diff --git a/libpcsxcore/new_dynarec/new_dynarec.c b/libpcsxcore/new_dynarec/new_dynarec.c
-index f1005db..ebd1d4f 100644
+index 300a84c8..e4343533 100644
--- a/libpcsxcore/new_dynarec/new_dynarec.c
+++ b/libpcsxcore/new_dynarec/new_dynarec.c
-@@ -235,7 +235,7 @@ static struct decoded_insn
- int new_dynarec_hacks_old;
- int new_dynarec_did_compile;
+@@ -345,7 +345,7 @@ static struct compile_info
+ #define stat_clear(s)
+ #endif
-- #define HACK_ENABLED(x) ((new_dynarec_hacks | new_dynarec_hacks_pergame) & (x))
-+ #define HACK_ENABLED(x) ((NDHACK_NO_STALLS) & (x))
+- #define HACK_ENABLED(x) ((ndrc_g.hacks | ndrc_g.hacks_pergame) & (x))
++ #define HACK_ENABLED(x) ((NDHACK_NO_STALLS|NDHACK_NO_COMPAT_HACKS) & (x))
- 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;
+ /* registers that may be allocated */
+ /* 1-31 gpr */
+@@ -626,6 +626,7 @@ static int cycle_multiplier_active;
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)
+ int m = cycle_multiplier_active;
+ int s = (x >> 31) | 1;
+ return (x * m + s * 50) / 100;
+@@ -837,6 +838,9 @@ static noinline u_int generate_exception(u_int pc)
+ static void noinline *get_addr(struct ht_entry *ht, const u_int vaddr,
+ enum ndrc_compile_mode compile_mode)
{
+#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;
-@@ -6248,7 +6252,7 @@ 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;
+@@ -7421,7 +7425,7 @@ static noinline void pass2b_unneeded_regs(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);
-@@ -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++)
+@@ -8574,6 +8578,7 @@ static noinline void pass5a_preallocate1(void)
+ // to use, which can avoid a load-use penalty on certain CPUs.
+ static noinline void pass5b_preallocate2(void)
+ {
++ return;
+ int i, hr, limit = min(slen - 1, MAXBLOCK - 2);
+ for (i = 0; i < limit; i++)
{
- if (!i || !dops[i-1].is_jump)
-@@ -8950,6 +8955,7 @@ int new_recompile_block(u_int addr)
- }
- }
- }
-+#endif
-
- /* Pass 6 - Optimize clean/dirty state */
- clean_registers(0,slen-1,1);
-@@ -9204,6 +9210,11 @@ int new_recompile_block(u_int addr)
- load_regs(regs[i].regmap_entry,regs[i].regmap,INVCP,INVCP);
+@@ -9602,6 +9607,10 @@ static int noinline new_recompile_block(u_int addr)
- ds = assemble(i, ®s[i], ccadj[i]);
-+#ifdef DRC_DBG
-+ // 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 (dops[i].is_ujump)
- literal_pool(1024);
-@@ -9439,6 +9450,10 @@ int new_recompile_block(u_int addr)
- }
#ifdef ASSEM_PRINT
fflush(stdout);
+#endif
+printf("new_recompile_block done\n");
+fflush(stdout);
#endif
+ stat_inc(stat_bc_direct);
return 0;
- }
diff --git a/libpcsxcore/new_dynarec/pcsxmem.c b/libpcsxcore/new_dynarec/pcsxmem.c
-index bb471b6..8f68a3b 100644
+index 98e2c6be..edba031e 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)
