-diff --git a/libpcsxcore/new_dynarec/linkage_arm.S b/libpcsxcore/new_dynarec/linkage_arm.S
-index d32dc0b..e52dde8 100644
---- a/libpcsxcore/new_dynarec/linkage_arm.S
-+++ b/libpcsxcore/new_dynarec/linkage_arm.S
-@@ -442,7 +442,7 @@ 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
-@@ -530,7 +530,7 @@ 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,
-@@ -551,7 +551,7 @@ 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
-
-@@ -561,7 +561,7 @@ 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
-
-@@ -570,7 +570,7 @@ 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
-
-@@ -687,7 +687,7 @@ 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
-
-@@ -722,7 +722,7 @@ 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]
-@@ -750,7 +750,7 @@ 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/new_dynarec.c b/libpcsxcore/new_dynarec/new_dynarec.c
-index 6d7069d..586a6db 100644
+index 74f32ee3..4eec8a83 100644
--- a/libpcsxcore/new_dynarec/new_dynarec.c
+++ b/libpcsxcore/new_dynarec/new_dynarec.c
-@@ -38,10 +38,10 @@ static int sceBlock;
- #include "../psxhle.h" //emulator interface
- #include "emu_if.h" //emulator interface
+@@ -325,7 +325,7 @@ static struct compile_info
+ 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__
-@@ -362,6 +362,9 @@ static u_int get_vpage(u_int vaddr)
+ extern int cycle_count; // ... until end of the timeslice, counts -N -> 0 (CCREG)
+ extern int last_count; // last absolute target, often = next_interupt
+@@ -603,6 +603,7 @@ static int cycle_multiplier_active;
+
+ static int CLOCK_ADJUST(int x)
+ {
++ return x * 2;
+ int m = cycle_multiplier_active;
+ int s = (x >> 31) | 1;
+ return (x * m + s * 50) / 100;
+@@ -808,6 +809,9 @@ static noinline u_int generate_exception(u_int pc)
// This is called from the recompiled JR/JALR instructions
- void *get_addr(u_int vaddr)
+ static void noinline *get_addr(u_int vaddr, int can_compile)
{
+#ifdef DRC_DBG
+printf("get_addr %08x, pc=%08x\n", vaddr, psxRegs.pc);
+#endif
- u_int page=get_page(vaddr);
- u_int vpage=get_vpage(vaddr);
- struct ll_entry *head;
-@@ -4403,13 +4406,15 @@ 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=(int)out;
-- emit_jns(0);
-+// emit_jns(0);
-+emit_jmp(0);
- }
- else
- {
- emit_cmpimm(HOST_CCREG,-CLOCK_ADJUST(count+2));
- jaddr=(int)out;
-- emit_jns(0);
-+// emit_jns(0);
-+emit_jmp(0);
- }
- add_stub(CC_STUB,jaddr,idle?idle:(int)out,(*adj==0||invert||idle)?0:(count+2),i,addr,taken,0);
- }
-@@ -4884,7 +4889,8 @@ 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].is32,branch_regs[i].dirty,-1);
- #ifdef USE_MINI_HT
- if(rs1[i]==31) {
-@@ -5034,7 +5040,8 @@ void cjump_assemble(int i,struct regstat *i_regs)
- else if(nop) {
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),cc);
- int jaddr=(int)out;
-- emit_jns(0);
-+// emit_jns(0);
-+emit_jmp(0);
