Notes on R6RS benchmarking ========================== 3 September 2009 Petite Chez Scheme v7.9.3 runs all benchmarks successfully. PLT Scheme v4.2.1 runs all benchmarks successfully. Ypsilon 0.9.6-update3 runs all benchmarks successfully. On Pentium machines, Larceny v0.97 runs all benchmarks successfully. On SPARC machines, Larceny fails on the following benchmarks: compiler (bug in assembler; code segment is too large) bv2string (bug in assembler; code segment is too large) Petit Larceny v0.97 fails on the following benchmark: compiler (looks like it may be a bug in gcc) Ikarus 0.0.3 appears to go into an infinite loop on equal. Ikarus fails on the following benchmarks: dderiv (make-hashtable primitive not supported yet) mbrotZ (exception incorrectly raised on complex number syntax) cat3 (utf-16-codec not supported yet) read3 (utf-16-codec not supported yet) bibfreq2 (make-hashtable primitive not supported yet) equal (infinite loop?) normalization (string-normalize-nfkd primitive not supported yet) hashtable0 (make-eqv-hashtable primitive not supported yet) Mosh 0.2.0 appears to go into an infinite loop on earley. Mosh fails on the following benchmarks: sumfp (1e6 etc incorrectly read as exact) sum1 (1e6 etc incorrectly read as exact) read0 (incorrect results, then fails with unhandled exception) earley (infinite loop?) ray (1e6 etc incorrectly read as exact) equal (segmentation fault) normalization (incorrect results) bv2string (incorrect results) listsort (segmentation fault)