| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
| |
Data section helps serialization and cloning of a ir_variable. This
patch includes the helper bits used for read only ir_variables.
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Paul Berry <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, when loop_controls analyzed a loop and found that it had a
fixed bound (known at compile time), it would remove all of the loop
terminators and instead set the loop's normative_bound field to force
the loop to execute the correct number of times.
This made loop unrolling easy, but it had a serious disadvantage.
Since most GPU's don't have a native mechanism for executing a loop a
fixed number of times, in order to implement the normative bound, the
back-ends would have to synthesize a new loop induction variable. As
a result, many loops wound up having two induction variables instead
of one. This caused extra register pressure and unnecessary
instructions.
This patch modifies loop_controls so that it doesn't set the loop's
normative_bound anymore. Instead it leaves one of the terminators in
the loop (the limiting terminator), so the back-end doesn't have to go
to any extra work to ensure the loop terminates at the right time.
This complicates loop unrolling slightly: when deciding whether a loop
can be unrolled, we have to account for the presence of the limiting
terminator. And when we do unroll the loop, we have to remove the
limiting terminator first.
For an example of how this results in more efficient back end code,
consider the loop:
for (int i = 0; i < 100; i++) {
total += i;
}
Previous to this patch, on i965, this loop would compile down to this
(vec4) native code:
mov(8) g4<1>.xD 0D
mov(8) g8<1>.xD 0D
loop:
cmp.ge.f0(8) null g8<4;4,1>.xD 100D
(+f0) if(8)
break(8)
endif(8)
add(8) g5<1>.xD g5<4;4,1>.xD g4<4;4,1>.xD
add(8) g8<1>.xD g8<4;4,1>.xD 1D
add(8) g4<1>.xD g4<4;4,1>.xD 1D
while(8) loop
(notice that both g8 and g4 are loop induction variables; one is used
to terminate the loop, and the other is used to accumulate the total).
After this patch, the same loop compiles to:
mov(8) g4<1>.xD 0D
loop:
cmp.ge.f0(8) null g4<4;4,1>.xD 100D
(+f0) if(8)
break(8)
endif(8)
add(8) g5<1>.xD g5<4;4,1>.xD g4<4;4,1>.xD
add(8) g4<1>.xD g4<4;4,1>.xD 1D
while(8) loop
Reviewed-by: Ian Romanick <[email protected]>
|
| |
|
|
|
|
|
|
| |
This value is now redundant with
loop_variable_state::limiting_terminator->iterations and
ir_loop::normative_bound.
Reviewed-by: Ian Romanick <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, the sole responsibility of loop_analysis was to find all
the variables referenced in the loop that are either loop constant or
induction variables, and find all of the simple if statements that
might terminate the loop. The remainder of the analysis necessary to
determine how many times a loop executed was performed by
loop_controls.
This patch makes loop_analysis also responsible for determining the
number of iterations after which each loop terminator will terminate
the loop, and for figuring out which terminator will terminate the
loop first (I'm calling this the "limiting terminator").
This will allow loop unrolling to make use of information that was
previously only visible from loop_controls, namely the identity of the
limiting terminator.
Reviewed-by: Ian Romanick <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
| |
When loop_control_visitor::visit_leave(ir_loop *) is analyzing a loop
terminator that acts on a certain ir_variable, it doesn't need to walk
the list of induction variables to find the loop_variable entry
corresponding to the variable. It can just look it up in the
loop_variable_state hashtable and verify that the loop_variable entry
represents an induction variable.
Reviewed-by: Ian Romanick <[email protected]>
|
| |
|
|
|
|
|
|
|
| |
These fields were part of some planned optimizations that never
materialized. Remove them for now to simplify things; if we ever get
round to adding the optimizations that would require them, we can
always re-introduce them.
Reviewed-by: Ian Romanick <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch replaces the ir_loop fields "from", "to", "increment",
"counter", and "cmp" with a single integer ("normative_bound") that
serves the same purpose.
I've used the name "normative_bound" to emphasize the fact that the
back-end is required to emit code to prevent the loop from running
more than normative_bound times. (By contrast, an "informative" bound
would be a bound that is informational only).
Reviewed-by: Jordan Justen <[email protected]>
Reviewed-by: Ian Romanick <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously, when visiting a variable dereference, loop analysis would
only consider its effect on the innermost enclosing loop. As a
result, when encountering a loop like this:
for (int i = 0; i < 3; i++) {
for (int j = 0; j < 3; j++) {
...
i = 2;
}
}
it would incorrectly conclude that the outer loop ran three times.
Fixes piglit test "vs-inner-loop-modifies-outer-loop-var.shader_test".
Reviewed-by: Jordan Justen <[email protected]>
Reviewed-by: Ian Romanick <[email protected]>
|
| |
|
|
|
|
|
| |
This function is about to get more complex.
Reviewed-by: Jordan Justen <[email protected]>
Reviewed-by: Ian Romanick <[email protected]>
|
| |
|
|
|
|
| |
This field was neither initialized nor used. It was just dead memory.
Reviewed-by: Kenneth Graunke <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously the loop_state was allocated in the loop_analysis
constructor, but not freed in the (nonexistent) destructor. Moving
the allocation of the loop_state makes this code appear less sketchy.
Either way, there is no actual leak. The loop_state is freed by the
single caller of analyze_loop_variables.
Signed-off-by: Ian Romanick <[email protected]>
Cc: Dave Airlie <[email protected]>
Reviewed-by: Matt Turner <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57753
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
While ~loop_state() is already freeing the loop_variable_state objects
via ralloc_free(this->mem_ctx), the ~loop_variable_state() destructor
was never getting called, so the hash table inside loop_variable_state
was never getting destroyed.
Fixes a memory leak in any shader with loops.
NOTE: This is a candidate for stable release branches.
Signed-off-by: Kenneth Graunke <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Function calls may have side effects that alter variables used inside
the loop. In the fragment shader, they may even terminate the shader.
This means our analysis about loop-constant or induction variables may
be completely wrong.
In general it's impossible to determine whether they actually do or not
(due to the halting problem), so we'd need to perform conservative
static analysis. For now, it's not worth the complexity: most functions
will be inlined, at which point we can unroll them successfully.
Fixes Piglit tests:
- shaders/glsl-fs-unroll-out-param
- shaders/glsl-fs-unroll-side-effect
NOTE: This is a candidate for release branches.
Signed-off-by: Kenneth Graunke <[email protected]>
Reviewed-by: Ian Romanick <[email protected]>
|
| |
|
|
|
| |
Shaves 1.6% (+/- 1.0%) off of ff_fragment_shader glean texCombine time
(n=5).
|
| |
|
|
|
|
|
|
|
| |
This increases the chance that GLSL programs will actually work.
Note that continues and returns are not yet lowered, so linking
will just fail if not supported.
Signed-off-by: Ian Romanick <[email protected]>
|
| | |
|
| | |
|
| |
|
|
| |
This is the next step on the road to loop unrolling
|
|
|
This is the first step eventually leading to loop unrolling.
|
