Correct, my initial suspicion being either driver/firmware related involved the controllers of the USB keys of which I'm fairly confident kingston did some bait and switch gimping the lineup. There is no fing way someone else can post such decent speeds for these drives and not be relatively close for others benchmarks unless the drivers used are gimped in some way which being USB I highly doubt even though it would be worth a try. Windows in crystal diskmark does report pretty gimped speeds for 4k random as well. I highly suspect Kingston did a bait and switch on the controllers/Nand they use in the drives. I'll be sticking to samsung (the 128gb in the batch I bought is much better) until the joke competition catches up for flash drives. For SSD's even some chinese brands are actually on par for performance I tried a EDILOCA 4tb which is 30% cheaper and actually out performs most pci gen 5 ssd's (hopefully it lasts just as long as my samsungs). Tomorrow will also confirm if the initial drive having issues was a bait and switch or just a terrible controller/nand for random r/w.Run CrystalDiskMark on a PC and bench the drives there (plugged into a USB2 port, then a USB3 port) and post the outputs. I suspect you just have drives with terrible random write performance, so any writes during boot block reads for ages.
Thanks to everyone but I'm certain it is just the drive itself and not anything pi related.
Samsung BAR Plus 128GB
Sequential write speed 61134 KB/sec (target 10000) - PASS
Random write speed 6170 IOPS (target 500) - PASS
Random read speed 2905 IOPS (target 1500) - PASS
Even a Lexar D40E 128GB actually performed worse and failed the random writes. All are the same specs/generation.
Sequential write speed 28487 KB/sec (target 10000) - PASS
Random write speed 344 IOPS (target 500) - FAIL
Random read speed 1773 IOPS (target 1500) - PASS
I'll just stick with the samsung since the Nand is fresh and likely to last.
Statistics: Posted by solace50 — Thu Sep 11, 2025 5:10 am