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
The T2 can't be deactivated any more than the PMIC or SMC on previous Mac Pro's can be deactivated.
The T2 plays a SSD controller and SMC roles on the system. Those can't be turned off (deactivated).
The T2 also protects and validates the boot firmware. If the chip was deactivated the firmware wouldn't even launch and the system would not even boot at all.
https://www.apple.com/mac/docs/Apple_T2_Security_Chip_Overview.pdf ( PDF file ).
There are option settings in the EFI firmware to adjust "secure boot". The T2 chip has a contributing role ( in that it gets the firmware going) but those options aren't its role. ( the firmware is being run may the main Intel CPU. )
[doublepost=1551416925][/doublepost]
That's a bit of handwaving. The T2 validates Apple's EFI implementation before handing it off to the Intel CPU to be run. That isn't going to brick any hardware. What it will stop is rogue hacks to the core firmware in non volatile memory. Those might brick hardware but it highly unlikely Apple is going to brick their own systems with their own firmware.
Apple's firmware may not 'talk to' some 3rd stuff but that isn't bricked. However, the T2 doesn't "inject" code into any other 3rd party piece of hardware. That isn't its job at all.
All of this falls apart once the handwaving above crumbles. The primary thing the T2 would do is eliminate the default boot storage drive from being replaced. That's about all of the "memory" it would have major impact on. Impact on RAM? No. The CPU? Only if it need rogue modifications done to the Apple firmware.
GPU? Integration with Thunderbolt infrastructure and/or warring over Metal support aren't particularly highly coupled to the T2 either.
One of the primary tings the T2 covers is being a SSD controller and implementing a SSD drive. The notion that all Macs will have a SSD drive shouldn't be surprise. Apple has been on that track for 3-4 years now.
The huge issue difference between these other Mac models and the Mac Pro is that it doesn't necessarily have the same internal volume limitations. There have been no internal 2nd drives in Mac laptops for quite a long period of time. The mini is frozen in volume while cranking up the power and CPU 'horsepower' ... so more internal volume got allocated to fans. iMac Pro ... same issue ( HDD pragmatically tossed for 2nd fan ). [ There are two NAND daughter cards though so volume is at range where could have 2nd drive if could fit 4TB onto a single card for the T2 drive. They also could crammed two SSD into the Mac Pro 2013 on the "Compute GPU" if not for thermal and bandwidth limitations. The outline for the SSD socket was there on the card. Volume wasn't the issue. ]
Whether the next Mac Pro has slots/sockets for more than one internal drive is up to Apple. It would be a pretty bonehead move to limit the Mac Pro to just one internal drive ( that's a corner case with the Mac Pro 2013 design they seem to have got at least a notion of the message on ). But internal volume space limitations have little directly to do with a T2. The T2 doesn't limit volume. Drive slots/sockets that are primarily targeted at being "data" working-space/storage/archive drives (not boot drives) have really about zero impact by the T2's indirect relationship with the firmware security setting options.
If Apple is maniacally stuck on making the Mac Pro a literal desktop ( and highly overlapping the iMac Pro in that aspect) then perhaps they are still stuck on this "one and only one internal drive" notion. If they have moved back to deskside being an target then volume for more than one storage drive shouldn't be an issue.
Similar issue with an empty x16 PCI-e slot. No internal volume allocated for it is extremely independent of the T2's primary functions.
Apple isn't going to build a container for everything possible. The iMac Pro also has a CPU socket and RAM in DIMMs. So the premise that they are going to gratuitously throw out every possible sockets in all possible cases is weak. How are they going to go on a vast Holy War again every possible GPU card on the planet and still support eGPUs ? They probably aren't.
Mac Pro 2019 will surely have a T3 chip and soldered in SSD. The question is whether Apple is going to make a big Mac Mini, or something closer to the cheesegrater with RAM slots, PCIe slots, M.2 slots, SATA bays? The new Mac Pro is supposed to be "modular" but we really don't know what that means and we probably won't until wwdc yet I still have hope!
soldered in SSD??? when the imac pro has cards?
also needs to be T3 or T2 on CPU PCI-E not PCH PCI-E
The iMac Pro may have card that fit into slots, but the T2 chip verifies that the firmware matches, so you can't upgrade the storage even if you do replace the chips. Accept it, the 7,1 will ship with a non-upgradable SSD, it is all part of Apple's security (and money-making) focus. Hopefully the drive be at least 512gb and not 128gb. I assume it will have similar specs as the iMac Pro so I guess 512gb/1TB/2TB/4TB.
