Resructure of main vcf_parse function.

This can speed things up by 6-7% on basic single-sample files.

The previous loop caused lots of branch prediction misses due to the
counter 'i' being used to do 8 different parts of code depending on
token number.  Additionally it's got better error checking now as
previously running out of tokens early just did a return 0 rather than
complaining about missing columns.

Author: jkbonfield

vcf.c

@@ -46,9 +46,29 @@ DEALINGS IN THE SOFTWARE.  */
 #include "htslib/khash_str2int.h"
 #include "htslib/kstring.h"
 #include "htslib/sam.h"
-
 #include "htslib/khash.h"
+
+#if 0
+// This helps on Intel a bit, often 6-7% faster VCF parsing.
+// Conversely sometimes harms AMD Zen4 as ~9% slower.
+// Possibly related to IPC differences.  However for now it's just a
+// curiousity we ignore and stick with the simpler code.
+//
+// Left here as a hint for future explorers.
+static inline int xstreq(const char *a, const char *b) {
+    while (*a && *a == *b)
+        a++, b++;
+    return *a == *b;
+}
+
+#define KHASH_MAP_INIT_XSTR(name, khval_t) \
+  KHASH_INIT(name, kh_cstr_t, khval_t, 1, kh_str_hash_func, xstreq)
+
+KHASH_MAP_INIT_XSTR(vdict, bcf_idinfo_t)
+#else
 KHASH_MAP_INIT_STR(vdict, bcf_idinfo_t)
+#endif
+
 typedef khash_t(vdict) vdict_t;
 
 KHASH_MAP_INIT_STR(hdict, bcf_hrec_t*)
@@ -1373,7 +1393,8 @@ bcf_hdr_t *bcf_hdr_init(const char *mode)
     for (i = 0; i < 3; ++i) {
         if ((h->dict[i] = kh_init(vdict)) == NULL) goto fail;
         // Supersize the hash to make collisions very unlikely
-        if (kh_resize(vdict, h->dict[i], 32768) < 0) goto fail;
+        static int dsize[3] = {16384,16384,2048}; // info, contig, format
+        if (kh_resize(vdict, h->dict[i], dsize[i]) < 0) goto fail;
     }
 
     bcf_hdr_aux_t *aux = (bcf_hdr_aux_t*)calloc(1,sizeof(bcf_hdr_aux_t));
@@ -3328,6 +3349,10 @@ static int vcf_parse_info(kstring_t *str, const bcf_hdr_t *h, bcf1_t *v, char *p
             // key=value;
             *next_equals = 0;
             from = val = next_equals+1;
+            // Prefetching d->keys[hash] helps here provided we avoid
+            // computing hash twice (needs API change), but not universally.
+            // It may be significant for other applications though so it's
+            // something to consider for the future.
         } else {
             // key;
             val = NULL;
@@ -3491,7 +3516,7 @@ static int vcf_parse_info(kstring_t *str, const bcf_hdr_t *h, bcf1_t *v, char *p
 
