Pylons

As far as I know no Macs support 20 Gbps USB 3.x 2x2, so the Pro model will be limited to about the same 1000 MBps as the non-pro.

Also the article states that the non-pro version is faster because of the upgrade “from USB 3.1 Gen 2 to USB 3.2 Gen 2”. This is not true. USB 3.1 and 3.2 are the same. (Thanks USB IF!) Speed is determined by the Gen 1 (5Gbps), Gen 2 (10 Gbps) or Gen 2 2x2 (20 Gbps).
Yes this is super-confusing, and adds to the need for journalists to explain more to readers (and complain about stupid naming).

(Yes the upgrade from SATA to NVMe reason for speed upgrade is correct, but the USB bus is now the limiting factor.)

The Xeon E3 v6 series uses socket 1151 so it tops out at 4 cores. Apple will not make a Mac Pro using this CPU as a 4-core entry model because then they would have to design a completely different motherboard for the models with more than 4 cores.
In fact Intel always launches their consumer CPUs earlier than the enterprise CPUs, and the Xeon E5 v4 was released just in March, and v5 (Skylake-EP) is rumoured for 1H 2017 (see e.g. http://wccftech.com/intel-skylake-e-lga-3647-hexa-channel-memory/) with a new socket LGA 3647. Jumping to v6 but being limited to 4 cores makes no sense.
Releasing an updated Mac Pro any time before 1H 2017 is probably not likely, since they would have to make a major redesign for the new socket in 1H 2017 anyway.

The current Mac Pro is a very unusual design choice because the space constraints limits the setup to one CPU. This would mean that the CPU model doesn't have to have  support for dual CPUs. Apple uses E5-1xxx v2 series for the lower end but Intel probably figures users with high performance requirements may want to use dual-CPU machines so they limited the E5-1xxx v2 to 8 cores, forcing Apple to use the E5-2xxx v2 series for the 10 and 12-core models.
The E5-1xxx v4 still has max 8 cores, while the E5-2xxx v4 goes all the way up to 22 cores. E5-2xxx v5 is rumoured to have up to 26 cores!
(It will also support 6 channels of DDR4 memory, which is completely unnecessary in the Mac Pro form factor.)

It will be interesting to see what 2017 brings.

I think I just found the reason for the confusion about the CPU models and cache numbers. It seems to be that Apple are using standard Intel models and not any special ones. It is just that Apple adds the amounts of L2 and L3 caches together (!). Intel ARK lists L3 cache, while there is 1 MB of L2 cache per core. Then the numbers add up to the following Intel models with their corresponding recommended prices:
I repeat the following from my earlier post in another article:
with the choices of 24 and 28-core models, Apple is assuming people who need the higher CPU performance also need more than 1 TB of RAM. There are extremely few use cases that need more than 128 GB of RAM, so I think there should be options for non-M versions too, in order to save those $3000.

Looking forward! Crossing fingers for models with more key travel and without Touch Bar, but I'm not holding my breath.
Has anyone kept track in the past of how long time has elapsed from these database listings to actual product launches? (Yes, the article states "in the coming months, or even sooner", but I was thinking of a table or statistical analysis.)

Oh please please Apple, give us a Pro-focused Mac mini with standard user-replaceable m.2 SSD and SO-DIMM DDR4. Having integrated non-upgradable SSD and RAM is ridiculous in anything but the thinnest ultrabook form factor.

The previous 27-inch iMac can also use 128 GB RAM, so that is not new. It is just that Apple did not offer to sell it with 4x 32 GB SO-DIMMs. You can buy 4x 32 GB DDR4 SO-DIMMs and install yourself, or have them installed by a third-party Mac seller.

Beats said:

Do you guys think Apple Silicon Macs wil come this year? If so, it would be wise to wait. I would love one for my recording studio.

I have no inside info, but my personal guess is that MB 12" or MBA 13" would be the first to go ARM, maybe late this year, and I don't expect iMac or Mac Pro until late 2021. But I hope I am wrong and we will see them earlier.

Regarding putting an ARM Mac in a recording studio I'd be very careful for the first year, until all software and plugins have been updated to natively support it. Unless you're used to always updating your software to the latest version as soon as it comes out. In my experience, from the 3 or 4 studios I have recorded in as a musician (and previously having a small home studio myself), studios are very reluctant to experiment with the latest and would rather prefer something that is already tried and tested. That said, music production is not as computationally heavy as for example video editing, and even advanced effects or VST instruments could potentially run well under Rosetta emulation.
Note that I am not saying that studios will prefer Intel Macs indefinitely. Of course they will also move to ARM, but as with anyone depending on a production environment for living, it may take a little longer for ARM Macs to get there. 

mcdave said:

So that’s the full line with a T-series chip? The end of the Hackintosh?

