mirrors/dynarmic
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity
1724 commits 1 branch 0 tags 34 MiB
Commit graph

1486 commits

Author SHA1 Message Date
Merry
d83eae2004 Merge pull request #406 from lioncash/mul 
constant_propagation_pass: Fold Mul32 and Mul64 cases where applicable
 2020-04-22 20:57:37 +01:00
Merry
73d9393300 Merge pull request #405 from lioncash/inst 
a64: Add ARMv8.4+ instructions encodings to the encoding table
 2020-04-22 20:57:37 +01:00
Lioncash
7ad6981437 constant_propagation_pass: deduplicate common 32/64 bit checking for results in folding functions 
It's common for an folding operation to apply to both the 32-bit and
64-bit variant of the same opcode, which leads to checking which kind of
result we need to store the value as. This moves it to its own function,
so that we don't need to duplicate it in various functions.
 2020-04-22 20:57:37 +01:00
Lioncash
f1a66c37ba a64: Add ARMv8.4+ instructions encodings to the encoding table 
Keeps the table up to date with the ARM specification.
 2020-04-22 20:57:37 +01:00
Lioncash
72daf37208 constant_propagation_pass: Fold Mul32 and Mul64 cases where applicable 
Multiplication operations can currently be folded if:

1. Both arguments are known constant values
2. Either operand is zero (in which case the result is also zero)
3. Either operand is one (in which case the result is the non-one
operand).
 2020-04-22 20:57:37 +01:00
Lioncash
43b2eb4688 constant_propagation_pass: Fold SignExtend{Type}ToLong opcodes if possible 2020-04-22 20:57:37 +01:00
Lioncash
2da2cf9058 constant_propagation_pass: Fold SignExtend{Type}ToWord opcodes if possible 2020-04-22 20:57:37 +01:00
Lioncash
0583d401e3 ir/value: Add IsSignedImmediate() and IsUnsignedImmediate() functions to Value's interface 
This allows testing against arbitrary values while also simultaneously
eliminating the need to check IsImmediate() all the time in expressions.
 2020-04-22 20:57:37 +01:00
Lioncash
c42f6ea184 constant_propagation_pass: Fold ZeroExtend{Type}ToLong opcodes if possible 
These are equivalent to the ZeroExtendXToWord variants, so we can
trivially do this as well.
 2020-04-22 20:57:37 +01:00
Lioncash
e3258e8525 ir/value: Add a GetImmediateAsS64() function 
Provides a signed analogue to GetImmediateAsU64() for consistency with
both integral classes when it comes to signed/unsigned..
 2020-04-22 20:57:37 +01:00
Lioncash
2274214ff0 constant_propagation_pass: Combine zero-extension folding code into its own function 
Separates the behavior from the actual switch statement and gets rid of
duplication, now that we can use the general GetImmediateAsU64()
function.
 2020-04-22 20:57:37 +01:00
Lioncash
4a3c064b15 ir/value: Add an IsZero() member function to Value's interface 
By far, one of the most common things to check for is whether or not a
value is zero, as it typically allows folding away unnecesary
operations (other close contenders that can help with eliding operations  are 1 and -1).

So instead of requiring a check for an immediate and then actually
retrieving the integral value and checking it, we can wrap it within a
function to make it more convenient.
 2020-04-22 20:57:37 +01:00
Merry
c649f11c0a Merge pull request #401 from lioncash/folding 
constant_propagation_pass: Fold &, |, ^, and ~ operations where applicable
 2020-04-22 20:56:01 +01:00
MerryMage
2524d536b0 A32/ir_emitter: Bugfix: ExceptionRaised was producing incorrect PC 
Use actual PC and not pipelined PC.
 2020-04-22 20:56:01 +01:00
Lioncash
c09f4cf28e constant_propagation_pass: Fold NOT operations 2020-04-22 20:55:50 +01:00
Lioncash
d69fceec55 value: Move ImmediateToU64() to be a part of Value's interface 
This'll make it slightly nicer to do basic constant folding for 32-bit
and 64-bit variants of the same IR opcode type. By that, I mean it's
possible to inspect immediate values without a bunch of conditional
checks beforehand to verify that it's possible to call GetU32() or
GetU64, etc.
