If it seemed pretty fast to you, having used a 9300 Pro, then I’m sure it was fine. You’d notice what I was talking about with the 8TB Samsung QLC… for a copy of any decent size at all, it would start start out looking like a reasonable copy time then quickly fall into the realm of a spinning hard disks.
Are you using this inside an M2 Mac Pro or in a Thunderbolt enclosure? If the latter, which one? And no issues at all? I’ve used 9300 Pros quite a bit (not boot) and usually they’re fine but I’ve gotten kernel panics on both an M2 Mac Pro and an M2 Mac Studio that only happen with a 9300 Pro attached. Is your system stable?
I'm using it the 2019 Mac Pro (Intel) which only has PCIe3 slots but with a
Highpoint 1580 Rocket card that is PCI4. The good news is even this old setup seems to wring out close to the max speed of this drive, and this drive is working faster than the 9300 Pro ever did. And yes, I'm using it inside the mac Pro with the
Sonnet J3i bracket holding the 15TB Micron 9300 Pro and the 30TB Solidigm D5-P5336 (as well as an 8TB SATA Samsung 870 drive for windows/bootcamp). The U.2 drives are housed in old velociraptor heatsink enclosures (thanks to
@tsialex ) suggestion/help
here.
There is a good chance it will be faster on the 2023 with PCIe4 slots.
For a point of comparison, when I tested the 30TB Micron 9400 Pro on my setup it got 5000MB/sec reads (although only around 3500MB/sec writes) in windows, but it didnt work in macOS. So it's nice to see that with the HighPoint PCI4 card, I'm getting similar speeds to what the Micron 9400 Pro was getting under windows. I think higher performance is possible even on my PCIe3 system, meaning with the right drive, the old 2019 Mac Pro, can probably get around 6000MB/s read/writes with a PCIe4 card despite it's relatively slow PCIe3 slots.
My system seems to be stable. I still have the 9300 Pro in there (which is now working as time machine drive, and it's glorious moving between the two drives). I regularly see 2.1GB/s real world throughput between the drives via iStat menus between those drives.
Here I did a few bench marks on this. Please note my system was doing a lot of work, including a lot of disk work, so the drive was decently busy during the test, but hopefully it gives you some sense of performance under macOS.
The atto test shows writes around 2GB/s and reads around 5930MB/sec. Not sure I trust it because most I saw on iStat was around 3GB/s reads. The Aja is somewhere in the middle, and the black magic (which only uses 5GB files) seems slowest. You can see the test with larger file sizes (64GB) seems to get the better throughput. If you deal with larger files, like multi terabyte files the performance could be different perhaps. Interestingly the settings on the AJA app says it disables read/write cache by default, so in theory it's using the raw throughput of the drive. All of the above tests are on a FileVault drive, so odds are it would be say 10% faster without the encryption.
Sorry to hear about your panics. What controller card are you using for your drives?