TRIM support?

Yes, the M4 is currently the most "popular" drive. It has TRIM support.
 
Did you check to see if the SSD was actually scheduled for defrag before you turned it off? The defrag service will still be scheduled, but if you check, the SSD will be de-selected on the drive list.
 
How's TRIM coming along for SSDs in RAID0? I'm considering a couple of 256GB M4s.
 
Raap said:
How's TRIM coming along for SSDs in RAID0? I'm considering a couple of 256GB M4s.
Click to expand...

Some controllers I think support pass through?

And the sandforce controllers do garbage collecting on their own, right...so no TRIM needed and RAID should function fine.
 
I was thinking of getting either 2 M4's or 2 C300s in RAID0... Would I be better off just getting a larger single one? I wasn't sure if they support TRIM or the garbage collection when in RAID.
 
JLAW904 said:
I was thinking of getting either 2 M4's or 2 C300s in RAID0... Would I be better off just getting a larger single one? I wasn't sure if they support TRIM or the garbage collection when in RAID.
Click to expand...

That's on the same Marvell chipset that the Intel 510s are on, right? Only Sandforce chipsets implement their own GC functionality I think. I don't know which, if any RAID chipsets support TRIM, but I would assume that RAID will not allow TRIM to run. until you find data from Intel or the chipset manufacturer that states otherwise.

I'm no expert here though.

Read here: http://www.bit-tech.net/news/hardware/2010/03/23/intel-releases-trim-for-raid/1
 
basically if you are holding off on a raid0 of ssds because of the lack of trim, you are making a huge mistake.

the drives performance will always be way more than a single drive that has trim. I don't care how much data you throw at a raid0, you will not degrade it to the point that a single drive with trim was better.
 
gjs278 said:
the drives performance will always be way more than a single drive that has trim. I don't care how much data you throw at a raid0, you will not degrade it to the point that a single drive with trim was better.
Click to expand...

You should not say always. There are some SSDs that perform significantly worse when no pages are marked invalid. For example, look at tweaktown's reviews, where you can see that the 75% full bar (light purple) is less than half of the "empty but dirty" bar (green) for a number of the SSDs that they have tested. They do not have a 100% full bar, but that would be significantly lower than the 75% full bar for many SSDs. And without TRIM, the drive will eventually be operating at 100% full all the time, even if you are currently only using a small amount of the RAID 0 space.

http://www.tweaktown.com/reviews/4179/corsair_force_3_120gb_solid_state_drive_review/index10.html
 
john4200 said:
You should not say always. There are some SSDs that perform significantly worse when no pages are marked invalid. For example, look at tweaktown's reviews, where you can see that the 75% full bar (light purple) is less than half of the "empty but dirty" bar (green) for a number of the SSDs that they have tested. They do not have a 100% full bar, but that would be significantly lower than the 75% full bar for many SSDs. And without TRIM, the drive will eventually be operating at 100% full all the time, even if you are currently only using a small amount of the RAID 0 space.

http://www.tweaktown.com/reviews/4179/corsair_force_3_120gb_solid_state_drive_review/index10.html
Click to expand...

I'm saying always. no user will ever cause their raid0 to be slower than a single drive with trim. even when you raid the drive you're already giving it half the amount of writes that the solo trim drive would have gotten, and garbage collection will idly pick up anything in the raid.

considering the raid drives will get half or even 1/3rd to 1/4th of the writes combined with garbage collection, a raid0 of two drives is not going to degrade to the point it will be slower than a solo drive with trim. the drive will never appear to be using 100% unless you are actually using 100% space, due to garbage collection eventually cleaning out the dirt. if you actually are using 100%, both drives would be just as slow.

even on those tests, the 75% fill dirty (worst case really) aren't even below half of a completely empty clean drive in terms of performance. considering your trim drive won't be empty, it'll at least be at 25 - 50% clean, the speeds will be greater on the raid0 and with more room to boot.
 
gjs278 said:
I'm saying always.
Click to expand...

