Sidebar Sidebar
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)

M

Mago

macrumors 68030
Original poster
Aug 16, 2011
2,789
912
Beyond the Thunderdome
goMac said:
Wasn’t part of that prediction that they would be completely custom slots and not PCIe compliant? And that they would be in an AMD Mac?
Click to expand...
I predicted it either a full custom pcie non STD, as well a STD pcie with some extension, the full AMD Mac was mostly (then) a willingness,, no cues, this is very different, we have evidence on Apple testing semi custom Zen2 from AMD, also cues on a New sub 6000$ "gaming" Mac (I want a trashcan re-boot) and everything Poppin telllus a MacBook pro and both new iMac / iMac Pro are coming next half year.


I don't expect an full AMD MacBook pro until year's end before next Xmas, the Mac more likely to debut AMD Apu/Zen2 are iMac and Mac mini, a true preduction here is not an "AMD" Mac, this it's just reading, my "prediction" is the iMac Pro 21 along the iMac Pro 27, and ECC RAM not STD but bto.

Next Mac: Intel based MBP 14 in March/April, then by next WWDC or in close date, the all new all AMD Macs (iMac & mini, maybe with Apple's Zen 2 based CPU - rebranded licensed and TSMC baked with apple's waffers)
 
CWallace

CWallace

macrumors G5
Aug 17, 2007
12,528
11,546
Seattle, WA
If the AMD's are as super-amazing as their supporters claim, there is no way Apple releases a Mac Pro on Intel only to release a Mac mini on AMD six months later that will be many times faster at one-sixth the price.
 
fuchsdh

fuchsdh

macrumors 68020
Jun 19, 2014
2,028
1,831
CWallace said:
If the AMD's are as super-amazing as their supporters claim, there is no way Apple releases a Mac Pro on Intel only to release a Mac mini on AMD six months later that will be many times faster at one-sixth the price.
Click to expand...
I mean Apple sort of already does (in that you can get an iMac that bests the pro Macs in single core for much less than half the cost, or you can get iPads that have on-paper superior benches at phenomenally lower prices) and that doesn’t much matter. If you want or need ECC memory, highest possible performance for long jobs, or need the most cores, you’re always getting a "worse" value proposition from a simple dollar standpoint. The 80/20 rule or some variant of it exists in almost every market. We just have people who don't understand "this is a bad deal for me" doesn't mean it's a bad deal for everyone (gestures wildly at the "who buys a $6K computer?!" threads.)

So I don't think Apple producing the Mac Pro now means they wouldn't come out with an AMD Mac mini this year. But there's been way more smoke about Apple switching to their own processors and next to none about imminently switching to AMD.
 
  • Like
Reactions: martyjmclean and Adult80HD
Blair Paulsen

Blair Paulsen

macrumors regular
Jun 22, 2016
211
157
San Diego, CA USA
I'd suggest even the most credible "smoke" is farther in the future than suggested.
I've always considered Apple/Intel a marriage of convenience, nothing more - so moving to AMD or their own chips is a solid prediction. That said, even if they've made good progress in prep, I'd guess early 2021 is the soonest Apple can put a stable configuration on the market.
 
  • Like
Reactions: OkiRun
M

Mago

macrumors 68030
Original poster
Aug 16, 2011
2,789
912
Beyond the Thunderdome
fuchsdh said:
Apple switching to their own processors
Click to expand...
What you mean as "apple's own processors" ?

FYI all Apple SOC actually aren't Product of apple, Apple just licensed ARM IP and customized it to their own products, often adding insanely big cache (which helps with benchmarks but not so In Real world).

FYI AMD also licenses Zen cores IP so Apple can buy this license customize it for the Mac and start manufacturing at TSMC as AMD also uses the same TSMC 7nm process and designed Zen2 around it, but also AMD just may sell some of the chiplets and Apple integrate it around they own PCH build around a T2 ir T3 chip, also AMD simple borrows some Ryzen Threadripper matrix and apple bakes they waffers with it as they need with total supply chain control.

