Commit Verlauf

Autor SHA1 Nachricht Datum
  Jay Shah a7cce59d25 adjust tolerances in test script for kv cache vor 2 Monaten
  Jay Shah d437d3dd5c remove smem usage for when rmem -> gmem epilogue is used vor 2 Monaten
  Ganesh Bikshandi e49cb5f77c passes except for hdim=256. vor 2 Monaten
  Ganesh Bikshandi dc2c952f37 compiles and builes. Not validates. vor 2 Monaten
  Ganesh Bikshandi a075e769fb handle gqa_parallel with rmem-to-gmem. Not validating yet. vor 2 Monaten
  Ganesh Bikshandi f0b49460ec changes to use tiledcopy (still not passing). vor 2 Monaten
  Ganesh Bikshandi 0085f04b6a add fp8 test case. vor 2 Monaten
  Ganesh Bikshandi 2b840ef32d fix the test case and re-factor too. vor 2 Monaten
  Jay Shah 2472e5e0b4 add 'in principle' fp8 kv cache support vor 2 Monaten
  Ganesh Bikshandi cd55fb3b5d set correct tolerance limit vor 2 Monaten
  Ganesh Bikshandi 70ff847363 all cases passed. vor 2 Monaten
  Ganesh Bikshandi 9a4941cb35 add variable seqlen case. vor 2 Monaten
  Ganesh Bikshandi c516d6349d Adding another test case. vor 2 Monaten
  Ganesh Bikshandi 24b4b4f7b5 add reference from python. vor 2 Monaten
  Jay Shah 3615696245 Add some more cases to test script and raise thresholds a bit for max error vor 2 Monaten
  Jay Shah 4a77193890 update test script to use heuristic vor 2 Monaten
  Ganesh Bikshandi ffa48eb8cc more refactoring. vor 2 Monaten
  Ganesh Bikshandi a30863fe27 recent version. vor 2 Monaten