For the 2 TB Samsung T7 Shield, I got over 800 MB/s on my M1 Mac mini, but over 1000 MB/s on my 2017 Core i5 iMac. (See
pics.) Random writes on the Intel Mac seemed to be better too. So, it seems the controller on the Intel Macs was much faster than on the M1. (I haven't benched the M4 with these drives.) The 1 TB Samsung T7 was a bit slower. My T5 runs considerably slower, but I don't remember the exact numbers.
Apple M1 + 2 TB T7 Shield:
Intel i5 + 2 TB T7 Shield:
So I just tested this on my M4. I didn't do a lot of testing, but I get the impression the rear Thunderbolt 4 ports
may be a bit faster than the front USB 3.2 ports, despite all of these drives being USB 3.2.
Samsung T7 Shield 2 TB - 865 MB/s write
Samsung T7 2 TB - 785 MB/s write
Samsung T5 1 TB - 510 MB/s write
Note however that the T7 Shield is 60% full, and after I would do stuff with it, sometimes it would slow right down, like RIGHT down, to under 400 MB/s and once even below 200 MB/s. This is my bigger concern with these drives. They don't seem to behave consistently under load, especially when they are partially full. They test well when empty though.
I've noticed occasional lagginess in Photos, and since my Photos Library is on this drive, I'm thinking it's the drive and not something else. Also, I wanted to put some more data on the external drive, so that's why I'm keen on going with a 4 TB NVMe drive. My
Qwiizlabs ES40UR giant heatsink type SSD enclosure should arrive early next week, and I'll pick up a decent NVMe drive once they go on sale. I think I've scrapped the DRAM-less idea once and for all and will go with either a WD SN850X or else a Samsung 990 Pro, depending upon what is well priced. Luckily the Black Friday sales have begun, so there should be some decent pricing in the next couple of weeks.
That is quite a difference from the YT short from today I found and reposted above??
Is that test from your MM with 24GB of Ram and the 256GB SSD?
No, that's from Reddit, but I've seen several posts here at MacRumors with similar numbers. However, the numbers really vary. Some of the M4 256 numbers are much slower, but other numbers are similar. None are much above 3000 though.
For the M4 512 numbers, they also really vary, with some well below 3000 or even below 2000, but several above 4000.
Given my very variable results with my drives above, I'm not surprised the reported numbers for the internal drive vary too. Plus I suspect some of them may have done the testing right after they set up their machines, before all the indexing or whatever has finishedd.