SSD Thread (Benchmarks, Brands, News, and Advice)

Watching some Micron 300 SSD videos now. Interesting.

 

4k random read from micron c300 is only 10% better than x-25m in sata 3G while x-25m has 10 channels vs c300's 8 channels.

 

what do u mean by JMicrons neww SSD?
Micron C300 is Marvell controller.

 

Yeah, I think the G2 Intel drives will still be relevant when the next generation of SSD's come out this year. Looks like the 6G SATA interface compatability will be included with most if not all of the next gens. I wonder if any of the new notebooks will have 6G? Judging from the new Marvell controller and the new one from Sandforce there will be substantially more overprovisioning and many comparisons to Intels random 4k performance.

 

ocz agility beats vertex in these tests: http://www.phoronix.com/scan.php?page=article&item=ocz_agility_120gb&num=3

 

What was I thinking, JMicron with a good SSD?

 

Man, I am starting to drool already over the next gen SSD's of just about any good brand. I hope I am rich by then!

http://www.dailytech.com/article.aspx?newsid=17300

 

http://www.legitreviews.com/article/1174/1/

 

c300 win 7 boot is not fast at all.

 

thanks, I ran this the first day i got my drive. just wondering how often I need to run this?

 

If you have a TRIM enabled system, from the BIOS settings, O/S (currently only Win 7), SATA driver (MS default or Intel IRST drivers) and of course a TRIM enabled drive, then you shouldn't really need to run this anymore.

Of course, if this was my system, I would set it to run once a week - just to be sure!

Cheers!

 

dude far as I know. If you TRIM when you're not supposed to. You'll kill the drive.. So uh. This isn't one of those "do it anyway just to be sure" situations..

 

you can't kill the drive. and actually, the manual trim doesn't run for more than a second or two, if you already have automatic trim.

 

Does the Intel X25-V SSDSA2M040G2XX 40GB support Trim?

 

it should. but i'm unsure so far, have to verify. if it does, it's a great drive. if not, not so much. it doesn't have much headroom.

it could be the major breaktrough in ssds for corporations. how much client-pcs for corporations need more than 40gb? most don't => cheap very fast systems for everyone. at least our corporation is considering it.

 

Phil,

Yes it does support TRIM and it's available too:

See:
http://www.dailytech.com/article.aspx?newsid=17301

 

davepermen is right, you can't kill it doing a TRIM (it will eventually do it anyways and if it doesn't need it - it doesn't).

Also, if you schedule the Intel SSD Toolbox to perform an Optimize - it only lets you schedule it once each week anyways (or, at least that is what it recommends).

Cheers!

 

one thing that is true, is, if trim works as it should, there should be zero need to manually trim once a week. i'm interested why they suggest that, actually.

 

davepermen,

for one thing, if you are using a notebook without a hibernation file and the battery gives out - there will be files that need to be 'synchronized' again between the O/S and the SSD's controller. Or, of course, just a plain old crash, whether inside one specific program, or, Windows itself (although I haven't seen that in a long while...).

Anytime that the O/S isn't in charge of the SSD and writes/moves or deletes are performed on it, the SSD Toolbox will ensure that TRIM-wise, the O/S and the SSD will be in-sync.

I think it is an 'easy' insurance policy to set it on weekly schedule and forget it. I certainly would run my Intel's like this. If there is no TRIM to be performed, then non will take place so it really is a win-win-win with no downsides as far as I can see.

 

yeah i guess. it should result, in theory, in a no-op every week. but in practice, it might clean up some little overseen things.

 

Some European stores have the Intel 40GB in stock: http://geizhals.at/eu/a489284.html

 

nice. and i just got two 80gb ssds haha

 

Supposedly, if your system supports S1 sleep and ACPI 2.0 the bitmap will not get out of sync between the OS and controller if you use S1 sleep. My system only supports S3 sleep and there are no BIOS options for either sleep or ACPI 2.0. I have tested it on my system and whenever I use sleepmode my winsat disk 16k random scores drop to 30-40mb/s and gradually recover to 110mb/s if I leave it to idle at the desktop.

You can check which sleep states are supported by typing powercfg.exe /a from an elevated command prompt.


I made this benchmark before using sleepmode...


This is immediately after resuming from sleep.

 

as always, nice info, thanks!

 

So it seems Intel's presser is at 4:30PM Pacific tomorrow. Think we'll get any new info?

 

at least clarification on what info is real

 

Hopefully Intel's next gen controller will be able to utilize lesser quality nand like the new Sandforce controller. SSD prices haven't been coming down because SSD's have been made from only the highest quality nand causing supply constraints. The manufacturers have, and are still reluctant to increase capacity because of the past few years economic environment.

I would like to see SSD prices come down so that PC's can be fully saturated with them and next gen OS's can be specifically written for flash storage without workarounds such as TRIM and the need for controllers to "mimick" mechanical HDD's. Then maybe tiller will agree that SSD's have reached, "primetime".

