Sata 3(6gb/s) controller

[zedekc]

I was going to buy the Asus U3S6, but the supply seems very scarce atm, and the only people with it available have jacked the price up?


Does anyone have a recommendation as far as a comparable or better sata 3(6gb/s) controller for around the same price?

 

[LordConrad]

Any SATA III card that does not use a Marvell controller should work fine. Marvell controllers have some issues at SATA III speeds.

 

[groberts101]

All the cheap cards will use the Marvell 6G chips so any SSD that goes beyond the limits of that chips 400/250 R/W speeds will see a bottlenck. The Marvell's internal PCI-E x1 archetecture is the limiting factor there regardless of what the card interface speeds are rated at(4x, etc).

Have to move to about $150+ raidcards to get away from them completely.

 

[LordConrad]

@Gene O
I can't point out any such cards because I haven't ever looked for them. I do know that I would not be willing to accept Marvell's limitations unless I got the card super cheap (or free). In my case, I have two SATA III ports on my motherboard courtesy of a Marvell chip which I don't complain about because they were essentially free.

 

[zedekc]

so with a lower end drive like the M4 it wouldn't be too badly bottlenecked since it has lower write speeds which im not too concerned about anyhow.

400mb read 250mb write is not too bad for me with a controller priced at 30 bucks. Too me I guess it would not be worth it to spend more on the controller to get a small boost in read performance.(which i really want great read performance for multitasking) , but for the price point in the drives i'm buying it doesn't make it worthwhile to ad that expensive of a controller to get past the internal 1x pci slot limitation.


So the only question i would have then is if its 1x internally that means i would get the same performance out of using it on a pci 2.0 1x slot then?

If i buy a stated 1x pci card as apposed to the 4x pcie stated slot like the Asus U3S6, they both are limited internally so the performance should be the same, its just a matter of getting the card to physically fit in the pci slot. I would be better off saving my good x8 pci 2.0 slot for something else that would require more bandwidth.



Why would asus do that for marketing purposes, I know they have to be smarter than that right? lol

 

[groberts101]

#1 so with a lower end drive like the M4 it wouldn't be too badly bottlenecked since it has lower write speeds which im not too concerned about anyhow.


#2 400mb read 250mb write is not too bad for me with a controller priced at 30 bucks. Too me I guess it would not be worth it to spend more on the controller to get a small boost in read performance.(which i really want great read performance for multitasking) , but for the price point in the drives i'm buying it doesn't make it worthwhile to ad that expensive of a controller to get past the internal 1x pci slot limitation.



#3 So the only question i would have then is if its 1x internally that means i would get the same performance out of using it on a pci 2.0 1x slot then?

#4 If i buy a stated 1x pci card as apposed to the 4x pcie stated slot like the Asus U3S6, they both are limited internally so the performance should be the same, its just a matter of getting the card to physically fit in the pci slot. I would be better off saving my good x8 pci 2.0 slot for something else that would require more bandwidth.



#5 Why would asus do that for marketing purposes, I know they have to be smarter than that right? lol

#1 In theory?.. yes.

#2 I've seen decent results for that card but would probably lean towards the Highpoint RcketRaid 620 for a tad more money. Have to take into account that drivers used can and do have slight impacts on total speeds.

#3 In theory.. yes.. though I doubt you will find any at x1 speeds anyways.

#4 same as above. Doubt you will find a x1 speed 6G card anyways. You would not need to use the x8 slot but should double check your board to be sure that the PCI-E slot runs at least up to the 2.0 spec. Some have had issue on certain ports only to realize that the board has them running at PCI-E 1.0 speeds causing heavy bottlenecks.

#5 you answered your own question there. lol Is just marketing and bigger numbers on the package.

 

[zedekc]

do pci-e 2.0 1x slot support up to 500mb/s i thought i read that somewhere. Well i guess it wouldn't matter for the marvell controller based sata controllers because someone already said(groberts101) they bottleneck at 400mb/s read 250mb/s or something like that.



what would be a good $150 controller to get past the cheap marvel internal controller bottleneck issue?

(and also do regular sata cables just go at the 6gb/s speed or do i need special 'sata 3' cables?)

 

[zedekc]

also I read in product reviews that the asus motherboards do not have enough bios cache or something to support alot of the sata controllers one was an esata controller rocketraid 622, one such person said

'Pros: Looks like a nice card. I have the HighPoint Rocket 620 in another machine and it works fine.

Cons: Doesn't work with my Phenom II x6 AM3 system using an Asus M488TD-V EVO/USB3 mobo. In the two full PCIe 2 slots it displays during boot but then the system hard-hangs before doing the final BIOS output. In the short PCIe 2 slot the card firmware never displays - it's like the card isn't in the system at all.

My M488TD-V EVO system is very plain vanilla with no other PCI or PCIe cards since I'm using onboard video and nothing but mouse, keyboard and monitor connected so I don't think this is a system configuration issue. It does have 16GB of memory but that shouldn't be a factor.

Other Thoughts: At first I thought I had a defective 622 but now I think that the Asus mobo doesn't have enough BIOS memory headroom to allow the firmware to initialize. This happened to me with another RAID controller in my Asus P6T6 mobo system. Ironically, I was able to use the Rocket 620 in that system.

This problem with Asus and BIOS memory is a known issue - just google it. Essentially don't expect to be able to add PCIe cards with significant firmware (eg controller cards) to Asus-based systems. They need to cut out the BIOS cruft like expressgate or move to EFI which has lots of memory to work with.

Only giving this 3 stars because I don't know if it actually works or is defective but I suspect the BIOS headroom issue'



the problem is i have a phenom ii (x4 though) and an Asus M4A79XTD EVO mobo. So i'm worried that with rocketraid cards that it will not work. Any thoughts? or opinions on this assertion?

 

[pat]

I had the same problem with an Asus motherboard and a rockrtRAID card. The problem was that I wanted to boot from a RAID array created with the AMD raid controller and if I installed the card in the computer, then it would not see tha AMD RAID and could not proceed to boot. The same setup was working right with a Biostar motherboard. So yes, Asus BIOS has a bug.

But I finally made it working by flashing the rocketRAID card with the option of not bootable. That prevented the BIOS to mount it as a bootable device, allowing me to normally boot fom the array.

So, the problem seems that you cannot have 2 RAID controller set as bootable at the same time. The lack of cache memory seem a plausible explanation for that, as it happen only with Asus motherboad (at least in my case).

There is ways to fix that behavior.

1) If you plan to boot from the rocketRAID card, just set the AMD controller to IDE so BIOS won't have to mount extra stuff and you'll be able to boot. Not mounting the AMD RAID/AHCI bios will leave room for the rocketRAID bios.

2) If you want to boot from the chipset controller, flash the rocketRAID card (mine was an old 2300 PCI-e controller, don't knw if the new one have this option) and disable the boot ability of the card.

3)Buy another not Asus motheboard...

 

[zedekc]

I had the same problem with an Asus motherboard and a rockrtRAID card. The problem was that I wanted to boot from a RAID array created with the AMD raid controller and if I installed the card in the computer, then it would not see tha AMD RAID and could not proceed to boot. The same setup was working right with a Biostar motherboard. So yes, Asus BIOS has a bug.

But I finally made it working by flashing the rocketRAID card with the option of not bootable. That prevented the BIOS to mount it as a bootable device, allowing me to normally boot fom the array.

So, the problem seems that you cannot have 2 RAID controller set as bootable at the same time. The lack of cache memory seem a plausible explanation for that, as it happen only with Asus motherboad (at least in my case).

There is ways to fix that behavior.

1) If you plan to boot from the rocketRAID card, just set the AMD controller to IDE so BIOS won't have to mount extra stuff and you'll be able to boot. Not mounting the AMD RAID/AHCI bios will leave room for the rocketRAID bios.

2) If you want to boot from the chipset controller, flash the rocketRAID card (mine was an old 2300 PCI-e controller, don't knw if the new one have this option) and disable the boot ability of the card.

3)Buy another not Asus motheboard...

thanks pat i appreciate your response!

 

[zedekc]

All the cheap cards will use the Marvell 6G chips so any SSD that goes beyond the limits of that chips 400/250 R/W speeds will see a bottlenck. The Marvell's internal PCI-E x1 archetecture is the limiting factor there regardless of what the card interface speeds are rated at(4x, etc).

Have to move to about $150+ raidcards to get away from them completely.

sorry i meant to direct that question to you robert:

'do pci-e 2.0 1x slot support up to 500mb/s i thought i read that somewhere. Well i guess it wouldn't matter for the marvell controller based sata controllers because someone already said(groberts101) they bottleneck at 400mb/s read 250mb/s or something like that.



what would be a good $150 controller to get past the cheap marvel internal controller bottleneck issue?

(and also do regular sata cables just go at the 6gb/s speed or do i need special 'sata 3' cables?)'