Need information on ASUS P7H55-M/USB3 Marvell Controller

[Patcy]

In my Motherboard I have a habit SATA2 3 GB / s ports and then there's the Marvel controller, which provides 2 ports for SATA3 6 GB / s.I recently purchased the OCZ Vertex 120 GB 3 (updated to the latest firmware 2.11) and connect the SSD at 6 GB / s Marvell port. To my surprise, ATTO reported extremely low rates in the benchmark, 207MB / s read and 125MB / s write .There is no trouble in the BIOS, wherever you set the SATA mode, and the Marvell controller mode for "Achi mode’

I have connected the SATA2 port Vertex3 native speeds and ATTO reported much better, 283MB / s read and 267MB / s write, which is quite in line with the performance SATA2.Does anyone have any idea why the Marvell controller speed is so poor? What is the point then with 2 of 6 GB / s ports, if you will never reach even close?

[Miguel Reyes]

I can only imagine, because I do not have one yet. If I remember the wonder PCI pass routes. Having speed issues on PCI lanes, always check that the negotiations track. CPUZ as main tab, you can see is 16 or 8 or 4 or 1. From the information provided would be the first thing I suspect. Other things could be like the caches are configured, ensuring that the caches and the device driver, and ensure that the restart after a change so that the best negotiation is done.

[Himalayan]

I can only imagine, because I do not have one yet. If I remember the wonder PCI pass routes. Having speed issues on PCI lanes, always check that the negotiations track. CPUZ as main tab, you can see is 16 or 8 or 4 or 1. From the information provided would be the first thing I suspect. Other things could be like the caches are configured, ensuring that the caches and the device driver, and ensure that the restart after a change so that the best negotiation is done.

Well, I'm told that the driver uses ASUS 6Gbps Marvell PCIe x1, which is, I think 2.5Gbitps be limited to performance. If that's the case, then how you do like 6 Gbps marketing is nowhere near.

[Battista]

The ATTO in this configuration is trying to check the disk itself "direct", I'm not sure that even the "direct" nothing more. Trying to ignore all wonderful (cough cough) "Improvements" that are, may give very different results? If a bank tries to communicate directly on a disk, maybe not doing it directly, and trying only makes it worse?

In terms of speed 6G, insurance marketing, but someday it might mean maybe some truth? Just before it is replaced with PSAT (parallel number) and marked up to 48 grams, or some other number of useless.

[Ramona19]

If I look at this link, your test reveals that they are receiving about 400 Mb / s read and write 132MB / s. I'm not even close to my table Marvell.

I'm getting 206MB / s read and write 125Mb / s. In fact, what you perceive, I'm getting half the speed they're getting in the readings, my writing is even closer.

[Gimbya]

Certainly that would not be the first to get much worse than the normal speed of the bad things. You have to connect something to the altitude. Check the drivers, check the link bandwidth, latency maybe mess with the bios for the bus and link speed items. Check out any of the write cache things? Could also be "held back" (my word), where an overclock is a bit high and some doesn't negotiate the same rate, or the transfer of data is each error and try again.

Also, while SATA3 can use the same cables, some of the cables supplied with things, and cables that worked very well for SATA2, could operate at a lower rate, instead of negotiating a full possible speed. so while SATA2 cables that work well, I could not believe that having large cables becomes more important when it comes to negotiating 2x speeds through the basic interface.

What else you could look smart properties, after running the unit in the port for a while. The heat slightly intelligent software and make a quick review of the few things that do not show the connection and wiring fault type.

[Cain]

Well, irritated with the Marvell controller. Now I have determined to upgrade to a P67 board (ASUS P8P67) with the i5-2500K. Atleast I will acquire pure 6Gbps speeds.

[Leslie f]

Well, I'm not convinced if this discussion does not have more than a few ASUS motherboards make use of a PLX bridge.

In my P55 board one generation of an x4 lane is divided into the USB3 and SATA III. It was almost certainly a trouble of sharing the lane.