 int vcf_parse(kstring_t *s, const bcf_hdr_t *h, bcf1_t *v)
 {
-    int i = 0, ret = -2, overflow = 0;
+    int ret = -2, overflow = 0;
     char *p, *q, *r, *t;
     kstring_t *str;
     khint_t k;
@@ -3528,79 +3553,125 @@ int vcf_parse(kstring_t *s, const bcf_hdr_t *h, bcf1_t *v)
     bcf_clear1(v);
     str = &v->shared;
     memset(&aux, 0, sizeof(ks_tokaux_t));
-    for (p = kstrtok(s->s, "\t", &aux), i = 0; p; p = kstrtok(0, 0, &aux), ++i) {
-        q = (char*)aux.p;
-        *q = 0;
-        if (i == 0) { // CHROM
-            vdict_t *d = (vdict_t*)h->dict[BCF_DT_CTG];
-            k = kh_get(vdict, d, p);
-            if (k == kh_end(d))
-            {
-                hts_log_warning("Contig '%s' is not defined in the header. (Quick workaround: index the file with tabix.)", p);
-                v->errcode = BCF_ERR_CTG_UNDEF;
-                if ((k = fix_chromosome(h, d, p)) == kh_end(d)) {
-                    hts_log_error("Could not add dummy header for contig '%s'", p);
-                    v->errcode |= BCF_ERR_CTG_INVALID;
+
+    // CHROM
+    if (!(p = kstrtok(s->s, "\t", &aux)))
+        goto err;
+    *(q = (char*)aux.p) = 0;
+
+    vdict_t *d = (vdict_t*)h->dict[BCF_DT_CTG];
+    k = kh_get(vdict, d, p);
+    if (k == kh_end(d)) {
+        hts_log_warning("Contig '%s' is not defined in the header. (Quick workaround: index the file with tabix.)", p);
+        v->errcode = BCF_ERR_CTG_UNDEF;
+        if ((k = fix_chromosome(h, d, p)) == kh_end(d)) {
+            hts_log_error("Could not add dummy header for contig '%s'", p);
+            v->errcode |= BCF_ERR_CTG_INVALID;
+            goto err;
+        }
+    }
+    v->rid = kh_val(d, k).id;
+
+    // POS
+    if (!(p = kstrtok(0, 0, &aux)))
+        goto err;
+    *(q = (char*)aux.p) = 0;
+
+    overflow = 0;
+    char *tmp = p;
+    v->pos = hts_str2uint(p, &p, 63, &overflow);
+    if (overflow) {
+        hts_log_error("Position value '%s' is too large", tmp);
+        goto err;
+    } else if ( *p ) {
+        hts_log_error("Could not parse the position '%s'", tmp);
+        goto err;
+    } else {
+        v->pos -= 1;
+    }
+    if (v->pos >= INT32_MAX)
+        v->unpacked |= BCF_IS_64BIT;
+
+    // ID
+    if (!(p = kstrtok(0, 0, &aux)))
+        goto err;
+    *(q = (char*)aux.p) = 0;
+
+    if (NOT_DOT(p)) bcf_enc_vchar(str, q - p, p);
+    else bcf_enc_size(str, 0, BCF_BT_CHAR);
+
+    // REF
+    if (!(p = kstrtok(0, 0, &aux)))
+        goto err;
+    *(q = (char*)aux.p) = 0;
+
+    bcf_enc_vchar(str, q - p, p);
+    v->n_allele = 1, v->rlen = q - p;
+
+    // ALT
+    if (!(p = kstrtok(0, 0, &aux)))
+        goto err;
+    *(q = (char*)aux.p) = 0;
+
+    if (NOT_DOT(p)) {
+        for (r = t = p;; ++r) {
+            if (*r == ',' || *r == 0) {
+                if (v->n_allele == UINT16_MAX) {
+                    hts_log_error("Too many ALT alleles at %s:%"PRIhts_pos,
+                                  bcf_seqname_safe(h,v), v->pos+1);
+                    v->errcode |= BCF_ERR_LIMITS;
                     goto err;
                 }
+                bcf_enc_vchar(str, r - t, t);
+                t = r + 1;
+                ++v->n_allele;
             }
-            v->rid = kh_val(d, k).id;
-        } else if (i == 1) { // POS
-            overflow = 0;
-            char *tmp = p;
-            v->pos = hts_str2uint(p, &p, 63, &overflow);
-            if (overflow) {
-                hts_log_error("Position value '%s' is too large", tmp);
-                goto err;
-            } else if ( *p ) {
-                hts_log_error("Could not parse the position '%s'", tmp);
-                goto err;
-            } else {
-                v->pos -= 1;
-            }
-            if (v->pos >= INT32_MAX)
-                v->unpacked |= BCF_IS_64BIT;
-        } else if (i == 2) { // ID
-            if (NOT_DOT(p)) bcf_enc_vchar(str, q - p, p);
-            else bcf_enc_size(str, 0, BCF_BT_CHAR);
-        } else if (i == 3) { // REF
-            bcf_enc_vchar(str, q - p, p);
-            v->n_allele = 1, v->rlen = q - p;
-        } else if (i == 4) { // ALT
-            if (NOT_DOT(p)) {
-                for (r = t = p;; ++r) {
-                    if (*r == ',' || *r == 0) {
-                        if (v->n_allele == UINT16_MAX) {
-                            hts_log_error("Too many ALT alleles at %s:%"PRIhts_pos,
-                                          bcf_seqname_safe(h,v), v->pos+1);
-                            v->errcode |= BCF_ERR_LIMITS;
-                            goto err;
-                        }
-                        bcf_enc_vchar(str, r - t, t);
-                        t = r + 1;
-                        ++v->n_allele;
-                    }
-                    if (r == q) break;
-                }
-            }
-        } else if (i == 5) { // QUAL
-            if (NOT_DOT(p)) v->qual = atof(p);
-            else bcf_float_set_missing(v->qual);
-            if ( v->max_unpack && !(v->max_unpack>>1) ) goto end; // BCF_UN_STR
-        } else if (i == 6) { // FILTER
-            if (NOT_DOT(p)) {
-                if (vcf_parse_filter(str, h, v, p, q)) goto err;
-            } else bcf_enc_vint(str, 0, 0, -1);
-            if ( v->max_unpack && !(v->max_unpack>>2) ) goto end; // BCF_UN_FLT
-        } else if (i == 7) { // INFO
-            if (NOT_DOT(p)) {
-                if (vcf_parse_info(str, h, v, p, q)) goto err;
-            }
-            if ( v->max_unpack && !(v->max_unpack>>3) ) goto end;
-        } else if (i == 8) {// FORMAT
-            return vcf_parse_format(s, h, v, p, q) == 0 ? 0 : -2;
+            if (r == q) break;
+        }
+    }
+
+    // QUAL
+    if (!(p = kstrtok(0, 0, &aux)))
+        goto err;
+    *(q = (char*)aux.p) = 0;
+
+    if (NOT_DOT(p)) v->qual = atof(p);
+    else bcf_float_set_missing(v->qual);
+    if ( v->max_unpack && !(v->max_unpack>>1) ) goto end; // BCF_UN_STR
+
+    // FILTER
+    if (!(p = kstrtok(0, 0, &aux)))
+        goto err;
+    *(q = (char*)aux.p) = 0;
+
+    if (NOT_DOT(p)) {
+        if (vcf_parse_filter(str, h, v, p, q)) {
+            goto err;
+        }
+    } else bcf_enc_vint(str, 0, 0, -1);
+    if ( v->max_unpack && !(v->max_unpack>>2) ) goto end; // BCF_UN_FLT
+
+    // INFO
+    if (!(p = kstrtok(0, 0, &aux)))
+        goto err;
+    *(q = (char*)aux.p) = 0;
+
+    if (NOT_DOT(p)) {
+        if (vcf_parse_info(str, h, v, p, q)) {
+            goto err;
         }
     }
+    if ( v->max_unpack && !(v->max_unpack>>3) ) goto end;
+
+    // FORMAT; optional
+    p = kstrtok(0, 0, &aux);
+    if (p) {
+        *(q = (char*)aux.p) = 0;
+
+        return vcf_parse_format(s, h, v, p, q) == 0 ? 0 : -2;
+    } else {
+        return 0;
+    }
 
  end:
     ret = 0;