Benchmark Results: Sequential Read And Write (CrystalDiskMark)
With NTFS compression enabled, the Samsung drive demonstrates slightly worse performance in sequential data reads. The drop in sequential write performance is very clear, however. The resulting fragmentation during compression by sparse clusters does not seem to be crucially affecting throughput, though the difference is measurable.
However, in the conclusion, it is stated that compression ends up writing more vs. uncompressed NTFS, thus consuming more PE cycles. Shouldn't the opposite be true? When writing to the file system, if a file is compressible it should take up less space and therefore conserve more PEs (though actually compressing the files for the first time should result in more writes).
Why does on-the-fly compression result in more writes even though the amount to be written is smaller?
Keep in mind, whatever storage option you use, you need room to install updates on top of installing the game, most especially for MMOGs. This means room to download the update AND install it.
Correctomundo. Compression involves replace repeated occurrences of data with references to a single copy of that data existing earlier in the input (uncompressed) data stream. That's why it's not right to think of a compressed archive as a container that stores any given file into a discrete space. If anything, the files kind of overlap in a big mixing pot.
When you compress on the fly, you have to completely decompress all the files in an archive and recompress it when you're done. Hence it's all random transfers for the most part.
BrightCandlePresumably this negatively impacts Sandforce based drives more than the Samsung by making the data stored compressed? Any chance you can do this with a Sandforce drive to see the impact?
It's not a sequential transfer. Plus it's already precompressed data. Nothing SandForce can do about it. SandForce, Samsung, it's not going to make a difference.