Only the NAND chips is on the cards in a iMac Pro. Those are not SSDs; just daugthercards with the NAND chips and a small communications buffer. The T2's SSD controller component ( whiich pragmatically is the SSD from the viewpoint of the rest of the system) is and extremely likely will be soldered to the motherboard. The daughter cards probably have two purposes. One, can go higher capacity ( more NAND chips). Two, can do a "pull and destroy" data from system ( also can be 'replace repaired' if NAND failure. ). Those probably should be mapped into the Mac Pro design also as there is at least as much if not more volume available as the iMac Pro.
The T2 does a lot more than just being a SSD controller. It also has System Management Controller (SMC) duties. Pragmatically that puts it next to PCH (and PMIC) since that is where SMC/PMIC typically tie into a system. Attaching ti to a PCH PCI-e lanes is a far better fit. A second SSD would be far better placed on the CPU's lanes. The boot storage drive bandwidth shouldn't necessarily be the highest drive bandwidth possible. There are several workload types where the working space storage bandwidth should be higher. Poor scaling is trying to saddle the boot drive as a "do everything" drive.
The skewed viewpoint here is that the T-series long term goal is to subsume the CPU. That is really not what Apple has done so far with the T-series. SSD , audio , video , fans controller, SMC duties ( possible future PMIC duties ). Those are all distinctly typically non CPU package duties.
Apple really shouldn't particularly need to do a T3. They need to clean up the T2 (and its firmware/software stack ) first. T3 as perhaps silicon bug fixes the T2 needs perhaps, but "more" and/or better "tech porn" specs if not what they need right now. The T2 also probably needs to also get more affordable ( which also probably won't happen if focused on throwing "more" onto its plate. )
[doublepost=1551451033][/doublepost]
Technically the storage could be upgraded by replacing the two daughter cards with bigger versions. That would involve doing a full T2 reset though. You'd have a new drive (and would need to restore from backup). You can't just pull the daughter cards and stick them somewhere else without having to completely erase them and start over from scratch.
Pragmatically, there probably won't be storage upgrades any more than there were GPU upgrades for the Mac Pro 2013. These NAND daughter cards are somewhat even more custom than the GPUs were. They are incomplete SSD, so it is extremely unlikely any 3rd party is going to made them and Apple is extremely unlikely to want to sell them detached from authorized repairs. (any more than they sell CPUs, )
A money grab would far more be there being only one storage drive. It is not so much that it would be soldered in as much as users are more so herded into a single very large capacity SSD. That's would be a money grab.
Likewise if they set the baseline entry SSD option relatively high for ( 1TB and up). The Mac Pro is unlikely to have the laptop and mini SSD entry points. If they are once again targeting $2,999 ( or up) as the base entry price then there is just too much in the base price to be a SSD that small ( 100-200 GB range ). Apple probably going to ship with one [ edit not two ] GPU so that the 2013 2nd GPU bill of materials cost can be offset with more SSD/RAM.
Last edited:
well they should route more pci-e lanes into the PCH (other server boards have that)
[doublepost=1551455317][/doublepost]
1TB base is to high at apples pricing. also 2 GPU's will suck down a lot of the pci-e lanes and kill upgrades of video.
You’re predicting a machine that is literally just doubling down on everything customers rejected about the last machine, and ignoring everything they actually want.
The DMI link from the CPU to the PCH is fixed at x4 worth of bandwidth. That doesn't get any better on the servers in the current designs. ( I think you are alluding back to the old designs where they put "extra" high end SAS links in the 'Cadillac' server PCH option. That doesn't exists anymore. Neither does high end SAS in the PCH. )
That's a type. It will probably ship with one instead of two 2GPU. That"s the trade-off being made: save on not having 2nd GPU cost and then assign more expensive to something else so overall cost doesn't go down.
