LS replicated 5 times

ofdock_taskpar_alt/Makefile

@@ -212,6 +212,14 @@ ENABLE_K35 = YES
 # bt ushort_float
 ENABLE_K36 = YES
 
+# prng ls4, ls5
+ENABLE_K37 = YES
+ENABLE_K38 = YES
+
+# krnl_ls4, Krnl_ls5
+ENABLE_K39 = YES
+ENABLE_K40 = YES
+
 
 ifeq ($(ENABLE_K1),YES)
 	K1 =-DENABLE_KERNEL1
@@ -435,7 +443,31 @@ else
 	K36 =
 endif
 
-ENABLE_KERNELS = $(K1) $(K2) $(K3) $(K4) $(K5) $(K6) $(K7) $(K8) $(K9) $(K10) $(K11) $(K12) $(K13) $(K14) $(K15) $(K16) $(K17) $(K18) $(K19) $(K20) $(K21) $(K22) $(K23) $(K24) $(K25) $(K26) $(K27) $(K28) $(K29) $(K30) $(K31) $(K32) $(K33) $(K34) $(K35) $(K36)
+ifeq ($(ENABLE_K37),YES)
+	K37 =-DENABLE_KERNEL37
+else
+	K37 =
+endif
+
+ifeq ($(ENABLE_K38),YES)
+	K38 =-DENABLE_KERNEL38
+else
+	K38 =
+endif
+
+ifeq ($(ENABLE_K39),YES)
+	K39 =-DENABLE_KERNEL39
+else
+	K39 =
+endif
+
+ifeq ($(ENABLE_K40),YES)
+	K40 =-DENABLE_KERNEL40
+else
+	K40 =
+endif
+
+ENABLE_KERNELS = $(K1) $(K2) $(K3) $(K4) $(K5) $(K6) $(K7) $(K8) $(K9) $(K10) $(K11) $(K12) $(K13) $(K14) $(K15) $(K16) $(K17) $(K18) $(K19) $(K20) $(K21) $(K22) $(K23) $(K24) $(K25) $(K26) $(K27) $(K28) $(K29) $(K30) $(K31) $(K32) $(K33) $(K34) $(K35) $(K36) $(K37) $(K38) $(K39) $(K40)
 
 # =============================
 # Reproduce result (remove randomness)
@@ -461,9 +493,13 @@ FIXED_POINT_INTRAE=NO
 FIXED_POINT_LS1=YES
 FIXED_POINT_LS2=YES
 FIXED_POINT_LS3=YES
+FIXED_POINT_LS4=YES
+FIXED_POINT_LS5=YES
 #FIXED_POINT_LS1=NO
 #FIXED_POINT_LS2=NO
 #FIXED_POINT_LS3=NO
+#FIXED_POINT_LS4=NO
+#FIXED_POINT_LS5=NO
 
 SINGLE_COPY_POP_ENE=YES
 
@@ -505,6 +541,18 @@ else
 	FIPOLS3_FLAG=
 endif
 
+ifeq ($(FIXED_POINT_LS4), YES)
+	FIPOLS4_FLAG=-DFIXED_POINT_LS4
+else	
+	FIPOLS4_FLAG=
+endif
+
+ifeq ($(FIXED_POINT_LS5), YES)
+	FIPOLS5_FLAG=-DFIXED_POINT_LS5
+else	
+	FIPOLS5_FLAG=
+endif
+
 ifeq ($(SINGLE_COPY_POP_ENE), YES)
 	COPYPOPENE_FLAG=-DSINGLE_COPY_POP_ENE
 else	
@@ -517,7 +565,7 @@ else
 	SEP_FGRID_FLAG=
 endif
 
-FIPO_FLAG = $(FIPOCO_FLAG) $(FIPOIE_FLAG) $(FIPOIA_FLAG) $(FIPOLS1_FLAG) $(FIPOLS2_FLAG) $(FIPOLS3_FLAG) $(COPYPOPENE_FLAG) $(SEP_FGRID_FLAG) 
+FIPO_FLAG = $(FIPOCO_FLAG) $(FIPOIE_FLAG) $(FIPOIA_FLAG) $(FIPOLS1_FLAG) $(FIPOLS2_FLAG) $(FIPOLS3_FLAG) $(FIPOLS4_FLAG) $(FIPOLS5_FLAG) $(COPYPOPENE_FLAG) $(SEP_FGRID_FLAG) 
 
