What might make Windows switch a DVD drive to non-DMA mode?

[EddyH]

OK, this was something of an odd victory for my informed guesswork on what the problem was, but it's kinda annoying and baffling also.

New DVD writer in works PC. Was all fine and dandy for writing a couple of 4.2Gb-ish DVD+Rs at 8x (slow machine, didn't want to be too ambitious ) with ImgBurn, did oddly drop the write rate to 6x but didn't think much of it at the time as it all passed off OK in about 10 minutes for each. Was even able to multitask without much issue.

Come to try it out on DVD+RW and -RW .... all hell breaks loose. The +RW wrote at about 1.3x overall, and no more than about 1.5x - the -RW, about 1.0x (!) overall and a max of 1.4x ... also everything else on it slowed to an ABSOLUTE crawl. Have you seen those videos of crazy hackers convincing XP to run on a 20mb 486DX clocked down to 7mhz? Yeah - something like that. Couldn't realistically get anything else done on it for the 40 minutes to an HOUR each disc took to write, as each window needed about a minute to draw. The buffers kept failing and under-running also, though this seemed kind of random, arbitary and rapid rather than a continual and gradual fill-empty-fill-empty.

Thought at first it was a memory issue, as it's a little RAM starved, but even with all other programs shut and a not particularly high "Commit Charge" listed, it was dragging. Noticed the CPU was maxed when writing was in progress, and dropped for the few seconds the buffer took to refill (the only point at which any other program would run!) which gave me a clue...

Open up device manager, check old DVD and new DVD+/-/RAM drives, all ok... check the adaptor, everything fine on the first channel (HD only), but channel 2 confirmed the suspicion. Master device (old DVDROM) - "DMA if available" and DMA mode 4 listed; new Slave DVDRAM - "PIO modes only" ... arrgh! So the poor little Celery 2.8 was trying to shift all the data to the drive by itself, as well as get it off the hd (also the same partition the virtual memory is on, erk), shuttle it around memory, work with the other programs, etc - I'm surprised it made it up to 1800kb/s!
A very old problem that I never thought I'd see on a post-2000 machine!

(attached picture is a DVDInfo / ImgBurn readout showing the carnage and it's effect - luckily it at least forced the drive down to 2.4x even though it's a 4x "only" disc so it shouldnt be hurt quite so badly by burnproof gaps)



So - WHY in the name of all that's holy would this option be set that way? Everything else has defaulted to DMA, and if it's been done by the sysadmin... well, I have no way of determining why they'd do that. Unless it's some kind of tamper test that's gone awry
A virus? Just regular Microsoft idiocy? What???

Have set it back to DMA and will see if future RW burnings go ok (too late to try anything else tonight, particularly if it falls back to 1.0x and i can't get it to stop/shut down), or if it slips back and even plain +/-R's have difficulty.
Oddly leaving it on PIO didn't seem to affect an exploratory test done with Nero CD/DVDSpeed, though I expected it to start choking!

(was going to attach a screencap of that also, but it doesn't seem to have prtscr'd as i'd expected and i shut it already but it would have added fuel to the "DVD-RW longevity" debate, as it's clear to see the area where I previously burned a different, 2.5gb dataset on my own PC as a point where the PIE and to a lesser extent PIFs drop markedly. worrying.)

[redwudz]

The most common cause of reverting to PIO mode is read failures. If you have 6 consecutive read failures, the OS will revert the channel to PIO. It's a Windows 'feature' of questionable value. It does allow the drive to keep reading, but at a much slower rate. Unfortunately it won't reset automatically. PIO uses a lot of CPU power. You will see that in Task Manager when the drives tries to do a read.

Most of the time, you can just uninstall the channel the drive is on and reboot and it will restore it to DMA. But sometimes you need to do some registry changes to keep it from falling back to PIO by itself.

Here's a link to a fairly good explanation of it all and some fixes: http://winhlp.com/WxDMA.htm

[blinky88]

Check your IDE cable, you need an 80 conducter cable attached to the burner, I have found that a 40 conducter cable can cause the problem you are having. Also multitasking can create alot of problems, better off turning of as many applications running in the background off as well. I burn at 6x regardless of the speed of the media and a coaster is an extremely rare event. If your HDD is 100ATA that also should have an 80 conducter cable.

