Solid State Drives (SSD)

[Zalodine]

My Crucial M4 is at the newest firmware 000F by default, I installed HDTune to test and it only reads at ~270MB (transfer rate test). I'm using the Sata III 6Gb/s port on my Asus P6X58D board and a proper SATA III cable , why am I receiving Sata II speed?

[targetbsp]

The x58 chipset doesn't support sata 3.  You need a 6 series or newer board.

Some x58 boards have third party support for 'sata 3' but its limited to 380mb/sec.  if you're plugged in that port I'm not sure why you wouldn't be getting those speeds.

You don't want to be in that port anyway though.  You want to be in the Intel ports as they are more stable and have better small file performance which will give better real world performance for most people.

[Zalodine]

Thank you very much targetbsp, so the prob lies in my system, not the SSD itself, that helped the day. I'll try getting a newer board soon, silly me... Good day!

P.S: But still I don't understand that I can't get at least 350MB/s...

[targetbsp]

---

Zalodine wrote:

P.S: But still I don't understand that I can't get at least 350MB/s...

---

Neither do I!

All I can suggest is double checking the port it is in and make sure you have the latest Asus bios and if no luck, maybe someone on the Asus forum can help?

[Holt]

The problem is, that the populare SATA 6Gb/s controller like ASMedia 1061 and Marvell 912x only offer one PCIe Lane for connection and that lanes are less then 5Gb/s in most cases. I saw one fellow getting 440MB/s of an ASrock PCIe x1 Card in the first PCIe x16 Slot of his 790FX board but only 360MB/s with the board in a different slot of the same motherboard. Judge yourself how to blame!

 The best solution would be to have a controller with at least 2 PCIe Lanes like Marvells 88SE9220, 88SE9230 or 88SE9235, but it is rare to find them even on High-End Mainboards and I havn't seen any PCIe Card with one of them.

But don't warry, the everyday difference is less then 10% when using only a 3Gb/s connection because only few accesses extend the length where transferrates hit the SATA 3Gb/s limit. Most are just to short and looking at the about 25MB/s one can read at 4k access length cs 500MB/s when reading 256kB I hope you get the point. 

[targetbsp]

---

Holt wrote:

 The best solution would be to have a controller with at least 2 PCIe Lanes like Marvells 88SE9220, 88SE9230 or 88SE9235, but it is rare to find them even on High-End Mainboards and I havn't seen any PCIe Card with one of them.

---

I'm *still* waiting for someone to make me one of those in card form!

[parsec]

Yes, those chips look like they might be better, but they would be only under certain circumstances. Check Marvell's product brief for those chips:

Marvel 92xx Product Brief

They can be connected to one or two PCIe lanes. What if they use just one? Ridiculous, but who knows?

Next, those chips can provide two or four SATA ports, depending on the model. Say they use two PCIe lanes for connection, great right? What happens when two or more drives are connected at once? Is the bandwidth of two lanes always available to one drive? If two drives were connected and both active (which is rare, true) are we back to one PCIe lane per drive again? I would hope these chips would be better, but we won't know until they are tested.

Why aren't they being used? I bet they cost more than the single PCIe lane chips. Instead of $1 each in volume, they are $3 each (pure speculation), so the mother board manufactures don't use them, and go with the $1 ASMedia chips.

It really makes me wonder about mother board manufactures, and how really fancy their products actually are...

[targetbsp]

I've been waiting for the 9182 (not a typo) which is also 2 lane (as opposed to the 9128)

[parsec]

Yes, I am familar with that chip, which was used on a few high end boards. It seems to have disappeared from Marvell's list of chips. As I recall, it performed at a very good level, close to the Intel 6Gb/s chipsets. Yet another engima of Marvell's product offerings, as well as mother board manufactures.