OPSEL ASM Syntax for v_cvt_scalef32_pk_f32_fp4 : opsel:[x,y,z]
where, x & y i.e. OPSEL[1 : 0] selects which src_byte to read.
OPSEL ASM Syntax for v_cvt_scalef32_pk_fp4_f32 : opsel:[a,b,c,d]
where, c & d i.e. OPSEL[3 : 2] selects which dst_byte to write.
Co-authored-by: Pravin Jagtap <Pravin.Jagtap@amd.com>
OPSEL[1:0] collectively decide which byte to read
from src input.
Builtin takes additional imm argument which
represents index (with valid values:[0:3]) of src
byte read. Out of bounds checks will added in next
patch.
OPSEL ASM Syntax: opsel:[x,y,z]
where,
opsel[x] = Inst{11} = src0_modifier{2}
opsel[y] = Inst{12} = src1_modifier{2}
opsel[z] = Inst{14} = src0_modifier{3}
Note: Inst{13} i.e. OPSEL[2] is ignored in
asm syntax and opsel[z] is meaningless
for v_cvt_scalef32_f32_{fp|bf}8
Co-authored-by: Pravin Jagtap <Pravin.Jagtap@amd.com>
This was a bit annoying because these introduce a new special case
encoding usage. op_sel is repurposed as a subset of dpp controls,
and is eligible for VOP3->VOP1 shrinking. For some reason fi also
uses an enum value, so we need to convert the raw boolean to 1 instead
of -1.
The 2 registers are swapped, so this has 2 defs. Ideally the builtin
would return a pair, but that's difficult so return a vector instead.
This would make a hypothetical builtin that supports v2f16 directly
uglier.
These use a new VOP3PX encoding for the v_mfma_scale_* instructions,
which bundles the pre-scale v_mfma_ld_scale_b32. None of the modifiers
are supported yet (op_sel, neg or clamp).
I'm not sure the intrinsic should really expose op_sel (or any of the
others). If I'm reading the documentation correctly, we should be able
to just have the raw scale operands and auto-match op_sel to byte
extract patterns.
The op_sel syntax also seems extra horrible in this usage, especially with the
usual assumed op_sel_hi=-1 behavior.
Use a local pointer type to represent the named barrier in builtin and
intrinsic. This makes the definitions more user friendly
bacause they do not need to worry about the hardware ID assignment. Also
this approach is more like the other popular GPU programming language.
Named barriers should be represented as global variables of addrspace(3)
in LLVM-IR. Compiler assigns the special LDS offsets for those variables
during AMDGPULowerModuleLDS pass. Those addresses are converted to hw
barrier ID during instruction selection. The rest of the
instruction-selection changes are primarily due to the
intrinsic-definition changes.
This reverts commit
7792b4ae79.
The problem was a conflict with
e55d6f5ea2
"[AMDGPU] Simplify and improve codegen for llvm.amdgcn.set.inactive
(https://github.com/llvm/llvm-project/pull/107889)"
which changed the syntax of V_SET_INACTIVE (and thus made my MIR test
crash).
...if only we had a merge queue.
Reverts llvm/llvm-project#108173
si-init-whole-wave.mir crashes on some buildbots (although it passed
both locally with sanitizers enabled and in pre-merge tests).
Investigating.
This intrinsic is meant to be used in functions that have a "tail" that
needs to be run with all the lanes enabled. The "tail" may contain
complex control flow that makes it unsuitable for the use of the
existing WWM intrinsics. Instead, we will pretend that the function
starts with all the lanes enabled, then branches into the actual body of
the function for the lanes that were meant to run it, and then finally
all the lanes will rejoin and run the tail.
As such, the intrinsic will return the EXEC mask for the body of the
function, and is meant to be used only as part of a very limited pattern
(for now only in amdgpu_cs_chain functions):
```
entry:
%func_exec = call i1 @llvm.amdgcn.init.whole.wave()
br i1 %func_exec, label %func, label %tail
func:
; ... stuff that should run with the actual EXEC mask
br label %tail
tail:
; ... stuff that runs with all the lanes enabled;
; can contain more than one basic block
```
It's an error to use the result of this intrinsic for anything
other than a branch (but unfortunately checking that in the verifier is
non-trivial because SIAnnotateControlFlow will introduce an amdgcn.if
between the intrinsic and the branch).