+@@ -238,6 +238,8 @@ static void write_biu(u32 value)
return;
+ }
+extern u32 handler_cycle;
+handler_cycle = psxRegs.cycle;
- switch (value) {
- case 0x800: case 0x804:
- unmap_ram_write();
+ 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 b2cc07b..f916580 100644
+index 064c06b6..07e2afb5 100644
--- a/libpcsxcore/psxcounters.c
+++ b/libpcsxcore/psxcounters.c
-@@ -378,9 +378,12 @@ void psxRcntUpdate()
+@@ -455,9 +455,12 @@ void psxRcntUpdate()
/******************************************************************************/
_psxRcntWcount( index, value );
psxRcntSet();
-@@ -389,6 +392,7 @@ void psxRcntWcount( u32 index, u32 value )
+@@ -466,6 +469,7 @@ void psxRcntWcount( u32 index, u32 value )
void psxRcntWmode( u32 index, u32 value )
{
verboseLog( 1, "[RCNT %i] wmode: %x\n", index, value );
_psxRcntWmode( index, value );
_psxRcntWcount( index, 0 );
-@@ -400,6 +404,7 @@ void psxRcntWmode( u32 index, u32 value )
+@@ -477,6 +481,7 @@ void psxRcntWmode( u32 index, u32 value )
void psxRcntWtarget( u32 index, u32 value )
{
verboseLog( 1, "[RCNT %i] wtarget: %x\n", index, value );
rcnts[index].target = value;
-@@ -412,6 +417,7 @@ void psxRcntWtarget( u32 index, u32 value )
- u32 psxRcntRcount( u32 index )
+@@ -490,6 +495,7 @@ u32 psxRcntRcount0()
{
+ u32 index = 0;
u32 count;
+handler_cycle = psxRegs.cycle;
- count = _psxRcntRcount( index );
+ if ((rcnts[0].mode & 7) == (RcSyncModeEnable | Rc01UnblankReset) ||
+ (rcnts[0].mode & 7) == (RcSyncModeEnable | Rc01UnblankReset2))
+diff --git a/libpcsxcore/psxevents.c b/libpcsxcore/psxevents.c
+index 1e2d01f6..0ee15974 100644
+--- a/libpcsxcore/psxevents.c
++++ b/libpcsxcore/psxevents.c
+@@ -77,11 +77,13 @@ void irq_test(psxCP0Regs *cp0)
+ }
+ }
+
+- cp0->n.Cause &= ~0x400;
++ u32 c2 = cp0->n.Cause & ~0x400;
+ if (psxHu32(0x1070) & psxHu32(0x1074))
+- cp0->n.Cause |= 0x400;
+- if (((cp0->n.Cause | 1) & cp0->n.SR & 0x401) == 0x401)
++ c2 |= 0x400;
++ if (((c2 | 1) & cp0->n.SR & 0x401) == 0x401) {
++ cp0->n.Cause = c2;
+ psxException(0, 0, cp0);
++ }
+ }
+
+ void gen_interupt(psxCP0Regs *cp0)
+diff --git a/libpcsxcore/psxinterpreter.c b/libpcsxcore/psxinterpreter.c
+index 68d79321..50a38f8d 100644
+--- a/libpcsxcore/psxinterpreter.c
++++ b/libpcsxcore/psxinterpreter.c
+@@ -243,7 +243,7 @@ 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;
+ }
+
+@@ -440,7 +440,9 @@ static void doBranch(psxRegisters *regs, u32 tar, enum R3000Abdt taken) {
+ regs->CP0.n.Target = pc_final;
+ regs->branching = 0;
+
++ psxRegs.cycle += 2;
+ psxBranchTest();
++ psxRegs.cycle -= 2;
+ }
+
+ static void doBranchReg(psxRegisters *regs, u32 tar) {
+@@ -973,7 +975,7 @@ void MTC0(psxRegisters *regs_, int reg, u32 val) {
+ }
+ }
+
+-OP(psxMTC0) { MTC0(regs_, _Rd_, _u32(_rRt_)); }
++OP(psxMTC0) { MTC0(regs_, _Rd_, _u32(_rRt_)); psxBranchTest(); }
