Random model 600 discoveries while trying to test the new ram boards more rigorously.
Idea was to: Populate both slots, which makes a little over 220k available. Generate about 256k of random binary data on a modern machine. I just copied 256k of random stuff and used gpg on it a few times randomize it thoroughly. Xmodem the blob to the 600 until it aborts. Xmodem the blob back to the pc. Use dd to copy out a truncated excerpt from the original blob of exactly the same size as the blob that came back from the 600. Binary compare the two truncated blobs. I encountered 2 things: You can't xmodem more than 64k in one file. So I split the 256k blob into 4 64k blobs and discovered the next thing. You can't *actually* xmodem more than 64k *minus one byte*, in a single file. If you try to send a file over 64k and allow telcom to truncate it at 64k, or if you try to send a file that is exactly 65536 bytes, you will transfer 65536 bytes both ways, and neither side of the connection will issue any warnings or errors. Both TELCOM and the xmodem util on the modern machine will say everything went fine. But if you transfer a 65536 byte file from a pc, to the 600, and back, the final byte will be changed from whatever it was to a 0x20 space. All other bytes are preserved. If you transfer a file 65535 bytes or less, the file makes it both ways 100% accurate. I don't yet know when the change takes place. Is it when telcom receives the file from the pc, or when telcom sends the file to the pc? Both? There are a few ways to test the individual parts, just it's soooo sloooow. It takes about 10 minutes to transfer 64k. I think this mostly proves out the ram boards, as well as can be without an actual ram test program. To really be sure, I should swap the 2 boards between slots 1 & 2, and repeat the whole process. Ugh. -- bkw
