on cameras , phones, "thin slate" tablet, etc.
Apple had a dogma (since the edge thinning) that all iMac ports go into the back. That dogma ran into the even higher priority dogma on "even thinner , is even better". The Headphone jack and Ethernet jack moved too. That wasn't M1's Thunderbolt/USB internal implementation related. That was 100% the "thinnest" dogma.
Another substantive problem with the "these were suppose to be TB ports" is that these ports are imlemented in a substantively different fashion.
In the iFixit teardown they give Apple some positive "repairability points" for the USB-C (only) ports. Step 9 of the teardown.
iMac M1 24" Teardown
Is the new M1 iMac just a silly-sized M1 iPad, a detached-keyboard M1 MacBook, an M1 Mac Mini stretched out like pizza dough? We’ll only know by...
www.ifixit.com
The ports are on a separate board so if a user "burns out" the USB port the repair is straightforward. This is a draughtboard swap out and it is not a whole motherboard swap out.
Speaking of the motherboard..... The "thin" dogma has constrained the logic board to be the same approximate size of the chin. As much as folks complain about the chin, its size on modern "small" screen iMacs is not that big.
If Apple attaches two more RAM stacks and makes the die used 40-120% bigge rto create a bigger SoC, where do all the other chips around the M1 here go ? The "X" in the name being used here is primarily to denote "bigger die and substantailly more stuff". If the iMac 24 still had an historically sized motherboard that would about zero problem to take a bigger package. ( the board had space for a CPU + RAM DIMMs + GPU + VRAM.). Apple threw all of that away and painted themselves into a corner with a "chin size only" logic board.
[ the back of this logic board is pragmatically fully populated also. So "move some stuff on top to the bottom" isn't really an option either. ]
As a technological miniaturization exercise this is all pretty impressive. If look at an iMac motherboard from 4-5 years ago and look at this, it is truely a major leap. However, it is a dual edged leap. SoC size is going to be limited. RAM capacity size is going to be limited. Storage capacity size is going to be limited. If still had the old logic board area available all of those could be moved up an order of magnitude from what Apple has constrained themselves to here.
There is certainly enough here to be a "good enough" desktop system for a wide range of people. It can fill the original classic iMac mission. Moore's Law density improvements should allow them to put bigger caches , more specialized processors into the SoC. Also more capacity into the RAM stack and NAND stack chips.
The two fans here aren't really indicative of enabling a larger package (and TDP). It is much more likely what Apple is trying to do is keep the heat away from the LCD screen panel. The XDR monitor has fans to do the same thing with a lower board power consumption issue. If there aren't air movement across the board the heat from the board will rise straight up. Stright up from the logic board is the panel. You have now put an "Easy bake oven" under a LC panel... probably should have fans to get that heat out of there. the two fans also can spin slower than if just had one fan. So also an issue of keeping the iMac as quiet as possible. ( again because of the chin can't have a larger fan that spins slower, so forced into using smaller ones. Plus the speakers have be balanced so the a fan can't crowd the speakers out of the left ( or right) side. Everything has to be packed into this same single chin space. ).
Just 3 TB controllers would make sense if Apple is using full size Jade (10 CPU + 32 GPU ) dies to make Jade2C and Jade4C packages. the Jade2C would have 6 controllers with is already probably overkill ( but apple could reseerect the MP 2013 desing parameter. Probably to much hate from the same critics of the MP 2013 the last time). Jade4C would have gross overkill of 12 but that is incrementally better than a gross overkill of 16 .
Or since to get a close packed dies where the 4 corners meet layout , they'd need two "mirror' dies.
[T][P]
[P][T]
The lower right corner of the upper T would match the upper left coner of lower T when rotated 180 degrees. The lower left corner of P would match the right upper corner of lower P when rotated 180 degrees. The distance between those dies is minimized at that
One could drop the TB controller for PCI-e controllers in addition to adjusting the layout in mirror image. Would cost incrementally more but save space.
If they could make the "extra' TB pins outs from the SoC multiple functional those could be a could of x8 and x4 PCI-e v4 outputs in an optional configuration. Not going to make the dGPU add-in-card folks happy but there are range of storage, video out , audio , etc cards that would happy with a couple of x8's and x4's.
The x1 PCI-e v4 links are decent for an SD-Card. This new discrete solution only needs x1 or x2 PCI-e v3 worth of bandwidth. ( with v4 a future version wouldn't need x2 to get to "full speed". )
Similar issue with E cores. Jade having 8 P and 2 E looks "funny" . but if multiple by four 16 p and 8 E looks pretty reasonable. The single building block looks like it has "too few", but once you multiple by 2 and 4 you get to plenty.
"bogus 3D renders" or 3D renders as replacement for Apple take down requests ?
They will likely use the computer name as the way to differentiate them.
MacMini, MacMini Pro; iMac, iMac-Pro; MacBook Air, MacBook Pro, etc.
If Apple can somehow keep price for M1X/16/512 MBPR around 2000 USD, it would be a huge win. Assuming this model has MagSafe + two TB3 ports, no touchbar, reduced bezels, miniLED, and 1080p webcam there won't be much cheaper competition on portable ultrabooks market. Anything else like FaceID, MagSafe charger combined with Ethernet port (as they did in iMac) would put make this laptop the best MacBook ever made. I honestly think that the price for base MBPR14 will be around 2200 USD, given world components shortage. In this case I will probably wait till MBA update, would interesting to see rumored MBA15 version as well. I really have no complaints about M1 performance, so basic M2 with 10-15% performance increase would be enough for me.
LOL
I had a ZX80! - and threw it away many years ago....possibly worth some $$ now to a collector....
Incredible to think that you could actually do something useful with only 1 kByte of memory
LOL
I had a ZX80! - and threw it away many years ago....possibly worth some $$ now to a collector....
Incredible to think that you could actually do something useful with only 1 kByte of memory
Maybe meant a Z80. ZX80 was a computer while the Z80 was a CPU - and I had one of them in a DEC Rainbow.
“No one will need more than 637KB of memory for a personal computer. 640KB ought to be enough for anybody,”LOL
I had a ZX80! - and threw it away many years ago....possibly worth some $$ now to a collector....
Incredible to think that you could actually do something useful with only 1 kByte of memory
-Bill Gates allegedly
I mean that’s of course still substantially more than 1k but in modern standards both are minuscule. There are microcontrollers running all sorts of things still though on very little RAM. The Arduino UNO has just 2K of RAM and is used for all sorts of things, like blinking LEDs… Most people use them for blinking LEDs, at least initially