Printable View
A lot of SF makers are using the SF 5 Series fw which is fast but trim is broken...this applies to all 24nm toggle drives as they require 5 Series base fw. Smarter makers like Ocz didn't sacrifice trim for a lil extra speed...they stayed on the previous SF base fw.
Really good read here:
http://www.tweaktown.com/articles/48...ted/index.html
Uhh,Quote:
Originally Posted by Zaxx
Thanks Zaxx for posting that. It affects more than just Toggle flash though. I really hope everyone here reads it because it is an amazing article :)
It definitely shines the light on some things, like why the latest SF versions werent that hot. interesting, cant wait for the fixed firmware to filter down :)
Couldn't agree more...it's a great article. Chris doesn't skip the details...that's for sure. Also, the series 5 fw is required to run the new tosh 19nm toggle as well as the 24nm. With the fw sorted, I think we can expect to see additional SF based 24nm toggle products launched as well as the new 19nm. If I had to guess I'd say the 24nm equipped drives will be more likely to carry a 5 year warranty vs a probable 3yr for the less durable 19nm. We'll prolly see the Ocz V3 MaxIOPs go to 19 or 24 toggle and I'm sure we'll see a V4 MaxIOP launch as well.Quote:
Originally Posted by Highendtoys
Just wait till you guys see what Chris is doing with power measurement:) Did you know that some SSDs have 2x the idle power as others and some use 4x the power when writing random data? :)
Yes to the first (doesn't everyone?). As for the second, that is irrelevant and misleading. What you need to measure is something directly proportional to energy per byte written. The physics is that power is energy per unit time. Write speed is, of course, bytes per unit time. So if you report the ratio (power_consumption / write_speed) you will have energy / byte. If power_consumption is in Watts and write_speed is in MB/sec, then it will be Joules/MB.Quote:
Originally Posted by Highendtoys
The reason just looking at power consumption is misleading is that some SSDs can write several times faster than others. For example, SSD B may write 3 times faster than SSD A, and B may use twice the power when writing. But to write, say 100MB, SSD B is clearly the more energy efficient, since B will use 2/3 the energy as A uses to write that 100MB.
I'm already way ahead of you:) That is why I don't just rush in when I come up with new benchmarks. I need some time to validate my data and to collect enough of it to make sure other variables are not being measured at the same time.
The Plextor M5P already breaks this rule, with 19nm flash and 5 year warrenty.Quote:
Originally Posted by Zaxx
In reality, with 19nm having a specified life of 3K write cycles, endurance will still be far more then needed on the desktop.
Cool...glad to hear that and kudos to Marvell for excellent nand wear characteristics. Will be interesting to see how well the other makers' new 9187s stack up against Plextor's great quality and fast fw.