-diff --git a/libpcsxcore/new_dynarec/linkage_arm.S b/libpcsxcore/new_dynarec/linkage_arm.S
-index fcb4e1a..e2547c4 100644
---- a/libpcsxcore/new_dynarec/linkage_arm.S
-+++ b/libpcsxcore/new_dynarec/linkage_arm.S
-@@ -439,7 +439,7 @@ FUNCTION(cc_interrupt):
- str r1, [fp, #LO_pending_exception]
- and r2, r2, r10, lsr #17
- add r3, fp, #LO_restore_candidate
-- str r10, [fp, #LO_cycle] /* PCSX cycles */
-+@@@ str r10, [fp, #LO_cycle] /* PCSX cycles */
- @@ str r10, [fp, #LO_reg_cop0+36] /* Count */
- ldr r4, [r2, r3]
- mov r10, lr
-@@ -519,7 +519,7 @@ FUNCTION(jump_syscall_hle):
- mov r1, #0 /* in delay slot */
- add r2, r2, r10
- mov r0, #0x20 /* cause */
-- str r2, [fp, #LO_cycle] /* PCSX cycle counter */
-+@@@ str r2, [fp, #LO_cycle] /* PCSX cycle counter */
- bl psxException
-
- /* note: psxException might do recursive recompiler call from it's HLE code,
-@@ -540,7 +540,7 @@ FUNCTION(jump_hlecall):
- str r0, [fp, #LO_pcaddr]
- add r2, r2, r10
- adr lr, pcsx_return
-- str r2, [fp, #LO_cycle] /* PCSX cycle counter */
-+@@@ str r2, [fp, #LO_cycle] /* PCSX cycle counter */
- bx r1
- .size jump_hlecall, .-jump_hlecall
-
-@@ -550,7 +550,7 @@ FUNCTION(jump_intcall):
- str r0, [fp, #LO_pcaddr]
- add r2, r2, r10
- adr lr, pcsx_return
-- str r2, [fp, #LO_cycle] /* PCSX cycle counter */
-+@@@ str r2, [fp, #LO_cycle] /* PCSX cycle counter */
- b execI
- .size jump_hlecall, .-jump_hlecall
-
-@@ -559,7 +559,7 @@ FUNCTION(new_dyna_leave):
- ldr r0, [fp, #LO_last_count]
- add r12, fp, #28
- add r10, r0, r10
-- str r10, [fp, #LO_cycle]
-+@@@ str r10, [fp, #LO_cycle]
- ldmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, ip, pc}
- .size new_dyna_leave, .-new_dyna_leave
-
-@@ -676,7 +676,7 @@ FUNCTION(new_dyna_start):
- \readop r0, [r1, r3, lsl #\tab_shift]
- .endif
- movcc pc, lr
-- str r2, [fp, #LO_cycle]
-+@@@ str r2, [fp, #LO_cycle]
- bx r1
- .endm
-
-@@ -711,7 +711,7 @@ FUNCTION(jump_handler_read32):
- mov r0, r1
- add r2, r2, r12
- push {r2, lr}
-- str r2, [fp, #LO_cycle]
-+@@@ str r2, [fp, #LO_cycle]
- blx r3
-
- ldr r0, [fp, #LO_next_interupt]
-@@ -739,7 +739,7 @@ FUNCTION(jump_handler_write_h):
- add r2, r2, r12
- mov r0, r1
- push {r2, lr}
-- str r2, [fp, #LO_cycle]
-+@@@ str r2, [fp, #LO_cycle]
- blx r3
-
- ldr r0, [fp, #LO_next_interupt]
-diff --git a/libpcsxcore/new_dynarec/linkage_arm64.S b/libpcsxcore/new_dynarec/linkage_arm64.S
-index 060ac48..e3007d3 100644
---- a/libpcsxcore/new_dynarec/linkage_arm64.S
-+++ b/libpcsxcore/new_dynarec/linkage_arm64.S
-@@ -131,7 +131,7 @@ FUNCTION(cc_interrupt):
- str wzr, [rFP, #LO_pending_exception]
- and w2, w2, rCC, lsr #17
- add x3, rFP, #LO_restore_candidate
-- str rCC, [rFP, #LO_cycle] /* PCSX cycles */
