[00:05] cutting one instruction off will slow the launch down by 4 cycles, and increase the vulnerable footprint [00:05] (not field-tested, but i have a hunch) [00:05] i still don't quite see how to do it though [00:05] i assume you'd get rid of a dat [00:06] and rearrange the instructions [00:06] replace dat's with jmps [00:06] spl *1, #i12 [00:06] then again i'm only coming at this in between runs at work [00:06] djn.f @0, #-4 [00:06] should be spl @1 [00:07] oh, of course [00:07] here i was thinking along those lines and i got fuddled by indirection again :P [00:07] oh wait, nevermind.. i misunderstood [00:07] interesting [00:08] i like it [00:19] 80,69194497tbl jmp j12, j11 [00:19] 80,69194497 jmp j10, j9 [00:19] 80,69194497 jmp j8, j7 [00:19] 80,69194497 jmp j6, j5 [00:19] 80,69194497l spl 1, j4 [00:20] 80,69194497 spl 1, j3 [00:20] 80,69194497 spl 2, j2 [00:20] 80,69194497 spl @1, j1 [00:20] 80,69194497 djn.f @-1, #l-1 [00:20] it stutters 6 times instead of 3 [00:20] well, 3 times for 2 instructions [00:21] it should be djn.f @0, #l-1 [00:21] no [00:22] if you do that, the first djn will jump to the wrong place [00:22] you want it to jump to j1 [00:22] er wait [00:22] wtf [00:22] why is it working the way i wrote it? [00:22] i did 0 first and had a bug [00:22] i guess djn jumps to the value before it got dec'd? [00:23] gotta fieldtest, writing down [00:24] i did field test [00:25] http://www.nomorepasting.com/getpaste.php?pasteid=24271 [00:25] it's not a pre-decrement [00:25] i don't know the details of how mars works technically [00:25] but doesn't it copy the instruction into a temporary place, and then modify the core but not the copied instruction? [00:25] something like that anyway, i'm pretty sure it's jumping to -1 the first time, not -2 [00:27] another in-register/in-memory case, ugh [00:27] hehe [00:27] anyway, it seems to me that [00:27] one less vulnerable instruction in the launcher [00:27] isn't worth three more vulnerable instructions in core [00:28] or rather, three twice-as-vulnerable instructions [00:28] i wouldnt exchange one instruction into hassle like this [00:32] yeah, plus you can use up a spare field pretty easy [04:48] Join: X-Scale2 joined #corewars [04:51] MSG: Ping timeout: 240 seconds [05:47] MSG: Quit: mov.i #1,1 [06:02] MSG: Ping timeout: 240 seconds [10:47] Join: Whoa_dude joined #corewars [10:47] Hey! [11:37] MSG: Quit: CGI:IRC [15:38] Join: Metcalf joined #corewars [15:38] Hi :-) [15:42] If you've got a pointer into core, how many instructions does it take to exchange the number in @pointer and the cell after it? [16:05] define after it [16:11] If pointer contains 40, then it would need to exchange numbers at pointer+40 and pointer+41 and afterwards leave pointer containing 40 [16:18] so the question is not how many instructions it takes, but if it can be done with one [16:18] dejavu [16:20] It can't be done with on instruction [16:20] E.g. 5: [16:20] mov >ptr, temp [16:20] move @ptr, add #1,ptr [16:20] mov temp, @ptr [16:21] sub #1,ptr [16:22] me goes back to dinner [16:22] had really annoying 2 hours, i have two computers that i need to fix and neither co-operate [16:27] I bet I've had a worse day! [16:27] probably [16:38] Join: Fizmo joined #corewars [16:38] hi hi [16:38] cool [16:39] my first new warrior after months make a pretty nice 8 pts lead on the hill :-P [16:39] Hi Fizmo [16:39] Oh, going to look [16:39] never expected [16:39] 94nop? [16:39] lp [16:40] Without p-space? [16:40] yes [16:40] Very nice [16:41] I did two new lp-warriors [16:41] second is in the queue [16:44] Your lead is better now your second warrior entered the hill :-) [16:45] I think I'll update the LP HoF [16:46] cool [17:42] Mind has gone blank :-( [17:43] How do I prove mul #CORESIZE-1, #CORESIZE-xxx always gives xxx in the b-field, regardless of CORESIZE? [17:46] Can anyone recommend a good netbook? [17:51] Or even recommend one to avoid? [18:17] netbook? [18:20] Yes, I'd like one [18:21] Preferably with Windows, a 10" screen and 12 hours battery life! [19:41] MSG: Quit: ChatZilla 0.9.84 [Firefox 3.0.6/2009011913] [20:15] Join: fiveop joined #corewars [20:16] Hi fiveop [21:13] MSG: Quit: mov.i #1,1 [21:25] hrm. [22:41] MSG: Quit: humhum [23:14] Join: Mannerisk joined #corewars [23:16] hrm. [23:17] bleh. [23:17] screw windows met! [23:17] get a netbook with linux.. come to the dark side.. you can run my code then. [23:20] "Ill never run your code!" [23:22] lol [23:30] get one of the solid state netbooks tho met.. they're sweet.. no chance of blowing your disk up either.