Merge branch 'master' of github.com:fangq/mcxcl

fangq · Jun 12, 2024 · 2cd9c9e · 2cd9c9e
2 parents 4609451 + 2b91626
commit 2cd9c9e
Show file tree

Hide file tree

Showing 4 changed files with 13 additions and 10 deletions.
diff --git a/src/mcx_core.cl b/src/mcx_core.cl
@@ -960,8 +960,11 @@ int launchnewphoton(float4* p, float4* v, float4* f, short4* flipdir, FLOAT4VEC*
 #if defined(MCX_SRC_PLANAR) || defined(MCX_SRC_PATTERN) || defined(MCX_SRC_PATTERN3D) || defined(MCX_SRC_FOURIER) || defined(MCX_SRC_PENCILARRAY) /*a rectangular grid over a plane*/
         float rx = rand_uniform01(t);
         float ry = rand_uniform01(t);
-        float rz;
+
 #if defined(MCX_SRC_PATTERN3D)
+
+        float rz;
+
         rz = rand_uniform01(t);
         p[0] = (float4)(p[0].x + rx * gcfg->srcparam1.x,
                         p[0].y + ry * gcfg->srcparam1.y,
@@ -1169,8 +1172,8 @@ int launchnewphoton(float4* p, float4* v, float4* f, short4* flipdir, FLOAT4VEC*
             } else { // odd number length, interpolate between neigboring values
                 ang = fmin(rand_uniform01(t) * (GPU_PARAM(gcfg, nangle) - 1), GPU_PARAM(gcfg, nangle) - 1 - EPS);
                 sphi = ang - ((int)ang);
-                cphi = ((1.f - sphi) * (((__local float*)(sharedmem))[((int)ang >= GPU_PARAM(gcfg, nangle) - 1 ? GPU_PARAM(gcfg, nangle) - 1 : (int)(ang)) + GPU_PARAM(gcfg, nphaselen)]) +
-                        sphi * (((__local float*)(sharedmem))[((int)ang + 1 >= GPU_PARAM(gcfg, nangle) - 1 ? GPU_PARAM(gcfg, nangle) - 1 : (int)(ang) + 1) + GPU_PARAM(gcfg, nphaselen)]));
+                cphi = ((1.f - sphi) * (((__local float*)(sharedmem))[((uint)ang >= GPU_PARAM(gcfg, nangle) - 1 ? GPU_PARAM(gcfg, nangle) - 1 : (int)(ang)) + GPU_PARAM(gcfg, nphaselen)]) +
+                        sphi * (((__local float*)(sharedmem))[((uint)ang + 1 >= GPU_PARAM(gcfg, nangle) - 1 ? GPU_PARAM(gcfg, nangle) - 1 : (int)(ang) + 1) + GPU_PARAM(gcfg, nphaselen)]));
             }
 
             cphi *= ONE_PI; // next zenith angle computed based on angleinvcdf
@@ -1351,7 +1354,7 @@ __kernel void mcx_main_loop(__global const uint* media,
         barrier(CLK_LOCAL_MEM_FENCE);
     }
 
-    if (idx >= gcfg->threadphoton * (get_local_size(0) * get_num_groups(0)) + gcfg->oddphoton) {
+    if ((uint)idx >= gcfg->threadphoton * (get_local_size(0) * get_num_groups(0)) + gcfg->oddphoton) {
         return;
     }
 