 # =============================
 # AOC flags

ofdock_taskpar_alt/device/Krnl_Conform.cl

@@ -135,6 +135,14 @@ while(active) {
 			case 0x03:
 				fl_tmp = read_channel_altera(chan_LS2Conf_LS3_genotype);
 			break;
+
+			case 0x04:
+				fl_tmp = read_channel_altera(chan_LS2Conf_LS4_genotype);
+			break;
+
+			case 0x05:
+				fl_tmp = read_channel_altera(chan_LS2Conf_LS5_genotype);
+			break;
 		}
 		
 		if (i > 2) {

ofdock_taskpar_alt/device/Krnl_IGL_Arbiter.cl

@@ -19,6 +19,8 @@ while(active) {
 	bool LS1_end_valid = false;
 	bool LS2_end_valid = false;
 	bool LS3_end_valid = false;
+	bool LS4_end_valid = false;
+	bool LS5_end_valid = false;
 
 	bool Off_active;
 	bool IC_active;
@@ -26,6 +28,8 @@ while(active) {
 	bool LS1_end_active;
 	bool LS2_end_active;
 	bool LS3_end_active;
+	bool LS4_end_active;
+	bool LS5_end_active;
 
 	while (
 		(Off_valid     == false) &&
@@ -33,7 +37,9 @@ while(active) {
 		(GG_valid      == false) && 
 		(LS1_end_valid == false) &&
 		(LS2_end_valid == false) &&
-		(LS3_end_valid == false) 
+		(LS3_end_valid == false) &&
+		(LS4_end_valid == false) &&
+		(LS5_end_valid == false) 
 	){
 		Off_active     = read_channel_nb_altera(chan_IGLArbiter_Off,     &Off_valid);
 		IC_active      = read_channel_nb_altera(chan_GA2IGL_IC_active,   &IC_valid);
@@ -41,11 +47,13 @@ while(active) {
 		LS1_end_active = read_channel_nb_altera(chan_LS2Arbiter_LS1_end, &LS1_end_valid);
 		LS2_end_active = read_channel_nb_altera(chan_LS2Arbiter_LS2_end, &LS2_end_valid);
 		LS3_end_active = read_channel_nb_altera(chan_LS2Arbiter_LS3_end, &LS3_end_valid);
+		LS4_end_active = read_channel_nb_altera(chan_LS2Arbiter_LS4_end, &LS4_end_valid);
+		LS5_end_active = read_channel_nb_altera(chan_LS2Arbiter_LS5_end, &LS5_end_valid);
 	}
 
 	uchar bound_tmp = 0;
 	active = Off_valid ? 0x00 : 0x01;
-	char mode [3];	// mode for all LS
+	char mode [5];	// mode for all LS
 
 /*
 	float genotypeICGG  [ACTUAL_GENOTYPE_LENGTH]; 
@@ -54,110 +62,96 @@ while(active) {
 
 	// get genotype from IC, GG, LS1, LS2, LS3
 	if (active == 0x01) {
-		//#pragma ivdep
-/*
-		for (uchar i=0; i<DockConst_num_of_genes; i++) {
-*/
-			if (IC_valid == true) {
-				bound_tmp++;
-/*
-				if (i == 0) {bound_tmp++; }
-				genotypeICGG [i] = read_channel_altera(chan_IC2Conf_genotype);
-*/
-			}
-			else if (GG_valid == true) {
-				bound_tmp++;
-/*
-				if (i == 0) {bound_tmp++; }
-				genotypeICGG [i] = read_channel_altera(chan_GG2Conf_genotype);
-*/
-			}	
-			else{
-/*
-				float genotype1;
-				float genotype2;
-				float genotype3;
-*/
 