- add_stub(CC_STUB,jaddr,(int)out,0,i,start+i*4+8,NOTTAKEN,0);
- }
- else {
-@@ -5300,7 +5307,8 @@ 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);
- int jaddr=(int)out;
-- emit_jns(0);
-+// emit_jns(0);
-+emit_jmp(0);
- add_stub(CC_STUB,jaddr,(int)out,0,i,start+i*4+8,NOTTAKEN,0);
- emit_storereg(CCREG,HOST_CCREG);
- }
-@@ -5309,7 +5317,8 @@ void cjump_assemble(int i,struct regstat *i_regs)
- assert(cc==HOST_CCREG);
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),cc);
- int jaddr=(int)out;
-- emit_jns(0);
-+// emit_jns(0);
-+emit_jmp(0);
- add_stub(CC_STUB,jaddr,(int)out,0,i,start+i*4+8,likely[i]?NULLDS:NOTTAKEN,0);
- }
- }
-@@ -5419,7 +5428,8 @@ void sjump_assemble(int i,struct regstat *i_regs)
- else if(nevertaken) {
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),cc);
- int jaddr=(int)out;
-- emit_jns(0);
-+// emit_jns(0);
-+emit_jmp(0);
- add_stub(CC_STUB,jaddr,(int)out,0,i,start+i*4+8,NOTTAKEN,0);
- }
- else {
-@@ -5628,7 +5638,8 @@ 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);
- int jaddr=(int)out;
-- emit_jns(0);
-+// emit_jns(0);
-+emit_jmp(0);
- add_stub(CC_STUB,jaddr,(int)out,0,i,start+i*4+8,NOTTAKEN,0);
- emit_storereg(CCREG,HOST_CCREG);
- }
-@@ -5637,7 +5648,8 @@ void sjump_assemble(int i,struct regstat *i_regs)
- assert(cc==HOST_CCREG);
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),cc);
- int jaddr=(int)out;
-- emit_jns(0);
-+// emit_jns(0);
-+emit_jmp(0);
- add_stub(CC_STUB,jaddr,(int)out,0,i,start+i*4+8,likely[i]?NULLDS:NOTTAKEN,0);
- }
- }
-@@ -5833,7 +5845,8 @@ void fjump_assemble(int i,struct regstat *i_regs)
- emit_loadreg(CCREG,HOST_CCREG);
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),HOST_CCREG);
- int jaddr=(int)out;
-- emit_jns(0);
-+// emit_jns(0);
-+emit_jmp(0);
- add_stub(CC_STUB,jaddr,(int)out,0,i,start+i*4+8,NOTTAKEN,0);
- emit_storereg(CCREG,HOST_CCREG);
- }
-@@ -5842,7 +5855,8 @@ void fjump_assemble(int i,struct regstat *i_regs)
- assert(cc==HOST_CCREG);
- emit_addimm_and_set_flags(CLOCK_ADJUST(ccadj[i]+2),cc);
- int jaddr=(int)out;
-- emit_jns(0);
-+// emit_jns(0);
-+emit_jmp(0);
- add_stub(CC_STUB,jaddr,(int)out,0,i,start+i*4+8,likely[i]?NULLDS:NOTTAKEN,0);
- }
- }
-@@ -6463,7 +6477,7 @@ void unneeded_registers(int istart,int iend,int r)
+ u_int start_page = get_page_prev(vaddr);
+ u_int i, page, end_page = get_page(vaddr);
+ void *found_clean = NULL;
+@@ -7213,7 +7217,7 @@ static noinline void pass2_unneeded_regs(int istart,int iend,int r)
// R0 is always unneeded
- u|=1;uu|=1;
+ u|=1;
// Save it
- unneeded_reg[i]=u;
+ unneeded_reg[i]=1;//u;
- unneeded_reg_upper[i]=uu;
gte_unneeded[i]=gte_u;
/*
-@@ -9676,6 +9690,7 @@ int new_recompile_block(int addr)
-
- // This allocates registers (if possible) one instruction prior
- // to use, which can avoid a load-use penalty on certain CPUs.
