vertex 3 or c300 raid 0

[Midnight]

im looking to spend $300 on a ssd solution. i think i have a few different options in terms of next gen drives but the vertex 3 seems to be the best choice.

so i can get a 128gb vertex 3 for $300 or 2x 64 gb 2nd gen drives like c300s or g2s.

i know about lack of trim on raid discs.

it just seems like the raided solution will yield a faster solution?

will be running 6gb sata.

http://thessdreview.com/our-reviews...ssd-review-pcmark-vantage-comparison-testing/

that review tho says the vertex 3 256gb is faster but im not getting that one. it would be the 128gb version.

thanks.

 

[SickBeast]

IMO RAID is a PITA, plus you're losing TRIM which is a big deal IMO.

I say go for the Vertex 3 or even the C400 if you can wait a bit.

 

[Zap]

You don't want to use a Micron C300 or Micron C400 in RAID because the firmware isn't tuned for running without Trim.

 

[Concillian]

I think for most users the garbage collection (GC) issue of the Crucial / Micron drives are a non-issue. Most people don't run their drives FULL, and then write random data at a 32 queue depth for 20 minutes. Even if we assume this is only being done at 20 MB / sec, that's 3x more data than Anand admitted to writing on an average day let alone 20 minutes.

Think about what these torture tests are doing, they really are torture. Fill the drive FULL, then write random data as fast as possible for 20 minutes. This starts at over 100 MB / sec on most of these drives. You're talking about writing 25 to 200 GB of data (depending on the drive) in 20 minutes. This is a week's worth of writes for most people, without a weeks worth of time to run garbage collection.

If someone knows they're going to be fairly light on writes, and not fill their drive totally full, it should be preferable to put off GC until you aren't really doing anything else, as your performance when you are doing something should be more consistent. Certain users will be doing LOTS of random writes, and for the most part, they know who they are and will skip this kind of drive. For the average power user though, who does ~10 GB of writes a day like Anand does, and has 10 GB of free space available, this kind of garbage collection is not a game breaker, as they'll very likely have time to garbage collect that 10 GB per day, even with low priority GC like the C300 / M4.

Old drives that had NO garbage collection, where secure erase was the only solution was a major problem. I don't see low priority GC as the same issue. It will be a problem for some users, but probably not most users. I think this is an issue where the torture test is very much NOT a real world scenario. It's testing a corner case for a specific (high write usage) scenario.

At least that's my thought, am I wrong?

I saw this link, he did give it time to run it's GC before testing, though:
http://thessdreview.com/Forums/ssd-benchmark-locker/214.htm

 

[nanaki333]

IMO RAID is a PITA, plus you're losing TRIM which is a big deal IMO.

I say go for the Vertex 3 or even the C400 if you can wait a bit.

i used my G1 in raid0 for 2 years and just ran tony trim every so often. cleaned up my drives really nicely and kept the performance up.

 

[ShadowVVL]

what about corsair preference 3 drives are they any good?

I was thinking about trying one but i haven't found a bench or review other then ces coverage.

 

[Midnight]

well for my question it doesnt even have to be the c300. how about vertex 2 x2 vs 1 vertex 3?

2x 60gb vs 1x 120gb

thanks.

 

[Old Hippie]

Old drives that had NO garbage collection,

ALL SSDs have GC.

If not they would slow to a crawl after the cells were written to (used) the first time and never recover.

But you're correct about the GC issue with these drives.

I've run Intel G1s and G2s in RAID0 and have been running a Crucial on an LSI6Gb/s expansion card all with no TRIM.

TRIM/slowness/recovery has never been an issue but I don't even run my drives at 50% full.

Seems Crucial set-up GC in a little different manner but it's pretty much a moot point for most desktop users.