Did anyone ever have a problem with onboard Marvell controllers?

[BonzaiDuck]

Specifically, I'm interested in an answer to that question as pertains to the ASUS Z68 motherboards -- including the "Gen3" Z68's.

More than a year ago, I was having trouble with a very intermittent instability -- unannounced resets and BSOD's about once every seven to ten days -- once in a while after a month of 24/7 running.

I was troubleshooting for two months on the sig-rig. Eventually, after disabling the onboard Asmedia USB3, using the XMP profiles to configure RAM to essentially the same settings, and disabling the onboard Marvell SATA controller, the problem just disappeared and never, ever, ever returned.

Now I'm refining a similar system with a 2700K processor, the Gen3 version of the same motherboard, and SLI.

Looking at my event logs, the same problem has now surface on this second system! Every . . . seven to ten days!!

the one thing the two rigs have in common besides near-identical processors and motherboards: I was using the Marvell controller for one HDD on the sig-rig before I chose to disable it, and I'm using the onboard Marvell on the newer rig -- for hot-swap and connection to my front-panel eSATA port. I don't even have HDDs connected, but the controller is "Enabled."

 

[Schmide]

My exp with those MBs (Gigabyte z68xp ud3). They need a bit of uncore voltage to keep an overclock well.

 

[ronbo613]

The Marvell controller on my Gigabyte motherboard never performed as advertised and has caused a few hardware problems.

 

[Z15CAM]

As you know I run an i7 2700K/ASUS Z68-V Pro Gen3 however I have no need for the Marvel SATA Controller so I Disabled it in BIOS.

If you suspect the Marvel SATA Controller maybe causing your issue I would note the Chips number and go directly to the Marvel site and download their driver.

 

[BonzaiDuck]

My exp with those MBs (Gigabyte z68xp ud3). They need a bit of uncore voltage to keep an overclock well.

I've tweaked the VCORE's Offset and Extra Turbo Voltage -- the latter so I can keep the Offset from being unnecessarily high for an overclock. It works for the sig rig. VCCIO -- maybe VCSSA on newer boards -- needs a bump up for filling all four memory sockets and possibly for running RAM at Cmd-rate=1 when it's XMP profile defaults to 2.

I can reset to stock in troubleshooting this if it occurs again, but there are a lot of causes for 124 Stop Code and they can easily be software and drivers. Ah'm working on it! I just can't replicate the problem in an hour's time.

That's why I'm interested in ronbo613's answer and Z15CAM's as well. The very same trouble I'd had before went away about 9 months ago on the sig-rig after disabling the Marvell.

Ironically, I use some other PCI-E SATA-III controllers that use a different Marvell chip, offer up Marvell's Hyper-Duo SSD caching, and have proven rock-solid. But I also believe you can build the best computer, with the capacity you want, and with a minimum of controllers (preferably just one).

Z15CAM doesn't have any problems -- possibly but not conclusively because he doesn't use the Marvell controller -- he doesn't know, and I don't know. But ronbo613 reports an onboard marvell controller that threw up "hardware problems."

So I have to suspect it, as I suspected it last year. Since I'd intended to use it exclusively for hot-swap or eSATA ports, I can easily disable it to see if the problem disappears -- as may have occurred with a near-identical board last year.

The board-makers provide all these extras, and that's nice. It's just a form of Noob Over-Enthusiasm to try and load up a system with a fan in every vent, a drive in every bay, every controller enabled and turned on. You don't buy these boards because you intend to use every feature. The features just provide some options and some options may not work that well.

 

[BonzaiDuck]

So, if Z15CAM suggested disabling the Marvell controller, why did I enable it or attempt to use it if I suspected it as a destabilizing factor on a similar system?

The motherboard on this system was a refurbished offering from Ascendtech -- a corporate-IT "asset-handler" with two store fronts. I had been initially pleased with the board's performance -- a Z68/Gen3. Of course, it's easy to cast suspicion on the board itself if it wasn't factory-new in shrink-wrap when I bought it. It had cost me roughly 35% of the original retail price when resellers still carried it.

So when I initially tested two of the Intel SATA-II ports, they didn't seem to work. The other two SATA-II ports worked, no less than the Intel SATA-III ports. I suspected something like a solder-joint ring-crack. After all -- it was a "refurbished" board from Ascendtech.

Apparently, the SATA cables for those two ports had never been plugged in all the way. With one connection testing this time, it suddenly "works" properly. This of course means that I can disable to Marvell controller because I "just don't need it."

I still suspect the Marvell controller as a "trouble-maker," and only time will tell. But it does suggest a sort of first principle that can still be "flexible." "Simple is Best." So for the Noobs out there, start with a "simple" hardware configuration; don't load up your system with six or seven hard drives if you don't need to. And don't enable motherboard features you don't plan to use.