Data Corruption Problem Solved

Per Hansson · Apr 23, 2003

It now seems like Abit has finally managed to fix the Data Corruption problem on the NF7-S mainboard, this is their reply to Nordichardware. They have also replied to the thread on their forums which I posted about here two days ago.

"Here's the latest update for SATA RAID-0 broken issue on NF7-S mainboards. After co-working with Silicon Image for one week, we finally found the SATA RAID-0 broken issue was caused by some improper register values. We already tested latest beta driver from Silicon Image; which fixes the wrong register values and had been confirmed to solve SATA RAID-0 broken issue in ABIT lab. Silicon Image will provide us another official release driver and ABIT will release it to customers in a couple of days." UPDATE: Asus has a driver which fixes the problem, it works on the NF7-S also (confirmed by myself plus others on Abit forums) you can download the driver here (before downloading please read my note below)

Tarkus · Apr 27, 2003

I sure hope it works and the final fix has no performance hit. I just ordered a pair of Raptor 10K RPM drives for RAID. We'll see.

Per Hansson · Apr 29, 2003

NOTE 04/29: While this update does solve the problem it comes at a hefty performance degradation:
(I.e. we are still waiting for the final fix from Abit and Silicon Image.)

Originally posted by biosmaster at Abit's forums
It is true that corruption has stopped.
Unfortunately it is true that in terms of sustained or random writes to the hard drive we have moved back to pre PIO4 days.

i.e Two SATA 80gig 8mb cache drives maximum sustained writes in Raid0 config 32K clusters 16.7 mb second , for comaprison on the same controller on an Intel Granite Bay board, same clusters same drives= sustained writes of 87mb second.

On the Nvidia IDE port one old Seagate 40 gig drive with only 512kb cache and a 5400 spindle gives sustained writes of 27 mb sec, AND THATS A SINGLE DRIVE!!!

PC Mark 2002 1120 v 2250 n the INTEL.

So while it has stopped corruption you are better off with a single drive, and video capture is going to suffer.

Drive read rates are still high at over 97mb per sec which is why Sandra doesnt suffer too badly in File Benchmark, I get 52000 in Sandra which initially lead me to believe all was not too bad.

But run either Atto or PTest 4.0 and you will see what I mean..

Unregistered · Apr 29, 2003

Work the DesternDigital Raptor on RAID 0 ?

Unregistered · Apr 29, 2003

WesternDigital

Per Hansson · Apr 29, 2003

I am not sure if I understand your question...

Yes, the Serial ATA harddrive "Raptor" from Western Digital works with the board however it sufferrs from the same data corruption issues as all other harddrives do when connected to the Silicon Image serial ATA connectors....

Per Hansson · Apr 29, 2003

This is now posted on the Silicon Image website:
------------------------------------------------------------------
SiI3112A: nForce RAID 0 Data Corruption
April 24, 2003

At this time we are aware of the issue and are working with nVidia to develop a solution to this data corruption problem. Some of the motherboard suppliers may have posted an SiI3112A driver on their websites, but it needs to be clarified that those drivers were intended to be an interim solution that just prevents the data corruption from occurring. There is a significant trade-off in performance when using these drivers so it is not likely to be the final solution.

As soon as nVidia and Silicon Image have an acceptable solution to the issue, we will post it on our websites as well as provide it to our motherboard customers for posting.

Thank you for your understanding.
------------------------------------------------------------------

Arris · May 2, 2003

Silicon Image reports data corruption fix available!!!

===========================================
For those of you who have email us with questions regarding this issue, this isjust to let you know that we have identified the issue with nVidia regarding theissue with RAID 0 data corruption with the nForce-based motherboards. This wasidentified as a problem with the PCI bridge in the corelogic that affected the way data was transferred to our SATA controller chip.

A system BIOS fix has been posted by nVidia and is available from our support site below, from nVidia, or your respective motherboard manufacturers. The fix is in system BIOS only and does not require any revisions to existing SiI3112A drivers.

Those of you using the beta "patch" driver posted several days ago should revert to your old drivers or download the latest SiI3112A drivers from our Knowledgebase. The patch driver was only meant as a stopgap measure to stop the data corruption from occurring, but had a significant performance trade off.

http://12.24.47.40/?cid=2&c=12&cpc=...pP4SveI6dt2WJi7

Best Regards,

Silicon Image Technical Support

Data Corruption Problem Solved

Per Hansson

Posts: 1,978 +234

Tarkus

Posts: 615 +2

Per Hansson

Posts: 1,978 +234

Unregistered

Unregistered

Per Hansson

Posts: 1,978 +234

Per Hansson

Posts: 1,978 +234

Arris

Posts: 4,719 +451

Similar threads

Latest posts