home *** CD-ROM | disk | FTP | other *** search
- Path: ix.netcom.com!JAMESCHA
- From: jamesch@ix.netcom.com (James Chamblin)
- Newsgroups: comp.sys.cbm
- Subject: Re: Serial SPEED
- Date: 12 Jan 1996 12:48:25 GMT
- Organization: Netcom
- Message-ID: <4d5lap$snb@ixnews8.ix.netcom.com>
- References: <821430808@p71.f411.n201.z2.ftn>
- NNTP-Posting-Host: ix-orl1-16.ix.netcom.com
- X-NETCOM-Date: Fri Jan 12 4:48:25 AM PST 1996
- X-Newsreader: News Xpress Version 1.0 Beta #4
-
- In article <821430808@p71.f411.n201.z2.ftn>,
- Pontus Berg <Pontus.Berg@p71.anet.canit.se> wrote:
- >Howdy!
- >
- >I just had a chat with Rowdy and we came up with a kewl idea that I'd
- like to
- >present on how to speed up normal serial transfer of files. (While
- writing I got
- >an even better one that I'll mail in the next mail after this one!)
- >
- >The idea is based on a REU being available;
- >
- >Recieve:
- >
- >* Set up the REU for transfer C64->REU with fixed address in the C64
- end
- >- Fixed address set to $DD01 (the serial bus isn't it?) * Poll the
- port for new
- >bit
- >* Do STA $FF00 to execute the storage in the REU
- >
- >This way you "sample" the serial bus and the overhead for each bit is
- kept to a
- >minimum. You store 8 times the data you need but you do it FAST!
- >
- >Of course you'd need to deciper what you have got in the REU but I
- imagine this
- >to be a faster process when you can do it in bigger hunks than on each
- byte and
- >you only have to do it after every (reu size/8) bytes.
- >
- >Aslo the disk I/O is faster if you process it in bigger hunks!
- >
- >The other way around is also possible; you set up the bytes and
- prepare the
- >bytes, place them in the REU and there you go banging them on the
- port.
- >
- >Oh, well. Perhaps this planeted ideas in someones head... Off for the
- better
- >one!
- >
- >
- >/Pontus Berg, Bacchus@FairLight.COM
- >
- >.... MicroSoft, as good as AIDS!
- > (Pontus)
-
- Actually, I've seen this idea used for sampling data, except it used a
- chip that put out the digital signal in parallel and the data was read
- off the user port. You get 1 sample every machine cycle. playing the
- sample would be reversed using the SID volume as the destination
- register.
-
- You know, this idea could have alot of potential, like mabey a digital
- scope or frequency counter.
-