mirrors/Bochs
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Optimization of RCPSS/RCPPS functions

Browse Source
This commit is contained in:
Stanislav Shwartsman 2003-11-19 20:27:58 +00:00
parent 9e6e7609e3
commit a6c1bdbbb2
2 changed files with 13 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
bochs/cpu/sse_pfp.cc
View File

@ -41,9 +41,9 @@ void BX_CPU_C::check_exceptionsSSE(int exceptions_flags)
}
}
static void mxcsr_to_softfloat_status_word(softfloat_status_word_t &status, bx_mxcsr_t mxcsr)
static void mxcsr_to_softfloat_status_word(softfloat_status_word_t &status, bx_mxcsr_t mxcsr, unsigned precision = 32)
{
status.float_precision = 32; // affects only float32 and float80 operations
status.float_precision = precision; // affects only float32 and float80 operations
status.float_detect_tininess = float_tininess_before_rounding;
status.float_exception_flags = 0; // clear exceptions before execution
status.float_nan_handling_mode = float_first_operand_nan;
@ -53,21 +53,15 @@ static void mxcsr_to_softfloat_status_word(softfloat_status_word_t &status, bx_m
(mxcsr.get_flush_masked_underflow() && mxcsr.get_UM()) ? 1 : 0;
}
static void approx_prepare_softfloat_status_word(softfloat_status_word_t &status)
BX_CPP_INLINE Float32 convert_to_QNaN(Float32 op)
{
status.float_precision = 12;
status.float_detect_tininess = float_tininess_before_rounding;
status.float_exception_flags = 0; // clear exceptions before execution
status.float_nan_handling_mode = float_first_operand_nan;
status.float_rounding_mode = float_round_nearest_even;
status.flush_underflow_to_zero = 1;
return op | 0x00400000;
}
// approximate reciprocal of scalar single precision FP
static Float32 approximate_reciprocal(Float32 op)
{
softfloat_status_word_t status_word;
approx_prepare_softfloat_status_word(status_word);
float_class_t op_class = float32_class(op);
static const Float32 one = 0x3F800000;
@ -75,7 +69,7 @@ static Float32 approximate_reciprocal(Float32 op)
if (op_class == float_NaN)
{
result = op | 0x00400000;
return convert_to_QNaN(op);
}
else {
if (op_class == float_denormal)
@ -98,6 +92,9 @@ static Float32 approximate_reciprocal(Float32 op)
* of the current rounding mode.
*/
mxcsr_to_softfloat_status_word(status_word,
bx_mxcsr_t(MXCSR_FLUSH_MASKED_UNDERFLOW | MXCSR_UM), 12);
result = float32_div(one, op, status_word);
}
@ -1540,8 +1537,8 @@ void BX_CPU_C::RCPSS_VssWss(bxInstruction_c *i)
read_virtual_dword(i->seg(), RMAddr(i), &op);
}
BX_WRITE_XMM_REG_LO_DWORD(i->nnn(),
approximate_reciprocal(op));
Float32 result = approximate_reciprocal(op);
BX_WRITE_XMM_REG_LO_DWORD(i->nnn(), result);
#else
BX_INFO(("RCPSS_VssWss: required SSE, use --enable-sse option"));

3
bochs/cpu/xmm.h
View File

@ -154,6 +154,9 @@ struct bx_mxcsr_t
{
Bit32u mxcsr;
bx_mxcsr_t (Bit32u val)
: mxcsr(val) {}
#define IMPLEMENT_MXCSR_ACCESSOR(name, bitmask, bitnum) \
int get_##name () const { \
return (mxcsr & (bitmask)) >> (bitnum); \