That’s a large CPU performance hike, Apple must be confident its own silicon can deliver more.

Quite likely the beginning of the end of the Hackintosh yes. But I guess Apple won't cut macOS compatibility for non-T machines until they are all classified as vintage.

Regarding the performance hike, I think perhaps AppleInsider may have wrongly interpreted the "65% more Amp Designer plug-ins" from Apple's page as "65% faster performance", which is not quite true. This comparison is also based on
"Testing conducted by Apple in July 2020 using preproduction 3.6GHz 10-core Intel Core i9-based 27‑inch iMac systems with 128GB of RAM and shipping 3.6GHz 8-core Intel Core i9-based 27‑inch iMac systems with 64GB of RAM. Tested using Logic Pro X 10.5.1 with project consisting of multiple tracks, each with an Amp Designer plug-in instance applied. Individual tracks were added during playback until CPU became overloaded. Performance tests are conducted using specific computer systems and reflect the approximate performance of iMac."
where actually the double amount of RAM could have played a role.

Finally, I find it a little bit strange that 1 GbE is still standard, when 2.5 GbE would cost about $2-3 more. It is becoming very common on AMD B550/B570 and Intel Z490 boards.
On the other hand, $100 to upgrade to 10GbE is actually a very competitive price. Kudos to Apple for that! (That goes for the Mac mini as well.)

Thanks for the article! I've been following the eGPU scene quite a bit but have yet to make a purchase. Currently leaning towards the Razer Core X because it's widely available through regular retail channels also outside the US. An eGPU is a great way to make any Mac a lot more upgradable and future-proof. My worry however is that Apple may not be taking eGPUs seriously enough, for example taking long time to fix stability issues like ones mentioned by ShapeshiftingFish. For more reading I can also recommend taking a look at the Buyer's Guide at eGPU.io, https://egpu.io/best-egpu-buyers-guide They also have a large community (as already mentioned in the article) with lots of helpful users in the forums.

macgui said:

I bought a popular 2880 m.2 NVMe stick and  a three different enclosures, two 3.2/G1 and one 3.2/G2. I also got a SanDisk Ultra Extreme Pro Whiz Bang NVMe external drive.

They all seem to perform as advertised, and that disappoints me.  None of these drives come anywhere near the theoretical maximum bus protocols, to wit:

USB 2.0, up to 480Mbps
USB 3.0, up to 5Gbps 
USB 3.1, up to 10Gbps
USB 3.2, up to 20Gbps (Gen 2)
USB 4.0, up to 40Gbps

There's been some musical chairs with the USB naming convention, and I want to avoid that as I don't think it germane to my question.

The real world speeds seem to be about half the max or less. Even counting for overhead, the throughput never seems to come close to saturating the bus. To get greater throughput the practice seems to be create another protocol, get another type of drive, connector, and external case, and get a fraction of the new theoretical performance limit. Not to mention none of this equals the speed of Apple internal flash memory (no doubt part of that is due to being internal).

The tested WD drive above seems to get <1Gbps, nowhere near the TM, but that's fast? Compared to a spinner, sure that's true. But it seems like getting an external that comes close to TM is like getting 50,000hrs out of an LED or CFL. (Of course, someone here has done that.)

I'm not prone to conspiracy theories, so can someone point out the whys and wherefores of science for my missing the point? Is it a consumer/prosumer vs commercial user thing? Allowing for scary RAID performance? Off for some Tylenol. I don't know if I even got my questions right. 

Stating that USB 3.2 is up to 20 Gbps (Gen 2) is not completely right.
USB 3.2 Gen 2 (which is actually 2x1) is 10 Gbps, just like USB 3.1 Gen 2.
I found the table at https://en.wikipedia.org/wiki/USB_3.0#USB_3.2 very helpful.
What is 20 Gbps is USB 3.2 Gen 2x2. It is worth noting that Apple does not have any device supporting this, so USB 3.1 Gen 2 at 10 Gbps is the fastest USB you'll get with any Mac.
(USB-IF has really messed up the naming! It was simplified to "Superspeed USB 10 Gbps" and "Superspeed USB 20 Gbps" 2019, but hasn't caught on yet.)

auxio said:

From what I understand, a lot of it has to do with the overhead of the USB protocol itself.  If you want to get into the technical details, there's an old discussion here about USB 2.0.  The TL;DR answer at the end states:

> Where is the bottleneck in today's implementations?

As one can see above, there is no botleneck, all raw bit-time space is eaten by protocol overhead.

At least with USB 3.1 Gen 2 and 3.2 Gen 2 (and 3.2 Gen 2x2, jeez this is a mess) the encoding is more efficient at 128b/132b instead of 8b/10b.