Readspeed + Spinrite = Speed Boost

[Alasala]

Hi all, first time poster here.

So I have 3 drives attached to my motherboard directly, and I also have a hotswap caddy that I have 4 other drives that I can switch out for backups etc. I was really curious about Readspeed so I downloaded it and initially ran it 4 times, so I got 4 sets of data for my 3 main drives, and 1 set of data for each of the swapable drives.

I was taken by how consistent the numbers were across each of the runs, but there were two obvious things jumping out at me. My SSD was much slower at the start of the drive, and two identical Samsung HD753LJ 750GB drives were about 5-6MB/s different. So I ran Readspeed a couple more times on each of these.

I then ran Spinrite at level 2 on my SSD and then re-ran Readspeed - the performance at the start of the drive now matches the rest of the drive, increasing the speed at the front by 100MB/s but appears to have decreased at the end of the drive. I seem to recall @Steve saying something about this on security now, that the 540 MB/s will be because the drive is actully marked as blank there so gets max speed indicated, but after running Spinrite there now is actual data written there hence the decrease? Does that sound right?

Crucial SSD BX300 240GB
Run 1 240GB CT240BX300SSD1                432.5   529.2   532.5   540.8   540.8
Run 2 240GB CT240BX300SSD1                432.3   529.1   532.6   540.9   540.8
Run 2 240GB CT240BX300SSD1                432.7   529.0   532.6   540.8   540.8
Run 4 240GB CT240BX300SSD1                432.6   529.1   532.7   540.9   540.9

After Spinrite at level 2
Run 1 240GB CT240BX300SSD1                532.3   529.7   528.8   532.8   532.4
Run 2 240GB CT240BX300SSD1                532.1   529.5   528.6   532.6   532.3
Run 3 240GB CT240BX300SSD1                532.1   529.5   528.6   532.6   532.3
Run 4 240GB CT240BX300SSD1                532.3   529.6   528.7   532.7   532.4

My other 2 main drives are 4TB and 2TB spinners and nothing really jumped out at me for either of these.

Then my 2 identical Samsungs, here are the results. Again consistent numbers, albeit the first half of the drive was 5/6MB/s different between the two drives, the second half of the drives are pretty similar. I ran the slower drive through Spinrite at level 4, and now the numbers are pretty much identical at the front of the drive, still a 6 MB/s discrepancy at the 25% point, the rest of the drive showing no difference.

Faster Drive
Run 1 750GB SAMSUNG HD753LJ                86.5    80.3    76.5    62.5    46.7
Run 2 750GB SAMSUNG HD753LJ                86.6    80.3    76.5    62.5    46.7
Run 3 750GB SAMSUNG HD753LJ                86.8    80.3    76.5    62.5    46.7

Slower Drive
Run 1 750GB SAMSUNG HD753LJ                80.7    74.3    77.2    65.2    46.0
Run 2 750GB SAMSUNG HD753LJ                81.0    74.6    77.1    65.5    46.0
Run 3 750GB SAMSUNG HD753LJ                81.5    74.1    77.2    65.2    46.0

Slower drive after Spinrite at level 4
Run 1 750GB SAMSUNG HD753LJ                86.6    74.8    77.1    65.4    46.0

I think this limited test shows pretty conclusively that Readspeed is highly accurate and consistent across multiple runs, and that a maintenance run of Spinrite gives a clearly observable boost to the read speed of the SSD drive and even for the start of the Samsung 750GB spinning drive.

I can't wait for 6.1 so I can Spinrite my other drives.

 

[AlanD]

I then ran Spinrite at level 2 on my SSD and then re-ran Readspeed - the performance at the start of the drive now matches the rest of the drive, increasing the speed at the front by 100MB/s but appears to have decreased at the end of the drive. I seem to recall @Steve saying something about this on security now, that the 540 MB/s will be because the drive is actully marked as blank there so gets max speed indicated, but after running Spinrite there now is actual data written there hence the decrease? Does that sound right?

That would seem to make sense. Before you ran SR, the drive knew that the last part was empty, so may well have returned immediately after a mapping table scan. Once SR has run and written to every sector, the disk is no longer "unused", it is full of empty sectors.