The intrinsic is lowered to a SI_INIT_WHOLE_WAVE pseudo, which for now
is expanded in si-wqm (which is where SI_INIT_EXEC is handled too);
however the information that the function was conceptually started in
whole wave mode is stored in the machine function info
(hasInitWholeWave). This will be useful in prolog epilog insertion,
where we can skip saving the inactive lanes for CSRs (since if the
function started with all the lanes active, then there are no inactive
lanes to preserve).
Use GCNPat instead of Custom Lowering to select instructions for
intrinsic llvm.fptrunc.round. "SupportedRoundMode : TImmLeaf" is used as
a predicate to select only when the rounding mode is supported.
"as_hw_round_mode : SDNodeXForm" is developed to translate the round
modes to the corresponding ones that hardware recognizes.
isExtractHiElt should return new source register instead of returning
instruction that defines it. Src = MI.getOperand(0).getReg() is not
correct when MI(for example G_UNMERGE_VALUES) defines multiple registers.
Refactor existing code to work with source registers only.
This regresses the arbitrary address space pointer case. Ideally
we could write a pattern that matches a pointer based only on
its size, but using iPTR/iPTRAny seem to not work for this.
This is still relying on the manual code for splitting 64-bit
constants, and handling pointers.
We were missing some of the tablegen patterns for all immediate types,
so this has some side effect DAG path improvements. This also reduces
the diff in the 2 selector outputs.
In order to ensure the correctness of ptr addrspace(7) lowering, we need
a backwards-compatible way to flag buffer intrinsics as volatile that
can't be dropped (unlike metadata).
To acheive this in a backwards-compatible way, we use bit 31 of the
auxilliary immediates of buffer intrinsics as the volatile flag. When
this bit is set, the MachineMemOperand for said intrinsic is marked
volatile. Existing code will ensure that this results in the appropriate
use of flags like glc and dlc.
This commit also harmorizes the handling of the auxilliary immediate for
atomic intrinsics, which new go through extract_cpol like loads and
stores, which masks off the volatile bit.
This is used to select the source modifier (neg) from the immediate
operand. After a follow up commit this will no longer be DOTIU specific.
Co-authored-by: Changpeng Fang <changpeng.fang@amd.com>
For scratch load/store, our hardware only accept non-negative value in
SGPR/VGPR. Besides the case that we can prove from known bits, we can
also prove that the value in `base` will be non-negative: 1.) When the
ADD for the address calculation has NonUnsignedWrap flag. 2.) When the
immediate offset is already negative.
For the f64 case, this gives us a cheaper to materialize 32-bit
constant. It's less obviously a win for f32 and f16. It forces us to
use a VOP3 encoding so it's a neutral code size change.
GlobalISel cases don't work because of the constant-is-copy-to-vgpr
problem.
https://reviews.llvm.org/D157111
This isn't always folded to fneg for a freestanding fsub depending on
the denormal mode. When matching source modifiers, we're implicitly
canonicalizing the input so we can fold it here.
Doesn't bother handling the VOP3P case since it's only relevant with
DAZ, which nobody really uses with f16.
For f64, tests show an existing bug where DAGCombiner tries to respect
the denormal mode for fsub -0, x, but not after it's lowered to fadd
-0, (fneg x). Either the fold is wrong or we shouldn't restrict the
fsub case based on the denormal mode.
https://reviews.llvm.org/D155652
Use the new matchtable-based combiner backend for all AMDGPU combiners.
This drop-in from the user's perspective; there are no test changes, the new combiner behaves exactly like the old one.
