diff --git a/README.txt b/README.txt
index d4af1d1..be5e293 100644
--- a/README.txt
+++ b/README.txt
@@ -4,7 +4,7 @@ So don't use it.
 
 Example programs:
 - return.bin: returns -1094647826 (puts it to register 10)
-- return2.bin: returns a different value with more steps
+- return2.bin: returns a different value with more steps. if it returns DADD, something's wrong
 
 Emulator exit codes:
 - <0 = OK
diff --git a/src/cpu.c b/src/cpu.c
index f8714a1..a6fb4e9 100644
--- a/src/cpu.c
+++ b/src/cpu.c
@@ -23,6 +23,7 @@ typedef struct CPU_s CPU;
 CPU create_cpu(AddressSpace *addressSpace) {
     CPU cpu = cpu_default;
     cpu.addressSpace = addressSpace;
+    cpu.registers[2] = addressSpace->romSize + addressSpace->ramSize - 1; // stack pointer
     
     return cpu;
 }
diff --git a/src/instruction_executor.c b/src/instruction_executor.c
index f9538c7..6a81984 100644
--- a/src/instruction_executor.c
+++ b/src/instruction_executor.c
@@ -56,14 +56,14 @@ int execute_instruction_on_cpu(CPU *cpu, uint32_t instruction) { // TODO conside
                 printf("BLT: x%d %d < x%d %d?", rs1, registers[rs1], rs2, registers[rs2]);
                 branch = (int32_t)registers[rs1] < (int32_t)registers[rs2];
             } else if (funct3 == 0b101) { // BGE
-                printf("BGE: x%d %d > x%d %d?", rs1, registers[rs1], rs2, registers[rs2]);
-                branch = (int32_t)registers[rs1] > (int32_t)registers[rs2];
+                printf("BGE: x%d %d >= x%d %d?", rs1, registers[rs1], rs2, registers[rs2]);
+                branch = (int32_t)registers[rs1] >= (int32_t)registers[rs2];
             } else if (funct3 == 0b110) { // BLTU
                 printf("BLTU: x%d %u < x%d %u?", rs1, registers[rs1], rs2, registers[rs2]);
                 branch = registers[rs1] < registers[rs2];
             } else if (funct3 == 0b111) { // BGEU
-                printf("BGEU: x%d %u > x%d %u?", rs1, registers[rs1], rs2, registers[rs2]);
-                branch = registers[rs1] > registers[rs2];
+                printf("BGEU: x%d %u >= x%d %u?", rs1, registers[rs1], rs2, registers[rs2]);
+                branch = registers[rs1] >= registers[rs2];
             } else {
                 // TODO illegal instruction
             }
@@ -83,7 +83,7 @@ int execute_instruction_on_cpu(CPU *cpu, uint32_t instruction) { // TODO conside
 
             switch (funct3) {
                 case 0b000: // ADDI
-                    printf("ADDI - x%u %d + %d = %d -> x%u", rs1, registers[rs1], imm, registers[rs1] + imm, rd);
+                    printf("ADDI - x%u 0x%X (%d) + 0x%X (%d) = 0x%X -> x%u", rs1, registers[rs1], registers[rs1], imm, imm, registers[rs1] + imm, rd);
                     registers[rd] = registers[rs1] + imm;
                     break;
                 default:
@@ -128,6 +128,7 @@ int execute_instruction_on_cpu(CPU *cpu, uint32_t instruction) { // TODO conside
             uint8_t rs1 = instruction >> 15 & 0x1F;
             uint8_t rs2 = instruction >> 20 & 0x1F;
             int32_t imm = ((int32_t)instruction >> 20 & 0xffffffe0) | rd;
+            uint32_t og = registers[rs1]; // for log only
             uint32_t addr = registers[rs1] + imm;
 
             if (funct3 == 0b000) { // SB
@@ -135,18 +136,19 @@ int execute_instruction_on_cpu(CPU *cpu, uint32_t instruction) { // TODO conside
                 uint8_t val = registers[rs2] & 0xFF;
                 write_address_space(addressSpace, addr, 1, &val);
 
-                printf("SB: x%d %u -> %u", rs2, val, addr);
+                printf("SB (8bit): x%d 0x%X (%u) -> 0x%X + %d = 0x%X", rs2, rs2, val, og, imm, addr);
             } else if (funct3 == 0b001) { // SH
                 // SH stores a 16-bit value from the low bits of register rs2 to memory.
-                uint8_t val = registers[rs2] & 0xFFFF;
+                uint16_t val = registers[rs2] & 0xFFFF;
                 write_address_space(addressSpace, addr, 2, &val);
 
-                printf("SH: x%d %u -> %u", rs2, val, addr);
+                printf("SH (16bit): x%d 0x%X (%u) -> 0x%X + %d = 0x%X", rs2, rs2, val, og, imm, addr);
             } else if (funct3 == 0b010) { // SW
                 // SW stores a 32-bit value from the low bits of register rs2 to memory.
-                write_address_space(addressSpace, addr, 4, registers + rs2);
+                uint32_t val = registers[rs2];
+                write_address_space(addressSpace, addr, 4, &val);
 
-                printf("SW: x%d %u -> %u", rs2, registers[rs2], addr);
+                printf("SW (32bit): x%d 0x%X (%u) -> 0x%X + %d = 0x%X", rs2, rs2, val, og, imm, addr);
             } else {
                 // TODO illegal funct3
             }