Disappointing (Poor Write Performance)

Congrats on  your return!  I really wish I had know about this in time to get a refund on my own drive.  This SanDisk has become my most regretable hardware purchase in at least a dozen years.

I think I found out what caused the low 4ks on my previous benchmarks. After disabling the pagefile (moved it to ASRock XFast Ram) it scores as it should. The rest still needs to be improved a lot with the new firmware, though.

715 on AS SSD for a SandForce drive is great actually. It uses all incompressible data, so the write performance takes a big hit on SandForce SSDs.

I did some testing recently on a 120GB Extreme that turned out interesting. I once had an AS SSD run fail on me when I had OC’d the CPU to high, and got a BSOD. Later I found the AS SSD test data file left over, exactly 1GB worth. Saved that for testing, copied it multiple times into a few folders, 10GB, 20GB, etc.

I started with an empty 120GB EX, and ran an AS SSD benchmark, and saved the results. Then I filled it up with those data files, until it had less than 1GB free. I then ran AS SSD again, since I knew it would fail, it needs at least 1GB of free space to run its test. After the failure, that SSD had ~50MB free, according to Windows.

I next quick formatted that SSD. NOT a secure erase, just a quick format in Windows. Not a delete and create volume, just the format. A format will trigger TRIM commands to be sent. I thought that if TRIM was not working at all on this SSD, GC would not have a chance to clean up if I ran another AS SSD run right after formatting. Formatted and ran AS SSD immediately.

The result was… the same as the first, empty SSD benchmark. Actually, a bit better.

I’m still thinking about this, but it seems that TRIM is working at least to a degree on these SSDs?

Common guys, you just bench and conclude your SSD is slower than a HDD?

Never compare a SandForce based SSD > 60GB with any HDD. Your HDD will never be able to beat it.

Sandisk did produce a great SSD, but they are using a SandForce controller in which they had no control or very small control on the firmware. The issue with TRIM is not due to Sandisk.

To the one that replaced his SSD with an Intel 330, you know it’s the same controller ? Just a little bit different firmware.

If you want to replace your SSD, at least go with another controller…

“Sandisk did produce a great SSD, but they are using a SandForce controller in which they had no control or very small control on the firmware. The issue with TRIM is not due to Sandisk.”

it is arguable that they released a “great SSD” when they released a competitive SSD and then chose to poorly support it.  Sandisk is wholy responsible for the drives and could modify the firmware directly if they wanted, or not use Sandforce controllers.  Whne a fuel pump goes in a car and causes engine damage, people don’t go “Don’t blame Volkswagon, they had no control over it, blame Bosch.”

“To the one that replaced his SSD with an Intel 330, you know it’s the same controller ? Just a little bit different firmware.”

Trim and GC failure complaints on the Intel drives are close to 0.  Intel doesn’t seem affected by the 5.0.1 issues and have their own firmware developers who possibly caught and already fixed the issues before release.

Also I could write to my Velociraptor HDD faster than I could write to the Sandisk Extreme before I sent it back. 

If sequential writes is the only thing that matters for you, then you should stick with HDD.

Once you do random (ex: 2 files at onces), your vraptor is out of the league.

I think SandForce does not allow manufacturers to look at their code. And by the way, this guy is telling that TRIM works in 5.0.1: http://forums.sandisk.com/t5/SanDisk-Extreme-SSD/TRIM-EFFICIENCY-OF-240GB-SANDISK-EXTREME-WHEN-USED-AS-A/td-p/283036

That article specifically notes that TRIM is “crippled”