Number: 1843 Date: 15-Aug-84 14':15':51 Submitter: Sannella.PA Source: JonL.pa Subject: Figure out **WHY** FOO had the longest ATOMHASH#PROBES number Assigned To: Jonl Attn: Release Status: Fixed In/By: Problem Type: Performance Impact: Moderate Difficulty: Frequency: Priority: Perhaps System: Language Support Subsystem: Storage Formats/Mgt Machine: Disk: Lisp Version: Source Files: Microcode Version: Memory Size: File Server: Server Software Version: Disposition: I take the following as an adequate explanation of the observed phenomenon': After a Full.sysout is created, there are over 18000 litatoms; after a few other packages, and the ABC environment, are loaded, one quickly gets up to about 26000 litatoms --- this is roughly 80% occupancy for the litatom hash table, and the algorithm (not a particularly "bright" one, but at least it''s fast) just doesn''t distribute atoms well when the table is nearly full. See also AR 1440 for extensive discussion of the topic.' ["JonL.pa" " 4-Sep-84 21':56':12" Assigned% To': Attn': Status':(Open->Fixed) Disposition':] Description: ' mjs': dammit, there MUST be some reason besides the fact that it was introduced after all of the packages were loaded. I refuse to believe that it is coincidence.' ' -----' ' Date': 27 Jul 84 03':28 PDT' From': JonL.pa' Subject': FOO! in LOADFULL.CM' To': Masinter, Sannella' cc': JonL.pa' ' All right, dammit, who removed the (QUOTE FOO) from LOADFULL.CM??? There''s no damn reason at all why FULL.sysout shouldn''t have FOO already existing -- its the difference between an ATOMHASH#PROBES of about 50 and one of about 200 *** for the most common meta variable that we all use ***. More importantly, it costs 15-30 seconds of page swapping every 5 minutes or so of debugging.' ' PLEASE PUT IT BACK ASAP.' ' -- JonL --' ' ' Workaround: Test Case: Edit-By: JonL.pa Edit-Date: 4-Sep-84 21':56':12