• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

K7D Master-L high I/O load problems

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

emboss

Member
Joined
Nov 10, 2003
Location
Canberra, Down Under
The one time when I try to illustrate the benefits of a SMP machine, the K7D lets me down :|

My setup has a PIO mode-3 CD writer (HP8100 internal, sitting alone on IDE1), and a WD800JB sitting alone on IDE0. Two L5'd XP2500's, 512MB of RAM, a 9800Pro, and a red K7D Master-L complete the guts of the package. The system is several-days prime95 stable, so it's not a general stability problem. The OS is Windows XP Pro, SP1 with the drivers being the latest (with the exception of using 4.4 cats as 4.6's seem to have a few problems on my machine). The problem occurs when the I/O of the system is heavily loaded by ripping tracks from the CD drive.

If I rip tracks from the drive using CDEX and playback a previously encoded MP3 using any player (using the onboard sound), it will spontaniously stop playing the MP3 anywhere from 1 second to 20 seconds into the song. For example, Winamp will not completely freeze up (can still operate the controls) but no more sound will be produced until the program is restarted. My guess is that bus contention issues are causing something to be dropped, and the audio device is getting confused, since everything else continues to work correctly.

Has anyone else experienced anything like this?
 
What soundcard do you use?
What version of drivers?

Have you set both IDE channels to "DMA if possible" in Device Manager?

Do you have the same problem, if using a different program, such as WM9?

What IRQ's are being shared in your system?

Did you install the AMD EIDE drivers?
 
Soundcard: onboard sound (as mentioned :) )

Drivers:
AC'97 Driver for AMD-768 Controller (version 1.0.0.0 supposedly, dated 27/09/2001)
AMD-768 Bus Master IDE Controller V1.43 (version 1.43)

Both IDE ports are set to DMA if available. HDD (primary master) is using UDMA mode 5 according to WinXP, CDROM (secondary master) is not using UDMA (not surprising, since it only supports PIO mode 3).

All sound-producing applications do this. Winamp, WMP9, games, etc. Games usually completely lock up.

Only shared IRQ is IRQ17: AC'97, DFE530TX (main net card), onboard Intel LAN (unused), NEC USB2.0 card, Radeon 9800.

Note that everything works fine if I don't use a PIO CDROM drive. I dropped a new (borrowed) 52x writer in that runs at UDMA mode 4 IIRC, did the same tests, and everything works fine. So it's almost certainly to do with how the CDROM drive is being accessed (PIO stresses port I/O extremely heavily).
 
emboss said:
Soundcard: onboard sound (as mentioned :) ).

Doh! Don't know how I missed that...

PIO drives have to use the CPU to transfer data to memory. DMA drives, on the other hand, can transfer to memory, directly.

Your issue is definitely because of the PIO, but really it still shouldn't happen. Your CPU ~should~ be able to do both, simultaneously.

I would recommend you download and install the Realtek drivers for that soundcard, instead of the crummy AMD AC97 drivers. The Realtek drivers are MUCH better. You can find links to them in various threads about that board, here in the SMP forum. They are especially useful for Windows Server 2003, because the AC97 drivers don't work under that OS. The first thread they were mentioned in was a thread about sound drivers for the K7D for Windows Server 2003. (let your Search fingers do the walkin...)
 
I agree with CMC on the drivers, but would recommend temporarily changing the CDROM out for a model that uses UDMA for a short test. I have seen many issues with PIO mode devices on later OS and applications.
 
I know all about PIO/bus-mastering stuff, having written a small OS :) Hence my wondering if anyone else had experienced it with other port-I/O-heavy stuff (though most I/O heavy stuff uses bus mastering nowadays), especially as the problems went away with a UDMA drive.

I'll give the realtek drivers a go and see if that helps things (though I agree, there should not be issues in the first place, but who knows ...).
 
Back