Usually you can update the firmware.
TRIM was added to Linux 2.6.33 (Feb 2010).
Please test again changing the i/o scheduler from cfq to noop...
cfq is good for servers with mechanical drives, noop for SSDs and deadline for desktop machines using mechanical drives (much better system/desktop responsiveness). Ubuntu uses cfq by default...
Also very important is to get rid of atime, meaning add noatime to the mount options in /etc/fstab, And no, relatime won't do, it will still write on every read, just a little later.
I've been thinking of getting a Vertex 2, but how reliable are they? Not the quoted numbers, but what are people's actual experiences? I know people close to me have had failures inside of six months (one of them had two go bad), so I'm a little worried that it might not just be freak coincidence. Anyone else heard anything like that? (Is it just what happens in Australia? )
I tried looking at customer feedback on Egg and Amazon, but it's just too sparse for my tastes (fewer than 100 just isn't even close to a good sample, IMO, much less fewer than 25).
do not buy! corrupts filesystem on suspend/resume
I bought the OCZ Vertex 2 120GB SSD drive after reading this positive review. Unfortunately, for laptops, the Vertex 2 series drives (and also these other drives that use SandForce controllers) have a major bug in their SandForce firmware (both v1.10 and v1.11): filesystem corruption upon resume from suspend to RAM!
When you resume your laptop from suspend to RAM, it wakes up to a corrupted filesystem. In my case, I was greeted with a blank X screen with mouse cursor (still responding to movement) but nothing else worked: I couldn't even switch out to a Linux virtual terminal (e.g. Ctrl-Alt-F1). The only course of action was to power down the system.
This problem has been reported on various forums and is known to occur in both Windows (people get BSOD) and Linux (what I described above).
Bottom line: Do not buy SSD drives with SandForce if you want to use it in laptops!
Just get an Intel SSD. Those are the only ones really worth the money.
Originally Posted by sunaku
That is not true IMHO.
Originally Posted by thefirstm
Intel SSD's cost about 2.3-2.5 EUR/GB, which is pretty high compared to Crucial RealSSD which costs 2.0-2.1 EUR/GB. The reading speed of the Crucial RealSSD is even considerable faster than Intel SSD's (writing access time is a bit slower than Intels).
Corsair Nova is a bit slower, but costs only about 1.7-1.8 GB/EUR.
I read a few reviews and benchmarks of SSD's and the Crucial RealSSD and Corsair Nova are the most attractive for the price. But I'll probably wait until next year with my first SSD drive purchase.
Corsair Nova uses a Indilinx Controller and Crucial RealSSD uses Marvell Controller. I don't know if these work flawlessly under Linux.
If you consider active & idle power consumption, the Intel X25 drives (idle: 0.075W, active: 0.150W) are much more power-efficient than the OCZ drives (idle: 0.5W, active: 2W) and Crucial drives (idle: 0.092W, active: 1.7W read, 3.1W write).
Originally Posted by Fenrin
Note: I got these numbers from Newegg.com product descriptions.
For me, as a laptop user, the most important things in an SSD are mobility (proper support for S3 sleep and resume) and battery life. I already stated how the Sandforce-based OCZ drives caused filesystem corruptions when resuming from S3 sleep state. And considering the power consumption of the current crop of SSDs, I agree with thefirstm that right now the Intel SSDs are the only ones worth the money. YMMV.
Well, this is a little bit to absolute:
1. product describtions are not the ultimate truth, websides testing those drives reported different results, even if Intel is usually coming out as "good" at power consuption
2. even on a low power notebook (for example 10 Watts idling), the power diff is not that hight at all, even if we take the debatable values from product describtions it is only up to 0.5 Watts during normal operations (idle).
3. other aspects can be important too, customers may seek for durability, product/vendor/storage size diversity or performance/throughput, etc even for laptops.
4. Intel has had its firmware-bugs to, nobody knows which defects appear or will be solved in the future.
Or just don't use suspend to ram... should boot in about 5 seconds if you tweak it just a little.
Originally Posted by sunaku