Depends on D153757
NOTE: This would land iff D153757 (RFC) lands too.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D153758
The inverse ballot intrinsic takes in a boolean mask for all lanes and
returns the boolean for the current lane. See SPIR-V's
`subgroupInverseBallot()` in the [[ https://github.com/KhronosGroup/GLSL/blob/master/extensions/khr/GL_KHR_shader_subgroup.txt | GL_KHR_shader_subgroup extension ]].
This allows decision making via branch and select instructions with a manually
manipulated mask.
Implemented in GlobalISel and SelectionDAG, since currently both are supported.
The SelectionDAG required pseudo instructions to use the custom inserter.
The boolean mask needs to be uniform for all lanes.
Therefore we expect SGPR input. In case the source is in a
VGPR, we insert one or more `v_readfirstlane` instructions.
Reviewed By: nhaehnle
Differential Revision: https://reviews.llvm.org/D146287
Values in SGPR register are treated as unsigned by hardware.
When value in 32-bit SGPR base can be negative calculate offset using
32-bit add instruction, otherwise use sgpr base(unsigned) + offset.
Does not affect case where whole offset comes from SGPR register
(immediate offset is 0).
LoopStrengthReduce.cpp changes offsets to negative and in some
iterations value in SGPR register could be negative.
Differential Revision: https://reviews.llvm.org/D144955
The matching for V_FMA_MIX was partially implemented with a C++
matcher (for fmas with 32 bit results and 16 bit inputs) and partially
in tablegen (for fmas with 16 bit results). Move the C++ matcher logic
into tablegen to make this more consistent and so we can remove the
duplication between SDAG and GISel.
Differential Revision: https://reviews.llvm.org/D144612
v_permlane16_b32 and v_permlanex16_b32 should not set abs and neg src
modifiers on any input, but they can set op_sel on src0 or src1 to
represent fi or bc when desired. The ISel patterns were setting
the src_modifier bits to -1, effectively setting abs and neg as well,
whenever it was intended to set op_sel, due to an error in ISel. ISel
should now correctly only set the op_sel bits.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D144519
We were only allowing these med3 patterns if the operands were known
to not be NaN, but we should also allow it if the calls to max/min
have the `nnan` or `fast` flags.
Differential Revision: https://reviews.llvm.org/D139506
When selectVOP3PMadMixModsImpl fails, it can still create new copy instr
via selectVOP3ModsImpl. When selectG_FMA_FMAD gives up, new copy instr
will remain dead but will not be automatically removed.
InstructionSelect does not check if instructions created during selection
are dead.
Such dead copy doesn't have register class on dst operand and causes crash.
Fix is to build copy when operands are being added to selected instruction.
Differential Revision: https://reviews.llvm.org/D138044
Adds support for selecting the following instructions using GlobalISel:
- v_mad_mix/v_fma_mix
- v_mad_mixhi/v_fma_mixhi
- v_mad_mixlo/v_fma_mixlo
To select those instructions properly, some additional changes were
needed which impacted other tests as well.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D134354
Preparation patch for D134354 to make V2S16 G_BUILD_VECTOR legal.
Also removes RegBankInfo's scalarization of small BUILD_VECTORs,
replacing it with InstructionSelector logic instead.
This allows for V2S16 BUILD_VECTOR instructions to survive
all the way to ISel so we can select FMA/MAD_MIX instructions
in D134354.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D134433
Remove manual selection for atomic fadd from global-isel.
Stop pre-isel translation to AtomicLoadFAdd/G_ATOMICRMW_FADD
which corresponds to llvm-ir's atomicrmw fadd instruction.
global and flat atomic fadd patterns changes:
Split rtn/no-rtn patterns
Add missing patterns or fix predicates
Remove atomicrmw patterns for v2f16 (atomic rmw doesn't support vectors).
Patterns now check addrspace of pointer, added patterns for flat intrinsic.
with global addrspace pointer that selects into global atomic instruction.
buffer atomic fadd patterns changes:
Rdit patterns to import into global-isel.
Remove gfx6/gfx7 _addr64 and _offset patterns.
Remove patterns that can't be reached (same pattern but different feature).
Differential Revision: https://reviews.llvm.org/D130579