[TI-ASM] The CRC: Killer or just overkill?

DigiTan · Post by **DigiTan** » Fri 27 Oct, 2006 4:34 am

I was considering the noise caused by (un)plugging the calc-to-calc cable and turned my attention to Milos Bazelides 16-bit CRC stuff. I want to be able to verify calc datastreams up to 1500 bytes (maybe in installments of 250) using either an oldskool checksum or his CRC, but only if the price is right. Speed will be a big factor and I don't even know if noise is a real issue on these cables. I was hoping some of the CLAP guys or somebody could shed some light on how bad link corruption can get.

Does CRC sound reasonable?

KermMartian · Post by **KermMartian** » Fri 27 Oct, 2006 6:14 am

In my opinion, it's complete overkill. All you need is a decent 8-bit checksum; the amount of noise on a calculator cable, if it's of a decent length and not constantly being hotplugged, is negligable.

Post by **tr1p1ea** » Fri 27 Oct, 2006 8:33 pm

I think that CRC might be a bit of an overkill. I guess the only way to determine if its needed is to do some testing. Its possible that there isnt much noise or that there is a very low failure rate as far as calc-to-calc cables are concerned.

It could however just be one of those 'for the sake of it' kind of things

.

Timendus · Post by **Timendus** » Sat 28 Oct, 2006 11:25 am

In my experience, there is pretty much no noise at all on the calculator port. So I'd agree it's overkill.

KermMartian · Post by **KermMartian** » Sun 29 Oct, 2006 12:22 am

Timendus wrote:In my experience, there is pretty much no noise at all on the calculator port. So I'd agree it's overkill.

O.o We agreed on something. pwnsome.

Timendus · Post by **Timendus** » Sun 29 Oct, 2006 1:50 pm

Surprised me too, you used to be the one shouting that more checksums is t3h 0wnz0rs, versus me shouting it kills performance and adds redundant code...

Anyway, Digitan, what are you coding? You've made me curious

DigiTan · Post by **DigiTan** » Mon 30 Oct, 2006 11:30 pm

Right now it's preliminary link stuff for the next Robot War installment. Hopefully with true plug-n-play so users won't have to halt things to startup multiplay anymore.

KermMartian · Post by **KermMartian** » Tue 31 Oct, 2006 4:42 pm

/me points to CLAP or Cn2...

DigiTan · Post by **DigiTan** » Tue 31 Oct, 2006 9:35 pm

Actually CLAP has enough similarities that it's probably already compatible with RW1. Or very close. I might go with CLAP to keep from having to reinvent the wheel.

Timendus · Post by **Timendus** » Tue 31 Oct, 2006 9:59 pm

Do you plan on having two calculators communicating or more? If the answer is two, you might want to wait just a bit for the new library I'm working on that's specifically intended for use in fast games. If the answer is more, be warned that it is all highly experimental

DigiTan · Post by **DigiTan** » Wed 01 Nov, 2006 5:10 am

Party of 2 in this version. There's a some code to tell when the link isn't idling anymore. After that, it's in need of code to detect when another RW2 is on the link, codes to transfer 50-1500 bytes, and codes to detect a disconnect.

Timendus · Post by **Timendus** » Wed 01 Nov, 2006 7:15 pm

It's all pretty much there, except for the detection of a disconnect. It can detect a problem with data exchange though, so you could use that. Would require a few little hacks though

Anyway, I'd better just get the damned thing working again so I can release it and you can play with it a bit.