Awesome that you fixed this measurement error (ahem.. sorry to be boasting but... I told you so :P), but you're still not quite there. Look at the datasheet. Now look at bench. See the giant difference?
That's DIPM. Enable it and you'll have entirely correct measurements.
(failing an edit function, I'm replying to my own post)
Edit: I see that Anand said there will be DIPM-only testing in 2013. Great. Just keep in mind that all power consumption measurements you have now don't really mean anything to desktop and laptop use cases. They only apply to server workloads where DIPM doesn't exist, or workloads where the SSD is constantly hammered at maximum speed.
Ahem. This is a weird place to do this, but I'm a huge fan of your work. It's truly awe-inspiring. Can't wait for the MADPSU! (And more updates in English, please. Sometimes Google Translate just doesn't cut it ;) )
While performance per watt is surely there, these all fall short in absolute terms of the best 5400 RPM regular hard drives, which come in around 1.5W under load. (And regular hard drives don't care if you're writing incompressible or not.) I'd like to think the technology will get there eventually, but as with cars over the last couple of decades there seems to be a never-ending focus on (horse)power.
Any thoughts if the 240GB 525 + mSATA>2.5" adapter a better choice over a 520 or Sammy 840 for an efficient ultrabook ? With that in mind, does anyone know if: a) the adapter degrades speed (are data+control lines direct connection ?), b) which make of adapter Anand used, c) if choice of adapter affects compatibility ? The Intel datasheet states typ Idle 250mW, Active 300mW; any explanation why measured Idle is double the Intel spec ? Do the measured active consumptions, eg. 3.47W during Random Wr, accord with (am guessing the 10:1 disparity due to Intel using avg power over time) the Intel active 300mW 'typ' spec ?
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
8 Comments
Back to Article
RU482 - Wednesday, February 20, 2013 - link
bravo on your follow through on this one!ssj3gohan - Thursday, February 21, 2013 - link
Awesome that you fixed this measurement error (ahem.. sorry to be boasting but... I told you so :P), but you're still not quite there. Look at the datasheet. Now look at bench. See the giant difference?That's DIPM. Enable it and you'll have entirely correct measurements.
ssj3gohan - Thursday, February 21, 2013 - link
(failing an edit function, I'm replying to my own post)Edit: I see that Anand said there will be DIPM-only testing in 2013. Great. Just keep in mind that all power consumption measurements you have now don't really mean anything to desktop and laptop use cases. They only apply to server workloads where DIPM doesn't exist, or workloads where the SSD is constantly hammered at maximum speed.
asphyxxiate - Thursday, February 21, 2013 - link
Ahem. This is a weird place to do this, but I'm a huge fan of your work. It's truly awe-inspiring. Can't wait for the MADPSU! (And more updates in English, please. Sometimes Google Translate just doesn't cut it ;) )ssj3gohan - Friday, February 22, 2013 - link
Uh... thanks :) but yeah, this really isn't the place. I'll update my blog very soon about MADPSU.[/completely offtopic interlude]
ABR - Friday, February 22, 2013 - link
While performance per watt is surely there, these all fall short in absolute terms of the best 5400 RPM regular hard drives, which come in around 1.5W under load. (And regular hard drives don't care if you're writing incompressible or not.) I'd like to think the technology will get there eventually, but as with cars over the last couple of decades there seems to be a never-ending focus on (horse)power.phillyry - Sunday, February 24, 2013 - link
But these race to sleep sooner and therefore use less power for the same work (see the iPhone 5 review for an overview of this concept).So, not only will the SSD use less power but so will the system.
will2 - Sunday, April 7, 2013 - link
Any thoughts if the 240GB 525 + mSATA>2.5" adapter a better choice over a 520 or Sammy 840 for an efficient ultrabook ?With that in mind, does anyone know if: a) the adapter degrades speed (are data+control lines direct connection ?), b) which make of adapter Anand used, c) if choice of adapter affects compatibility ?
The Intel datasheet states typ Idle 250mW, Active 300mW; any explanation why measured Idle is double the Intel spec ? Do the measured active consumptions, eg. 3.47W during Random Wr, accord with (am guessing the 10:1 disparity due to Intel using avg power over time) the Intel active 300mW 'typ' spec ?