Asmjit FI

RPCS3 · Jan 23, 2024 · db251fd · db251fd
1 parent d3942b6
commit db251fd
Show file tree

Hide file tree

Showing 2 changed files with 34 additions and 18 deletions.
diff --git a/rpcs3/Emu/Cell/SPUASMJITRecompiler.cpp b/rpcs3/Emu/Cell/SPUASMJITRecompiler.cpp
@@ -3983,7 +3983,35 @@ void spu_recompiler::CEQB(spu_opcode_t op)
 void spu_recompiler::FI(spu_opcode_t op)
 {
 	// Floating Interpolate
+	const XmmLink& va = XmmGet(op.ra, XmmType::Float);
 	const XmmLink& vb = XmmGet(op.rb, XmmType::Float);
+	const XmmLink& vb_base = XmmAlloc();
+	const XmmLink& ymul = XmmAlloc();
+	const XmmLink& temp_reg = XmmAlloc();
+
+	c->movdqa(vb_base, vb);
+	c->movdqa(ymul, vb);
+	c->movdqa(temp_reg, va);
+
+	c->pand(vb_base, XmmConst(v128::from32p(0x007ffc00u)));
+	c->pslld(vb_base, 9);
+
+	c->pand(temp_reg, XmmConst(v128::from32p(0x7ffff))); // va_fraction
+	c->pand(ymul, XmmConst(v128::from32p(0x3ff)));
+	c->pmulld(ymul, temp_reg);
+
+	c->movdqa(temp_reg, vb_base);
+	c->psubd(temp_reg, ymul);
+	c->psrld(temp_reg, 9);
+
+	c->pcmpgtd(vb_base, ymul);
+	c->pand(vb_base, XmmConst(v128::from32p(1 << 23)));
+	c->paddd(temp_reg, vb_base);
+
+	c->pand(vb, XmmConst(v128::from32p(0xff800000u)));
+	c->pand(temp_reg, XmmConst(v128::from32p(~0xff800000u)));
+	c->por(vb, temp_reg);
+
 	c->movaps(SPU_OFF_128(gpr, op.rt), vb);
 }
 

diff --git a/rpcs3/Emu/Cell/SPULLVMRecompiler.cpp b/rpcs3/Emu/Cell/SPULLVMRecompiler.cpp
@@ -6073,24 +6073,6 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
 
 	void FI(spu_opcode_t op)
 	{
-		// TODO
-		if (g_cfg.core.spu_xfloat_accuracy == xfloat_accuracy::accurate)
-		{
-			set_vr(op.rt, get_vr<f64[4]>(op.rb));
-			// const auto [a, b] = get_vrs<f64[4]>(op.ra, op.rb);
-
-			// const auto mask_se = splat<s64[4]>(0xfff0000000000000ull);
-			// const auto mask_bf = splat<s64[4]>(0x000fff8000000000ull);
-			// const auto mask_sf = splat<s64[4]>(0x0000007fe0000000ull);
-			// const auto mask_yf = splat<s64[4]>(0x0000ffffe0000000ull);
-
-			// const auto base = bitcast<f64[4]>((bitcast<s64[4]>(b) & mask_bf) | 0x3ff0000000000000ull);
-			// const auto step = fpcast<f64[4]>(bitcast<s64[4]>(b) & mask_sf) * fsplat<f64[4]>(std::exp2(-13.f));
-			// const auto yval = fpcast<f64[4]>(bitcast<s64[4]>(a) & mask_yf) * fsplat<f64[4]>(std::exp2(-19.f));
-			// set_vr(op.rt, bitcast<f64[4]>((bitcast<s64[4]>(b) & mask_se) | (bitcast<s64[4]>(base - step * yval) & ~mask_se)));
-			return;
-		}
-
 		register_intrinsic("spu_fi", [&](llvm::CallInst* ci)
 		{
 			const auto a = bitcast<u32[4]>(value<f32[4]>(ci->getOperand(0)));
@@ -6102,6 +6084,12 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
 			return bitcast<f32[4]>((b & 0xff800000u) | (bitcast<u32[4]>(fpcast<f32[4]>(bnew)) & ~0xff800000u)); // Inject old sign and exponent
 		});
 
+		if (g_cfg.core.spu_xfloat_accuracy == xfloat_accuracy::accurate)
+		{
+			const auto r = eval(fi(a, b));
+			set_vr(op.rt, r);
+		}
+
 		if (g_cfg.core.spu_xfloat_accuracy == xfloat_accuracy::approximate)
 		{
 			register_intrinsic("spu_re", [&](llvm::CallInst* ci)