Update... looking at event logs and listening, it seems to be an issue when the
drive needs to spin up. Drives in there now use much less power than the orig
one. Maybe not a power issue?

Leaving them spinning seems to not exhibit the problem. Is this a general SATA
behavior? XP fault? Certain hw designs badly done?

Vol ID doesn't seem to be an issue. Speed negotiation possibly (if they negotiate)?

BTW, is there a tool to manually send a spin down cmd to the disk for XP?
Searched but no luck. It might help diagnosing if I can tie spin up/down
definitively to the event, rather than waiting for the kernel to decide.

-Skip

On 3/30/2010 11:31 AM, Dr Skip wrote:
> I've got what may be a hardware or architecture issue that I'm at a loss
> for even where to start looking.
> 
> I'll start with some questions. Hopefully the answers will show me the
> error of my ways.
> 
> 1) Will SATA drives and interfaces work with whatever speeds are
> available and negotiate? In other words, if a drive (or 2) are different
> speeds than the speed of the controller (capability) will it still work
> well (via negotiation), or do they have to match.
> 
> 2) Will a SATA controller support more than 2 channels, or is it
> possible to design it such. That isn't possible with PATA master/slave.
> 
> 3) With XP, if one used Linux (clonezilla for instance) to clone the
> first drive (C), then installed it as a second (D), both vol IDs would
> be the same. This works (in other words, it boots and reads/writes) but
> will the system exhibit transient problems, conflicts, etc?
> 
> In a nutshell, the system in question has 2 SATA drives of a faster
> variety than orig. It appears to have 4 channels on the primary
> controller (CD is on it too), and all works well until the second cloned
> drive is on for a while. Eventually some access causes it to drop one or
> both drives out of DMA mode to PIO mode. This indicates windows got an
> error reading. I've just tested #3 by changing the drive ID, but no joy.
> Perhaps someone has greater insights on this?
> 
> It could be a limitation of the architecture of the particular hardware
> design perhaps, a speed issue, a vol ID issue, or something else. It
> might also be a spinup issue. Both new drives are from the same mfg as
> the orig, but are 3Gb/s rather than the orig 1.5Gb/s.
> 
> BTW, DMA mode gives about 100+MB/s performance, PIO <8MB/s. It isn't an
> option...
> 
> Does this sound familiar to anyone? ANY pointers as I grope around
> without bus analyzers or driver source code?
> 
> Thanks in advance.
> 
> -Skip
> 
> 
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist