Post AssZA2855D5JX8R2Xo by [email protected] | |
More posts by [email protected] | |
Post #Ass7g9LD0YYIRz6LrM by [email protected] | |
0 likes, 0 repeats | |
Why is static single assignment form weird?Because, for the sake of mapping to … | |
Post #Ass7z5nqO31mylWnU8 by [email protected] | |
0 likes, 1 repeats | |
@lritter Finally, it is COMEFROM's redemption arc.https://en.wikipedia.org/… | |
Post #Ass7z5wLsPpjP9VakK by [email protected] | |
0 likes, 0 repeats | |
@GyrosGeier exactly. give comefrom a condition, and it's complete. (you nee… | |
Post #Ass9pQ9u38Er7kMUvg by [email protected] | |
0 likes, 0 repeats | |
This change aligns all edges of the graph to face backwards in time.Secondly, w… | |
Post #AssBlrkbilMLRtxkJs by [email protected] | |
0 likes, 0 repeats | |
Is being a wild forest of edges the only weirdness observed in single static as… | |
Post #AssDHe0ETGR21zzJrM by [email protected] | |
0 likes, 0 repeats | |
Our revised single static assignment form has a problem, though: it is possible… | |
Post #AssDhgISZ41efuvfLk by [email protected] | |
0 likes, 0 repeats | |
When we now, analogous to continuation-passing style or Thorin's control fl… | |
Post #AssDonQsQrnEpjG7CC by [email protected] | |
0 likes, 0 repeats | |
@lritter Sounds like Sea-of-Nodes, right? | |
Post #AssEHnkCLvaYCK7vw8 by [email protected] | |
0 likes, 0 repeats | |
@bars no. how? | |
Post #AssFAikyb6P0elg8Nk by [email protected] | |
0 likes, 0 repeats | |
@lritter How do you eval when blocks don't point to their successors? | |
Post #AssFAirMDNVSyYfEKO by [email protected] | |
0 likes, 0 repeats | |
@xavier topologically order all blocks, giving you an array of blocks. Then, co… | |
Post #AssIYrT4A2gCUCVIqO by [email protected] | |
0 likes, 0 repeats | |
@lritter okay fair enough 🤔 pretty good for data locality too | |
Post #AssIYrYNqGvukgzY8G by [email protected] | |
0 likes, 0 repeats | |
@xavier it's at this point an emulator. but you can reverse the transformat… | |
Post #AssU7LSIH0pwpfmvxY by [email protected] | |
0 likes, 0 repeats | |
@lritter Now that I read your document on sr.ht, I see how off I was.Initially,… | |
Post #AssV5lpBMWx9zRJ6i8 by [email protected] | |
0 likes, 0 repeats | |
@bars i guess in that sense they're related, but it diverges where NOIR pre… | |
Post #AssVzM4EJYXiK0qmae by [email protected] | |
0 likes, 0 repeats | |
@lritter > NOIR prescribes no dominator treewait, how do you place phi nodes… | |
Post #AssWI1JpgFUbTS352e by [email protected] | |
0 likes, 0 repeats | |
@bars from the same document.note how here the optional type extension naturall… | |
Post #AssXqqF6x9arDB0Zbk by [email protected] | |
0 likes, 0 repeats | |
@lritter sorry, I'm getting somewhat lost here... if you don't mind hol… | |
Post #AssY4ttIBcGea1VSfA by [email protected] | |
0 likes, 0 repeats | |
@lritter I'd love to lookup the answers to these from the current implement… | |
Post #AssYZseqov8ljWWYM4 by [email protected] | |
0 likes, 0 repeats | |
@bars let's discuss `or` as the more generic case of phi here.in this case,… | |
Post #AssYnrTU16Cg7KsDVQ by [email protected] | |
0 likes, 0 repeats | |
@bars so, e.g. phi {block_a, %a}, {block_b, %b}, {block_c, %c} turns into%block… | |
Post #AssZA2855D5JX8R2Xo by [email protected] | |
0 likes, 0 repeats | |
@bars ah, it's because it has left the pilot stage. i need to fix that: htt… | |
Post #AssZBGbSkrkGiGlZtQ by [email protected] | |
0 likes, 0 repeats | |
@lritter > when i get the SSA program to translate, phi nodes are already in… | |
Post #AssZhRMYHqDDLk8L7w by [email protected] | |
0 likes, 0 repeats | |
@bars you can also generate NOIR from Structured Control Flow. it's not unl… | |
Post #Ast1sNBqwDYoJK89eC by [email protected] | |
0 likes, 0 repeats | |
@lritter I’d be interested in writing on what this IR is like to lower to ass… | |
Post #AstLx3bdB3FC6PhOi0 by [email protected] | |
0 likes, 0 repeats | |
@trishume i have a backend that compiles to C. that's low enough. |