[doublepost=1551457729][/doublepost]
Actually not. I'll fix the typo. The savings in 2nd GPU cost (because dropped) will probably be put into raising RAM/SDD capacity.
P.S. There will probably be some BTO configuration that has two. Not having having some BTO configuration with two would be missing the point about just as much if they completely disallowed it. The problem was "all Mac Pro users get two" not so much that there were one or two mandated.
Last edited:
super micro
X11DAi-N
X11SPi-TF
and others do it check the block maps in the manuals.
The thing is though, Mac Pro users do NOT want this. So if this is Apple's plan then they're shooting themselves in the foot right off the bat. You gotta make products for your target audience!
I'm a Mac Pro user and it wouldn't bother me in the slightest if the boot drive was locked away behind the T2 chip -- as long as it's limited to the boot drive. Give me a secure boot volume and some m.2 slots (or equivalent) and I'm totally happy.
OK. In the SP level chipsets. I was thinking of just the servers with Intel W. ( which is too narrow since not in zone of what Apple would probably select. ). C422/X299 there is no "PCI uplink" feature on the chipset. The Xeon SP specific chipsets would mean being saddled with those CPU options also. I highly doubt Apple is going that way. The 620 series TDP is about twice as high and cost more. The SP socket is bigger. The SP prices are substantive higher ( unless toss base clock speed out the window).
Second it isn't clear from the manuals but this PCI-e Uplink seems to be more targeted so at offload SATA ( like it was for SAS on the older ones). [ manuals mention don't get QAT on the uplink which is presumably Intel's Quick Assist Technology ) If so, that carries the presumption that Apple is going to put SATA device bays in the next Mac Pro. I wouldn't bet a large sum on that. If multiple SSD drives the Uplink may not buy much versus just directly hanging the 2nd SDD directly off the CPU without the PCH as some sort of extra "middle man". The link to the embedded SATA controller in the PCH would be a different value prop.
The X11DAi-N is a 621 which in only buying a x1 link. If there are >5 HDD SATA drives hooked up that makes a value add difference. If there is 0-2 drives probably not so much. Doubtful Apple will be targeting HDDs at all. Perhaps they'll look at 2.5" drives as an SSD alternative form ( and a secondary HDD usage). Those SuperMicro boards each have 10 SATA connectors. The 2006-2012 models didn't have 10. That number isn't going up and extremely likely going significantly down.
[ later update with an edit]
P.S. it ended up I had some time after posting this to dig out the Intel 620 spec pdf ( somehow didn't find it on initial quick search.). The PCI-e uplinks are aimed at primarily aimed at two targets : Intel's Quick Assist Technology (QAT) and the fact the the 622 (and up) members of the series have 10GbE ports that need to deliver data. There is some access to some low bandwidth boot/system management stuff also. ( the usb , pcie lanes , sata aspects of the PCH are all on DMI only, ). QAT is Intel's crypto engine (on a segregated PCI-e data stream for 'normal' DMI traffic). While the T2 does some encryption stuff it isn't a generic crypto offload engine. You send data there to be encrypted ( and the CPU doesn't have to do that work) but it isn't a general service for random processes. It is doubtful Apple wants to be hard bound to Intel's crypto API long term.
For the 10GbE (and 1GbE) ports... Apple already had to write drivers for a discrete 10GbE controller. They are unlikely to use Intel's solution there either. ( hooking the discrete 10GbE controller(s) directly to the CPU works just fine. ). Apple probably isn't going to use Intel's Wifi that they are weaving into their PCHs either, because they have done drivers for what they want in that space also.
The 620 series chipset PCI-e uplink doesn't buy anything special for Apple's Mac systems (not generic bandwidth). It is far more useful if buying 100% into Intel specific solutions. If Apple is considering kicking Intel to the curb then it has even less "value add". ( not necessarily ARM but AMD). If Intel had a 2 year manufacturing process lead over all the other options then maybe but at this point in time that "Uplink" feature is at least as much about lock-in as bandwidth upside.
The C422 is far better fit for what Apple is probably going to be doing.