Then you are obviously wrong. Could not be more obviously wrong.

And you clearly do not understand garbage collection. Hopefully no one reading your post will be confused by the nonsense you have posted about garbage collection.

Once all the LBAs have been written at least once, even if you later delete ALL the data on the drive, as far as GC is concerned the drive is 100% full (except for the reserved space, obviously). Without TRIM, there is no way for garbage collection to know that you have deleted files. So, as I said, 100% full eventually (or whatever the highwater mark is for usage of your SSD over time). And the speed will be well under half the speed of an equivalent TRIMmed SSD for quite a few models of SSDs.

So, a single TRIMmed SSD can be faster than two SSDs in RAID 0, for a number of SSDs as shown in the link I posted previously.
 
Last edited:
john4200 said:
Then you are obviously wrong. Could not be more obviously wrong.

And you clearly do not understand garbage collection. Hopefully no one reading your post will be confused by the nonsense you have posted about garbage collection.

Once all the LBAs have been written at least once, even if you later delete ALL the data on the drive, as far as GC is concerned the drive is 100% full (except for the reserved space, obviously). Without TRIM, there is no way for garbage collection to know that you have deleted files. So, as I said, 100% full eventually (or whatever the highwater mark is for usage of your SSD over time). And the speed will be well under half the speed of an equivalent TRIMmed SSD for quite a few models of SSDs.

So, a single TRIMmed SSD can be faster than two SSDs in RAID 0, for a number of SSDs as shown in the link I posted previously.
Click to expand...

there is not a single user on this planet with a raid0 array of two drives that are slower than the solo drive with trim, filled with the same amount of data. you're trying to compare two drives that are filled 100% capacity to a 0% drive that's completely clean. it's not a valid comparison.

assuming you fill even 25% of that solo drive, you have just lowered it into performance levels that are always slower than two drives that are 75% full, even according to your chart. for any of the drives. and the drives will eventually kick in garbage collection to have even less of a gap. you can do the math right there, the 75% dirty drive is not even 50% of the 0% clean drive, the solo drive will never double it in terms of performance.

if you wrote 100% of the drive and then deleted everything, the garbage collection would kick in during an idle time and reduce that 100% down. a drive is not permanently ruined by just filling it up one time as you are trying to imply if it's in a raid0 array. if you seriously believe that, go take two force 3 drives (one of the worst cases according to that chart). secure erase them. run the bench of the raid0. fill them 100%, delete everything, and let them sit a week doing nothing. run the bench of the raid0.

secure erase the drives. take a solo drive, run a bench. now fill it 100%, delete everything and let it sit a week with trim and garbage collection. run the bench on the solo drive.

the solo drive will not be faster than the raid0 at any point during your testing. go ahead and then fill both drives 100%, delete 50% of that, and let the garbage collection kick in. run the benches. wipe it down to a solo, fill it 100%, delete 50%, run the bench. raid0 will still be faster.

also I'd just like to note how hilarious your chart is. according to this thing a new force 3 that is completely clean will perform *worse* than a drive that is dirty at 0%. that doesn't even make any sense. I guess all of the force 3 users should dirty up their drives right away to get better numbers.
 
Only Sandforce chipsets implement their own GC functionality I think.
Click to expand...
Please quit spreading this incorrect rumor.

ALL SSDs have a built-in GC mode.....EVERY ONE.

I find it hard to believe that there's still some people that call themselves "enthusiasts" and don't know the basics.
 
JLAW904 said:
I was thinking of getting either 2 M4's or 2 C300s in RAID0... Would I be better off just getting a larger single one? I wasn't sure if they support TRIM or the garbage collection when in RAID.
Click to expand...

any ssd's in raid will never have trim support unless all the manufactures come together and create a new standard.

trim support for raid passthough is when you run a single ssd over a raid controller where you have a regular hard drives in raid. For example, a single ssd boot drive with dual hdd's in raid1.

every ssd has a slightly different implementation of garbage collection. some drives might have aggressive garbage collection, while other drives might have really slow garbage collection which makes it less suitable for raid. this is why the sandforce drives are recommended more often then the crucial c300/m4 drives to put in raid.
 
gjs278 said:
you're trying to compare two drives that are filled 100% capacity to a 0% drive that's completely clean.
Click to expand...

