Intel X25-M Raid 0 Alignment?

[gemsurf]

I just dumped my two Titan 128GBs and bought two Intel 80GB X25-M's ---- again! I previously had two when they first came out at $600, but I only paid $450 each thanks to MS Live CB on Ebay. I couldn't stand the thought of having that much wrapped up in drives, so I sold them for $550 each after a month and a half.

Well, as Anand points out, the real test of good SSD performance is to have them for awhile and then go back to a spindle drive. Anyway, I bought two Titans when they first came out and they really really sucked after any length of time in the system so I'm back to the Intel drives.

Has anyone played with the alignment on these? I tried it and all the other tweaks I've found over at OCZ, with mixed results on the Titans. Seemed to help on single drive setups but did nothing in Raid 0. I will be using these Intels in Raid 0, Vista 64bit and wonder if there are any suggestions out there?

 

[ochadd]

Return them and get three of these:
Adata SLC- Samsung SLC

That way you don't have to deal with the nonsense of having an $800 storage setup go to hell on you.

 

[taltamir]

what? why would he want that?

 

[gemsurf]

Originally posted by: taltamir
what? why would he want that?

You're right, I don't. I do full image backups periodically and keep movies, and other large files etc... on a couple 640gb WD Black drives. The two Intels are just for OS, Apps and working files.

 

[ochadd]

Originally posted by: taltamir
what? why would he want that?

... so you don't have to deal with the nonsense of having an $800 storage setup go to hell on you.

 

[IBDoomed]

Originally posted by: ochadd

Originally posted by: taltamir
what? why would he want that?

... so you don't have to deal with the nonsense of having an $800 storage setup go to hell on you.

I think you're being overly negative. Is there evidence to suggest that these intel drives will not last the typical 5 years and the samsungs will?

 

[taltamir]

oh he is talking about THAT (SLC has 10x the write life of MLC)...

ok he is just spouting FUD. there is no way in hell that is an issue. it will take a home user 10+ years to run out of writes, and it doesn't go to hell, t just becomes write only. And it is not going to be a big deal even if it happens 3 years down the line...

To put things in perspective, the X25-E is MLC so it will last 5-15 years in a home user machine. the SLC drives like X25-E and the samsung SLC will last 10 times longer... so 50-150 years...

ofcourse those values will be DRASTiCALLY reduced if you enable the trim command when it becomes available. but still the argument is fud... one year from now this 800$ setup will be worth 300$, two years from now it will be worth 100$. three years from now it will be worthless (since each drive is extremely small and it will be really slow by than)

 

[magreen]

@ochadd: you're just spreading FUD. You were asked to state the reason for your criticism of the intel drives and you did nothing but repeat your empty statement.

You're doing nothing but trolling.

 

[ochadd]

Originally posted by: magreen
@ochadd: you're just spreading FUD. You were asked to state the reason for your criticism of the intel drives and you did nothing but repeat your empty statement.

You're doing nothing but trolling.

It's a simple conclusion that Anand himself saw as well. Eventually, it's gauranteed, that MLCs will decrease in performance once all the cells have been written to. They will still be faster than hard drives and I would still love a pair of X25 units if they were given to me. I would not however spend that kind of cash on drives that will deteriorate in performance when there are alternatives out there.

The SLC drives in a RAID setup will have the same latency and access performance in three years that they do when you pull them out of the box. They will beat the Intel MLC units in write and come damn close to matching the read performance once the Intel units are "broke" in.

 

[taltamir]

that is complete and utter bullshit. more specifically, FUD.

SLC and MLC both have a ONE TIME performance drop once all cells are written to, and that drop is due to internal fragmentation, because only 512KB can be deleted at once but as little as 4kb written.

The exact value of the drop varies slightly depending on the exact distribution of fragmentation, but it is about the same once it happened.

There is no "slowdown" of the cells, the cells write as fast as the first day you made them, its simply that if the drive is full that fragmentation causes a performance loss... this is what the TRIM command is going to fix.

the term for this issue is steadystate... ALL SSDs, MLC, SLC, whatever, they all have it... SOME ship in steadystate from the factory, others do not. But anandtech and pcper and other quality review sites bring all drives to steadystate before benchmarking them. And a steady state intel X25-M wipes the floor with the samsung SLC... true the intel loses almost 30% from steadystate while the samsung loses 7% (IIRC). that is because one if MLC and one is SLC, but that is a one time loss that does not result in further degradation, and even after it the intel is STILL faster.

