Yes, Sandforce may be slow for in compressible data but it is good if you want your drive to last long. It has the lowest data entropy which increases the life of the drive by 2x or even more. No other controller is even close to .5 write amplification levels
Any users with 0.5x? Like, real users, as studied by 3rd parties, not LSI-selected test cases? I've seen reports close to as low as 0.7x, but 0.5x seems to be a marketing value. FI, how are those GBs of photos coming off that SD card going to only be half written to the NAND? And, how compressible are typical page file writes (that one I really don't know)? Most common data is either very small, or already compressed, except for web and email, IME, and email typically is for businesses.
IoW, I'd like to see SF's controllers get 0.5x write amplification, with some methodology and data sets that can be sufficiently reviewed. Small low-QD writes are going to be difficult to compress, and I can think of very few desktop cases where large writes will not commonly include incompressible data.
Here's a good thread on the topic. Small sample size, but there is actual data, from actual work:
http://www.xtremesystems.org/forums/showthread.php?272545-Sandforce-Life-Time-Throttling/page23
I'll believe 0.5-0.6x is typical for OL(T/A)P servers all day long, but desktops and notebooks, I'm not so sure about, and I have yet to see close to that from any non-benchmarking users taking the time to report their WA.