Raid 0 Performance

[diamondmines]

OC, rendering and games. Anything you can think about. Even transferring larger junks of files around.

[kalniel]

OC, rendering and games. Anything you can think about. Even transferring larger junks of files around.

Overclocking will not benefit from RAID 0 SSDs, Rendering will not benefit. Games will not benefit.

Only the last one might be affected, but it would be limited by your target/source destination. Transferring from one drive to another is just about the same speed regardless of whether those drives are in a RAID or not.

[diamondmines]

So what really is the benefit of raid 0 ?

[kalniel]

So what really is the benefit of raid 0 ?

Very little really. It was of some benefit when you had slow mechanical drives and you were doing lots of storage limited tasks or have lots of users accessing files at the same time, but most people don't use computers as file servers etc. so there's not much need for it on the desktop, especially not since SSDs.

[Agent]

So what really is the benefit of raid 0 ?

When you a high data transfer date. As Kalniel has pointed out, this is rarely the case for home users. You'd need to be moving a *lot* of files around *very* often for RAID0 to be a solution in a modern system.

There is a golden, but old rule when it comes to RAID: If you don't know that you need RAID, it's almost certain you don't need RAID.

[Zak33]

may I ask something please, about the alleged increase in life span.

Surely the part of the ssd that will wear our eventually will be the (dunno word for it?) table of contents bit.... cos it's accessed all the time

so putting half as much data onto certain chips on each ssd is irrelevent, no? COs the TOC (or whaever it's called) still gets overwritten as many times.

[Agent]

Surely the part of the ssd that will wear our eventually will be the (dunno word for it?) table of contents bit.... cos it's accessed all the time

so putting half as much data onto certain chips on each ssd is irrelevent, no? COs the TOC (or whaever it's called) still gets overwritten as many times.

Accessing data on a SSD doesn't inherently wear it. One of the wonderful things with NAND is that when you can't toggle a bit, it gets 'stuck'. This means even when NAND has been worn through, you can still read your data off it - just not change it.

If a part gets updated a lot - wear levelling will kick in (and there are many other features of SSDs which keep their life nice and long). Accessing one bit on NAND chip X is the same speed as accessing it from NAND chip Y.

[Zak33]

so does raid lengthen life then or not?

[diamondmines]

But I have noticed windows boot times to load faster, programs to load faster and multi-tasked operations to be much faster with raid-0. So there is a benefit. I think I'll just stick to 2x raid-0 because it doesn't look it's a good idea to go for more drives.

Thanks again.

[diamondmines]

Can I raid-0 two Seagate drives; one is 250GB and the other is 500GB. They are both 7200RPM drives.

Thanks

[Oswald]

So what really is the benefit of raid 0 ?

Bragging rights in benchmarks... For every day use, I'm sure most people will hardly notice any difference at all. The seconds that are won in everyday computer use by going to a raid 0 SSD setup don't make up for the hours lost by googling for SSD reviews, tweaking, benching, ...

[tfboy]

Can I raid-0 two Seagate drives; one is 250GB and the other is 500GB. They are both 7200RPM drives.

Thanks

Yes you can, but the capacity will be the smallest drive's capacity multiplied by the number of drives so 250GB x 2 = 500GB in your case - the remaining 250GB of the 500GB drive will not be used.

[diamondmines]

Thanks.

[tfboy]

You're welcome. BTW, there's a "Thanks" button on the bottom left of each post you can click for thanking people

[watercooled]

so does raid lengthen life then or not?

Realistically, I'd say it's very unlikely, if anything the opposite because you'll may be increasing write amplification for smaller writes, depending on stripe size and alignment.

Drives with different capacities actually do have different performance in many cases, especially for writes because of more physical devices to spread them over, but reads may saturate the controller even on smaller capacity drives. Most drives at a given time will use the same size NAND dies, higher capacity drives may use more dies per package and/or more packages.

A decent quality controller should never allow a load of writes to wear out a given area of the drive because of wear levelling, which is why defragmentation is not only bad because of unnecessary writes, it's also utterly useless because the drive constantly rearranges where data is stored on the NAND anyway.

As for the ~1GB/s results on Youtube, they must simply be the result of caching somewhere, 2x200≠1000.

[DanceswithUnix]

Interesting article on Anadtech on a Sandforce firmware bug which may explain your lack of performance:

http://www.anandtech.com/show/6107/c...s-240gb-review

as I believe your drives are SF based?