-#if 0
-				if (LS1_end_valid == true) {
-					//printf("%-15s %5s\n",   "LS1_valid: ", "reading genotypes");
-					if (i == 0) {bound_tmp++; }
-					genotype1 /*[i]*/ = read_channel_altera(chan_LS2Conf_LS1_genotype);
-				}
-#endif
-/*
-				if (LS2_end_valid == true) {
-					genotype2 = read_channel_altera(chan_LS2Conf_LS2_genotype);
-				}
-				if (LS3_end_valid == true) {
-					genotype3 = read_channel_altera(chan_LS2Conf_LS3_genotype);
+		if (IC_valid == true) {
+			bound_tmp++;
+		}
+		else if (GG_valid == true) {
+			bound_tmp++;
+		}	
+		else{
+			// Reorder the mode & genotype coming from LS
+			if (LS1_end_valid) {mode[0] = 0x01; bound_tmp++;
+				
+				if (LS2_end_valid) {mode[1] = 0x02; bound_tmp++;
+
+					if (LS3_end_valid) {mode[2] = 0x03; bound_tmp++;
+						if (LS4_end_valid) {mode[3] = 0x04; bound_tmp++;
+							if (LS5_end_valid) {mode[4] = 0x05; bound_tmp++;}
+						}
+						else {
+							if (LS5_end_valid) {mode[3] = 0x05; bound_tmp++;}
+						}
+					}
+					else { // LS1: yes, LS2: yes, LS3: no
+						if (LS4_end_valid) {mode[2] = 0x04; bound_tmp++;
+							if (LS5_end_valid) {mode[3] = 0x05; bound_tmp++;}
+						}
+						else {
+							if (LS5_end_valid) {mode[2] = 0x05; bound_tmp++;}
+						}
+					}
 				}
-*/
-				// Reorder the mode & genotype coming from LS
-				if (LS1_end_valid) {
-					mode[0] = 0x01; bound_tmp++;
-/*
-					genotype1 = read_channel_altera(chan_LS2Conf_LS1_genotype);
-					if (i == 0) {mode[0] = 0x01; bound_tmp++;}
-					genotype[0][i & MASK_GENOTYPE] = genotype1; 
-*/					
-					if (LS2_end_valid) {
-						mode[1] = 0x02; bound_tmp++;
-/*
-						if (i == 0) {mode[1] = 0x02; bound_tmp++;}
-						genotype[1][i & MASK_GENOTYPE] = genotype2;
-*/
+				else { // LS1: yes, LS2: no
+					if (LS3_end_valid) {mode[1] = 0x03; bound_tmp++;
+						if (LS4_end_valid) {mode[2] = 0x04; bound_tmp++;
+							if (LS5_end_valid) {mode[3] = 0x05; bound_tmp++;}
+						}
+						else {
+							if (LS5_end_valid) {mode[2] = 0x05; bound_tmp++;}
+						}
+					} 
+					else { // LS1: yes, LS2: no, LS3: no
+						if (LS4_end_valid) {mode[1] = 0x04; bound_tmp++;
+							if (LS5_end_valid) {mode[2] = 0x05; bound_tmp++;}
+						}
+						else {
+							if (LS5_end_valid) {mode[1] = 0x05; bound_tmp++;}
+						}
 
-						if (LS3_end_valid) {
-							mode[2] = 0x03; bound_tmp++;
-/*
-							if (i == 0) {mode[2] = 0x03; bound_tmp++;} 
-							genotype[2][i & MASK_GENOTYPE] = genotype3;
-*/
+					}
+				}
+			}
+			else { // LS1: no
+				if (LS2_end_valid) {mode[0] = 0x02; bound_tmp++;
+					if (LS3_end_valid) {mode[1] = 0x03; bound_tmp++;
+						if (LS4_end_valid) {mode[2] = 0x04; bound_tmp++;
+							if (LS5_end_valid) {mode[3] = 0x05; bound_tmp++;}
+						}
+						else {
+							if (LS5_end_valid) {mode[2] = 0x05; bound_tmp++;}
 						}
 					}
-					else {
-						if (LS3_end_valid) {
-							mode[1] = 0x03; bound_tmp++;
-/*
-							if (i == 0) {mode[1] = 0x03; bound_tmp++;}
-							genotype[1][i & MASK_GENOTYPE] = genotype3;
-*/
+					else { // LS1: no, LS2: yes, LS3: no
+						if (LS4_end_valid) {mode[1] = 0x04; bound_tmp++;
+							if (LS5_end_valid) {mode[2] = 0x05; bound_tmp++;}
+						}
+						else {
+							if (LS5_end_valid) {mode[1] = 0x05; bound_tmp++;}
 						}
 					}
 				}
-				else {
-					if (LS2_end_valid) {
-						mode[0] = 0x02; bound_tmp++;
-/*
-						if (i == 0) {mode[0] = 0x02; bound_tmp++;}
-						genotype[0][i & MASK_GENOTYPE] = genotype2;
-*/
-
-						if (LS3_end_valid) {
-							mode[1] = 0x03; bound_tmp++;
-/*
-							if (i == 0) {mode[1] = 0x03; bound_tmp++;}
-							genotype[1][i & MASK_GENOTYPE] = genotype3;
-*/
+				else { // LS1: no, LS2: no
+					if (LS3_end_valid) {mode[0] = 0x03; bound_tmp++;
+						if (LS4_end_valid) {mode[1] = 0x04; bound_tmp++;
+							if (LS5_end_valid) {mode[2] = 0x05; bound_tmp++;}
+						}
+						else {
+							if (LS5_end_valid) {mode[1] = 0x05; bound_tmp++;}
 						}
 					}
-					else {
-						if (LS3_end_valid) {
-							mode[0] = 0x03; bound_tmp++;
-/*
-							if (i == 0) {mode[0] = 0x03; bound_tmp++;}
-							genotype[0][i & MASK_GENOTYPE] = genotype3;
-*/
+					else { // LS1: no, LS2: no, LS3: no
+						if (LS4_end_valid) {mode[0] = 0x04; bound_tmp++;
+							if (LS5_end_valid) {mode[1] = 0x05; bound_tmp++;}
 						}
+						else {
+							if (LS5_end_valid) {mode[0] = 0x05; bound_tmp++;}
+						}
+
 					}
 				}
-					
-			}
-/*
-		} // End of for-loop for (uchar i=0; i<DockConst_num_of_genes; i++) { }
-*/
+			}			
+		}
 	} // End if (active == true)
 