@@ -1424,8 +1427,8 @@ __kernel void mcx_main_loop(__global const uint* media,
                 if (GPU_PARAM(gcfg, nphase) > 2) { // after padding the left/right ends, nphase must be 3 or more
                     tmp0 = rand_uniform01(t) * (GPU_PARAM(gcfg, nphase) - 1);
                     theta = tmp0 - ((int)tmp0);
-                    tmp0 = (1.f - theta) * ((__local float*)(sharedmem))[(int)tmp0   >= GPU_PARAM(gcfg, nphase) ? GPU_PARAM(gcfg, nphase) - 1 : (int)(tmp0)  ] +
-                           theta * ((__local float*)(sharedmem))[(int)tmp0 + 1 >= GPU_PARAM(gcfg, nphase) ? GPU_PARAM(gcfg, nphase) - 1 : (int)(tmp0) + 1];
+                    tmp0 = (1.f - theta) * ((__local float*)(sharedmem))[(uint)tmp0   >= GPU_PARAM(gcfg, nphase) ? GPU_PARAM(gcfg, nphase) - 1 : (int)(tmp0)  ] +
+                           theta * ((__local float*)(sharedmem))[(uint)tmp0 + 1 >= GPU_PARAM(gcfg, nphase) ? GPU_PARAM(gcfg, nphase) - 1 : (int)(tmp0) + 1];
                     theta = acos(tmp0);
                     stheta = MCX_MATHFUN(sin)(theta);
                     ctheta = tmp0;

diff --git a/src/mcx_utils.c b/src/mcx_utils.c
@@ -3191,7 +3191,7 @@ void mcx_loadseedjdat(char* filename, Config* cfg) {
             uint dims[3] = {1, 1, 1};
             float* ppath = NULL;
             char* type;
-            History his = {0};
+            History his = {{0}};
 
             cJSON* vsize = cJSON_GetObjectItem(seed, "_ArraySize_");
 

diff --git a/src/mcx_utils.h b/src/mcx_utils.h
@@ -160,7 +160,7 @@ typedef struct MCXConfig {
     float4* detpos;               /**<detector positions and radius, overwrite detradius*/
 
     unsigned int maxgate;         /**<simultaneous recording gates*/
-    unsigned int respin;          /**<number of repeatitions (if positive), or number of divisions (if negative)*/
+    int respin;                   /**<number of repeatitions (if positive), or number of divisions (if negative)*/
     unsigned int printnum;        /**<number of printed threads (for debugging)*/
     int gpuid;                    /**<the ID of the GPU to use, starting from 1, 0 for auto*/
 

diff --git a/test/testmcx.sh b/test/testmcx.sh
@@ -119,12 +119,12 @@ if [ -z "$temp" ]; then echo "fail to save photon seeds"; fail=$((fail+1)); else
 
 echo "test photon replay flag -E ... "
 rm -rf replaytest.*
-temp=`("$MCX" --bench cube60 -s replaytest -q 1 -S 0 $PARAM && "$MCX" --bench cube60 -E replaytest_detp.jdat -S 0 $PARAM) | sed 's/\x1b\[[0-9;]*m//g' | grep -o -E '(simulated|detected)\s+[0-9.]+ photons' | tail -2 | sed -e 's/^[a-z ]*//g' | sort | uniq -c | grep '^\s*2\s*\d*'`
+temp=`("$MCX" --bench cube60 -s replaytest -q 1 -S 0 $PARAM && "$MCX" --bench cube60 -E replaytest_detp.jdat -S 0 $PARAM) | sed $'s/\x1b\[[0-9;]*m//g' | grep -o -E '(simulated|detected)\s+[0-9.]+ photons' | tail -2 | sed -e 's/^[a-z ]*//g' | sort | uniq -c | grep '^\s*2\s*\d*'`
 if [ -z "$temp" ]; then echo "fail to run photon replay -E"; fail=$((fail+1)); else echo "ok"; fi
 
 echo "test photon replay ... "
 rm -rf replaytest.*
-temp=`("$MCX" --bench cube60 -s replaytest -q 1 -S 0 $PARAM && "$MCX" --bench cube60 -E replaytest_detp.jdat -S 0 $PARAM) | sed 's/\x1b\[[0-9;]*m//g' | grep -o -E 'absorbed:.*3[0-8]\.[0-9]+%'`
+temp=`("$MCX" --bench cube60 -s replaytest -q 1 -S 0 $PARAM && "$MCX" --bench cube60 -E replaytest_detp.jdat -S 0 $PARAM) | sed $'s/\x1b\[[0-9;]*m//g' | grep -o -E 'absorbed:.*3[0-8]\.[0-9]+%'`
 if [ -z "$temp" ]; then echo "fail to run photon replay"; fail=$((fail+1)); else echo "ok"; fi
 
 echo "test heterogeneous domain ... "