Crucial M4 ATTO speed?

[samuel1]

Howdy everyone,

I am running a Crucial M4 (128GB) SSD on a SATA III controller card (PCI-E) and it seems the highest speed it can achieve is 200MB/s read and ~175MB/s write.

Does this seem particularly "slow" to anyone or is it just me?

Many thanks for your solicited opinions 😛

 

[AkumaX]

link to controller?

 

[beginner99]

See thread

http://forums.anandtech.com/showthread.php?t=2227791&highlight=marvel+sata


and following review:

http://www.anandtech.com/show/2973/6...-intel-x58-p55


Simply said, non-native SATA-3 controller are in general pretty much useless. Sorry to tell you, your card is worthless unless you need additonal SATA ports but for speed, stick to native Intel SATA-2 ports.

I made a similar mistake. back then bough an expensive mobo with Marvel SATA-3 controller only to figure otu later it is a) very slow to POST b) does not reach SATA-3 speeds and c) no TRIM on ssd.

 

[bryanW1995]

That seems slow to me, I was around 505 read and 270 write on each of my 256gb m4's before putting them in RAID.

 

[groberts101]

It looks to be throttled down to revision 1.0 spec's and placing the card in another PCI-E slot should allow the reads to nearly double. If your board allows the card to run at revision 2 spec's, that is. May want to double check the cards spec's too. It needs to be at least x1 speed with revision 2 spec's.

If you can get rid of that bottleneck.. the writes will go up to that chips miserable 250MB/s limit too.

 

[sequoia464]

If you can get rid of that bottleneck.. the writes will go up to that chips miserable 250MB/s limit too.

That 128 gb M4 tops out at 175 on the writes. At least according to the specs.
I'm getting close to 390 MB's with two of them in raid on an AMD board.
Are they supposed to be faster with the new firmware?

 

[groberts101]

oops.. my bad. If you say the 128GB M4 reaches that ceiling?.. then who am I to argue that point. I owned mine for all of 3 weeks before returning it. lol

Then it would take the larger capacity version to hit that marvell write cap in that case.

My main point of the OP being bottlenecked by older PCI-E revison 1.0 spec's still stands though.

 

[sequoia464]

I wasn't clear, what i meant is it takes both of mine in raid0 to hit 380 on the writes. Singly they are only around 170 - 180.

For what I do, the reads are much more important so they work fine for me. I'd love to move into something a bit faster but not in the finances just yet.

 

[samuel1]

Wow -- thanks everyone for your thoughtful replies!

I am running an eVGA 750i (non-FTW) motherboard (which only has SATA II slots) and so I purchased the ORICO "PCI-E SATA III PAS3064-2S2E PAS3062-2E/PAS3062-2S" card.

It is completely unclear what the actual model number is -- it's super generic and uber-Chinese.

So -- I am not clear, if it's an SATA III card plugged into my PCI-E 1x slot, why is this a problem? Is the bandwidth insufficient?

Are all SSDs slow with small files? I honestly, for some reason, expected much faster performance when upgrading from my Raptor X (circa 2006 which is still running as my secondary drive).

Thanks again everyone!

 

[VirtualLarry]

Very few SATA controllers are truely efficient with bandwidth. The Intel SATA3 native controllers on the Z68/Z77 chipset are probably the best.

Look at it this way. A PCI-E 2.0 x1 slot only has 500MB/sec of bandwidth. SATA 6G has 600MB/sec of bandwidth. Most controller card have two SATA ports on them. So even in the case of best efficiency, that's 500MB of bandwidth to split two ways, and cannot come near supporting two 600MB/sec SATA ports.

 

[beginner99]

Wow -- thanks everyone for your thoughtful replies!

I am running an eVGA 750i (non-FTW) motherboard (which only has SATA II slots) and so I purchased the ORICO "PCI-E SATA III PAS3064-2S2E PAS3062-2E/PAS3062-2S" card.

It is completely unclear what the actual model number is -- it's super generic and uber-Chinese.

So -- I am not clear, if it's an SATA III card plugged into my PCI-E 1x slot, why is this a problem? Is the bandwidth insufficient?

Are all SSDs slow with small files? I honestly, for some reason, expected much faster performance when upgrading from my Raptor X (circa 2006 which is still running as my secondary drive).

Thanks again everyone!

Seriously, read the review I linked earlier...and what other people posted.

 

[bryanW1995]

I wasn't clear, what i meant is it takes both of mine in raid0 to hit 380 on the writes. Singly they are only around 170 - 180.

For what I do, the reads are much more important so they work fine for me. I'd love to move into something a bit faster but not in the finances just yet.

That info sounds correct to me, mine is 256gb and a bit faster at writes, which makes sense.

Very few SATA controllers are truely efficient with bandwidth. The Intel SATA3 native controllers on the Z68/Z77 chipset are probably the best.

Look at it this way. A PCI-E 2.0 x1 slot only has 500MB/sec of bandwidth. SATA 6G has 600MB/sec of bandwidth. Most controller card have two SATA ports on them. So even in the case of best efficiency, that's 500MB of bandwidth to split two ways, and cannot come near supporting two 600MB/sec SATA ports.

This is way better than the typical "just use the native sata II ports" comment. @OP: you probably wasted your money buying the controller. Try to return it if you can.

 

[Coup27]

I'm surprised nobody has taken these add-on card manufacturers to task over this, they're woefully inadequate to perform the task they're advertised to do.

 

[samuel1]

Hilarious. Sounds like I'd have better performance if I just stick to the native SATA II ports, then!

I'll give that a try and see if I get any faster performance.


Thank you everyone! :biggrin: