Testing / comparing : Intel D975XBX2 / Asus P5B DX ***56K WARNING***

[jschoef1]

A little of subject but it could be good for the guide. I have built 4 computers using identical parts using the BadAxe2 with Raid 1 on the Intel SATA controller and I have some interesting results.

First is an incompatibility with Intel's Raid 1 and Vista ( I didn't try XP). When the computer is shut down improperly (reset button, power outage, etc) the computer will not reboot 99% of the time. It will just hang on the vista scroll bar for a long time then you might get a black screen with a mouse arrow. I called Intel about this and they said that it is normal behavior for Raid and that it will boot but could take up to 36 hours.....real good solution. I then asked if it would work on the Marvell side, they said it would do the same thing. I obviously didn't believe them. What happens is the Matrix Storage Manager detects a dirty shutdown and goes into verify mode but it doesn't work properly. The only work around I found to get your computer working that day...is to shut it off, unplug a drive, boot it up, it then starts right up. After its up and running I attach the unhooked hard drive and it goes into rebuild mode. It works fine in rebuild mode.

I have now switched 3 out of the 4 computers to the Marvell controller and it works ALMOST perfectly. the only problem is sometimes on reboot it gives me an Apache error which you have to cancel out of 2 times then everything is fine (I prefer to keep the Apache monitor running to check on things).

Intel has leaked out there new Matrix Storage Manager #7 on the web. I installed this on 2 of the computers and the bug still exists...it also exists on a customers Dell. Interestingly enough, with #7 on the BadAxe and last months Bios update, it seems to be fixed, but only with the latest Bios. It will boot up slow but still will boot in about 5 minutes after a dirty shutdown which is much more acceptable then 36 hours. It will then verify for hours and be back up to speed (but still very usable in verify mode, just not quite as fast).

I cannot tell a real difference in real world speed tests with regular programs but one computer running 2x 320gig raid 1 Seagate drives on the Intel side gives me a burst speed of up to 210. The Marvell Burst on the other computer is 155. Average speed is within 1 number of each other on both controllers with a reading of 67-68. Random access is 11 on Intel and 13 on Marvell.

I decided to test 2 brand new drives (they are a little slower then the 320 gig drives but are SATA 2) I keep in stock on the same board but different controllers. This is in Raid 1 setup. I used 2 unformatted 160gig Seagate SATA Drives for the test. First one with the faster burst speed in the Intel controller. Second one is the Marvell.