 

[gemsurf]

I have a ton of faith in Anand, and this is one of the things that prompted me to pick up a couple of these again-

Quoted from Anand in his last SSD update-

Remember what I wrote in my last SSD piece: ?I?d venture a guess that Intel would not leave its most loyal customers out in the cold after spending $400 - $800 on a SSD. I can?t speak for Intel but like I said, stay tuned.?

 

[ochadd]

Originally posted by: taltamir
that is complete and utter bullshit. more specifically, FUD.

SLC and MLC both have a ONE TIME performance drop once all cells are written to, and that drop is due to internal fragmentation, because only 512KB can be deleted at once but as little as 4kb written.

The exact value of the drop varies slightly depending on the exact distribution of fragmentation, but it is about the same once it happened.

There is no "slowdown" of the cells, the cells write as fast as the first day you made them, its simply that if the drive is full that fragmentation causes a performance loss... this is what the TRIM command is going to fix.

the term for this issue is steadystate... ALL SSDs, MLC, SLC, whatever, they all have it... SOME ship in steadystate from the factory, others do not. But anandtech and pcper and other quality review sites bring all drives to steadystate before benchmarking them. And a steady state intel X25-M wipes the floor with the samsung SLC... true the intel loses almost 30% from steadystate while the samsung loses 7% (IIRC). that is because one if MLC and one is SLC, but that is a one time loss that does not result in further degradation, and even after it the intel is STILL faster.

I've yet to read a single account of sata SLC drives experiencing slow downs. If there is a slow down it doesn't appear to be detectable by the user.

The trim command is great, where is it? Where is this magical tool that OCZ, Intel, and others are all betting on solving all their issues? Last I read on the OCZ forums it will be compiled into an OS or firmware of the future. Dropping $800 on drives that need a yet released tool to continue operating at peak performance is pretty rough.

I'll conceded here to a point because formatting the drives and using whatever "erase" tool Intel has isn't that big of deal for most users buying these drives. I wipe my OS drive clean every year or two and taking the extra step isn't that big of deal. For $400++ a piece I expect them to take care of this in firmware or automatically through the OS.

There are plenty of people out there that do not have the knowledge to do things like this. People with money and little computer knowledge who drop large amounts of cash expecting top performance for the life of the machine. Guess they are also people who keep people like me employed...

 

[taltamir]

Originally posted by: ochadd
I've yet to read a single account of sata SLC drives experiencing slow downs. If there is a slow down it doesn't appear to be detectable by the user.

Read ANY of the SSD articles by Anandtech and by pcper.com (the guys who discovered the issue first).
Until you do stop spreading your fud

The trim command is great, where is it? Where is this magical tool that OCZ, Intel, and others are all betting on solving all their issues? Last I read on the OCZ forums it will be compiled into an OS or firmware of the future. Dropping $800 on drives that need a yet released tool to continue operating at peak performance is pretty rough.

I'll conceded here to a point because formatting the drives and using whatever "erase" tool Intel has isn't that big of deal for most users buying these drives. I wipe my OS drive clean every year or two and taking the extra step isn't that big of deal. For $400++ a piece I expect them to take care of this in firmware or automatically through the OS.

It cannot take care of it in firmware because it doesn't KNOW what is empty space and what is not... thats what the TRIM command does... and all it needs is for an OS to implement it, MS promises that the release version of win7 will have it... however I am NOT buying an SSD until they do and its tested to work.

There are plenty of people out there that do not have the knowledge to do things like this. People with money and little computer knowledge who drop large amounts of cash expecting top performance for the life of the machine. Guess they are also people who keep people like me employed...

Knowledge to do what? there is nothing for the user to do, doing a clear will be useless because you will only have improved speed for a week or two before you reach steadystate again...
you make it sound like the drives ground to a halt, they dont!... anand has tested them IN the steadystate condition to STILL be faster than any other drive ever made...