Last edited:
If you show me a pattern, I’ll show you the exception. All the macs in your pattern were never discussed before they were announced. The new Mac Pro was discussed before it was announced. That already breaks the pattern. So it's also possible that the new Mac Pro breaks the pattern in many other ways.
But anyway. The T2 (or T3) would be a good thing if used correctly, so I'd be happy to see your pattern being right.
Good. I'm very happy to hear that. And you point out largely the same problems as I did, so that makes it likely to be an accurate estimate of the situation.
Precisely.
Not really true. The iMac Pro wasn't called out by a specific name but it was definitely mentioned as a focal point during the April 2017 pow wow about professional Macs ( that meeting solely was not about the Mac Pro). They said they were working hard on putting the iMac more firmly into the Pro space ..... which ended up being an "iMac Pro".
Apple paid about $400-500M for a SSD controller company several years ago and development after acquisition. Apple is very much in the SSD controller and SSD making business for their own devices. Every Apple iOS , tvOS , Mac is going to come with a SSD going forward. Apple's "Flash SSDs are the future" mantra is ever bit as highly visible as a couple of other of their core objectives. The Mac Pro is extremely unlikely to be detached from that drumbeat.
Like "security" is a big Apple strategic focus point at the moment. Mac Pro probably isn't going to be substantively less secure than the rest of the Mac line up. Apple isn't likely going to say "we are the security company except for this product".
In 2019 an Apple SSD for a Mac looks like a T2. The other designs that don't are just old.
Apple is not really talking about the new Mac Pro much at all. They are talking about what didn't work so well with the last one. And they are trying to tap dance around the hole in the line up that is beyond stale while not talking about exactly how/when to fill it.
That is far more about Apple's colossal screw up in Mac Pro product management and not so much about a strategic objectives they have. That really isn't part of any "pattern" in the feature composition of the other products. (besides some of those also having SNAFU product management too with long Rip van Winkle gaps in updates).
I think it would help Apple tremendously if they had more Drive Endurance and MTBF numbers to go alone with their T2. I think most folks that are probably targeting will be happier with more than one internal drive. But there is a decent number also who are get fixated on the T2 drive in "failure" mode aspect. If the T2 drive is likely to last as long as the supported system service window then even fewer folks will care about the limitation of the controller being attached to the board.
The article doesn't exactly say that this was Intel.
"...which if one recent European retailer is to be believed, could mean an increase on the top end processor by almost double over Skylake, with a listing showing a retail price for an 8280L (?) .."
If someone is putting a placeholder into their inventory system then may just make up a price.
Similarly, Xeon W 3175x
Circa Janurary around CES time about '$8K'
https://www.anandtech.com/show/13788/intels-unlocked-28-core-xeon-w-3175x-oem-tells-us-around-8k-usd
Released around $3K
https://www.anandtech.com/show/13748/the-intel-xeon-w-3175x-review-28-unlocked-cores-2999-usd
Not sure if Intel is playing "rope a dope" but if Intel can tempt AMD to raise their prices and then Intel doesn't raise prices, Intel will have blunted the blow of being stuck on process for another year or so in this space.
It could also be a sales tactic with customer's to set expectation higher and then lower so it appears getting a discount.
[doublepost=1551538260][/doublepost]
I haven't see how Intel is going to name the AP versions.
https://www.anandtech.com/show/13586/intel-offers-more-cascade-lakeap-performance-numbers
If the 'L' at the end was "low clocked and doubled" then this could be dual die package. However, I would expect them to use the 9xxx to indicate that those are a different socket. Again it may be a inventory placeholder entry so the product name may not be correct, but perhaps the price is in that range. If Intel is putting two XCC dies into a package it probably will be in the range of double one XCC die with same die core count. The base clock will go done so it would be less than double ( gets the lower clock 'discount' but also getting two. )
Last edited:
Yet noone would call the iMacs or new Minis upgradable , even if they are in some respect .
After the tcMP fiasco , Apple can either introduce a trashcan 2.0 / Mini Plus / xMac - or a container for everything possible . I think it's that black and white , they already covered all shades of grey .
