Barker Bites Back
A look at some newsy stuff and interesting bits as well as those hopefully amusing byways of technology.
Friday 9 January 2009, 4:43 PM
SanDisk speeds up SSDs
The only issue I have with an SSD comes from trying to understand just where the speed improvements SanDisk claims for its latest SSDs lie.
What does work for me though are comments like, the latest SanDisk SSDs, the G3 series, are "more than five times faster than the fastest 7,200 RPM hard disk drives" and, they are " more than twice as fast as SSDs shipping in 2008". To enter the virtual world for a moment, they clock up speeds of 40,000 vRPM,, when vRPM means virtual revolutions per minute. In other words they are as fast as a 40,000 RPM disk drive in the real world, which seems fast.
According to SanDisk the "anticipated sequential performance of 200MB/s read and 140MB/s write.
Designed as drop-in replacements for hard-disk drives in notebook PCs, the full members in the SanDisk G3 family are SSD C25-G3 and SSD C18-G3 in the standard 2.5in and 1.8in form factors respectively available in capacities of 60, 120 and 240GB and prices at $149 (£97), $249 (£163) and $499 (£327).
PreviousComments on this post
You've hit on an issue I have been experiencing somewhat with Compact Flashes being used as hard drive replacements in the company's product.
Speed of access being touted in terms or specifications that are meaningful to the end-user. I fear though that specifying "virtual RPMs" will only cloud up the specs in the future. In the short-term it might provide a visual picture to the end-user but actual performance depends a lot more on what's going on inside the firmware inside the SSD and its interaction with the data being either written or read.
SanDisk in particular has been a bit cagey in describing their solid state media. I don't think they have been trying anything fast and loose but actually think they have been trying to be responsibly conservative in their reporting on speeds of access. At least in the Compact Flash arena the SanDisk specifications have been more conservative than experience has shown me.
Wear leveling algorithms and error correction for multi-level cells have a direct impact on the speed of the SSD system. In addition, the external operating system attempting to continue to support CHS (cylinders-heads-sectors) or the upgrade LBA puts restraints on access speed that might not be there if a new model or mode of access is created.
I agree that this is the issue. I did take up the use of "virtual RPMs" with some hesitation. On the one hand, it is potentially a vey useful metric to consider when looking to buy SSDs for commercial use.
I think it is one to watch and to report back on as we get some real idea of their performance.