 2020-04-22 20:55:50 +01:00
Lioncash
8013548bbb constant_propagation_pass: Fold OR operations 2020-04-22 20:55:50 +01:00
MerryMage
ca603c1215 reg_alloc: Emit AVX instructions where able 
Smaller codesize.
 2020-04-22 20:55:50 +01:00
Lioncash
898d096e39 constant_propagation_pass: Fold AND operations 2020-04-22 20:55:50 +01:00
MerryMage
e2358af5ef abi: Emit AVX instructions where able 
Smaller codesize.
 2020-04-22 20:55:50 +01:00
Lioncash
f40fcda1f6 ir/value: Add member function to check whether or not all bits of a contained value are set 
This is useful when we wish to know if a contained value is something
like 0xFFFFFFFF, as this helps perform constant folding. For example the
operation: x & 0xFFFFFFFF can be folded to just x in the 32-bit case.
 2020-04-22 20:55:50 +01:00
MerryMage
7c0378f56d a64_exclusive_monitor: Loosen memory ordering requirements 
It is not necessary to be as strict as it was.
 2020-04-22 20:55:50 +01:00
Lioncash
0ea99b7d59 constant_propagation_pass: Fold EOR operations 
It's possible to fold cases of exclusive OR operations if they can be
known to be an identity operation, or if both operands happen to be known
immediates, in which case we can just store the result of the
exclusive-OR directly.
 2020-04-22 20:55:50 +01:00
MerryMage
f0920c0ded Fix VShift terminology 
An arithmetic shift is by definition a signed shift, and a logical shift is by definition an unsigned shift.

- Rename VectorLogicalVShiftS* -> VectorArithmeticVShift*
- Rename VectorLogicalVShiftU* -> VectorLogicalVShift*
 2020-04-22 20:55:50 +01:00
MerryMage
b51dae790d emit_x64_vector: AVX512 implementation of EmitVectorLogicalVShiftS16 2020-04-22 20:55:50 +01:00
MerryMage
bd47f2ca8f emit_x64_vector: AVX512 implementation of EmitVectorLogicalVShiftS64 2020-04-22 20:55:50 +01:00
MerryMage
3bf183d7e8 emit_x64_vector: AVX2 implementation of EmitVectorLogicalVShiftS32 2020-04-22 20:55:50 +01:00
MerryMage
94f9d402eb emit_x64_vector: AVX512 implementation of EmitVectorLogicalVShiftU16() 2020-04-22 20:55:50 +01:00
MerryMage
6d9639e3b0 emit_x64_vector: AVX2 implementation of EmitVectorLogicalVShiftU64() 2020-04-22 20:55:50 +01:00
MerryMage
bbc066a266 emit_x64_vector: AVX2 implementation of EmitVectorLogicalVShiftU32() 2020-04-22 20:55:50 +01:00
Lioncash
da2e7fad87 emit_x64_vector: SSSE3 variant of EmitVectorCountLeadingZeros8() 
pshufb lyfe
 2020-04-22 20:55:50 +01:00
VelocityRa
c30b8dbe99 decoders: Cast to correctly-sized type before shifting 
Fixes decoding for 64-bit instructions

Does not help/apply to any currently supported ARM versions (since
all are 32-bit length or below), it's for future-proofing should
such an arch be supported.
 2020-04-22 20:55:50 +01:00
MerryMage
238f2f2cd0 a64_emit_x64: Lowercase PAGE_SIZE 
PAGE_SIZE is defined as a macro by musl.
