srenblad
/
artiq
forked from M-Labs/artiq
1
0
Fork 0
Code Pull Requests Activity
2,733 Commits 63 Branches 34 Tags 19 MiB
Commit Graph

449 Commits

Author SHA1 Message Date
whitequark ff0ab736e9 Commit missing parts of 8aa34ee9. 2015-12-31 22:36:25 +08:00
whitequark 79d020dd3a transforms.artiq_ir_generator: handle terminated try body. 2015-12-31 22:36:25 +08:00
whitequark 8aa34ee952 compiler: don't require exceptions to inherit ARTIQException. 2015-12-31 21:54:54 +08:00
whitequark 0b69e488e6 transforms.llvm_ir_generator: compare exn typeinfo using strcmp. 
This is necessary to correctly catch exceptions from RPCs.
 2015-12-31 19:53:28 +08:00
whitequark 9ed6b54249 transforms.cfg_simplifier: remove. 
Specifically, the following idiom broke our CFGSimplifier:

b1:
  branch_if %cond, %b2, %b3
b2:
  branch %b3
b3:
  %v = phi [ %b2 => 0, %b3 => 1 ]

This is very annoying to properly handle in general case, and also
this (specifically %b2) is exactly what broke LocalAccessValidator
in the first place. So, remove CFGSimplifier altogether.

