annala@neuro.usc.edu (A J Annala) (02/14/90)
There is a serious design defect in the IBM PC/AT that makes it very difficult to use for real time data acquisition ... the fixed disk controller was implemented using programmed i/o instead of DMA for transfering data between memory and the hard disk drive. The lack of hardware DMA support for the hard disk means that if some other DMA channel goes active (e.g. the 176.2 KHz parallel i/o channel on my 8 channel digital audio tape deck) the 80286 processor is slowed and the disk write request aborts. I would really appreciate any advice you can offer about how to get an IBM PC/AT hard disk controller that either uses DMA or has enough internal memory buffering to be immune to the presence of another active DMA channel.
phil@pepsi.amd.com (Phil Ngai) (02/15/90)
In article <22875@usc.edu> annala@neuro.usc.edu (A J Annala) writes: |I would really appreciate any advice you can offer about how to get |an IBM PC/AT hard disk controller that either uses DMA or has enough |internal memory buffering to be immune to the presence of another |active DMA channel. The WD1006 series have a track buffer, would that do? -- Phil Ngai, phil@amd.com {uunet,decwrl,ucbvax}!amdcad!phil When guns are outlawed, only governments will have guns.
pipkins@qmsseq.imagen.com (Jeff Pipkins) (02/15/90)
Before you pursue this much further, it may be a good idea to check out the January 1990 issue of Dr. Dobb's Journal. There is an article in there on real-time data aquisition, complete with source code for programming the DMA channel.