Initial support for ARM NEON on Cortex-A9 and A15 CPUs. All but ~10 tests
pass, and all examples compile and run correctly. Most of the examples
show a ~2x speedup on a single A15 core versus scalar code.
Current open issues/TODOs
- Code quality looks decent, but hasn't been carefully examined. Known
issues/opportunities for improvement include:
- fp32 vector divide is done as a series of scalar divides rather than
a vector divide (which I believe exists, but I may be mistaken.)
This is particularly harmful to examples/rt, which only runs ~1.5x
faster with ispc, likely due to long chains of scalar divides.
- The compiler isn't generating a vmin.f32 for e.g. the final scalar
min in reduce_min(); instead it's generating a compare and then a
select instruction (and similarly elsewhere).
- There are some additional FIXMEs in builtins/target-neon.ll that
include both a few pieces of missing functionality (e.g. rounding
doubles) as well as places that deserve attention for possible
code quality improvements.
- Currently only the "cortex-a9" and "cortex-15" CPU targets are
supported; LLVM supports many other ARM CPUs and ispc should provide
access to all of the ones that have NEON support (and aren't too
obscure.)
- ~5 of the reduce-* tests hit an assertion inside LLVM (unfortunately
only when the compiler runs on an ARM host, though).
- The Windows build hasn't been tested (though I've tried to update
ispc.vcxproj appropriately). It may just work, but will more likely
have various small issues.)
- Anything related to 64-bit ARM has seen no attention.
This forces all vector loads/stores to be done assuming that the given
pointer is aligned to the vector size, thus allowing the use of sometimes
more-efficient instructions. (If it isn't the case that the memory is
aligned, the program will fail!).
Cilk (cilk_for), OpenMP (#pragma omp parallel for), TBB(tbb::task_group and tbb::parallel_for)
as well as a new pthreads-based model that fully subscribes the machine (good for KNC).
With major contributions from Ingo Wald and James Brodman.
We now have two ways of approaching gather/scatters with a common base
pointer and with offset vectors. For targets with native gather/scatter,
we just turn those into base + {1/2/4/8}*offsets. For targets without,
we turn those into base + {1/2/4/8}*varying_offsets + const_offsets,
where const_offsets is a compile-time constant.
Infrastructure for issue #325.
When "break", "continue", or "return" is used under varying control flow,
we now always check the execution mask to see if all of the program
instances are executing it. (Previously, this was only done with "cbreak",
"ccontinue", and "creturn", which are now deprecated.)
An important effect of this change is that it fixes a family of cases
where we could end up running with an "all off" execution mask, which isn't
supposed to happen, as it leads to all sorts of invalid behavior.
This change does cause the volume rendering example to run 9% slower, but
doesn't affect the other examples.
Issue #257.
It can sometimes be useful to know the general place we were in the program
when an assertion hit; when the position is available / applicable, this
macro is now used.
Issue #268.
The intent of this was to indicate whether it was safe to run code
with an 'all of' mask on the given target (and then sometimes be
more flexible about e.g. running both true and false blocks of if
statements, etc.)
The problem is that even if the architecture has full native mask support,
it's still not safe to run 'uniform' memory operations with the mask all
off. Even more tricky, we sometimes transform masked varying memory operations
to uniform ones during optimization (e.g. gather->load and broadcast).
This fixes a number of the tests/switch-* tests that were failing on the
generic targets due to this issue.
The decl.* code now no longer interacts with Symbols, but just returns
names, types, initializer expressions, etc., as needed. This makes the
code a bit more understandable.
Fixes issues #171 and #130.
There's now a SOA variability class (in addition to uniform,
varying, and unbound variability); the SOA factor must be a
positive power of 2.
When applied to a type, the leaf elements of the type (i.e.
atomic types, pointer types, and enum types) are widened out
into arrays of the given SOA factor. For example, given
struct Point { float x, y, z; };
Then "soa<8> Point" has a memory layout of "float x[8], y[8],
z[8]".
Furthermore, array indexing syntax has been augmented so that
when indexing into arrays of SOA-variability data, the two-stage
indexing (first into the array of soa<> elements and then into
the leaf arrays of SOA data) is performed automatically.
There are two related optimizations that happen now. (These
currently only apply for gathers where the mask is known to be
all on, and to gathers that are accessing 32-bit sized elements,
but both of these may be generalized in the future.)
First, for any single gather, we are now more flexible in mapping it
to individual memory operations. Previously, we would only either map
it to a general gather (one scalar load per SIMD lane), or an
unaligned vector load (if the program instances could be determined
to be accessing a sequential set of locations in memory.)
Now, we are able to break gathers into scalar, 2-wide (i.e. 64-bit),
4-wide, or 8-wide loads. Further, we now generate code that shuffles
these loads around. Doing fewer, larger loads in this manner, when
possible, can be more efficient.
Second, we can coalesce memory accesses across multiple gathers. If
we have a series of gathers without any memory writes in the middle,
then we try to analyze their reads collectively and choose an efficient
set of loads for them. Not only does this help if different gathers
reuse values from the same location in memory, but it's specifically
helpful when data with AOS layout is being accessed; in this case,
we're often able to generate wide vector loads and appropriate shuffles
automatically.
When the --fuzz-test command-line option is given, the input program
will be randomly perturbed by the lexer in an effort to trigger
assertions or crashes in the compiler (neither of which should ever
happen, even for malformed programs.)
