pouët.net

__sam__ information 187 glöps

• general:
• level: user
• personal:
• first name: Samuel
• last name: Devulder

• 128b Thomson PIrat by PULS [web]
• yes, and maybe the trailing 4999... is just a bug. Who knows? :D
• isokadded on the 2015-11-16 01:40:15
• demo Windows Synopsis by Onslaught [web]
• Great! I always enjoy space.
• rulezadded on the 2015-11-15 22:19:09
• 128b Thomson PIrat by PULS [web]
• Vidéo in 10x of the fast version ;)
  
  128bytes version is too slow for me to record, even in 10x (takes more than 10 hours).
• isokadded on the 2015-11-15 17:03:59
• 128b ZX Spectrum z80pi (spigot) by Serzhsoft
• Quote:

  yea, 128b category reached, congrats! :)

  
  I've made a new pouet entry for this: http://www.pouet.net/prod_nfo.php?which=66603 :D
  
  Beware, the 128bytes version is EXTREMELY slow... this is possibly the slowest implemented version ever of the SPIGOT algorithm :P
• isokadded on the 2015-11-15 16:43:29
• 4k Windows koma::[sutra] by Rift
• Cool!
  
  How much horsepower is needed for this effect? I wish it could on the brave old amiga.
• rulezadded on the 2015-11-15 11:55:04
• 128b ZX Spectrum z80pi (spigot) by Serzhsoft
• Ok. I've found you add "ADD HL,DE" twice in the code right after calling L_A073. This is probably a size-optimization becasue factorizing these into the end of L_A073 will prevent you to use this compact call-twice structure.
• isokadded on the 2015-11-13 17:19:24
• 128b ZX Spectrum z80pi (spigot) by Serzhsoft
• digits ok and [128 bytes version :)
  
  I've noticed that your division doesn't do a final "ADC HL,DE" to compensate for the fact that when the carry is raied, one have subtracted one time too much. Is that compensated elsewhere in the code ?
• isokadded on the 2015-11-13 17:16:22
• 128b ZX Spectrum z80pi (spigot) by Serzhsoft
• [EDIT] 141 bytes, but I must check if all the produce digits are ok yet.
• isokadded on the 2015-11-13 09:06:45
• 128b ZX Spectrum z80pi (spigot) by Serzhsoft
• Yeah!
  
  Looking at your code, it seems you and I devised the very same 17bits division algorithm :) 6809 opcodes generally took 2 bytes, so the total size is 143bytes.. still above 128 :-( In addition the naive-multiplication is really much much slower. The program takes ages to finish (it used to take half an hour with a longer but faster division using the "mul" opcode).
• isokadded on the 2015-11-13 08:53:36
• 128b ZX Spectrum z80pi (spigot) by Serzhsoft
• Answering to myself: naive multiplciation is much slower. But, after benchmarking, it appears that on the 6809 this 17bit naive-division is faster than the standard one using shift/substract when running the spigot algorithm. The cause is in the size of the involved number.
  
  This is likely to be the same for the Z80.
• isokadded on the 2015-11-13 02:05:55