[EddyH]

Gah, sunuva...
Kind of glad I have a USB burner for my own PC, now... so it re-installs itself every time I power on

So all it needs is a slightly scratched disc and it all goes to hell? That would explain the way the problem went - burnt a couple of DVD+Rs successfully at so-called 8x speed (actually 6x!?!), or so we thought, something went wrong with ImgBurn on the second one though and it wouldn't read in any computer... probably generating way more than 6 read errors in the process.

Don't know what cable is in the computer as I didn't see the new drive being put in, but I may have chance when the guy in charge whips out the old DVDROM sometime next week (it's completely shot, only reads pressed CDROMs, under duress, and nothing else now). However, in the course of my investigation, it did report both the hard disc and the existing DVD at Mode 4 (which the new one switched to after fixing it), and the heavily abused HD doesn't seem to suffer any ill effects from the interface running 2x the recommended speed for a 40-conductor cable, so i'd suspect it's an 80.

Sure know PIO uses a lot of CPU... i was there for DMA becoming common and the vast improvement of speed (and also witness to the now unimaginably glacial HD pace of our older pc's in the cupboard/my old laptop that can't even access an IDE-repurposed 80x 2gb flashdrive at any decent pace) and I've got an A+ Core exam coming up tomorrow The effect it had on the PC was literally crippling.

Not sure what you mean about "uninstalling" it, but I'll definitely try out the registry-locking / fixes under that link. Many thanks!


.... stupid microsoft. Why would it improve read performance when the vast, vast majority of read errors would be from rubbish media (which a slow interface wouldn't fix), as EIDE/ATAPI drives are expected to be able to negotiate their own transmission speed?

[redwudz]

'Uninstall' is an menu option when you right click on the IDE channel. It resets the controller by reinstalling it after a reboot and usually fixes a DMA>PIO problem. But sometimes you need to do a registry change on the 'filters'. This is more common when you change to a new drive.

Windows doesn't always report the DMA/PIO status correctly. If in doubt, uninstall the channel.

I wish MS would just set up the OS do the 'uninstall' automatically after a reboot when PIO mode is detected.

[EddyH]

ahh, got ya. would it stop it falling back, or would you have to do that every time? (if so i'd as soon manually set it back from PIO to DMA as it's be more certain and take about as many mouse clicks)

will it have dropped to a lower DMA mode first (udma-3 / 33mb, mode 2 / 16mb, mode 1 / 11mbps...), or just gone straight to PIO, as PIO mode 3 & 4 are the same speeds (allegedly..) as dma-1 and 2, and therefore might be equal in MS's eyes?

also, interestingly, from that page

some Dell computers have DMA disabled in their BIOS by default for the second hard disk

maybe Acer run to the same inexplicable philosophy as Dell. No internal drive made in the last 5, possibly even 10 years should be DMA-incompatible... I never checked what Primary-Slave was set to, but Primary and Secondary Master were DMA, and Sec Slave was PIO. Hmm.

[redwudz]

Well, no drives should be PIO. I don't know why they would set one that way unless it was a unused channel. The OS is supposed to step down in speeds from DMA 4 > 3 > 2, but I've never noticed that. Mine always seem to go directly to PIO mode.

If they seem to end up in PIO mode fairly often, I think that article discusses a registry change that might help. I've only seen them revert to PIO after I have a particularly hard time reading a DVD disc. And usually the disc turns out to be corrupted or unplayable. You can usually hear it clattering away as it's 'seeking' the track. That can't be good for the player, either, so it may be a good think that it reverts to PIO. But it still seems to make the 'seeking' noise even in that mode.

I'm wondering what a SATA DVD drive does when it has a read problem.

[guns1inger]

I had a dodgy IBM HDD (is there any other kind of IBM HDD ?) on the same channel as a burner, and started having PIO fallback issues. Replaced the drive and the problems went away.

[bendixG15]

Here's Microsoft's side of the story .. For WinXP

http://www.microsoft.com/whdc/device/storage/IDE-DMA.mspx