 

ssd prices from intel should be lower than what they are right now. but it's not their fault. if you can sell at that price, why go lower?

and intels next bigger ssd roundup will be 3bit mlc, so that's sort of "lower quality", too.

and no, i don't want the os to go lower level than it is right now. i prefer to see it go higher level instead. there's only one fault that the os does right now wrong:

it does send the ssd data, and tells, when it doesn't need it anymore.

it (the os) gives the adress it wishes it to be placed. that address is actually arbitary (it would be the position on the hdd), and is just sort of a lookup value to get back to the file. instead, i'd like to see a more database like approach:

if the os wants to write say 1gb worth of data, it does so by sending the ssd, and gets an id back. that id then gets noted in the file system for the os to be able to access that 1gb worth of data later, or even free it when not needed anymore.

it's called CRUD (create, read, update, delete). it should NOT use addresses/indexes/ids anymore that the OS defines. instead, the ssd should be able to chose on it's own how it wants to place the data onto it. it could then do wear leveling how it knows it's best for itself, move around cells, how ever it want. and the os still has full access to all it's data.

it would clean up all the issues we have right now. but it would need a full rewrite of ntfs and other file systems.

remember WinFS? it would have solved that.. sad it didn't worked out

 

http://www.anandtech.com/storage/showdoc.aspx?i=3702&p=14

I was skeptical when I read this and still am because previous pricing expectations have missed the mark but it does make sense that if they can utilize lesser quality nand, prices should come down. I tried finding some forecasts for nand futures pricing from some commodities sites but it's all pay per view.
http://www.dramexchange.com/

 

lesser quality... that's all we need... joy

i thought 3bit/cell memory had a much lower write tolerance?

 

I have my eye on the New Crucial SSD. See how it holds up under actual usage. I do not want any lesser NAND than is in my Intel G2 ever! After all, it is my DATA AND BUSINESS we are talking about here.

 

see I'm not the only who would rather pay a premium for data security.

 

Count me in on that, too.

 

I'm wondering if these next gen engineering sample SSD's Anand and such are benchmarking have the good stuff then they do the old bait and switch to cheap NAND to reach saturation price points.

 

I would pay for data security...

I'll be very curious to see what the new generation SSDs will do.

 

Micron have 2-bit MLC with 30 000 write cycles, so if they go for 3-bit MLC it shouldn't be less then 10 000 write cycles and with larger capacity I think no one should even bother about data security.

 

spam 10char

 

http://www.digitimes.com/news/a20091229PD216.html
"Sources claim 3-bit MLC NAND produced by Korea-based supplier problematic......those shipped to a US customer have been returned due to unstable product quality...."

 

ocz roadmap:
</IMG>
</IMG>

 

CES: Toshiba Planning 32nm Update to SSD Product Line, Mulls Price Cuts to OEMs

 

at least intel won't deliver something that doesn't deliver, and is not reliable. so why should you bother?

 

while intel's track record is impressive that doesn't mean they haven't made mistakes, I'm worried that this could be one of them.

if only I could find a fast 32/64GB SLC drive for less than 500$

 

it won't be one of them. selling flash, and knowing how well flash lasts is one of their main business actually. since long before their first ssds.

you don't need an SLC, they don't give you anything an MLC of today can't deliver. i have slc, and intel mlc. i prefer the intel mlc.

 

I still want an SLC when they become fast and affordable! I believe that they are by definition more reliable. But then again, I am not a flash NAND engineer

 

well I worry about reliability, I two have an intel mlc and it's great, but I've already had a mlc fail on me(non-intel). it was a jmicron G.skill but, it still shouldn't have up and died. no to mention I've had quite a number of hard drive drop dead on my to, so if nearly everything goes 3bit/cell and write tolerance is drastically affected that could be a big deal.

 

they didn't fail because they're mlc. mlc doesn't matter. it might matter after 10 years of usage, yes, but not yet.

 

so 10 years on 2bit/cell... what about 3bit? see what I'm getting at? if they maintain reliability then I'm fine with it. so 3bit... let me quess that it will last 4years... what next 4bit? how long will it last? 1year?

 

i know what you're getting at, and i still say the same: it won't matter. they will verify it will last for, say, 5 years, and you can trust them at it.

the current ones from intel are internally estimated to run for 80 years on ordinary desktop user configs. so even a factor 10 lower would lead to 8 years still (and that would be at 4bit flash..)

i know what you mean. i just say forget it, don't worry. other problems are bigger problems.

 

controller went dead or mlc chips?

 

Seems Toshiba announced a 512 GB SSD drive way back at the end of 2008. We're now 2010, and consumers still can't buy 512 GB SSD drives for sub-$1000 prices. Why not? What's the hold up? Where's Intel? Where's Samsung? Anyone know? I need more storage!