Not really. The Mac Pro 2013 was talked about before E5 1600 v2 system was introduced. Apple made indirect references to a "future Intel Xeon" CPU. They could do exactly the same thing in April. Intel is already shipping Cascade Lake SP to the major cloud players now. By April there will be a more than a few "Intel confidential" slides about Cascade Lake SP info , graphics, roadmaps , etc. so it will be a fairly poorly kept 'secret' at that point for Intel. Shipping thousands of the product and having hundreds of different end user customers using the product off of cloud installations ... that is probably not a secret product anymore. Intel would be able to take Apple to task for revealing a super classified secret that there is another Xeon W generation coming.
It is unlikely Apple is going to wait until they are about to ship in volume to say "something". They may say essentially the same something at WWDC, but that isn't say much.
The only quirky thing is not about these Cascade Lake SP timetables, but about where the Intel W derivatives are at. So far it is mainly the presumption that the W derivatives will ship about the same time ( and that Intel skipped the Skylake-X Refresh ... that really didn't do much for W anyway. ). I wasn't expecting that stuff to start to leak until March-April anyway.
Apple waiting until WWDC to talk about the next Mac Pro is far more indicative that Cascade Lake is not what the hang up is. That would be far more likely that they did something dubious like bet the farm on Ice Lake (and now Cooper Lake) , Navi, and/or extremely goofy weirdo tech like next version of Thunderbolt which has no initiative to come soon at all. Announcing at WWDC is far more likely indicative that Apple has completely screwed this up worse than it looks like now. They'd be betting on the Cirque de Soleil extravaganza aspects of WWDC mask the odor coming off the Mac Pro product management. That the blunder would be masked in dazzle and a new variation of "can't innovate my ass" declaration.
Apple talked around/at the next Mac Pro in April 2017 and April 2018. So it is unexpected that they will talk at the next Mac Pro in April 2019. They have established a two year track record on this. Over the last two years WWDC has had NOTHING specifically on the Mac Pro. zip. nada. So this year they are going to flake on April and move it to WWDC? Really? The "broken analog clock" track record of Mac Pro and WWDC.
2011 : nothing
2012 : something (weak refresh of what they had. Please ignore all those new E5 workstations announced. )
2013 : something ( end of the year.)
2014 : nothing
2015 : nothing
2016: nothing
2017" almost nothing ( how about a iMac Pro instead? )
2018: nothing
The workstation motherboard vendors will probably be highly chatting up Cascade Lake with examples the week before WWDC at CompuTex. Dell , HP , Lenovo too may have some previous boxes the week before. But that makes Apple waiting until after all the more a screw up. If they are going to be arriving substantially later than those competitors then they need to say so earlier not after.
[doublepost=1551548086][/doublepost]
The Mac Pro 2006-2012 wasn't a container for everything possible. They never were in that particular business.
Go back in the archives and there were always folks around (a small few of the same folks now ) who regularly pointed to the additional space they filled in a super deluxe top end HP/Dell/IBM workstation. The longer card , more drives, more vaired tape back-up options , bigger power supply, more dangling SATA power cables available , PCI (not PCI-e ) slots for ancient cards , not enough DIMM sockets ( 8 per CPU socket) , etc.
The grey (in between) is exactly where Apple hasn't gone. A more even mix of a subset of Slots/sockets that Apple criteria has a better functional fit to analog with some mainstream options.
default GPU and boot SDD and 1-2 PCI-e slots ( a x16 and perhaps a x4 empty PCIe physical standard slot.), 1-2 M.2 slots, 8 DIMMs per CPU socket, along with four Thunderbolt ports. They haven't done that with any Mac to date. Somewhere closer to the middle. Not trying to cover everything, but covering some more options.
Priced in the >$2,499 range it won't pragmatically be an xMac as price is one of the primary issues with the core xMac crowd. It wouldn't be anything like a Mini Plus either. More than one fan not a "cylinder" ("trashcan").