-+#if 0
+ printf("ur (%d,%d) %x: ",istart,iend,start+i*4);
+@@ -8355,6 +8359,7 @@ static noinline void pass5a_preallocate1(void)
+ static noinline void pass5b_preallocate2(void)
+ {
+ int i, hr;
++ return;
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&&itype[i-1]!=FJUMP))
-@@ -9832,6 +9847,7 @@ int new_recompile_block(int addr)
- }
- }
- }
-+#endif
+ if (!i || !dops[i-1].is_jump)
+@@ -9380,6 +9385,10 @@ static int new_recompile_block(u_int addr)
- /* Pass 6 - Optimize clean/dirty state */
- clean_registers(0,slen-1,1);
-@@ -10149,6 +10165,11 @@ int new_recompile_block(int addr)
- case SPAN:
- pagespan_assemble(i,®s[i]);break;
- }
-+
-+#ifdef DRC_DBG
-+ if(itype[i]!=UJUMP&&itype[i]!=CJUMP&&itype[i]!=SJUMP&&itype[i]!=RJUMP)
-+ wb_dirtys(regs[i].regmap,regs[i].is32,regs[i].dirty);
+ #ifdef ASSEM_PRINT
+ fflush(stdout);
+#endif
- if(itype[i]==UJUMP||itype[i]==RJUMP||(source[i]>>16)==0x1000)
- literal_pool(1024);
- else
-@@ -10256,7 +10277,7 @@ 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)
-@@ -10370,6 +10391,10 @@ int new_recompile_block(int addr)
- }
- expirep=(expirep+1)&65535;
- }
+#ifdef DRC_DBG
+printf("new_recompile_block done\n");
+fflush(stdout);
-+#endif
+ #endif
+ stat_inc(stat_bc_direct);
return 0;
+diff --git a/libpcsxcore/new_dynarec/pcsxmem.c b/libpcsxcore/new_dynarec/pcsxmem.c
+index f4b1d90e..d3975ceb 100644
+--- a/libpcsxcore/new_dynarec/pcsxmem.c
++++ b/libpcsxcore/new_dynarec/pcsxmem.c
+@@ -258,6 +258,8 @@ static void write_biu(u32 value)
+ return;
+ }
+
++extern u32 handler_cycle;
++handler_cycle = psxRegs.cycle;
+ memprintf("write_biu %08x @%08x %u\n", value, psxRegs.pc, psxRegs.cycle);
+ psxRegs.biuReg = value;
+ }
+diff --git a/libpcsxcore/psxcounters.c b/libpcsxcore/psxcounters.c
+index d0d45ec5..4ed03c40 100644
+--- a/libpcsxcore/psxcounters.c
++++ b/libpcsxcore/psxcounters.c
+@@ -428,9 +428,12 @@ 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();
+@@ -439,6 +442,7 @@ 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 );
+@@ -450,6 +454,7 @@ 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;
+
+@@ -463,6 +468,7 @@ u32 psxRcntRcount0()
+ {
+ u32 index = 0;
+ u32 count;
++handler_cycle = psxRegs.cycle;
+
+ if ((rcnts[0].mode & 7) == (RcSyncModeEnable | Rc01UnblankReset) ||
+ (rcnts[0].mode & 7) == (RcSyncModeEnable | Rc01UnblankReset2))
+diff --git a/libpcsxcore/psxinterpreter.c b/libpcsxcore/psxinterpreter.c
+index f6ff2e8b..2f7147c3 100644
+--- a/libpcsxcore/psxinterpreter.c
++++ b/libpcsxcore/psxinterpreter.c
+@@ -245,7 +245,7 @@ static inline void addCycle(psxRegisters *regs)
+ {
+ assert(regs->subCycleStep >= 0x10000);
+ regs->subCycle += regs->subCycleStep;
+- regs->cycle += regs->subCycle >> 16;
++ regs->cycle += 2; //regs->subCycle >> 16;
+ regs->subCycle &= 0xffff;
+ }
+
+@@ -1348,8 +1348,15 @@ static void intShutdown() {
+ // single step (may do several ops in case of a branch or load delay)
+ // called by asm/dynarec
+ void execI(psxRegisters *regs) {
++ extern int last_count;
++ extern u32 next_interupt;
++ void do_insn_cmp(void);
++ printf("execI %08x c %u, ni %u\n", regs->pc, regs->cycle, next_interupt);
++ last_count = 0;
+ do {
+ execIbp(psxMemRLUT, regs);
++ if (regs->dloadReg[0] || regs->dloadReg[1])
++ do_insn_cmp();
+ } while (regs->dloadReg[0] || regs->dloadReg[1]);
}