REVISITED:
Hi Everyone,
Well I thought it time to update everyone on the SATA and IDE HDD corruption issues. It seems that Via's latest SATA driver 220d has fixed the corruption problems once and for all. I won't go into any detail but the info below is for the most redundant and outdated. The process to follow to get a stable system is as follows.
1. Download and install the latest Via SATA drivers. This post will point to the download page as pointing the file will date with time. At the time of writing it is version 220d and it seem really solid.
http://www.viaarena.com/?PageID=310
2. Bios version 1006 final is rock solid as well. There is a new beta bios on th Asus site 1007.003 beta which has an updated raid bios as well but I have not tested this yet.
3. Via 4in1 drivers. I am running 4.49 as they are stable with the above bios and SATA drivers. I find the latest 4.51 unstable on my setup but that just might be me. Anyway you can get the latest 4in1 or Hyperion drivers as they are now called here.
http://www.viaarena.com/?PageID=300 If you want older drivers then you can get all versions of the older drivers at
www.guru3d.com
4. I am still running the MS Win Xp hotfix for large drives which can be found here.
http://support.microsoft.com/default.aspx?scid=kb;[LN];331958 It might not be needed now Via have got their act together but I am running it and it is stable. It updates the atapi.sys file which when updated allows MS Win XP to work with disks larger than 137GB. I have a 120GB hdd with a 8mb onboard cache. The hotfix forces the flush cache command to happen on the disk before shutdown and hibenation etc so in my books it is worth installing as the fix wont be perminant until SP2 comes out sometime this year.
Last words. I am finally happy to say that my 2 systems are now rock solid and never crash since the latest SATA 220d drivers have come out. Others are finding the same as you can see by reading the official VIA forums who were experiencing the same problems and finding the 220d drivers fixing all. See
http://forums.viaarena.com/messagevi...E=&STARTPAGE=2
I will post this elsewhere where relevant.
Enjoy SATA and IDE data corruption free computing from here on in. By the way that doesn't mean that if you are OC that it won't still happen. My system is running stock.
Shaun
Hi All,
A number of people are having data corruptions on these forums and all over the web actually. You can follow the posts on the problem here
http://www.amdforums.com/showthread....hreadid=268068
I thought George's post was worthy of it's own section so I have posted it here. At the bottom of George's email is my intial quetion to him if you care to read. I have asked George to keep us in touch with any further news on the subject and given him this link.
I am on a mission to get the bottom of this. I wrote to the famous George Breese see
http://www.georgebreese.com/net/software/#PCI who you might remember wrote a little program called "pci latency patch" which fixed the problem people were having with Soundblaster live cards, and data corruption on the early Kt133 mb's. He also has written speed enhancements for raid boards up to the KT400. Anyway I mailed him and asked his advice on the problem and the gentleman he is he wrote a detailed explaination of the problem and a workaround. Read and enjoy.
(I am going to publish this as a separate post as George might add to it over time as he does further testing.)
Hi Shaun
Before I begin… I need to offer a quick comment. My 0.20b21 patch doesn’t recognize the VT8237 on your A7V600 as a new-generation chip, so the patch will treat it as an old one. This could cause problems. I’ll eventually have time to post an update to the patch for the sake of the VT8237, but until then I think you should be very careful using my current stuff.
48-bit LBA is an extension to the LBA scheme that’s used in all current IDE disk drives. LBA allows the OS to refer to disk sectors by a simple number, whereas the OS used to calculate the physical location in terms of head number and track number on the disk platter. When disks over 137GB in size became available, the LBA numbering scheme didn’t support a large enough number to represent all the sectors on the disk. By extending the LBA numbers to 48 bits, the OS can now ask for sector numbers between 0 and two-to-the-47th-power. (I may have a digit incorrect here or there…)
I believe that I’ve seen the problem you’re describing, and on much smaller hard disks. The problem arises when a “SCSI” disk driver is running the IDE controller. Windows 2000 and XP support two models for disk drivers. One model is specifically for PC/AT compatible IDE controllers, usually built into the motherboard’s Intel/VIA/SIS/Nvidia chipset. The other model was originally meant for SCSI controllers, but non-PC/AT-compatible IDE and IDE/RAID must use this model as well because it’s the only other model available.
In the PC/AT compatible IDE driver model, Microsoft supplies almost all of the drivers. The only thing that VIA’s driver does in this case is to select which DMA mode to use for each drive. Since all of the rest of the drivers in this model are Microsoft-supplied and they all talk to each other, Microsoft has taught the OS to directly tell the drives to spin down and up. They can flush a drive and spin it down before finishing the Windows shutdown process, which means that no data is lost… Except when an LBA bug appears. J
In the SCSI driver model, the OS doesn’t know how to drive the controller hardware directly. The OS passes generic requests to the driver, and it’s up to the driver to convert the request into something appropriate. But, the driver might not convert all types of requests. I’ve taken advantage of this fact before. For example, Windows 2000 Server sends commands to some hard disks to disable their onboard write-caching feature. The hard disk’s performance suffers as a result. But I’ve worked around this by installing Promise IDE controllers, whose driver doesn’t handle Microsoft’s write-cache-disable command.
Now that I’ve written all of that…
I’ve seen cases where disks would corrupt data at shutdown. The disks were on IDE controller cards that used the SCSI driver model. Either Windows failed to send a flush-and-power-down request to the driver, or else the driver didn’t convert this request to a suitable IDE command.
The real solution would be to rewrite the IDE driver. If the driver isn’t converting the flush-and-power-down commands, it should be rewritten to do so. If Microsoft doesn’t send such commands to disk drives on SCSI-model drivers, then the driver could hook into the OS to discover when the OS was shutting down, and issue the appropriate commands directly.
I’ve been told that there is a workaround that can be used now, although I haven’t had reason to try it yet. Disabling ACPI power management in Windows is supposed to solve the problem. I just installed a Silicon Image SATA-RAID controller and 10K RPM drive in my spare nForce2 PC last night, so I may need to test this workaround very soon!
There are kazillions of discussions on the Internet, describing how to disable ACPI, so I won’t go into it here. I would suggest trying to run with ACPI disabled, if it’s convenient for you (in other words, if your data is backed up).
I hope this helps. -- George
Hi George,
I used your pci latency patch to solve data corruption on my old A7V133 but I am sad to say I am having massive problems on to near identical systems. Can I give your patch a go once again even though I am using the A7V600. You seem to know a great deal about via chipsets and would really appreciate any advice. My findings so fare are below and on the amdfourms link at the bottom. I realise you are a busy person but any response would be appreciated.
Shaun Potts
I have had a nightmare with two new systems using the Asus A7V600 systems using Seagate SATA 120Gb and 160Gb HDD’s. The systems are perfectly stable when running but when I turn them off the HDD often corrupts. I eventually tracked down the cause and MS faulty solution to the problem. Even though the hotfix says for HDD’s bigger than 137gb the flush hdd cached problem effects both disks as they have 8mb of cache on the HDD’s. My guess is that the HDD doesn’t finish writing to the MFT and other systems files before the power turns off.
Here is MS links to the hotfix and so far they sort of work. WHAT IS 48bit logical addressing in plain English? Anything else I can do to stop corruption of system files. It is not a hardware problem as I have replaced MB, HDD and Ram which is the only hardware in the machine.
“The ATAPI driver for Windows XP Service Pack 1 (SP1) does not use 48-bit Logical Block Addressing (LBA) when it writes memory dump files or hibernation files. Additionally, the flush cache command is not issued to a large hard disk that has 48-bit LBA enabled when Windows XP enters standby or hibernation.”
http://support.microsoft.com/default.aspx?scid=kb;[LN];331958 Hard Disk May Become Corrupted When Entering Standby or Hibernation or When Writing a Memory Dump
http://support.microsoft.com/default.aspx?scid=kb;[LN];331958 The FLUSH CACHE Command Is Not Issued and the Hard Disk May Become Corrupted When You Enter Standby or Hibernate or (shutdown)
Link to my a forum at
http://www.amdforums.com/showthread....hreadid=268068 There are a great deal of people experiencing this problem so any advice explanation would be appreciated and added to the forum.
Shaun
Processors
Motherboards
Chipsets
Memory
Graphics Cards
Storage
Cases and Cooling
Mobile
Systems
Displays
Shows and Expos
Linear Mode