-+## str rCC, [rFP, #LO_cycle] /* PCSX cycles */
- # str rCC, [rFP, #LO_reg_cop0+36] /* Count */
- ldr w19, [x3, w2, uxtw]
- mov x21, lr
-@@ -254,7 +254,7 @@ FUNCTION(new_dyna_start):
- FUNCTION(new_dyna_leave):
- ldr w0, [rFP, #LO_last_count]
- add rCC, rCC, w0
-- str rCC, [rFP, #LO_cycle]
-+## str rCC, [rFP, #LO_cycle]
- ldp x19, x20, [sp, #16*1]
- ldp x21, x22, [sp, #16*2]
- ldp x23, x24, [sp, #16*3]
-@@ -272,7 +272,7 @@ FUNCTION(new_dyna_leave):
- /* w0 = adddr/data, x1 = rhandler, w2 = cycles, x3 = whandler */
- ldr w4, [rFP, #LO_last_count]
- add w4, w4, w2
-- str w4, [rFP, #LO_cycle]
-+## str w4, [rFP, #LO_cycle]
- .endm
-
- .macro memhandler_post
diff --git a/libpcsxcore/new_dynarec/new_dynarec.c b/libpcsxcore/new_dynarec/new_dynarec.c
-index e7b55b6..caa06d0 100644
+index f1005db..ebd1d4f 100644
--- a/libpcsxcore/new_dynarec/new_dynarec.c
+++ b/libpcsxcore/new_dynarec/new_dynarec.c
-@@ -43,10 +43,10 @@ static int sceBlock;
- #define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
- #endif
+@@ -235,7 +235,7 @@ static struct decoded_insn
+ int new_dynarec_hacks_old;
+ int new_dynarec_did_compile;
--//#define DISASM
--//#define assem_debug printf
-+#define DISASM
-+#define assem_debug printf
- //#define inv_debug printf
--#define assem_debug(...)
-+//#define assem_debug(...)
- #define inv_debug(...)
+- #define HACK_ENABLED(x) ((new_dynarec_hacks | new_dynarec_hacks_pergame) & (x))
++ #define HACK_ENABLED(x) ((NDHACK_NO_STALLS) & (x))
- #ifdef __i386__
-@@ -424,6 +424,9 @@ static int doesnt_expire_soon(void *tcaddr)
+ extern int cycle_count; // ... until end of the timeslice, counts -N -> 0
+ extern int last_count; // last absolute target, often = next_interupt
+@@ -471,6 +471,7 @@ int cycle_multiplier_old;
+
+ static int CLOCK_ADJUST(int x)
+ {
++ return x * 2;
+ int m = cycle_multiplier_override && cycle_multiplier == CYCLE_MULT_DEFAULT
+ ? cycle_multiplier_override : cycle_multiplier;
+ int s=(x>>31)|1;
+@@ -522,6 +523,9 @@ static int doesnt_expire_soon(void *tcaddr)
// This is called from the recompiled JR/JALR instructions
void noinline *get_addr(u_int vaddr)
{
u_int page=get_page(vaddr);
u_int vpage=get_vpage(vaddr);
struct ll_entry *head;
-@@ -4221,13 +4224,15 @@ void do_cc(int i,signed char i_regmap[],int *adj,int addr,int taken,int invert)
- }
- emit_addimm_and_set_flags(cycles,HOST_CCREG);
- jaddr=out;
-- emit_jns(0);
-+ //emit_jns(0);
-+ emit_jmp(0);
- }
- else
- {
- emit_cmpimm(HOST_CCREG,-CLOCK_ADJUST(count+2));
- jaddr=out;
-- emit_jns(0);
-+ //emit_jns(0);
-+ emit_jmp(0);
- }
- add_stub(CC_STUB,jaddr,idle?idle:out,(*adj==0||invert||idle)?0:(count+2),i,addr,taken,0);
- }