Licensing ARM implies a painful macOS architecture migration, resigning to pcie4 and thunderbolt(as nowadays there's no tb3 or pcie4 ARM implementation), support a new experimental architecture with no apps, while licensing or buying AMD Zen do not require changes to macOS, neither lose pcie4/thunderbolt, and ads few more selling points as AMD CPU are not as vulnerable as Intel, also allow RAM encryption (likely tailor made for the T3 Chip), are the faster, more efficient and powerful CPU available now.

I don't understand why some people here can't see the logic here and consider Intel CPU or ARM or nothing, it's not right, Intel is lost period, ARM still years away to be a desktop replacement for x86, meanwhile switching to ARM implies fill macOS rewrite (and a long fight with bugs) while AMD CPU are here are the best and faster now holding all the records, are compatible with 7nm process which enable Apple to bake they own chips using excedent capacity from they secured waffers at TSMC etc.

Further, there is no macOS leak suggesting Apple is testing RightNow macOS on ARM or even testing some ARM-only GPU, no Apple spend most 2019 testing AMD APUs (CPU+GPU) as macOS drivers witnessed.

Embrace the new all AMD era.

I don't think the new Zen iMac won't arrive due the Mac Pro freshness, Apple doesn't cares their user platform devaluation, most of the niche Mac pro will be already ordered before WWDC meanwhile those requiring afterburner and multiple GPU won't consider a Zen Mac despite how many cores or faster single thread or multi thread, for some applications there's no rival to multiple GPU despite CPU cores those rendering on 4 Vega ii won't jump to an iMac Just coz it has 64 core and pcie4, this is a niche where the iMac by definition is segregated.
 
Last edited:
  • Like
Reactions: OkiRun and ssgbryan
fuchsdh

fuchsdh

macrumors 68020
Jun 19, 2014
2,028
1,831
Because I'm a glutton for punishment, let's take a look back at Mago's predictions, shall we?

*Said BTO options for Mac Pro would arrive in November, max RAM would be 768GB — wrong.
*More doubling down on wrong date for Mac Pro
*Nvidia getting signed drivers by September — wrong.
*Right before WWDC, the expected Mac Pro — wrong in basically every single detail (even the color!)
*Earlier the month before, we were getting a bigger Mac mini — wrong.

That just takes us back to May 2019, but to start off this thread, way back in 2016 we had a theoretical Mac Pro which not only was nowhere close to what got shipped and when, we know from the April 2017 roundtable never existed in any real form. It was complete nonsense.

Let's see where we end up.
 
  • Like
  • Haha
Reactions: Adult80HD, tsialex, sirio76 and 3 others
Nugget

Nugget

Contributor
Nov 24, 2002
2,168
1,468
Tejas Hill Country
Mago said:
Happy New year 2020 or 20"
Click to expand...

Dates are generally shortened similar to a contraction, with an apostrophe to replace the eliminated digits. So, for example, 1989 is abbreviated to '89. The apostrophe takes the place of the missing "19".

Full quotation marks aren't appropriate -- especially as a suffix -- because then you've written 20 inches and not 2020. Neither is 20' accurate, since that would mean 20 feet, although humorously it is technically accurate as a contraction if you just imagine that the writer has dropped the trailing "20" and not the leading "20" from "2020." I suppose perhaps this would refer to the entire century from 2000 to 2099.

I'm not going anywhere near the decade discussion and whether or not we wait another year for that. :)

Happy new year, in any event!
 
Last edited:
  • Like
Reactions: OkiRun
M

ManuelGomes

macrumors 68000
Dec 4, 2014
1,617
354
Aveiro, Portugal
Anyone got a hold on the PCIe lane allocation yet? I'm curious to see which slots are hanging directly from the CPU's 64 lanes (assuming at least slot 1) and which are switched, and what's on the PCH.
 
askunk

askunk

macrumors 6502a
Oct 12, 2011
547
430
London
I believe it goes at 80 Gpbs. They said it's 4x the top USB speed (which is 20 Gbps in the 3.2 v2 version).
 
