Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.os.linux.hardware > #1273

Re: HOW2 copy IDE to SATA?

Cross-posted to 2 groups.

Show all headers | View raw

larry moe 'n curly wrote:
> On Jun 11, 4:33 am, no.top.p...@gmail.com wrote:
>> I read:
>>
>>> You  can  buy  integrated motherboards that allow a computer to use
>>> both  SATA and IDE hard drives. However, only one of the two may be
>>> used  at  any  given  time  and both cannot be used simultaneously.
>> So if they can't run on the same PC at the same time, how do you
>> copy your old IDEs to a SATA?
> 
> I've never had a problem doing that using the built-in SATA IDE and
> PATA IDE ports of a motherboard together, but sometimes a plug-in PCI
> card SATA or PATA controller didn't get along with the motherboard's
> built-in controller, and this included the two most popular types of
> PCI controllers, based on the VIA VT6421A or Silicon Image SiL3x12
> chip.  However I was told my problem may have been due to Windows XP
> not handling interrupts well.
> 
> The controller cards selling for $2-10 on Ebay are the same as those
> costing $20-30 from other dealers.

In the "good ole days", you'd fire up Device Manager, and view the
IRQ assignments. And, at the same time, review the "sharing" table
from the motherboard manual.

http://img835.imageshack.us/img835/2391/irqcity.gif

The example there, is my current motherboard P5E Deluxe, with
my usual junk in it.

Looking at the table on the right, a card stuffed into "PCI Slot #2",
would only share with PCI Express x1 slot #2. As long as I didn't place
a card in the PCI Express slot, I could avoid physically sharing the
interrupt line by using PCI Slot #2. So if I bought a SIL3112,
and suspected a problem caused by sharing, that's the PCI slot I'd plug
it into.

If the APIC is enabled, IRQs go above 16. And the slots tend to be
assigned above 16. If cards physically share, you cannot "move" them.
Attempts to move them, would cause physically sharing ones, to move together.

Now, what's difficult to work out in my particular table (or for
a lot of modern mixed PCIe/PCI motherboards), is whether physical
wires are involved. I thought PCI Express interrupts were done with
inband signalling, and I'm kinda surprised the table shows their
status the way it does. So I must be missing something there. On the
one hand, PCI Express is "made to look" like PCI, from a config
space point of view. But for that to happen, there would have to be
a mechanism to map from an inband signalled interrupt, to an IRQ,
and it could be that they don't really physically share, and the IRQs
could be reordered somehow.

Anyway, the value of that kind of table (the right-hand one, from the
motherboard manual), is to show it's easier to give unique interrupt
characteristics by using PCI Slot #2, than by using PCI Slot #1. So
if I was having trouble, I'd put that new card in the bottom slot.
I've never seen a bit of trouble with interrupts on this machine, and
I've never even looked at those tables since owning this thing. On past
older systems (pure PCI based, AGP video), I'd probably be spending the
odd day, trying to improve scratchy audio, and moving my PCI sound card
around would be par for the course.

There are even more useless things you can play with, such as the
interrupt reservation table in the BIOS, but I never got much of
anything out of that exercise. And can't recommend it to someone
else. It never made enough difference, to waste the time on. It
wasn't an "aha!" moment, like all the audio glitches would go away.
It was down in the "placebo effect" level of improvement :-)

There are a couple issues with SIL3112. There is the bug when
handling 1TB or larger drives. The INT 0x13 BIOS code on the
card has a bug in it, which can be fixed if you run into the
bug, by flash upgrading. A more serious issue from long ago,
was running one of those in RAID mirror mode, and finding out
when one of your disks dies, that the other one isn't a
mirror image :-( So if I was going to run one of those,
I'd want a modern non-RAID BIOS file loaded into it. And
a PMC flash chip on the card, so the flasher would work.
(The flasher apparently only worked with certain brands
of EEPROM on the SIL3112 card. Good card manufacturers would
make sure they used a chip that end-users could re-flash.)

I don't know any good stories about the VIA... Just the fun
of hunting down the right driver.

    Paul

Back to comp.os.linux.hardware | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

HOW2 copy IDE to SATA? no.top.post@gmail.com - 2012-06-11 11:33 +0000
  Re: HOW2 copy IDE to SATA? Richard Kettlewell <rjk@greenend.org.uk> - 2012-06-11 12:45 +0100
  Re: HOW2 copy IDE to SATA? terryc <newsninespam-spam@woa.com.au> - 2012-06-11 21:54 +1000
  Re: HOW2 copy IDE to SATA? Paul <nospam@needed.com> - 2012-06-11 08:44 -0400
    Re (2): HOW2 copy IDE to SATA? no.top.post@gmail.com - 2012-06-14 06:45 +0000
  Re: HOW2 copy IDE to SATA? John McGaw <Nobody@Nowh.ere> - 2012-06-11 10:19 -0400
    Re: HOW2 copy IDE to SATA? Jon Danniken <jonSPAMdanniken@yaSMPAhoo.com> - 2012-06-11 07:41 -0700
      Re: HOW2 copy IDE to SATA? Grinder <grinder@no.spam.maam.com> - 2012-06-11 10:53 -0500
      Re: HOW2 copy IDE to SATA? J G Miller <miller@yoyo.ORG> - 2012-06-11 17:33 +0000
      Re: HOW2 copy IDE to SATA? Anssi Saari <as@sci.fi> - 2012-06-13 10:51 +0300
    Re: HOW2 copy IDE to SATA? Anssi Saari <as@sci.fi> - 2012-06-13 10:51 +0300
  Re: HOW2 copy IDE to SATA? "larry moe 'n curly" <larrymoencurly@my-deja.com> - 2012-06-12 00:35 -0700
    Re: HOW2 copy IDE to SATA? Paul <nospam@needed.com> - 2012-06-12 04:28 -0400
  Re: HOW2 copy IDE to SATA? Joe Pfeiffer <pfeiffer@cs.nmsu.edu> - 2012-06-12 10:16 -0600
    Re: HOW2 copy IDE to SATA? Winniethepooh@100acrewoods.net (GMAN) - 2012-06-13 03:56 +0000
    Re: HOW2 copy IDE to SATA? Gordon <Gordon@clear.net.nz> - 2012-07-06 05:34 +0000

csiph-web