The notion Mac Pro 2013 design was the only option in the "grey" area is grossly myopic. A desire that retreating all the way back to cloning a HP Z8 or back to the almost exact driving constraints on the Mac Pro of the last decade isn't the only possible option. HDD aren't the dominate storage option they once were. ODD even more so for "sneaker-net" data transfer. The max cores in a single CPU package has very substantively changed. The max memory that one CPU package can directly drive has substantially changed.
Over a decade of technological change ( 2005-6 --> 2017-18 ) extremely likely widens the design options... not narrows them if not dogma devoted to form over function.
This video has a 2019 Mac Pro mock up design that seems like it could maybe turn out to be reality.
NEW Mac Pro & Apple 6K Display (2019)
(Skip to 5:33 to skip the intro):
NEW Mac Pro & Apple 6K Display (2019)
(Skip to 5:33 to skip the intro):
Last edited:
That design strikes me as exactly what the pessimists expect Apple to deliver - taking everything that made the 2013 model unsustainable and dialing it up to 11, if not 12.
In other words, I expect the actual 2019 Mac Pro to only share the "2019" and "Mac Pro" with this concept.
In other words, I expect the actual 2019 Mac Pro to only share the "2019" and "Mac Pro" with this concept.
Conjecture,conjecture,conjecture,conjecture,extrapolate,extrapolate,extrapolate,extrapolate...
I certainly think that the days of ordering a standard graphics card are gone (at least on the Mac front). Apple has already dabbled/partnered with the egpu what would make anyone think that they wouldn’t do a module/capsule type thing. Where you can’t exactly buy the cares but by the module and install it? Just taking it Andy putting it back in the box and maintains a healthy up cost to consumers. It’s the subscription of owning a Mac.
Well said .
True . The limited GPU offerings in particular have always been a hotly debated issue as I recall .
As for everything possible - of course not , that was just hyperbole .
That would require an entire line of MP models to cover every possible user's needs, even excluding the extremes .
But the cMP came close to being the perfect universal workstation computer at the time ; with the modifications and upgrades available ( or rather supported/ usable ) today, it would have been perfect back then .
The same concept with updated tech would be perfect today .
Here I disagree .
I believe the entire Mac line is located in that grey area , and covering pretty much all of it .
Maybe with the exception of the MBA .
Grey area, meaning it's just crippled enough to not be what pretends to be and falling behind in upgradability, options, pricing or long term usability .
Like a Netflix Original - it almost doesn't suck .
Just a few more options/ slots will not move a future MP into a grey area not already populated by Apple, I think .
An updated, smaller and lighter cheesegrater will do that .
Well, I'm certainly not rooting for 3.5" spinners or optical bays, but technological change only really works if the costs adjust at the same rate .
The only external drive tech evolution in a decade, that doesn't come at a premium, has been USB 3.x .
Everything TB is still silly expensive, so are high quality, silent (!) external drive enclosures .
Hence , a modicum of affordable internal storage is very much a factor in a workstation computer - and to this day that means at least some 2.5 SATA bays .
True . The limited GPU offerings in particular have always been a hotly debated issue as I recall .
As for everything possible - of course not , that was just hyperbole .
That would require an entire line of MP models to cover every possible user's needs, even excluding the extremes .
But the cMP came close to being the perfect universal workstation computer at the time ; with the modifications and upgrades available ( or rather supported/ usable ) today, it would have been perfect back then .
The same concept with updated tech would be perfect today .
Here I disagree .
I believe the entire Mac line is located in that grey area , and covering pretty much all of it .
Maybe with the exception of the MBA .
Grey area, meaning it's just crippled enough to not be what pretends to be and falling behind in upgradability, options, pricing or long term usability .
Like a Netflix Original - it almost doesn't suck .
Just a few more options/ slots will not move a future MP into a grey area not already populated by Apple, I think .
An updated, smaller and lighter cheesegrater will do that .
Well, I'm certainly not rooting for 3.5" spinners or optical bays, but technological change only really works if the costs adjust at the same rate .
The only external drive tech evolution in a decade, that doesn't come at a premium, has been USB 3.x .
Everything TB is still silly expensive, so are high quality, silent (!) external drive enclosures .
Hence , a modicum of affordable internal storage is very much a factor in a workstation computer - and to this day that means at least some 2.5 SATA bays .
