Hi, I can't confirm this problem on my system. The hardware is a 8 Port SE MIDI interface that connects to the parallel port. Neither when used with the MME library nor with the DirectMusic API do I see an increase in the use of non-paged kernel memory. I have written a little utility that might be useful to you. It tests MIDI hardware by sending all types of MIDI messages from an output and reading them back in from an input. If the driver for the Creative Labs cards has problems you might be able to detect it. You can download it here (it's free): http://earthvegaconnection.com/products/miditest/index.html Evert van der Poll -----Original Message----- From: wdmaudiodev-bounce@xxxxxxxxxxxxx [mailto:wdmaudiodev-bounce@xxxxxxxxxxxxx]On Behalf Of Ulf Jaenicke-Roessler Sent: Thursday, February 19, 2004 12:57 PM To: wdmaudiodev@xxxxxxxxxxxxx Subject: [wdmaudiodev] Soundcard midi driver Hi, I'm writing an MTC (midi timecode) object which works quite well with USB Midi interfaces. On soundcards however (actually a specific CreativeLabs card [CT4750, PCI128 or so] - could not test others so far), I found a serious problem which I cannot handle in my application. In generator mode the MTC works fine, but in slave mode a lot of non-pageable kernel memory is consumed until finally Windows cannot do anything which requires such memory (i.e. not network traffic a.s.o.). After closing my application the kernel memory is freed and everything is fine again. AFAICT the reason are the very frequent QuarterFrames that my application receives. In task manager I can see how the used kernel memory increases - about 4k per second. My application simply opens the MidiPort, adds a buffer to the interface (for SysEx messages) and starts listening via MidiInProc. Every short message is analysed and the function returns. There's nothing I ever could release for kernel memory, right? Can anybody confirm this to be a (general) bug? I'm using XP with either its native support for the soundcard or the latest driver from CreativeLabs. Are there cards which surely don't show such behaviour? And finally - is there anything I can do to prevent my "application" running out of nonpageable kernel memory? Thanks in advance Ulf ****************** WDMAUDIODEV addresses: Post message: mailto:wdmaudiodev@xxxxxxxxxxxxx Subscribe: mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=subscribe Unsubscribe: mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=unsubscribe Moderator: mailto:wdmaudiodev-moderators@xxxxxxxxxxxxx URL to WDMAUDIODEV page: http://www.wdmaudiodev.de/ ****************** WDMAUDIODEV addresses: Post message: mailto:wdmaudiodev@xxxxxxxxxxxxx Subscribe: mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=subscribe Unsubscribe: mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=unsubscribe Moderator: mailto:wdmaudiodev-moderators@xxxxxxxxxxxxx URL to WDMAUDIODEV page: http://www.wdmaudiodev.de/