notes on CS parallel performance

pollen/private/ts.rktd

@@ -1 +1 @@
-1588531708
+1588547974

pollen/scribblings/raco.scrbl

@@ -128,9 +128,9 @@ As a rule of thumb, parallel rendering works best if you do @exec{raco setup} fi
 > raco pollen render -p 
 }
 
-
 @italic{Warning}: In all cases, the newly rendered output file will overwrite any previous output file.
 
+@margin-note{As of mid-2020, Pollen's parallel-processing performance under the CS (= Chez Scheme) variant of Racket is worse than ordinary Racket. If you use Racket CS, you may get better results using @exec{-j 4} (which will limit the operation to four cores) than @exec{-p} (which will use all available cores).}
 
 @bold{Directory mode}: @racket[raco pollen render _directory] renders all preprocessor source files and then all pagetree files found in the specified directory. If none of these files are found, a pagetree will be generated for the directory (which will include all source files) and then rendered. If the @racket[_directory] argument is omitted, the command defaults to the current directory.
 
@@ -181,6 +181,8 @@ The alternative @exec{--jobs <count>} or @exec{-j <count>} switch does the same
 > raco pollen setup -j 4
 }
 
+@margin-note{As of mid-2020, Pollen's parallel-processing performance under the CS (= Chez Scheme) variant of Racket is worse than ordinary Racket. If you use Racket CS, you may get better results using @exec{-j 4} (which will limit the operation to four cores) than @exec{-p} (which will use all available cores).}
+
 The optional @exec{--dry-run} or @exec{-d} switch prints the paths that would be compiled by this command without actually doing so.