 2020-04-22 20:55:50 +01:00
MerryMage
7162f6f254 emit_x64_vector_floating_point: SSE4.1 implementation of EmitFPVectorToFixed 2020-04-22 20:55:50 +01:00
MerryMage
e7a5592699 emit_x64_vector_floating_point: EmitFPVectorRoundInt: Use FCODE 2020-04-22 20:55:50 +01:00
MerryMage
b8fde48732 emit_x64_vector: AVX implementation for EmitVectorCountLeadingZeros8 2020-04-22 20:55:50 +01:00
MerryMage
fd37b637aa emit_x64_vector: SSE implementation of EmitVectorCountLeadingZeros16 2020-04-22 20:55:50 +01:00
MerryMage
09bf273bc8 A64: Implement SCVTF, UCVTF (vector, fixed-point), scalar variant 2020-04-22 20:55:06 +01:00
MerryMage
03ad2072a7 emit_x64_floating_point: Reduce fallback LUT code in EmitFPToFixed 2020-04-22 20:55:06 +01:00
MerryMage
f9129db6fd A64: Implement FCVTZS, FCVTZU, UCVTF, SCVTF (vector, fixed-point), vector variant 2020-04-22 20:55:06 +01:00
Lioncash
48df9b9a7d A64: Implement UQSHL's vector immediate and register variants 2020-04-22 20:55:06 +01:00
Lioncash
d426dfe942 ir: Add opcodes for unsigned saturating left shifts 2020-04-22 20:55:06 +01:00
Lioncash
ab60720418 A64/translate/impl: Make signatures consistent for unimplemented by-element SIMD variants 
Makes them all consistent, so it isn't necessary to change the
prototypes over when implementing them.
 2020-04-22 20:55:06 +01:00
Lioncash
6b5ea6ee66 A64: Implement BRK 
Currently, we can just implement this as part of the exception
interface, similar to how it's done for the A32 interface with BKPT.
 2020-04-22 20:55:06 +01:00
Lioncash
b915364c16 A64/imm: Add full range of comparison operators to Imm template 
Makes the comparison interface consistent by providing all of the
relevant members. This also modifies the comparison operators to take
the Imm instance by value, as it's really only a u32 under the covers,
and it's cheaper to shuffle around a u32 than a 64-bit pointer address.
 2020-04-22 20:55:06 +01:00
MerryMage
02150bc0b7 IR: Add fbits argument to FPVectorFrom{Signed,Unsigned}Fixed 2020-04-22 20:55:06 +01:00
MerryMage
027b0ef725 A64: Implement SCVTF, UCVTF (scalar, fixed-point) 2020-04-22 20:55:06 +01:00
MerryMage
8051f60db0 opcodes.inc: Align columns to a tabstop of 4 2020-04-22 20:55:06 +01:00
MerryMage
90193b0e3d IR: Add fbits argument to FixedToFP-related opcodes 2020-04-22 20:55:06 +01:00
Lioncash
616a153c16 A64: Implement SQSHL's vector immediate variant 2020-04-22 20:55:06 +01:00
Lioncash
e8b0f25dff A64: Implement SQSHL's vector register variant 2020-04-22 20:55:06 +01:00
Lioncash
b14eaaec46 ir: Add opcodes for left signed saturated shifts 2020-04-22 20:55:06 +01:00
Lioncash
da55ed7b31 branch: Make variables const where applicable 2020-04-22 20:55:06 +01:00
Lioncash
867b666285 move_wide: Make variables const where applicable 2020-04-22 20:55:06 +01:00
Lioncash
78024a9dc4 load_store_register_unprivileged: Make variables const where applicable 2020-04-22 20:55:06 +01:00
Lioncash
e45e5da610 load_store_register_immediate: Place conditional bodies on their own line 
Makes the conditionals visually consistent with the rest of the
codebase.
 2020-04-22 20:55:06 +01:00
Lioncash
b586cf3f56 load_store_load_literal: Make variables const where applicable 2020-04-22 20:55:06 +01:00
Lioncash
c3a3b9687e data_processing_logical: Move datasize declarations after early-exit conditionals 
While we're at it, make variables const where applicable.
 2020-04-22 20:55:06 +01:00
Lioncash
ed797e6540 data_processing_conditional_select: Make variables const where applicable 
Makes CSEL's function consistent with all of the others.