M

Mago

macrumors 68030
Original poster
Aug 16, 2011
2,789
912
Beyond the Thunderdome
askunk said:
I believe it goes at 80 Gpbs. They said it's 4x the top USB speed (which is 20 Gbps in the 3.2 v2 version).
Click to expand...
No, "TB4" is exactly the same USB4 specification (which already includes TB3 alt-mode), indeed it's more a marketing maneuver as it was i.link iee1394 and firewire, both the same thing.
 
D

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
askunk said:
I believe it goes at 80 Gpbs. They said it's 4x the top USB speed (which is 20 Gbps in the 3.2 v2 version).
Click to expand...

" ...UPDATE: Intel confirmed it referenced USB 3.1 in the presentation, meaning Thunderbolt 4 is in fact not faster than Thunderbolt 3. We updated the text accordingly. .."
www.tomshardware.com

What Is Thunderbolt 4? Tiger Lake Tech Isn't Faster, Thunderbolt 3 With a New Name

Thunderbolt 4 appears to be mostly a re-branding of Thunderbolt 3.
www.tomshardware.com www.tomshardware.com


Intel was not talking about a version of USB that has pragmatically gone no where. The product distribution of 3.2 2x2 is minute even relative to Thunderbolt v3 let alone 3.1 gen 2.

4 * 10 Gb/s ( the only relatively popular version of USB 3.1 gen 2 ) is 40Gb/s is the comparision being made here. Compare to something folks actually might have as opposed to something almost nobody has.

Thunderbolt v4 probably will have a change. It will be dragging USB 3.2 2x2 into actual wide distribution. There actually would be a wide variety of shipping product. But when in Thunderbolt mode there will not be an overall bandwidth difference.

There is a decent chance that Intel might implement better data flow control and sharing between encoded DisplayPort and PCI-e data traffic. ( so if the DP traffic goes down to zero that PCI-e ecoded traffic get closer to the full x4 PCI-e v3 bandwidth ).

There is a slimmer chance that could get a 80 Gb/s mode but that would be for "pass through" DisplayPort v2 ( which basically takes Thunderbolt but makes it pragmatically all unidirectional outbound data flow. ).

But buy and large TBv4 is going to be a USB 4 certified implementation. Technically there are some corner cases that the current TBv3 controllers don't cover in the USB4 spec. It is primarily just alignment with USB 4 both in market naming number and with USB 4 specs.
[automerge]1578586245[/automerge]
Mago said:
No, "TB4" is exactly the same USB4 specification (which already includes TB3 alt-mode), indeed it's more a marketing maneuver as it was i.link iee1394 and firewire, both the same thing.
Click to expand...

It is more than just marketing naming. While USB 4 doesn't make TBv3 alt-mode mandatory everywhere. It does so in a narrow subset of what USB 4 calls "routers" which TBV3 terminology would call docks. If there is a USB 4 router/dock that have more than one USB 4 port then they have to implement TBv3 ( that context isn't really an alternative mode; it is mandatory). However, a host system (e.g. PC) or a 'dead end' perhipheral don't technically have to have TBv3 ( still an optional alt-mode).

The other issue is that USB 4 pretty much says 3.2 2x2 needs to be around to. USB-IF is flogging that forward after not much traction so far ( passed in 2017 and still no significant product base).

The TBv4 branding is going to mean something pragmatically different than USB 4. It is going to mean that the product isn't making the underlying TB protocol mode not optional. It is a effectively more a primary mode not an "alt-mode" ( the others are present but this is a more complete implementation.). Users won't have to 'guess' if the varies primary alt-modes are present or not.

USB 4 will mean you have to look carefully at the label for what you actually got. (there are optional alt-modes that many vendors can drop and still claim to be USB 4).
 
Last edited:
askunk

askunk

macrumors 6502a
Oct 12, 2011
547
430
London
deconstruct60 said:
...UPDATE: Intel confirmed it referenced USB 3.1 in the presentation, meaning Thunderbolt 4 is in fact not faster than Thunderbolt 3. We updated the text accordingly. .."
Click to expand...

Intel is getting lost on its own. :D I wish so much Apple could ditch them.

I have been skeptical in the past but lately I am starting to think that the ARM transition will be quite slow and that Apple could actually ditch them. If Steve were alive, they wouldn't have had the "kind" treatment he reserved for Motorola...
 
