pitrou commented on code in PR #13768: URL: https://github.com/apache/arrow/pull/13768#discussion_r937551293
########## go/arrow/bitutil/bitmaps.go: ########## @@ -422,3 +423,81 @@ func CopyBitmap(src []byte, srcOffset, length int, dst []byte, dstOffset int) { dst[nbytes-1] &= ^trailMask dst[nbytes-1] |= lastData & trailMask } + +type bitOp struct { + opWord func(uint64, uint64) uint64 + opByte func(byte, byte) byte +} + +var ( + bitAndOp = bitOp{ + opWord: func(l, r uint64) uint64 { return l & r }, + opByte: func(l, r byte) byte { return l & r }, + } + bitOrOp = bitOp{ + opWord: func(l, r uint64) uint64 { return l | r }, + opByte: func(l, r byte) byte { return l | r }, + } +) + +func alignedBitmapOp(op bitOp, left, right []byte, lOffset, rOffset int64, out []byte, outOffset int64, length int64) { + debug.Assert(lOffset%8 == rOffset%8, "aligned bitmap op called with unaligned offsets") + debug.Assert(lOffset%8 == outOffset%8, "aligned bitmap op called with unaligned output offset") + + nbytes := BytesForBits(length + lOffset%8) + left = left[lOffset/8:] + right = right[rOffset/8:] + out = out[outOffset/8:] + for i := int64(0); i < nbytes; i++ { + out[i] = op.opByte(left[i], right[i]) + } +} + +func unalignedBitmapOp(op bitOp, left, right []byte, lOffset, rOffset int64, out []byte, outOffset int64, length int64) { + leftRdr := NewBitmapWordReader(left, int(lOffset), int(length)) + rightRdr := NewBitmapWordReader(right, int(rOffset), int(length)) + writer := NewBitmapWordWriter(out, int(outOffset), int(length)) + + for nwords := leftRdr.Words(); nwords > 0; nwords-- { + writer.PutNextWord(op.opWord(leftRdr.NextWord(), rightRdr.NextWord())) Review Comment: Well, I took a look at the C++ benchmark, and it's actually limited by the popcount that's done at the end :-) If I remove the popcount I get this: ``` BenchmarkBitmapAnd/32768/0 732 ns 732 ns 967465 bytes_per_second=41.6736G/s BenchmarkBitmapAnd/131072/0 3105 ns 3105 ns 229726 bytes_per_second=39.3198G/s BenchmarkBitmapAnd/32768/1 2913 ns 2913 ns 240233 bytes_per_second=10.4774G/s BenchmarkBitmapAnd/131072/1 11528 ns 11526 ns 60865 bytes_per_second=10.5912G/s BenchmarkBitmapAnd/32768/2 2924 ns 2924 ns 236873 bytes_per_second=10.4378G/s BenchmarkBitmapAnd/131072/2 11552 ns 11550 ns 60619 bytes_per_second=10.5691G/s ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: github-unsubscr...@arrow.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org