feat: add CUDA benchmark CLI and shared search algorithms

bench/test_cli.py

@@ -0,0 +1,65 @@
+#!/usr/bin/env python3
+import argparse,json,subprocess,tempfile
+from pathlib import Path
+HERE=Path(__file__).resolve().parent
+BIN=HERE/'bin/slime-bench'
+def run(*args,ok=True):
+    p=subprocess.run([str(BIN),*map(str,args)],capture_output=True,text=True,timeout=180)
+    if ok:
+        assert p.returncode==0,p.stderr
+        return json.loads(p.stdout)
+    assert p.returncode!=0,p.stdout
+
+def term(x,z):
+    def i32(v):return ((v+2**31)%2**32)-2**31
+    return i32(x*x*4987142)+i32(x*5947611)+i32(z*z)*4392871+i32(z*389711)
+def slime(seed,x,z):
+    s=((seed+term(x,z))^987234911^0x5DEECE66D)&((1<<48)-1)
+    while True:
+        s=(s*0x5DEECE66D+11)&((1<<48)-1);q=s>>17
+        if q<2147483640:return q%10==0
+
+def main():
+    ap=argparse.ArgumentParser();ap.add_argument('--gpu',action='store_true');args=ap.parse_args()
+    for radius in [0,1,2,31,1000000]:
+        r=run('--x',-123,'--z',456,'--radius',radius,'--size',6,'--dry-run')
+        assert r['anchors']==1+2*radius*(radius+1)
+        assert r['first_anchor']==[-123,456-radius] and r['last_anchor']==[-123,456+radius]
+    for bad in [[],['--radius','-1'],['--radius','1','--size','0'],['--radius','1','--size','7'],['--radius','abc'],['--radius','0','--x','2147483647'],['--radius','1','--repeat','0'],['--radius','1','--position-batch','0']]:run(*bad,'--dry-run',ok=False)
+    summary={'cli_scope_and_errors':True}
+    if args.gpu:
+        common=['--x',51069,'--z',419124]
+        reference=run(*common,'--radius',0,'--size',5,'--repeat',2,'--max-results',100)
+        expected={61692151016558,15959335054446}
+        assert all(r['matches']==24 and r['root_jobs']==84 for r in reference['rounds'])
+        samples=[set(map(tuple,r['sample'])) for r in reference['rounds']]
+        assert samples[0]==samples[1] and expected <= {s for s,x,z in samples[0]}
+        sliced=run(*common,'--radius',0,'--size',5,'--job-batch',17,'--max-results',100,'--cache-classes',0)
+        assert set(map(tuple,sliced['rounds'][0]['sample']))==samples[0]
+        partial=run(*common,'--radius',0,'--size',5,'--max-results',3)
+        assert partial['rounds'][0]['matches']==24 and len(partial['rounds'][0]['sample'])==3 and partial['rounds'][0]['sample_truncated']
+        for n in [4,6]:
+            r=run(*common,'--radius',0,'--size',n,'--max-results',1000)
+            for seed,x,z in r['rounds'][0]['sample']:assert all(slime(seed,x+dx,z+dz) for dx in range(n) for dz in range(n))
+            if n>=5:
+                wanted={t for t in samples[0] if all(slime(t[0],t[1]+dx,t[2]+dz) for dx in range(n) for dz in range(n))}
+                assert set(map(tuple,r['rounds'][0]['sample']))==wanted
+            summary[f'size_{n}']=r['rounds'][0]['matches']
+        diamond=run(*common,'--radius',2,'--size',5,'--max-results',1000,'--position-batch',3)
+        split=[]
+        for dz in range(-2,3):
+            for dx in range(-2,3):
+                if abs(dx)+abs(dz)>2:continue
+                r=run('--x',51069+dx,'--z',419124+dz,'--radius',0,'--size',5,'--max-results',1000)
+                split.extend(map(tuple,r['rounds'][0]['sample']))
+        assert sorted(split)==sorted(map(tuple,diamond['rounds'][0]['sample']))
+        assert diamond['anchors']==13
+        run('--radius',0,'--devices','0,0',ok=False)
+        with tempfile.TemporaryDirectory() as tmp:
+            out=Path(tmp)/'report.json'
+            r=run(*common,'--radius',0,'--output',out)
+            assert json.loads(out.read_text())==r
+            run(*common,'--radius',0,'--output',out,ok=False)
+        summary['gpu_known_matches']=24;summary['diamond_vs_individual_anchors']=True
+    print(json.dumps(summary,indent=2))
+if __name__=='__main__':main()