Crucial m4 128GB vs OCZ Vertex 3 MaxIOPs 120GB

[zalbard]

On an unrelated note...
Tony, what's the official stance on page file on SSDs? Would it noticeably accelerate the NAND wearing? Would it throttle the drive much? Do you recommend a small or a large page file to reduce the amount of written data (512MB, 1GB, 2GB, more)?

[Ao1]

is it so difficult for ocz to state what?

The drives have to throttle, simple fact, without it you would have nand burn in weeks if a few months...then you try and RMA??? LOL

Except the product is as it is, Duraclass and life write throttle is needed. If you want faster lobby the nand manufacturers to get 100K PE/c on cheaper MLC and we will build drives just how you want them.

The nand dictates how the drives work remember

Guys...you want faster here is how to do it.

Raid 0

lose 50% of the array to OP

live with the day to day speed...the drive will last for years

dead simple solution

I can understand the advantage of DuraClass now. Without it you have to set an arbitrary cap on write speeds to provide longevity over a given time frame, whereas DuraClass allows short bursts of "uncapped" write speeds. The down side being that with heavier use write speeds have to be slowed down to even out wear. So you have a potential advantage or disadvantage depending on your workload over an arbitrary cap.

Most desktop users don't write enough, so they only see the benefit of "uncapped" speeds.

I guess it's a bit like a cigarette lighter, you can turn the flame up but the gas runs out quicker. Turn it down you have a small flame, but the fuel lasts longer. Here DuraClass regulates the size of the flame to make sure you don't run out of fuel over a fixed time frame. Depending on your use you could have a flamethrower or you could have a candle.

I guess the "disappointment" is that by combining compression and "A" grade 34nm that throttling is needed with SF drives.

As NAND gets smaller and controllers get faster the solution is increase performance throttling? Sounds counterproductive.

Anyway, why not give customers the choice? Sell an un-throttled SSD, but the warranty is based on "X" amount of writes and not a period of time?

[Antioch]

On an unrelated note...
Tony, what's the official stance on page file on SSDs? Would it noticeably accelerate the NAND wearing? Would it throttle the drive much? Do you recommend a small or a large page file to reduce the amount of written data (512MB, 1GB, 2GB, more)?

I'm curious about this, too.

[Ao1]

According to MS:

...."most pagefile operations are small random reads or larger sequential writes, both of which are types of operations that SSDs handle well.

In looking at telemetry data from thousands of traces and focusing on pagefile reads and writes, we find that

• Pagefile.sys reads outnumber pagefile.sys writes by about 40 to 1,
• Pagefile.sys read sizes are typically quite small, with 67% less than or equal to 4 KB, and 88% less than 16 KB.
• Pagefile.sys writes are relatively large, with 62% greater than or equal to 128 KB and 45% being exactly 1 MB in size.

In fact, given typical pagefile reference patterns and the favorable performance characteristics SSDs have on those patterns, there are few files better than the pagefile to place on an SSD."

Personally I don't use a page file at all. It generates unnecessary writes and adds latency to my set up with 8GB of RAM and a SSD.

[johnw]

is it so difficult for ocz to state what?

The drives have to throttle, simple fact, without it you would have nand burn in weeks if a few months...then you try and RMA??? LOL

To make a clear statement on throttling settings, firmware related to throttling, and tools that can change the throttling setting.

The drives do not "have to" throttle -- the alternative is, as you say, to let the flash memory wear out as it will. It may be OCZ's policy that disabling throttling is not supported, and that OCZ will not acknowledge the existance of a method to disable throttling. If so, it is odd that OCZ does not make a clear statement about that. The [nod-nod-wink-wink] that a program exists to change the throttling settings without a clear statement on the subject is not the best way to handle it.

[Marc HFR]

Anand's test of GC recovery is not very realistic. He uses HD Tach to do a 100+ GB sequential write to the SSD and looks at how the write speed degrades during that humongous write. Not many people do that to their SSDs. Also, he does the same thing to Sandforce SSDs, which is absurd, since HD Tach writes zeros, and Sandforce ends up writing almost nothing to the flash when you give it 100+ GB of zeros.

If you give the m4 a more realistic workload (for example, IOMeter with a reasonable mix), the GC works well.

Sorry but i doubt that HD Tach is writing 100 GB when doing is bench.

Even in long bench it's quick and i think it's less than 10 GB. It's not a full write ... HD Tach is ok for HDD but for SSD it's a non-sense.

And it's 10 GB of heavy compressible data for SandForce. Perhaps what is shown as a "full write" in fact wrote 1 GB on the NAND on a SandForce SSD ...

Useless.

[Marc HFR]

The drives have to throttle, simple fact, without it you would have nand burn in weeks if a few months...then you try and RMA??? LOL

Sorry but there is something i don't understand.

When i use my SSD, if i have to write for example 50 GB it ... i have too. If it take 6 ("new state") or 9 minutes (throttling) for example at the end we still have 50 GB written on it. I will not stop to write on it if i find it's slower ... if i write file it's that i HAVE to. And i don't see why it will change the SSD life at the end since it will in both case use the same PE cycles.

[alfaunits]

The amount of PE cycles used differs with different writing patterns.
It also differs with time (as it essentially changes the patterns)

IMO, it's pure marketing BS to say this is done to save NAND life (not your choice, it's MY darn drive ), but I can see why throttle would be used.

[johnw]

Sorry but i doubt that HD Tach is writing 100 GB when doing is bench.