 2020-04-22 20:55:06 +01:00
Lioncash
c82fa5ec5a data_processing_addsub: Move datasize declarations after early-exit conditionals 
While we're at it, also make relevant variables const where applicable
 2020-04-22 20:55:06 +01:00
Lioncash
f4a66d2477 data_processing_bitfield: Move datasize variables after early-exit conditionals 
Moves the declaration of datasize to the scope that it's used within.
This also takes the opportunity to apply const where applicable, and
make early-exits all vertically consistent with one another.
 2020-04-22 20:55:06 +01:00
Lioncash
2e0fcd6161 A64: Implement CLS's vector variant 
Leverages CLZ like the integral variant does.
 2020-04-22 20:55:06 +01:00
Lioncash
a2cd643525 emit_x64_vector: Make EmitVectorUnsignedSaturatedAccumulateSigned() internally linked 
Given this is just an internal helper function, it can be marked static.
 2020-04-22 20:55:06 +01:00
Lioncash
c39ea2e3c9 perf_map: Use std::string_view instead of std::string for PerfMapRegister() 
We can just use a non-owning view into a string in this case instead of
potentially allocating a std::string instance.
 2020-04-22 20:55:06 +01:00
MerryMage
12243692f5 A64: Implement SQRDMULH (vector), vector variant 2020-04-22 20:55:06 +01:00
MerryMage
a9ffcf08b1 A64: Implement SQDMULL (vector), vector variant 2020-04-22 20:55:06 +01:00
MerryMage
3e447614c6 IR: Add VectorSignedSaturatedDoublingMultiplyLong 2020-04-22 20:55:06 +01:00
MerryMage
06b31448aa emit_x64_vector: Changes to VectorSignedSaturatedDoublingMultiply 
* Return both the upper and lower parts of the multiply if required
* SSE2 does not support the pmuldq instruction, do sign correction to an unsigned result instead
* Improve port utilisation where possible (punpck instructions were a bottleneck)
 2020-04-22 20:55:06 +01:00
MerryMage
08c0e017a5 IR: Implement Vector{Signed,Unsigned}Multiply{16,32} 2020-04-22 20:55:06 +01:00
Lioncash
b6df34cdde backend_x64/a64_interface: Re-enable the constant folding pass 
This was disabled for debugging, but never re-enabled. Just to be sure,
testing was done downstream in yuzu to make sure this didn't happen to
break anything (which seems to be the case).
 2020-04-22 20:55:06 +01:00
MerryMage
06ba397af2 emit_x64_vector_floating_point: Hardware FMA implementation for RSqrtStepFused 2020-04-22 20:55:06 +01:00
MerryMage
e553c4fe8d emit_x64_vector_floating_point: Hardware FMA implementation of FPVectorRecipStepFused 2020-04-22 20:55:06 +01:00
MerryMage
3caeb62ef1 emit_x64_floating_point: Hardware FMA implementation of FPRSqrtStepFused 2020-04-22 20:55:06 +01:00
MerryMage
344ee76aba emit_x64_floating_point: Hardware FMA implementation of FPRecipStepFused{32,64} 2020-04-22 20:55:06 +01:00
MerryMage
1492573267 emit_x64_vector: SSE implementation of VectorSignedSaturatedAccumulateUnsigned{8,16,32} 2020-04-22 20:55:06 +01:00
Lioncash
26df6e5e7b emit_x64_vector: Correct static asserts for < 64-bit type checks in saturated accumulate fallbacks 
I had initially meant to use BitSize() here, not sizeof()
 2020-04-22 20:55:06 +01:00
MerryMage
a4a26ac226 emit_x64_vector: EmitVectorSignedSaturatedAccumulateUnsigned64: SSE implementation 2020-04-22 20:55:06 +01:00
MerryMage
a7c66d2d28 emit_x64_vector: Simplify fpsr_qc related code 
Move the bool conversion into A64JitState::GetFpsr so we don't have to continuously
pay the cost of conversion for every saturation instruction.