 	uchar bound = active ? bound_tmp : 1;

ofdock_taskpar_alt/device/Krnl_InterE.cl

@@ -485,6 +485,12 @@ while(active) {
 
 		// LS 3
 		case 0x03: write_channel_altera(chan_Intere2StoreLS_LS3_intere, final_interE); break;
+
+		// LS 4
+		case 0x04: write_channel_altera(chan_Intere2StoreLS_LS4_intere, final_interE); break;
+
+		// LS 5
+		case 0x05: write_channel_altera(chan_Intere2StoreLS_LS5_intere, final_interE); break;
 	}
 	// --------------------------------------------------------------
  	

ofdock_taskpar_alt/device/Krnl_IntraE.cl

@@ -225,6 +225,12 @@ while(active) {
 
 		// LS 3
 		case 0x03: write_channel_altera(chan_Intrae2StoreLS_LS3_intrae, intraE); break;
+
+		// LS 4
+		case 0x04: write_channel_altera(chan_Intrae2StoreLS_LS4_intrae, intraE); break;
+
+		// LS 5
+		case 0x05: write_channel_altera(chan_Intrae2StoreLS_LS5_intrae, intraE); break;
 	}
 	// --------------------------------------------------------------
 

ofdock_taskpar_alt/device/Krnl_PRNG.cl

@@ -113,6 +113,7 @@ void Krnl_Prng_GG_float(unsigned int  Host_seed,
 // --------------------------------------------------------------------------
 // --------------------------------------------------------------------------
 
+#if 0
 __kernel __attribute__ ((max_global_work_dim(0)))
 void Krnl_Prng_LS123_ushort(unsigned int Host_seed1,
 			    unsigned int Host_seed2, 
@@ -197,6 +198,68 @@ void Krnl_Prng_LS123_ushort(unsigned int Host_seed1,
 */
 	} // End of while(active)
 }
+#endif
+
+__kernel __attribute__ ((max_global_work_dim(0)))
+void Krnl_Prng_LS123_ushort(unsigned int Host_seed1,
+			    unsigned int Host_seed2, 
+			    unsigned int Host_seed3,
+			    unsigned int Host_seed4,
+			    unsigned int Host_seed5, 
+		            unsigned int DockConst_pop_size){
+
+
+	uint lfsr[5];
+	lfsr[0] = Host_seed1;
+	lfsr[1] = Host_seed2;
+	lfsr[2] = Host_seed3;
+	lfsr[3] = Host_seed4;
+	lfsr[4] = Host_seed5;
+
+	bool valid = false;
+
+	while(!valid) {
+		bool active = true;
+		active  = read_channel_nb_altera(chan_Arbiter_LS123_ushort_off, &valid);
+
+		ushort tmp[5];
+		
+		#pragma unroll
+		for (uint i=0; i<5; i++){
+			uchar  lsb[5];
+			lsb [i] = lfsr[i] & 0x01u;
+			lfsr[i] >>= 1;
+			lfsr[i] ^= (-lsb[i]) & 0xA3000000u;
+			tmp [i] = (DockConst_pop_size/MAX_UINT)*lfsr[i];
+		}
+
+		// to avoid having same entities undergoing LS simultaneously
+		if (
+			(tmp[0] == tmp[1]) || (tmp[0] == tmp[2]) || (tmp[0] == tmp[3]) || (tmp[0] == tmp[4]) ||
+			(tmp[1] == tmp[2]) || (tmp[1] == tmp[3]) || (tmp[1] == tmp[4]) ||
+			(tmp[2] == tmp[3]) || (tmp[2] == tmp[4]) ||
+			(tmp[3] == tmp[4])
+		) {
+			tmp[1] = tmp[0] + 1;
+			tmp[2] = tmp[1] + 2;
+			tmp[3] = tmp[2] + 3;
+			tmp[4] = tmp[3] + 4;
+		}
+
+		bool success = false;
+		ushort8 tmp123;
+		tmp123.s0 = tmp[0];
+		tmp123.s1 = tmp[1];
+		tmp123.s2 = tmp[2];
+		tmp123.s3 = tmp[3];
+		tmp123.s4 = tmp[4];
+
+		if(!valid) {
+			success = write_channel_nb_altera(chan_PRNG2GA_LS123_ushort_prng, tmp123);
+		}
+
+	} // End of while(active)
+}
 