D

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
askunk said:
Intel is getting lost on its own. :D I wish so much Apple could ditch them.
Click to expand...


Intel isn't lost. USB-IF went from
USB 3.0
then to
USB 3.1 gen 1
then to
USB 3.2 gen 1

to name basically the same thing three different times with three different names. Hooking up with USB-IF has naming drama coming regardless. That isn't Intel's fault that USB-IF renamed the same thing that most folks are using so many times that the tech porn press is confused about what they are talking about when Intel makes a colloquial reference to "USB" without the three word explicit enumeration. The commonly fast "USB" right now is USB 3.1 gen 2 ( or USB 3.2 gen 2 if want to use the name that practically nobody but tech porn folks know about.).

Intel looking to match up USB 4 to TB 4 is simple and straightforward and extremely hard to mess up unless on some tech porn hard on quest. USB 4 leaves the Thunderbolt portion highly optional, so it makes zero sense for Intel to back off of the marketing campaign of Thunderbolt being the "complete" implementation of USB that they were running along side with USB 3.1 gen 2 Type implementations. Other Type-C ports could plug in and get solely just a USB 2 connection and there was entirely in compliance with the standard. It is a somewhat a game of 'what am i going to get' when plug into Type-C if don't carefully read all the hieroglyphs printed small on near the port. If someone says that all of these Type-C are Thunderbolt then there is no guessing about what going to get. Intel is just extending that to TB v4.

There is absolutely nothing "lost" about that. In fact, will probably help a sizable number of end users from getting lost ( versus the hodge podge that the USB-If labeling and optional this and optional that in that standard.)


The path on Thunderbolt is one of the things that Intel got right. It has kept Apple closer as a customer. (they may loose Apple in 2020 but Apple probably would have quit quicker without TB. And AMD shot themselves in the foot while there were trying to subvert Thunderbolt in terms of t Apple as a Mac CPU component design win. ). The only goofy move they might have made is perhaps making the next gen TB v4 discrete controller wedded to the 10nm process. That would be a stumble ( but a consistent one for them.) . GF 12nm or Samsung (if needed to go smaller than that) would probably be a decent place to send those if had to at the moment.

I have been skeptical in the past but lately I am starting to think that the ARM transition will be quite slow and that Apple could actually ditch them. If Steve were alive, they wouldn't have had the "kind" treatment he reserved for Motorola...
Click to expand...

Of course it will be slow. Apple has diddly poop viable in the upper MBP range or anything on the non entry desktop. They have sucked in the cellular modem for a giant bucket load of money, so now so they also have far more distractions on broader execution path. Walk and chew gum at the same time... probably not (e.g, Mac Mini and Mac Pro slumbers. The bigger and older the chip division gets, the more likely it is to slow down. ).

And AMD transition would make more sense than an ARM one for at least half of the Mac line up. However, AMD is way behind the curve on USB 4 also. But Apple is usually a bit slow on the USB uptick ( e.g. transition to USB 3 was slow).

Steve Jobs would have highly unlikely would have done some quick "big bang" to ARM either. The smaller Mac ecosystem that needs more wider set of CPU/GPUs is something he was unilkely to do. The jump to x86 was to a bigger, wider ecosystem that Apple could contribute R&D; not take over for very small quantities of products. Apple has shown extremely little signs either with Jobs or after Jobs in being some super dupe niche chip vendor. High volume products? Sure. Sub million unit per year products .. no.

The only thing different Jobs might have done would have been to start ripping into Intel when the first slightly stumbled on the 14nm transition enough to "scare" them not doubling down on playing extra aggressive reach on 10nm. If Jobs wasn't distracted with the "cracking TV" market. Most likely though Apple would just have different Mac problems now if Jobs was still around.


P.S. several years ago in this thread and the previous "Mac Pro coming" threads various folks have declared Thunderbolt Dead. USB was going to 'kill it'. Well that didn't happen at all. Still here. And if Thunderbolt successfully merges there is really no "who killed whom" endings. That pretty is pretty far from 'lost' as much as those folks chirping about TB death were from the truth.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom