IDE-AHCI-RAID... so far

[idata]

When I start my bios i can choose how to set up my SSD: AHCI,IDE,RAID

AHCI and IDE are similar (AHCI newer and more advanced, IDE more standard and compatible) AHCI, I guess, support the automatic TRIM in window7 while IDE don't but i am not sure about this...

RAID has two possibilities:

a) single SSD

Doesn't support automatic TRIM but support the TOOLBOX (correct me if wrong)

b) With RAID-0 SSD (2xSSD disk)

Doesn't support anything but go faster ^_^ (there is any way to trim or clean the empty space with this configuration?)

This said if you have just one SSD which one would you advice?

AHCI?

IDE?

RAID-mono disk?

[Steve_Lionel]

It is often the case that a PC cannot boot from an add-on storage adapter. The BIOS needs to have support for the adapter and this would be unusual for a plug-in card.

There are often restrictions on the add-on RAID chips on motherboards - for example, you can't attach a DVD drive to one of their connectors and have it seen by the BIOS.

[idata]

Here's how I understand this to work:

For Windows to boot from an add-in card, the card needs to have an Option ROM. The Option ROM is executed by the BIOS and implements the storage protocol supported by the card. The BIOS calls the Option ROM to detect disks, and if there are disks, the BIOS should at a minimum allow you to, in your boot order, select the disk that the card considers to be the boot disk, although there may variations on exactly how this appears in the BIOS.

Once the boot disk is selected, the BIOS issues I/Os to the Option ROM. The BIOS reads the appropriate boot sectors and executes them which leads to an OS loader running, which continues to use the BIOS to load the OS. Windows' loader loads things like the kernel, the registry and boot start drivers. Once they're loaded, the kernel starts the boot start drivers. One of them had better know how to control the boot storage controller and allow Windows to find the disk, or else you end up with a 0x7B blue screen - or what appears to be a reboot if you don't set the system to not reboot on error or use F8 to select that option for the current boot.

[GCatt]

Einride wrote:

The "No" from Rob @ ASUS does not match up with the rest of the text in Rob's message. Drives do not boot from drivers. Drives boot--this occurs when the machine bios hands off the start-up process to a selected disk--from the boot sector on the disk. Boot sector code then hands off the process to the selected operating system. Disk drivers are added to the equation by the operating system later in the overall boot process. Maybe Rob was responding to some other question--whether asked or not.

By the way, Rob, there is are drivers on Rapidshare for the Marvell Sata 6 controller.

I will be able to speak more knowledably in a few hours when my U3S6 arrives. Testing will be done on ASUS P55 boards.

Thanks, Einride. I'm very interested to learn the result of your test, as this will definitely prove the case for booting from the U3S6 Asus card.

[idata]

Here is some preliminary testing with a U3S6. The test system is an ASUS P7P55D, which is on the certified list for the U3S6. There is one disk in the system, an Intel 80GB G1 with Win7x64, in AHCI mode.

Device Manager showing devices by connection, looks like this with the U3S6 installed in the second PCIEx16 slot. (The specs on the PCIEx16 slots are: One x16 device runs at x16. Two X16 devices, both run at x8. One x16 device (in this case the video card) and one x4 device (the U3S6) is supposed to work at those rates.)

Starting with the second entry on the PCI bus you will see a PCI Express Root Port 1. This branches into a bridge, then two more secondary bridges. The first secondary bridge handles the U3S6 USB controller (note, no driver, as I had not installed the ASUS driver package at that time.) The second secondaary bridge talks to the U3S6 Sata6 controller, which shows two unocupied channels. Compare this setup to the motherboard Sata controller which appears on the list as the last direct PCI bus connection (note, the SSD is on port 0, which is not expanded in the pic.)

First, installing the U3S6 had no adverse effect on booting up an already running system. I did not observe any adverse effect on any other system function, although the time to find problems was very limited.

Next, I moved the Sata cable on the SSD from the motherboard connector to Port 0 on the S3S6. The system booted up the same as before (no change to bios config--there was only one disk in the system. The SSD then appeared in Device Manager on Channel 0 of the U3S6 with standard Win drivers. In fact, all of the drivers in the pic above related to the U3S6 are standard Win7 drivers, including the various bridges. Those bridges, I believe, represent hardware interfaces on the U3S6.