 // --------------------------------------------------------------------------
 // --------------------------------------------------------------------------
@@ -303,5 +366,73 @@ void Krnl_Prng_LS3_float(unsigned int  Host_seed,
 		}
 	} // End of while(active)
 }
+
+__kernel __attribute__ ((max_global_work_dim(0)))
+void Krnl_Prng_LS4_float(unsigned int  Host_seed,
+			 unsigned char DockConst_num_of_genes){
+
+	uint lfsr = Host_seed;
+	bool valid = false;
+
+	while(!valid) {
+		bool active = true;
+		active  = read_channel_nb_altera(chan_Arbiter_LS4_float_off, &valid);
+	
+		for(uchar i=0; i<DockConst_num_of_genes; i++) {
+			float tmp;
+			uchar lsb;
+			lsb = lfsr & 0x01u;
+			lfsr >>= 1;
+			lfsr ^= (-lsb) & 0xA3000000u;
+			tmp = (0.999999f/MAX_UINT)*lfsr;
+			bool success = false;
+
+			#if defined (FIXED_POINT_LS2)
+			fixedpt fixpt_tmp = fixedpt_fromfloat(tmp);
+			if(!valid) {
+				success = write_channel_nb_altera(chan_PRNG2GA_LS4_float_prng, *(float*) &fixpt_tmp);
+			}
+			#else
+			if(!valid) {
+				success = write_channel_nb_altera(chan_PRNG2GA_LS4_float_prng, tmp);
+			}
+			#endif
+		}
+	} // End of while(active)
+}
+
+__kernel __attribute__ ((max_global_work_dim(0)))
+void Krnl_Prng_LS5_float(unsigned int  Host_seed,
+			 unsigned char DockConst_num_of_genes){
+
+	uint lfsr = Host_seed;
+	bool valid = false;
+
+	while(!valid) {
+		bool active = true;
+		active  = read_channel_nb_altera(chan_Arbiter_LS5_float_off, &valid);
+	
+		for(uchar i=0; i<DockConst_num_of_genes; i++) {
+			float tmp;
+			uchar lsb;
+			lsb = lfsr & 0x01u;
+			lfsr >>= 1;
+			lfsr ^= (-lsb) & 0xA3000000u;
+			tmp = (0.999999f/MAX_UINT)*lfsr;
+			bool success = false;
+
+			#if defined (FIXED_POINT_LS2)
+			fixedpt fixpt_tmp = fixedpt_fromfloat(tmp);
+			if(!valid) {
+				success = write_channel_nb_altera(chan_PRNG2GA_LS5_float_prng, *(float*) &fixpt_tmp);
+			}
+			#else
+			if(!valid) {
+				success = write_channel_nb_altera(chan_PRNG2GA_LS5_float_prng, tmp);
+			}
+			#endif
+		}
+	} // End of while(active)
+}
 // --------------------------------------------------------------------------
 // --------------------------------------------------------------------------     

ofdock_taskpar_alt/prof.md

@@ -526,7 +526,12 @@ NO HW TEST, THIS IS A CLEANUP COMMIT
 >>> commit "removed unneeded kernels files"
 
 
-. Expand from 3LS to 5LS blocks
+165. Expand from 3LS to 5LS blocks: `Conform`, `GA`, `IGL_Arbiter`, `InterE`, `IntraE`, `PRNG`, `performdocking`
+
+Both run in Sauron (non-instrumented freq=172MHz , intrumented freq=179 MHz)
+Speedup vs i5 cpu core: 3ptb: 59/34 = 1.73x, 1stp: 84/58 = 1.44x 
+(instrumented version runs faster: 59/32 = 1.84x, 84/56=1.5x)
+>>> commit "LS replicated 5 times"
 
 
 XXX, Between Conform and InterE, IntraE create a wider channel: