|
Top-level
 so there is still data corruption occurring on the Thinkpad. only this time, write operations are affected. it's more or less random but much more frequent than the read issue i fixed earlier. an old data word is getting counted twice. logic analyzer time. 😑 10 comments
after connecting the logic analyzer and bringing some internal signals out to pins, it suddenly decides to start working 😅  the data ready handshaking flag is getting cleared in the wrong spot. test1 is the flag, and the green M1 cursor shows that it is getting cleared even though DMA hasn't started i was using a version of the "dma running" signal that was latched on the falling edge of CMD, so it was possible to trigger this clear signal even when DMA wasn't actually in progress. after that change, it seems to be passing both the write and the read test loops. i'll run them a bit longer just to make sure. i will say that it's the perfect time of year to be running this huge logic analyzer. it puts out a lot of heat! @tubetime is it Geiger counter on the RAM time? (Apologies if that was the first thing you tried) |
Gregory's Server
this has turned into a very challenging project!