Even in long bench it's quick and i think it's less than 10 GB. It's not a full write ... HD Tach is ok for HDD but for SSD it's a non-sense.

Good point. I do not know any details about how HD Tach works, I was just making a probably bad assumption based on the labeling of the graph it produces. It may very well write a long sequence, then skip forward 10 times that amount, then write again, etc. Even so, it is not a very realistic write pattern for an SSD, so I still do not think Anand's test is a good indication of how well GC works in realistic usage.

[Ao1]

LOL, the OCZ warranty does not even include wear and tear! What is the point of write throttling/ DuraClass?

Limitation of Warranty:

• Connection to a faulty power source
• Alteration, Modification, Disassembly or unauthorized repair
• Improper use of product
• Normal wear and tear
• User inflicted intentional or accidental damage
• Any other cause not resulting from defects in materials or workmanship.

[alfaunits]

No manufacturer warrants that. I.e. the NAND has a limited life-cycle, none will support an RMA if the NAND is dead as a result of enough PE cycles.

[Ao1]

Exactly - so why throttle performance when NAND burn out is not covered by the warranty anyway?

[johnw]

if i write file it's that i HAVE to. And i don't see why it will change the SSD life at the end since it will in both case use the same PE cycles.

Agreed. I think the only case where throttling might serve a purpose is if you computer has a virus or other malfunction that causes it to write to the SSD continuously. Then throttling would kick in and keep the SSD at a permanently slow state.

But I think that is a poor way to handle it, since it has the side effect of slowing down valid writes after heavy use. A better way to handle it would just be to have a software toolbox with an alarm that monitors the wearout indicator and gives a warning if the flash is being worn out too quickly.

[Ao1]

Looks like the Max IOP has just been maxed out by Corsair at 85K IOPS.

http://www.corsair.com/solid-state-d...ard-drive.html

[Anvil]

I noticed
I've already ordered one, ETD May 27th.

Still, the V3 Standard Edition will most likely be identical in 99.9% of the tasks, unless theyve gone a special/expensive route when selecting NAND.
My main reason for ordering the Corsair is that it hopefully will fit in my laptop, the V3 doesn't fit in the Lenovo W5XX series, it's to wide at the connectors end.

[Antioch]

Hahaha, wow!

And Anvil you have too much cash to burn, what is c300s, m4s, v3s, and not force 3's. Im jealous though!

[Marc HFR]

Looks like the Max IOP has just been maxed out by Corsair at 85K IOPS.

http://www.corsair.com/solid-state-d...ard-drive.html

Just on problem, it's noticed as a sequential performance

On Vertex 3 120, QD32, i can get with sequential access with compressible data :

79797,85 IOPS for read 4K
82222,12 IOPS for write 4K

[Ao1]

The Corsair Force Series 3 specs state Random Write 4K 85,000 IOPS (Aligned)

The OCZ MAX IO states Random Write 4KB (Aligned): 75,000 IOPS

Either way its useless marketing figure that doesn't mean anything in real life

[bluestang]

Not that I'm going to buy one but thought I'd share with you guys...$219 sale... http://www.ssdestore.com/VTX3MI-25SA...5SAT3-120G.htm

[Marc HFR]

The Corsair Force Series 3 specs state Random Write 4K 85,000 IOPS (Aligned)

Where did you read that it's random and that it's write ?

We only go that on Corsair website :

Sequential Read/Write (using ATTO Disk Benchmark)
550 MB/s sequential read
520 MB/s sequential write
85K IOPs (4K aligned)

And in the press release

"The Force Series 3 features the highest performance version of the latest-generation SandForce SF-2281 SSD Processor with out-of-box read and write speeds of up to 550 MB/s and 520 MB/s respectively, and an impressive 85000K IOPS for measurably faster system response, boot times and application load times than SATA 2 solid-state drives"

Edit : Finaly got it from Corsair by mail
"The 85K IOPs number is 4K, 100% random, 100% write, 4K sector-alignment, using IOmeter 2008. This was done using an Intel P67 system, using the latest Intel RST drivers."

[beast200]

hi i am thinking of getting a m4 or v3
would there be much gain over a x25 80g which i am using on sata III port

cheers

[cherrypik]

Not that I'm going to buy one but thought I'd share with you guys...$219 sale... http://www.ssdestore.com/VTX3MI-25SA...5SAT3-120G.htm

Anyone know the seller rating at ssdestore?

[bluestang]

Anyone know the seller rating at ssdestore?

I know there are listed Runcore's website. Not on ResellerRatings.com though, so not sure.

[Nizzen]

Where did you read that it's random and that it's write ?

We only go that on Corsair website :

Sequential Read/Write (using ATTO Disk Benchmark)
550 MB/s sequential read
520 MB/s sequential write
85K IOPs (4K aligned)

And in the press release

"The Force Series 3 features the highest performance version of the latest-generation SandForce SF-2281 SSD Processor with out-of-box read and write speeds of up to 550 MB/s and 520 MB/s respectively, and an impressive 85000K IOPS for measurably faster system response, boot times and application load times than SATA 2 solid-state drives"

Edit : Finaly got it from Corsair by mail
"The 85K IOPs number is 4K, 100% random, 100% write, 4K sector-alignment, using IOmeter 2008. This was done using an Intel P67 system, using the latest Intel RST drivers."

What QD and how many workers?

[Antioch]

hi i am thinking of getting a m4 or v3
would there be much gain over a x25 80g which i am using on sata III port

cheers

I think there should be. That older Intel drive has both lower random and sequential performance than the new SSDs (even in spite of being 22nm).