APFS Benchmarks?

[mikeboss]

you got that results becault quickbench is outdated and not optimised for fast (pci-e) ssd. the test file size is so small that the whole test finishes in 1-2 second (on my MBA 13" 2015 , it has 512GB 4x ssd that does 1.5GB/s r/w) , so it could not represend real speed.

doesn't matter at all if QuickBench is outdated or not. it just measures the read/write speed using different block-sizes. the results are 100% valid, trust me.

 

[djcristi]

i repeat, since the test file size is not big (for faster ssd) the results are bad, in the model i mentioned that does 1.5GB/s sequential, in QuickBench it does not pass 1GB/s, it's a known thing that for fast ssd the test file has to be bigger to show the real speed (so that the access time can not influence the test), Quickbench finishes so fast so that's why it's outdated.

Instead i use the disk test from Xbench, which is accurate.

 

[EugW]

i repeat, since the test file size is not big (for faster ssd) the results are bad, in the model i mentioned that does 1.5GB/s sequential, in QuickBench it does not pass 1GB/s, it's a known thing that for fast ssd the test file has to be bigger to show the real speed (so that the access time can not influence the test), Quickbench finishes so fast so that's why it's outdated.

Instead i use the disk test from Xbench, which is accurate.

Sequential speed is not really as relevant for most usage (although it is for some). The most important speed test for system responsiveness is random speed with relatively small file sizes. If you have a complaint about the test being too short, a way around that is to have the program do repeated testing with those same small file sizes.

Since I am not a movie editor for example, I'm far more interested in say 4K random write speeds as a gauge of system performance.

The good news is you can run both benches, because they provide different info.

 

[djcristi]

i know that 4k r/w is the most important , that's why i got to use this softwares, i had in the past to find an USB flash with fast random 4k write , so i can install os x on it and boot from it, the only flash that could do 10MB/s was Sandisk Extreme 64GB (sdcz80-064g) , others (over 30 models tested) were on the <1MB/s league. So OS X run perfect from that Sandisk.

 

[Lazeran]

"blackmagic speed test" only tells half of the truth (at best). a proper disk-benchmark needs to do measurements using different block-sizes. like QuickBench ftom Intech.

found this post a few weeks back ->
https://malcont.net/2017/07/apfs-and-hfsplus-benchmarks-on-2017-macbook-pro-with-macos-high-sierra/
[doublepost=1505980452][/doublepost]my own findings... APFS being a tad slower across the board. interestingly, APFS was a little bit faster when I measured an HDD instead of an SSD.
View attachment 719586
View attachment 719587

IMO it's in the margin of error. Looks like APFS does not change read write speed, kinda expected.

 

[victor frisbee]

Looking for someone with a similar setup to compare SSD drive speeds with. I chose AmorphousDiskMark because it is very through and it also closely resembles CrystalDiskMark used in the PC World. I wanted to be able to benchmark across platforms. I ran the 1GiB test on my Crucial M500 1TB SSD. macOS High Sierra 10.13.1. MacBook Pro, Early 2011 2.2Ghz, 16MB Corsair DDR# 1333 (PC3 10600) RAM.

Please share your results.