No, that is not what I am comparing. I am comparing a drive with TRIM to a steady-state drive without TRIM.

Again, you apparently do not understand how SSDs work. Once all the LBAs have been written at least once, the drive is basically full, even if the filesystem on the SSD contains no files.
 
af22 said:
this is why the sandforce drives are recommended more often then the crucial c300/m4 drives to put in raid.
Click to expand...

Actually, Sandforce has a very slow implementation of both GC and TRIM, as anyone who has tested the SSDs with a heavy workload knows. That is why the write speed of Sandforce SSDs tends to degrade significantly and stay low even for some time after TRIM. Without TRIM, it is even worse.
 
Old Hippie said:
Please quit spreading this incorrect rumor.

ALL SSDs have a built-in GC mode.....EVERY ONE.

I find it hard to believe that there's still some people that call themselves "enthusiasts" and don't know the basics.
Click to expand...

My understanding is that the sandforce controllers implement their own GC functionality that does not rely on TRIM. The Marvell controllers implement TRIM natively...which is not the same.

Post your source that the Marvell controller will do GC on it's own...because I find no data to support that. The best I can find is some mention of various marvell controllers attempting to implement IGC and being totally worthless while doing so...but no mention on which SSDs have it.

And as for your stab at me, I think I actually said that I was no expert on this area "I'm no expert here though." - maybe you need some new glasses...or to actually READ a thread before you jump in and accuse people.
 
john4200 said:
No, that is not what I am comparing. I am comparing a drive with TRIM to a steady-state drive without TRIM.

Again, you apparently do not understand how SSDs work. Once all the LBAs have been written at least once, the drive is basically full, even if the filesystem on the SSD contains no files.
Click to expand...

you're wrong. raid your ssds, ignore john4200, he doesn't know what he's talking about. he seriously believes a drive becomes dirty beyond repair when you write to it 100% one time in a raid array.

Again, you apparently do not understand how SSDs work. Once all the LBAs have been written at least once, the drive is basically full, even if the filesystem on the SSD contains no files.
Click to expand...
yes, and that's why they added garbage collection to fix that issue. and it works great!
 
Actually, Sandforce has a very slow implementation of both GC and TRIM, as anyone who has tested the SSDs with a heavy workload knows. That is why the write speed of Sandforce SSDs tends to degrade significantly and stay low even for some time after TRIM. Without TRIM, it is even worse.
Click to expand...

I've filled my sandforce drives several times over now. I haven't secure deleted them. they benchmark within 80% of the day I got them for reads/writes and the 4k read is exactly the same. the 4k write dropped from 79 to 71, and I can bet that's because I've actually filled them with real data now instead of benchmarking empty drives. why aren't my drives completely dead in the water when it comes to speed? they're on a raid controller, they're obviously not getting trim... man what could it be. oh right, me leaving my computer idle. and the garbage collection.
 
Last edited:
rampantandroid said:
My understanding is that the sandforce controllers implement their own GC functionality that does not rely on TRIM. The Marvell controllers implement TRIM natively...which is not the same.

Post your source that the Marvell controller will do GC on it's own...because I find no data to support that. The best I can find is some mention of various marvell controllers attempting to implement IGC and being totally worthless while doing so...but no mention on which SSDs have it.

And as for your stab at me, I think I actually said that I was no expert on this area "I'm no expert here though." - maybe you need some new glasses...or to actually READ a thread before you jump in and accuse people.
Click to expand...
Please think what you want.

I'm sorry.
 
gjs278 said:
he seriously believes a drive becomes dirty beyond repair when you write to it 100% one time in a raid array.
Click to expand...

