brianm

DuhSesame said:

melgross said:

DuhSesame said:

Mike Wuerthele said:

DuhSesame said:

Mike Wuerthele said:

DuhSesame said:

Mike Wuerthele said:

DuhSesame said:

That's one thing I'm concerned about the T2 because they bottleneck the SSD performance since both flash modules and controller keeps improving over time.  Judging on the iMac Pro, I'm sure those "SSDs" are just raw flash modules, whereas the T2 chip ties the controller within.  That limited any future performance improvement, but every computer with an M.2 running PCIe 3.0 have the potential to upgrade a faster SSD.

Maybe that's not a problem for a Mac Pro, but not the iMac Pro and MacBook Pros.

I'm not certain that they'll get that much faster. While the theoretical max speed of the PCI-E 3.0 x4 connector with a M.2 slot is 3600 megabytes per second or so, the reality is a bit less, perhaps 2900 megabytes per second.

I don't think that's the case.  x4 should run at 3.9GB/s maximum and M.2 is just another form of it.  Source?

 https://www.pugetsystems.com/labs/articles/Overview-of-M-2-SSDs-586/

Their numbers aren't including overhead for the PCI-E connection itself. Best case, that overhead is 10%, thus the 3600 megabytes per second.

It's actually around 1.54% for PCIe 3.0, so 985MB/s.  You got 3.9GB/s when it times four.

https://www.tested.com/tech/457440-theoretical-vs-actual-bandwidth-pci-express-and-thunderbolt/
https://www.overclock.net/forum/355-ssd/1489684-ssd-interface-comparison-pci-express-vs-sata.html

But whether it's 3.6 or 3.9, it left both the iMac Pro and MacBook Pros lots of room for improvement.

Huh, TIL.

Still, there is NVMe storage overhead too. I just don't think there's as much leeway for growth as you do, is all.

I've talked to a friend about this and he thinks the T2 is better optimized for APFS.  That might be true, but I'm not sure some software improvements will overcome hardware advancements.

Then again, I know nothing about SSD controllers.  All of them could have different architectures.

There are over a half dozen major SSD  controller architectures out there. New ones come around every year or two.

Right, which is why I think locking your SSD in the T2 isn’t a good idea.
1. No upgrade options, once obsolete, forever obsolete.
2. (Some) nearly impossible to remove, you’ll need to perform SMD soldering.

Change security settings after setting up the initial admin account so you can boot from "external drives" - which may allow booting from PCIe cards as well if the OS itself has built-in drives for that - worst case from an external Thunderbolt drive that could be using whatever the latest technology is.

If it's not the T2, it's another controller of some sort that handles the storage and whatever connector the motherboard comes with like M.2 that has limits as well.
in the case of the T2 though, it looks like whatever apple does is just a PCIe connection to the flash chips (in the MacMini or MacBook Pro's it's soldered instead) - the T2 handles all the stuff the "on-board" SSD chip functions would do plus a more integrated encryption solution.

benji888 said:

This is bogus:

1) the person trying to steal your passwords has to first have access to your Mac.
2) he then ran some app to get your passwords...I’m guessing all this app does is enter your Mac’s password for the keychain items automatically and then extracts them and displays them all in a list, so, again, back to 1).
3) you can also lock keychain so that it has to be opened with a password, so they’d need not only your Mac’s password, but keychain’s password...this is not the default for keychain.

1&2) It can be chained with other exploits - zero-day, or known for older MacOS versions which this exploit affects. 3) try this, and see how long you can run with keychain staying locked, especially while browsing sites that require logins, apps during startup, etc...

I just finished testing the i3, i5 and i7 versions of the Mac mini with a variety of tests including power & heat generation. These were all tested stock 8 GB ram, all with 256 GB storage - not opened, no change in thermal paste.

 I found that with the i7 during a first test, or beginning of a long CPU intensive task starts off fast and got a result of about 1200 in Cinebench, then with additional tests (or continued long run) it scales down power usage from peaks up to around 150 watts to in the 90 watt range when Cinebench scores dropped as low as 1173 - but after 30-60 seconds the fan ramps up, and the performance goes back up - along with electricity usage where it climbs back up to 110-120 watt range and performance to match. I actually had the highest Cinebench score after the CPU was hot, and the fun was running at high speed when it hit reliably in the 1200 range with a peak of 1228. It does pump out quite a bit of heat.
The i3 in the testing I was doing did not seem to have much of a change at all between tests, and even after running for a while I could barely hear the fan with barely warm air coming out.
The i5 had a little bit of a variance, more than the i3, but not nearly as much as the i7 - it did generate more heat than the i3, no where near as much as the i7 - fan also barely audible. Running the same tests with the old 2012 quad-i7 it's fans were for sure the loudest - even the base 2018 i3 is faster than the 2012 quad-i7 in all tests, with the GPU scores being over 4x faster in CineBench & GeekBench.

 The Cinebench GPU test was also faster on the i7 where it reliably gets 44fps while the i3 & i5 were only getting around 41 fps. It is strange with all of them having the same integrated video, I even went back and re-tested to verify when I noticed it. (running a more advanced 3D app like Unigine Heaven or Valley it was only around 4fps so useless for gaming without an eGPU - I did test the BlackMagic eGPU with Radeon Pro 580 it jumped the same Heaven & Valley tests to around 36 fps with Ultra or Extreme settings - over 60fps with medium settings. while using around 130 watts of total power for Mac mini plus eGPU. Heaven in particular did get a slightly higher frame rate with Ultra settings on the i7 with eGPU than the i3 or i5 models which were only getting 29 fps)
A better eGPU with an Nvidia card like 1080Ti or newer should be very usable not only for games, but things like video rendering with Premiere.

the wired one doesn't have 2 USB ports like apple's old wired model or most of the competition... otherwise it looks great.