Next, I pulled an SSD from another working system, connected it to Motherboard Sata Port 0, changed the boot order in the bios to boot from the U3S6, and that worked without incident.

Last, I swapped the boot order to boot from the Motherboard Sata, and no go. But there was a good reason--the added disk was removed from a system with a bios configured for Sata raid and Win7 set up to use the iStor7 Storage Controller, while the test system bios was configured for AHCI. Then I rembered that I was testing on the only system here that used the AHCI bios option. So I quit testing before I was too severely tempted to make a change that would require a complete reinstall of someone else's computer.

So, I answered the question to my satisfaction that booting with the U3S6--at least on P55 boards--was not going to be a problem. I should point out that ASUS makes two Intel-based boards, the P7P55D Premium and the P6X58D Premium, which have the Sata6 components of the U3S6 directly on the motherboard. This might indicate that all boards that support I7 or I5 processors will work with Win7 and the U3S6--or not.

There were two issues with using the U3S6. First, S3 Sleep was flaky. I could put the system to sleep by using the Sleep button on the keyboard or by using the Start Menu Sleep shut-down option. But, something was keeping the sleep timer from working correctly. Which is to say, with the timer set to one minute, sleep might occur at 45 minutes, or some other time, or never. This problem went away when I reconnected the SSD to the motherboard Sata connection. This is not conclusive that the U3S6 controller caused this problem, since moving a disk from one controller to another controller of a different type with a different bus connection might not be consistent with the original installation of Windows. I eventually tracked down a Marvell driver and installed it. This did not change the Sleep timer problem. But it did change the driver structure from an AHCI Sata controller to a Raid Storage Controller, just like using raid mode in the system bios changes the on-board Sata port controller. I downloaded a bios update for a raid version of the Marvell sata6 controller, and found many references to bug fixes for sleep problems in the changelog. While this bios update will not work with this Sata chip on the U3S6, there might be an update later from ASUS to deal with my sleep issue. If sleep timing is not something you care about, then, not a problem.

The second issue was performance of the SSD, which fell off a little in small block read-write tests. I perceived no noticeable change in ordinary usage, however. I personally do not like AS-SSD tests, but, since these test are used a lot on this forum, here they are.

In the pic above, the SSD is connected to on-board port 0, and the U3S6 card (no drive) is installed. This is essentially the same result as without the U3S6.

In the pic above, the disk is running on the U3S6 using Win7 drivers.

In the pic above, the disk is running on the U3S6 controller using Marvell drivers. I found recent generic drivers for Marvell chips at Stationdrivers.com, and these are what I used. ASUS has somewhat older drivers for use on their motherboards which have the Marvell controller onboard. I have not tried them yet, but they might be customized to ASUS and work better--or worse. I will report on these later.

Read into these test results what you will. I prefer HD Tune for testing, which provides R/W transfer testing at 512 bite blocks, and at every binary multiple up to 8MB blocks. Plus a lot more. And no mysterious "Score".

I have not got around to doing anything with the USB ports on the U3S6. I read somewhere that there can be a bottleneck on bandwidth if one uses USB3 devices with Sata6 devices simultaneously through the U3S6 on a PCIe x4 connection, but no test reporting yet.

[GCatt]

Good work, Einride! Your tests validate the ability to boot from the Asus U3S6, which I didn't think was possible after receiving negative info from Asus tech support.

I also have an Asus P7P55D motherboard, but my configuration is different than your test system. I currently have three hard disks on the on-board Intel SATA controller:

Disk 1 – Windows 7 boot disk

Disk 2 & Disk 3 – A Raid-1 set for data

So, you can see that my on-board Intel disk controller is in Raid mode, and my Windows 7 install was done in to a disk connected to the Intel controller in Raid mode. I would like to move my boot disk to a SSD connected to the U3S6, but I'm not sure if I would need to re-install Windows since the U3S6 is in AHCI mode and my current system disk is in Raid mode. What do you think? Ideally, I would just recover a system image of my boot disk to the SSD, disconnect my current boot disk, and boot from the SSD. But, I'm not sure if I can do that since I will be moving from a system disk in Raid mode to one in AHCI mode. Thanks!!