Not just in RAID. On any system without TRIM. And not without repair (secure erase or manual TRIM can freshen everything). Without TRIM, after all LBAs have been written to at least once, even if there are no files currently on the filesystem, then the performance will be the same as if the SSD was 100% full. This is after GC has had enough time to reach a steady state by collecting as much of the reserved space as the firmware allows.

This is obvious to anyone who knows even the basics of ATA sectors, and computer filesystems. ATA devices do not know the structure of the files on them, they only have a list of LBAs. Without TRIM, the ATA device simply cannot know that you deleted files. All it knows is that all LBAs have been written.

TRIM eliminates that problem by allowing the OS to inform the ATA device which LBAs are no longer used.

The fact that without TRIM, after all LBAs have been written at least once, the performance of the SSD is the same as if it was filled completely with data is well known. Anyone who has any familiarity at all with the development of SSDs over the past few years knows this. That problem is the main reason TRIM was developed.
 
Last edited:
rampantandroid said:
Post your source that the Marvell controller will do GC on it's own...because I find no data to support that. The best I can find is some mention of various marvell controllers attempting to implement IGC and being totally worthless while doing so...but no mention on which SSDs have it.
Click to expand...

Old Hippie is, of course, 100% correct. All SSDs implement some form of garbage collection. Yes, GC is different from TRIM. All recent SSDs have both GC and TRIM.

This has been discussed at length in this very forum in the past. No need for you to continue to post incorrect information.
 
Old Hippie said:
Please think what you want.

I'm sorry.
Click to expand...

Because I asked for a source and you instead respond trying to insult me. How the fuck can people find out new info when people like you are around?


john4200 said:
Old Hippie is, of course, 100% correct. All SSDs implement some form of garbage collection. Yes, GC is different from TRIM. All recent SSDs have both GC and TRIM.

This has been discussed at length in this very forum in the past. No need for you to continue to post incorrect information.
Click to expand...

And again. I LOOKED FOR INFO. Why? Because I WANT TO KNOW MORE. So instead of responding and telling me where TO FIND OUT MORE so that I might LEARN people further insult. This is great.
 
rampantandroid said:
And again. I LOOKED FOR INFO. Why? Because I WANT TO KNOW MORE. So instead of responding and telling me where TO FIND OUT MORE so that I might LEARN people further insult.
Click to expand...

What insult?

You posted incorrect information, and Old Hippie and I posted the correct information. And I informed you that it has been discussed at length here in the past. That is useful information, since you know you are not looking for something that does not exist.

Because I do not want to waste my time to do searches for you that you can do yourself certainly does not constitute an insult.
 
How (or does it even) garbage collection function with TRIM? Is GC even necessary if TRIM is supported? Is it even possibly detrimental in terms of added wear?

From what I understand some SSD controllers have more aggressive GC than others, are these solutions actually worse if TRIM is supported?
 
Sorry to clarify. I'm wondering if GC is redundant with TRIM support. And whether or not GC (depending on how aggressive it is) can actually have a detrimental effect due to extra writes on the drive (at least I believe this is side effect of how GC functions, not certain on this).

In terms of a picking a drive, basically would choosing one using a controller that has less aggressive GC be better if you plan on using TRIM?
 
limitedaccess said:
Sorry to clarify. I'm wondering if GC is redundant with TRIM support. And whether or not GC (depending on how aggressive it is) can actually have a detrimental effect due to extra writes on the drive (at least I believe this is side effect of how GC functions, not certain on this).

In terms of a picking a drive, basically would choosing one using a controller that has less aggressive GC be better if you plan on using TRIM?
Click to expand...

The write issue, at least, is nothing to worry about. All current and recent SSDs have more than enough write lifespan for it to not be a factor. The whole lifespan issue was pretty much overblown from the very beginning.
 
limitedaccess said:
Sorry to clarify. I'm wondering if GC is redundant with TRIM support. And whether or not GC (depending on how aggressive it is) can actually have a detrimental effect due to extra writes on the drive (at least I believe this is side effect of how GC functions, not certain on this).
Click to expand...

