Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (
More info?)
In article <7nqd5150nt09pr5ll6itthbioksu2p6ock@4ax.com>, fammacd=!
SPAM^nothanks@tellurian.com says...
> On Fri, 8 Apr 2005 12:56:17 +0100, GSV Three Minds in a Can
> <GSV@quik.clara.co.uk> wrote:
>
> >Bitstring <q3kb51latd6t4nm1a6t1tcgqqr7p3mrruc@4ax.com>, from the
> >wonderful person Robert Myers <rmyers1400@comcast.net> said
> >>On Thu, 7 Apr 2005 21:44:52 +0100, GSV Three Minds in a Can
> >><GSV@quik.clara.co.uk> wrote:
> >>
> >>>Bitstring <b6ra5190b0qjftmqq3ik663kkm3l7qtk2p@4ax.com>, from the
> >>>wonderful person chrisv <chrisv@nospam.invalid> said
> >>><snip>
> >>>>>The OS itself is multithreaded, which means I can have Photoshop image
> >>>>>processing a huge .tif file and still be able to browse the web (subject
> >>>>>to my disk channels not having choked and died).
> >>>>
> >>>>Can't we already do that? I've been multitasking since 1985 on Amiga.
> >>>
> >>>No, we can't =reliably= do that yet (and I've been multitasking since
> >>>1970, on a Titan) with a PC. There are still too many ways for the
> >>>supposed 'background' task to pull the rest of the machine down to
> >>>unacceptable crawl (lots, but not all, of which has to do with CPU
> >>>loading).
> >>
> >>Oh, let me be stupid and ask. On the whole, Linux seems to do much
> >>better than Windows. Both seem to get dragged down by "background"
> >>tasks that involve lots of disk activity, and I think the problem is
> >>interrupts. A second CPU is going to help with that?
> >
> >The problem appears to be system code (which may indeed be running as a
> >result of interrupts) and which is run at such high priority that
> >nothing else gets a look in. The actual interrupt handling these days is
> >usually very little code - it just queues up something for a WinXP
> >'service' to handle ASAP. The service is what then chokes the system.
> >
> >(e.g.) I've got an intermittent problem which seems to be caused by my
> >USB GPS handset being mis-recognised as a graphics device (possibly
> >caused by Photoshop Elements service which auto-starts when it thinks it
> >sees some sort of graphics activity)
> >
> >Event Type: Error
> >Event Source: Service Control Manager
> >Event Category: None
> >Event ID: 7011
> >Date: 28/01/2005
> >Time: 18:53:49
> >User: N/A
> >Computer: QUIK5
> >Description:
> >Timeout (30000 milliseconds) waiting for a transaction response from the
> >stisvc service.
> >
> >After than happens (thankfully rarely) the system runs like a 15Khz
> >(yes, Khz) cpu, with 100% CPU dedicated to one of the 73 'system'
> >threads. Not even task manager can get a look in (well, it can, if you
> >wait 20 minutes! Usually I just hit reset).
> >
> >That sort of cr&p doesn't usually happen on a dualie.
>
> So we need dual CPUs to cover the sins of USB? Maybe we do.
Personally
> I think USB is just broken and will never play right - every USB device I
> have misbehaves in one way or another: an HP scanner which worked
> wonderfully on WinNT with a SCSI connection; a Logitech joystick which
> sometimes doesn't get seen from a cold start; a digital camera "monitor"
> which sits in the taskbar instead of just starting as a resident process
> (does this on both Win98SE & WinXP). I suspect that the camera monitor is
> confused about the Joystick and doesn't complete its startup right.
I thought USB was "working" these days, at east that's what everyone
tells me. Mine was working, after a fashion, but now it's busted. It
hangs after printing a half page (Epson Photo-R300), the USB stick is
now on a different drive (shows up one-and-a-half times), and forget
the digital camera. To print another half-page I have to install
everything USB and then re-install the drivers. What a PITA!
> Then there's all the devices with cautions about this & that USB interface
> - hubs which are said to be "incompatible" with the shiney new doo-hickey
> you just got. It's a mess.
Wow! I thought it was just me.
> Have you tried running Taskinfo, at a high priority, before it goes
> comatose? Might tell you something and I suspect it'll be the Interrupts
> Time Placeholder and DPC Time Placeholder which are hogging the CPU...
> though what you'll do about it who knows? At least you might be able to
> reboot safely - how convenient!
In may case the CPU is still responsive, just the USB goes comatose.
--
Keith