-@@ -4635,7 +4640,8 @@ static void rjump_assemble(int i,struct regstat *i_regs)
- // special case for RFE
- emit_jmp(0);
- else
-- emit_jns(0);
-+ //emit_jns(0);
-+ emit_jmp(0);
- //load_regs_bt(branch_regs[i].regmap,branch_regs[i].dirty,-1);
- #ifdef USE_MINI_HT
- if(rs1[i]==31) {
-@@ -4737,7 +4743,8 @@ static void cjump_assemble(int i,struct regstat *i_regs)
- else if(nop) {
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),cc);
- void *jaddr=out;
-- emit_jns(0);
-+ //emit_jns(0);
-+ emit_jmp(0);
- add_stub(CC_STUB,jaddr,out,0,i,start+i*4+8,NOTTAKEN,0);
- }
- else {
-@@ -4924,7 +4931,8 @@ static void cjump_assemble(int i,struct regstat *i_regs)
- emit_loadreg(CCREG,HOST_CCREG);
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),HOST_CCREG);
- void *jaddr=out;
-- emit_jns(0);
-+ //emit_jns(0);
-+ emit_jmp(0);
- add_stub(CC_STUB,jaddr,out,0,i,start+i*4+8,NOTTAKEN,0);
- emit_storereg(CCREG,HOST_CCREG);
- }
-@@ -4933,7 +4941,8 @@ static void cjump_assemble(int i,struct regstat *i_regs)
- assert(cc==HOST_CCREG);
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),cc);
- void *jaddr=out;
-- emit_jns(0);
-+ //emit_jns(0);
-+ emit_jmp(0);
- add_stub(CC_STUB,jaddr,out,0,i,start+i*4+8,likely[i]?NULLDS:NOTTAKEN,0);
- }
- }
-@@ -5032,7 +5041,8 @@ static void sjump_assemble(int i,struct regstat *i_regs)
- else if(nevertaken) {
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),cc);
- void *jaddr=out;
-- emit_jns(0);
-+ //emit_jns(0);
-+ emit_jmp(0);
- add_stub(CC_STUB,jaddr,out,0,i,start+i*4+8,NOTTAKEN,0);
- }
- else {
-@@ -5188,7 +5198,8 @@ static void sjump_assemble(int i,struct regstat *i_regs)
- emit_loadreg(CCREG,HOST_CCREG);
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),HOST_CCREG);
- void *jaddr=out;
-- emit_jns(0);
-+ //emit_jns(0);
-+ emit_jmp(0);
- add_stub(CC_STUB,jaddr,out,0,i,start+i*4+8,NOTTAKEN,0);
- emit_storereg(CCREG,HOST_CCREG);
- }
-@@ -5197,7 +5208,8 @@ static void sjump_assemble(int i,struct regstat *i_regs)
- assert(cc==HOST_CCREG);
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),cc);
- void *jaddr=out;
-- emit_jns(0);
-+ //emit_jns(0);
-+ emit_jmp(0);
- add_stub(CC_STUB,jaddr,out,0,i,start+i*4+8,likely[i]?NULLDS:NOTTAKEN,0);
- }
- }
-@@ -5685,7 +5697,7 @@ void unneeded_registers(int istart,int iend,int r)
+@@ -6248,7 +6252,7 @@ void unneeded_registers(int istart,int iend,int r)
// R0 is always unneeded
u|=1;
// Save it
gte_unneeded[i]=gte_u;
/*
printf("ur (%d,%d) %x: ",istart,iend,start+i*4);
-@@ -8209,6 +8221,7 @@ int new_recompile_block(int addr)
+@@ -8794,6 +8798,7 @@ int new_recompile_block(u_int addr)
// This allocates registers (if possible) one instruction prior
// to use, which can avoid a load-use penalty on certain CPUs.