+
+ // no exception
+ static inline void psxNULLne(psxRegisters *regs) {
+@@ -1132,6 +1134,7 @@ OP(psxHLE) {
+ dloadFlush(regs_);
+ psxHLEt[hleCode]();
+ regs_->branchSeen = 1;
++ regs_->cycle -= 2;
+ }
+
+ static void (INT_ATTR *psxBSC[64])(psxRegisters *regs_, u32 code) = {
+@@ -1182,18 +1185,20 @@ static void intReset() {
+ static inline void execI_(u8 **memRLUT, psxRegisters *regs) {
+ u32 pc = regs->pc;
+
+- addCycle(regs);
++ //addCycle(regs);
+ dloadStep(regs);
+
+ regs->pc += 4;
+ regs->code = fetch(regs, memRLUT, pc);
+ psxBSC[regs->code >> 26](regs, regs->code);
++ psxRegs.cycle += 2;
++ fetchNoCache(regs, memRLUT, regs->pc); // bus err check
+ }
+
+ static inline void execIbp(u8 **memRLUT, psxRegisters *regs) {
+ u32 pc = regs->pc;
+
+- addCycle(regs);
++ //addCycle(regs);
+ dloadStep(regs);
+
+ if (execBreakCheck(regs, pc))
+@@ -1202,6 +1207,8 @@ static inline void execIbp(u8 **memRLUT, psxRegisters *regs) {
+ regs->pc += 4;
+ regs->code = fetch(regs, memRLUT, pc);
+ psxBSC[regs->code >> 26](regs, regs->code);
++ psxRegs.cycle += 2;
++ fetchNoCache(regs, memRLUT, regs->pc); // bus err check
+ }
+
+ static void intExecute(psxRegisters *regs) {
+@@ -1218,20 +1225,28 @@ static void intExecuteBp(psxRegisters *regs) {
+ execIbp(memRLUT, regs);
+ }
+
++ extern int last_count;
++ void do_insn_cmp(void);
+ static void intExecuteBlock(psxRegisters *regs, enum blockExecCaller caller) {
+ u8 **memRLUT = psxMemRLUT;
+
++ last_count = 0;
+ regs->branchSeen = 0;
+- while (!regs->branchSeen)
++ while (!regs->branchSeen || (regs->dloadReg[0] || regs->dloadReg[1])) {
++ do_insn_cmp();
+ execI_(memRLUT, regs);
++ }
+ }
+
+ static void intExecuteBlockBp(psxRegisters *regs, enum blockExecCaller caller) {
+ u8 **memRLUT = psxMemRLUT;
+
++ last_count = 0;
+ regs->branchSeen = 0;
+- while (!regs->branchSeen)
++ while (!regs->branchSeen || (regs->dloadReg[0] || regs->dloadReg[1])) {
++ do_insn_cmp();
+ execIbp(memRLUT, regs);
++ }
+ }
+
+ static void intClear(u32 Addr, u32 Size) {
+@@ -1263,7 +1278,7 @@ static void setupCop(u32 sr)
+ else
+ psxBSC[17] = psxCOPd;
+ if (sr & (1u << 30))
+- psxBSC[18] = Config.DisableStalls ? psxCOP2 : psxCOP2_stall;
++ psxBSC[18] = psxCOP2;
+ else
+ psxBSC[18] = psxCOPd;
+ if (sr & (1u << 31))
+@@ -1282,7 +1297,7 @@ void intApplyConfig() {
+ assert(psxSPC[26] == psxDIV || psxSPC[26] == psxDIV_stall);
+ assert(psxSPC[27] == psxDIVU || psxSPC[27] == psxDIVU_stall);
+
+- if (Config.DisableStalls) {
++ if (1) {
+ psxBSC[18] = psxCOP2;
+ psxBSC[50] = gteLWC2;
+ psxBSC[58] = gteSWC2;
+@@ -1365,8 +1380,12 @@ 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) {
++ printf("execI %08x c %u, ni %u\n", regs->pc, regs->cycle, regs->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]);
+ }