Search the Community
Showing results for tags 'DMA'.
Found 3 results
I've read that when ANTIC (pretty accurate name now, huh?) is disabled (memory adress 559 dec or $22f hex = 0 ), atari operates about 30% percent faster. But I didn't notice any differences with PLOT operations, i.e. after setting this memory area to 0 and then restoring value after image has been drawn (to remove that nasty "rendering" behavior), time spent plotting was roughly same, the only change was that screen was blank until "rendering" been finished. What I'm doing wrong? More importantly, is there any faster version of PLOT for ACTION! ?
This is a project idea we were chatting about on ##Atari today. The basic idea is that we make it possible for PBI devices to take over the system bus to perform DMA transfers at up to 1.7 megabytes per second. This is done by hooking into the way that ANTIC stops the CPU with a small riser board, and connecting /HALT and a new /DMA signal to two currently unused pins on the PBI. The device protocol would work like this: when a DMA transfer is desired, the device asserts /DMA. Every cycle that starts with /DMA asserted is a DMA cycle for the device *unless* /HALT was asserted before the cycle began, in which case the device has to relinquish to ANTIC. Devices can use the presence of /HALT at a high level during a read of the device's registers to detect whether DMA is present; if not, devices can fall back to programmed I/O. I whipped up a quick board layout for this; the board is also designed to generate a "clean" phi2 output as well as allowing the use of a W65C02 or W65C816 instead of 6502C. It does not support other 6502s or 65C02s which do not have the /BE input. Thoughts?
Looking for anyone who has replaced the DMA in their STE and can contribute the old one for investigated into this project. http://exxos.www.idnet.com/IMPULSE/atari/last/gigafile/index.htm http://exxos.www.idnet.com/IMPULSE/atari/last/DMAfix/index.htm