X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;f=lib%2Fsorter%2Fgovern.c;h=2a8bea0b27da7330c18f7b4f65d91fadcb96bc53;hb=7bddf3297ee2592efb3ca02b822683371e64e340;hp=c1ef332ca4d8c210cf2c598e4156b004ca31199a;hpb=a8933d0fae1331ed903bd314eb170754a6d1ddf7;p=libucw.git diff --git a/lib/sorter/govern.c b/lib/sorter/govern.c index c1ef332c..2a8bea0b 100644 --- a/lib/sorter/govern.c +++ b/lib/sorter/govern.c @@ -177,14 +177,13 @@ sorter_radix_bits(struct sort_context *ctx, struct sort_bucket *b) return 0; u64 in = sbuck_size(b); - u64 mem = ctx->internal_estimate(ctx, b); - if (in < mem) + u64 mem = ctx->internal_estimate(ctx, b) * 0.8; // FIXME: Magical factor for hash non-uniformity + if (in <= mem) return 0; - uns n; - for (n = sorter_min_radix_bits; n < sorter_max_radix_bits && n < b->hash_bits; n++) - if ((in >> n) < mem) - break; + uns n = sorter_min_radix_bits; + while (n < sorter_max_radix_bits && n < b->hash_bits && (in >> n) > mem) + n++; return n; } @@ -192,8 +191,9 @@ static void sorter_radix(struct sort_context *ctx, struct sort_bucket *b, uns bits) { uns nbuck = 1 << bits; - SORT_XTRACE(2, "Running radix sort on %s with %d bits of %d (expected size %s)", + SORT_XTRACE(2, "Running radix split on %s with hash %d bits of %d (expecting %s buckets)", F_BSIZE(b), bits, b->hash_bits, stk_fsize(sbuck_size(b) / nbuck)); + sorter_free_buf(ctx); sorter_start_timer(ctx); struct sort_bucket **outs = alloca(nbuck * sizeof(struct sort_bucket *)); @@ -229,8 +229,6 @@ sorter_run(struct sort_context *ctx) clist_init(&ctx->bucket_list); sorter_prepare_buf(ctx); - /* FIXME: Remember to test sorting of empty files */ - // Create bucket containing the source struct sort_bucket *bin = sbuck_new(ctx); bin->flags = SBF_SOURCE | SBF_OPEN_READ; @@ -239,7 +237,7 @@ sorter_run(struct sort_context *ctx) else bin->fb = ctx->in_fb; bin->ident = "in"; - bin->size = ctx->in_size; /* FIXME: Sizes should be either sh_off_t or u64, not both; beware of ~0U */ + bin->size = ctx->in_size; bin->hash_bits = ctx->hash_bits; clist_add_tail(&ctx->bucket_list, &bin->n); SORT_XTRACE(2, "Input size: %s", F_BSIZE(bin));