No, GC and TRIM provide different but complementary functions. If you took away one or the other, it would generally hurt performance.

Idle GC does result in writes, and more writes depending on how aggressive it is. But as Forceman already said, that is nothing to worry about.
 
Last edited:
raid your ssds, that is the final answer. performance will never drop below half. even john4200's source shows that the performance never dips under 50%. I don't know why he insists the raid0 will be slower, but if he wants to be misguided, let him live in a world of solo ssds.
 
gjs278 said:
raid your ssds, that is the final answer. performance will never drop below half. even john4200's source shows that the performance never dips under 50%.
Click to expand...

That is not correct. It is quite possible for a RAID-0 of two SSDs to be slower than an equivalent single SSD with TRIM. It depends on your SSD model and how full your SSD(s) are.
 
JLAW904 said:
I was thinking of getting either 2 M4's or 2 C300s in RAID0... Would I be better off just getting a larger single one? I wasn't sure if they support TRIM or the garbage collection when in RAID.
Click to expand...

Unless you have specific performance needs that you have not mentioned, you will almost certainly have a better experience with one large SSD.
 
john4200 said:
That is not correct. It is quite possible for a RAID-0 of two SSDs to be slower than an equivalent single SSD with TRIM.
Click to expand...

I'm sure you have plenty of drives lying around to reproduce this. go for it and post the results here.

Unless you have specific performance needs that you have not mentioned, you will almost certainly have a better experience with one large SSD.
Click to expand...

it's cheaper to buy two 64gb c300s when they are shellshocking on newegg and raid them than it is to buy a single 128gb c300. on top of that, the sequential on the 128gb c300 is only 355mb/s. you get a huge boost in sequential read when using the two 64's instead. you also get a 10mb/s write boost when each drive can do 75mb/s write and the solo 128gb can only do 140gb. it makes perfect sense to raid c300's.

if you're looking for 256gb of space, 4 64gb c300's is much much greater than the performance of a solo 256gb c300
 
Because I asked for a source and you instead respond trying to insult me. How the fuck can people find out new info when people like you are around?
Click to expand...
It was no insult and I'm sorry you took it as such but it gets pretty frustrating when you see the same incorrect information over and over.

This issue has been discussed many times before.

Here's a great article on the basics of SSDs and I've taken the liberty of starting you at the page The Cleaning Lady and Write Amplification.

The Cleaning Lady = GC.

The whole article along with the first one will give you an understanding on how SSDs work along with a time-line of the advances/failures that have happened along the way.
 
I LOOKED FOR INFO. Why? Because I WANT TO KNOW MORE. So instead of responding and telling me where TO FIND OUT MORE so that I might LEARN people further insult. This is great.
Click to expand...
Sorry, and I can understand your frustration.

I respect anyone that wants to increase their knowledge but you need to understand I also get frustrated when I see incorrect statements given as advice on an enthusiast computer message board by someone that should, but doesn't, know the basics. :)

The ugliest part is that some other novice doing a search for GC is going to come across your post and use it on another site as proof. LOL!

I can only but hope they'll continue reading to get the truth. :)
 
gjs278 said:
on top of that, the sequential on the 128gb c300 is only 355mb/s. you get a huge boost in sequential read when using the two 64's instead.
Click to expand...

Yes, the sequential IO speeds will usually be faster in RAID than with a single bigger SSD.

But the thing is, most people do not need sequential speeds in excess of 400MB/s read, and in fact will not notice the difference in everyday usage.

What most people WILL notice is the hassle of dealing with a RAID-0 volume as compared to a single large SSD. In addition, the RAID-0 volume will be a lot more likely to fail and lose data than the single large SSD.

So, as I wrote before, unless the OP has a specific need for the performance of the RAID-0 volume of SSDs, he will probably have a better experience with a single large SSD.
 
You must log in or register to reply here.
Back
Top