+#if 0
for(i=0;i<slen-1;i++)
{
- if(!i||(itype[i-1]!=UJUMP&&itype[i-1]!=CJUMP&&itype[i-1]!=SJUMP&&itype[i-1]!=RJUMP))
-@@ -8365,6 +8378,7 @@ int new_recompile_block(int addr)
+ if (!i || !dops[i-1].is_jump)
+@@ -8950,6 +8955,7 @@ int new_recompile_block(u_int addr)
}
}
}
/* Pass 6 - Optimize clean/dirty state */
clean_registers(0,slen-1,1);
-@@ -8659,6 +8673,11 @@ int new_recompile_block(int addr)
- case SPAN:
- pagespan_assemble(i,®s[i]);break;
- }
-+
+@@ -9204,6 +9210,11 @@ int new_recompile_block(u_int addr)
+ load_regs(regs[i].regmap_entry,regs[i].regmap,INVCP,INVCP);
+
+ ds = assemble(i, ®s[i], ccadj[i]);
+#ifdef DRC_DBG
-+ if(itype[i]!=UJUMP&&itype[i]!=CJUMP&&itype[i]!=SJUMP&&itype[i]!=RJUMP)
-+ wb_dirtys(regs[i].regmap,regs[i].dirty);
++ // write-out non-consts, consts are likely different because of get_final_value()
++ if (!dops[i].is_jump)
++ wb_dirtys(regs[i].regmap,regs[i].dirty&~regs[i].loadedconst);
+#endif
- if(itype[i]==UJUMP||itype[i]==RJUMP||(source[i]>>16)==0x1000)
+
+ if (dops[i].is_ujump)
literal_pool(1024);
- else
-@@ -8767,7 +8786,7 @@ int new_recompile_block(int addr)
- }
- }
- // External Branch Targets (jump_in)
-- if(copy+slen*4>(void *)shadow+sizeof(shadow)) copy=shadow;
-+ if(copy+slen*4>(void *)shadow+sizeof(shadow)) {copy=shadow;printf("shadow overflow\n");}
- for(i=0;i<slen;i++)
- {
- if(bt[i]||i==0)
-@@ -8882,6 +8901,10 @@ int new_recompile_block(int addr)
- }
- expirep=(expirep+1)&65535;
+@@ -9439,6 +9450,10 @@ int new_recompile_block(u_int addr)
}
+ #ifdef ASSEM_PRINT
+ fflush(stdout);
++#endif
+#ifdef DRC_DBG
+printf("new_recompile_block done\n");
+fflush(stdout);
-+#endif
+ #endif
return 0;
}
+diff --git a/libpcsxcore/new_dynarec/pcsxmem.c b/libpcsxcore/new_dynarec/pcsxmem.c
+index bb471b6..8f68a3b 100644
+--- a/libpcsxcore/new_dynarec/pcsxmem.c
++++ b/libpcsxcore/new_dynarec/pcsxmem.c
+@@ -272,6 +272,8 @@ static void write_biu(u32 value)
+ if (address != 0xfffe0130)
+ return;
+
++extern u32 handler_cycle;
++handler_cycle = psxRegs.cycle;
+ switch (value) {
+ case 0x800: case 0x804:
+ unmap_ram_write();
+diff --git a/libpcsxcore/psxcounters.c b/libpcsxcore/psxcounters.c
+index b2cc07b..f916580 100644
+--- a/libpcsxcore/psxcounters.c
++++ b/libpcsxcore/psxcounters.c
+@@ -378,9 +378,12 @@ void psxRcntUpdate()
+
+ /******************************************************************************/
+
++extern u32 handler_cycle;
++
+ void psxRcntWcount( u32 index, u32 value )
+ {
+ verboseLog( 2, "[RCNT %i] wcount: %x\n", index, value );
++handler_cycle = psxRegs.cycle;
+
+ _psxRcntWcount( index, value );
+ psxRcntSet();
+@@ -389,6 +392,7 @@ void psxRcntWcount( u32 index, u32 value )
+ void psxRcntWmode( u32 index, u32 value )
+ {
+ verboseLog( 1, "[RCNT %i] wmode: %x\n", index, value );
++handler_cycle = psxRegs.cycle;
+
+ _psxRcntWmode( index, value );
+ _psxRcntWcount( index, 0 );
+@@ -400,6 +404,7 @@ void psxRcntWmode( u32 index, u32 value )
+ void psxRcntWtarget( u32 index, u32 value )
+ {
+ verboseLog( 1, "[RCNT %i] wtarget: %x\n", index, value );
++handler_cycle = psxRegs.cycle;
+
+ rcnts[index].target = value;
+
+@@ -412,6 +417,7 @@ void psxRcntWtarget( u32 index, u32 value )
+ u32 psxRcntRcount( u32 index )
+ {
+ u32 count;
++handler_cycle = psxRegs.cycle;
+
+ count = _psxRcntRcount( index );