 2020-04-22 20:55:06 +01:00
Lioncash
112cff9ab9 A64: Implement CLZ's vector variant 2020-04-22 20:55:06 +01:00
Lioncash
e739624296 ir: Add opcodes for vector CLZ operations 
We can optimize these cases further for with the use of a fair bit of
shuffling via pshufb and the use of masks, but given the uncommon use of
this instruction, I wouldn't consider it to be beneficial in terms of
amount of code to be worth it over a simple manageable naive solution
like this.

If we ever do hit a case where vectorized CLZ happens to be a
bottleneck, then we can revisit this. At least with AVX-512CD, this can
be done with a single instruction for the 32-bit word case.
 2020-04-22 20:55:05 +01:00
MerryMage
d4c37a68a8 A64/translate: VectorZeroUpper for V(64) stores 
Ensures correctness.
 2020-04-22 20:55:05 +01:00
MerryMage
b8daa4feac simd_two_register_misc: FNEG (vector) with Q == 0 had dirty upper 2020-04-22 20:55:05 +01:00
Lioncash
5653e7637e emit_x64_vector: Remove unnecessary [[maybe_unused]] attributes 
These were unintentionally left in when introducing SUQADD and USQADD
 2020-04-22 20:55:05 +01:00
Lioncash
14e026a7f0 A64: Implement USQADD's scalar and vector variants 2020-04-22 20:55:05 +01:00
Lioncash
d4a76aaa04 ir: Add opcodes form unsigned saturated accumulations of signed values 2020-04-22 20:55:05 +01:00
Lioncash
18ad7f237d A64: Implement SUQADD's scalar and vector variants 2020-04-22 20:55:05 +01:00
Lioncash
6f911a26da ir: Add opcodes for signed saturated accumulations of unsigned values 2020-04-22 20:55:05 +01:00
Lioncash
9a3d38d2ee A64: Implement SMLAL{2}, SMLSL{2}, UMLAL{2}, and UMLSL{2}'s vector by-element variants 
We can simply modify the general function made for SMULL{2} and
UMULL{2}'s by-element variants to also handle the other multiply-based
by-element variants.
 2020-04-22 20:55:05 +01:00
Lioncash
6ccfbc9b39 A64: Implement UMULL{2}'s vector by-element variant 2020-04-22 20:55:05 +01:00
Lioncash
58e21f175c A64: Implement SMULL{2}'s vector by-element variant 2020-04-22 20:55:05 +01:00
Lioncash
134bb02e19 ir/value: Replace includes with forward declarations 
enum classes are still considered complete types when forward declared
(as the compiler knows the exact size of the type from the declaration
alone). The only difference in this case being that the members of the
enum class aren't visible. Given we don't use the members within this
header in any way, we can simply forward declare them here and remove
the inclusions.
 2020-04-22 20:55:05 +01:00
Lioncash
2c8e07e7d0 ir/cond: Migrate to C++17 nested namespace specifiers 2020-04-22 20:55:05 +01:00
Lioncash
c3b7819a55 CMakeLists: Add missing cond.h header to file listing 
Allows the file to show up within IDEs more easily.
 2020-04-22 20:55:05 +01:00
Lioncash
0a3976059f A64: Implement URSQRTE 2020-04-22 20:55:05 +01:00
Lioncash
b6e74fd17d ir: Add opcodes for performing unsigned reciprocal square root estimates 2020-04-22 20:55:05 +01:00
Lioncash
bd3582e811 A64: Implement URECPE 2020-04-22 20:55:05 +01:00
Lioncash
af83360f89 ir: Add opcodes for unsigned reciprocal estimate 2020-04-22 20:55:05 +01:00
Lioncash
740ffa52ae A64: Implement SQNEG's scalar and vector variant 2020-04-22 20:53:46 +01:00
Lioncash
fca7eddb9e A64: Add opcodes for signed saturating negations 2020-04-22 20:53:46 +01:00
Lioncash
f1ebbcd7bc emit_x64_vector: Simplify "position == 0" case for EmitVectorExtract() 
In the event position is zero, we can just treat it as a NOP, given
there's no need to move the data.
 2020-04-22 20:53:46 +01:00