Instead, teach LocalAccessValidator how to look through predecessors
in order to get a location.
 2015-12-31 17:07:36 +08:00
whitequark 787a1d34fe transforms.llvm_ir_generator: don't crash when quoting over than one list. 2015-12-30 16:22:26 +08:00
whitequark 25e2824709 validators.escape: infer correct region for arguments. 2015-12-30 16:19:35 +08:00
whitequark 72b6cca9c3 transforms.cfg_simplifier: correctly update phis. 2015-12-30 16:07:14 +08:00
whitequark 2a474b7166 ir: fix incoming_{blocks,values,value_for_block}. 2015-12-30 16:06:18 +08:00
whitequark ac3a170c8f transforms.llvm_ir_generator: don't crash when quoting empty lists. 2015-12-30 15:47:57 +08:00
whitequark 5dcf936f4e transforms.inferencer: don't crash on partially-inferred binary expressions. 2015-12-30 15:46:54 +08:00
whitequark 6938036b28 embedding: unify all derived attribute types. 2015-12-30 15:38:49 +08:00
whitequark f52d364806 Mollify PEP 0479. 2015-12-30 15:33:30 +08:00
whitequark 78fb3e1b7b compiler: validate local accesses after interleaving. 2015-12-30 15:26:30 +08:00
whitequark df91500f68 Fix ff3206be. 2015-12-30 15:16:05 +08:00
whitequark ff3206be07 compiler.embedding: treat RPC methods like any others (fixes #211). 2015-12-30 14:21:09 +08:00
whitequark 14484216ec transforms.cfg_simplifier: handle phi and branch_if properly. 2015-12-30 12:10:28 +08:00
whitequark 8822db089f transforms.cfg_simplifier: implement. 
This allows error reporting in LocalAccessValidator to proceed
even when the predecessor is an empty block with only a branch.
 2015-12-29 02:15:57 +08:00
whitequark 57ebd57ddb transforms.dead_code_eliminator: update doc. 2015-12-29 02:15:57 +08:00
whitequark 82ec76af3e compiler.types: fix TFunction internal field order (closes #208). 2015-12-26 18:08:51 +08:00
whitequark b931096ec3 transforms.artiq_ir_generator: fix While codegen (closes #207). 2015-12-26 17:24:05 +08:00
whitequark 502e570e86 compiler: embed host exception constructors as such (fixes #204). 2015-12-26 03:17:29 +08:00
whitequark 8051fe9592 Commit missing parts of 082e9e20dd. 2015-12-26 03:16:50 +08:00
whitequark 082e9e20dd compiler: do not associate SSA values with iodelay even when inlining. 
Fixes #201.
 2015-12-25 15:02:33 +08:00
whitequark 33c3b3377e ir: keep loc when copying. 2015-12-25 14:59:28 +08:00
whitequark 690b2fd034 transforms.artiq_ir_generator: fix optarg codegen (closes #205). 2015-12-25 12:40:45 +08:00
whitequark 25188f0ca9 transforms.interleaver: correctly handle degenerate `with parallel:` blocks. 2015-12-21 21:32:48 +08:00
whitequark f957be4e6f transforms.llvm_ir_generator: handle loop instruction (fixes #202). 2015-12-21 21:12:17 +08:00
whitequark e4615e7b37 transforms.int_monomorphizer: visit children of CallT. 2015-12-20 18:07:31 +08:00
whitequark 54aeb5030d transforms.int_monomorphizer: also monomorphize round() calls (#198). 2015-12-20 02:36:30 +08:00
whitequark 95af6daa28 transforms.inferencer: accept round(width=n) form (#198). 2015-12-20 01:02:57 +08:00
whitequark 4fcd6abc5e Commit missing parts of 0395efd4. 2015-12-19 06:33:29 +08:00
whitequark 4fb1de33c9 Initial invocation of a @kernel function can now return a value (fixes #197). 2015-12-19 05:26:18 +08:00
whitequark 0395efd479 compiler: give environment types in LLVM IR readable names. 2015-12-18 23:41:51 +08:00
whitequark baa986aa32 compiler.prelude: add @portable as an alias for @kernel. 2015-12-18 23:00:29 +08:00
whitequark 7b3ace2e30 transforms.inferencer: fix unsupported decorator diagnostic when embedding. 
decorator.loc points to the quoted object with a synthesized location,
and it's not really worth refactoring when we can just point at the @.
 2015-12-18 22:58:17 +08:00
whitequark 0755aa38ff transforms.iodelay_estimator: allow goto in zero-iodelay while statements. 2015-12-18 21:14:40 +08:00
whitequark 2759310662 transforms.iodelay_estimator: reject control flow in 'with parallel:' (fixes #195). 2015-12-18 21:02:53 +08:00
whitequark f4b19fee5c compiler.types: make TValue hashable. 2015-12-18 17:31:20 +08:00
whitequark 3fbee2707b analyses.domination: consider unreachable blocks dominated by any other. 
As a result, the dominator tree can now process arbitrary (reducible)
CFGs and we do not run DCE before analyses, risking loss of
correspondence to the AST, which would arbitrarily silence analyses.
 2015-12-18 16:39:52 +08:00
whitequark 8cb7844621 transforms.interleaver: unroll loops. 2015-12-17 00:52:22 +08:00
whitequark 5dd1fc993e transforms.dead_code_eliminator: also remove dead instructions. 2015-12-17 00:52:22 +08:00
whitequark 2570932264 transforms.interleaver: don't fail on delay-free loops/conditionals. 2015-12-16 19:26:17 +08:00
whitequark 69cdeaf706 transforms.interleaver: don't fail on degenerate parallel blocks. 2015-12-16 19:26:17 +08:00
whitequark f8eaeaa43f compiler: explicitly represent loops in IR. 2015-12-16 15:33:26 +08:00
whitequark 35acc33ef6 validators.escape: don't fail on quoted values in lhs. 2015-12-16 13:57:02 +08:00
whitequark 8751d2ee6c Delay.{expr→interval}. 2015-12-16 13:57:02 +08:00
whitequark e755e1ba9e validators.escape: support arrays in multiple assignment. 2015-12-14 16:31:03 +08:00
whitequark 25027f6907 compiler: fix tests. 2015-12-10 23:16:36 +08:00
whitequark 2d906daf7f compiler: implement 'with watchdog' support. 2015-12-10 23:11:00 +08:00
whitequark 64f19b84f2 compiler.testbench.llvmgen: fix. 2015-12-10 23:08:05 +08:00
whitequark 4888e897b9 compiler: refactor to use builtins.TInt{32,64}. (NFC) 2015-12-10 23:06:23 +08:00
whitequark 143bae4bd3 compiler.builtins: {fn→obj}_{serial,parallel}. 2015-12-10 23:03:28 +08:00
whitequark c660028abf transforms.llvm_ir_generator: fix reraise inside EH block. 2015-12-10 23:02:32 +08:00
whitequark 3ec9b677b2 Revert "transforms.artiq_ir_generator: treat builtins in 'with' syntactically." 
This reverts commit ccc993071b.
 2015-12-10 22:25:15 +08:00
whitequark ccc993071b transforms.artiq_ir_generator: treat builtins in 'with' syntactically. 2015-12-10 21:29:16 +08:00
whitequark 22dab0c7cc compiler: clarify confusing comment (NFC). 2015-12-10 20:50:28 +08:00
whitequark 8cb2215edb transforms.inferencer: only instantiate RPC function types, not regular. 2015-12-02 21:48:14 +08:00
whitequark 3c9b53b07b compiler.embedding: add fast path for inference for lists of scalars in a host object. 2015-11-27 19:06:04 +08:00
whitequark 237e983770 compiler.types: replace isinstance() with .__class__==. 
This provides a 50% speedup.
 2015-11-27 18:22:55 +08:00
whitequark 6122fd70ca compiler.types: TVar.find: improve path compression. 
After this change, the compiler's complexity is nearly linear
even when very large arrays are used.
 2015-11-27 18:08:46 +08:00
whitequark bf1a583fda compiler.types: fix TVar.find() for very large paths. 2015-11-27 18:02:27 +08:00
whitequark 00164390a1 compiler.types: fix TDelay.unify(TVar()). 2015-11-27 17:10:36 +08:00
whitequark 4b410ce720 compiler.testbench.perf_embedding: fix. 2015-11-27 17:08:45 +08:00
whitequark 16ae0fb6eb compiler.embedding: instantiate RPC method types (fixes #180). 2015-11-27 16:29:13 +08:00
whitequark 0a794fe7e4 compiler.embedding: instantiate type of even the first embedding of an RPC. 2015-11-27 14:55:47 +08:00
whitequark cde21bcd26 compiler.embedding: instantiate RPC function types (#180). 2015-11-27 14:29:15 +08:00
whitequark 3b529c6f90 compiler.types: implement map. 2015-11-26 11:16:39 +08:00
whitequark e2f7d1047d compiler.testbench: fix after e1cd2ccd. 2015-11-26 10:24:14 +08:00
whitequark d14ad6727a compiler.embedding: avoid showing wildly incorrect suggestions. 2015-11-24 17:50:46 +08:00
whitequark 14993e89e2 compiler.embedding: show suggestions for mistyped host object attributes. 2015-11-24 17:44:58 +08:00
whitequark e1cd2ccd40 compiler: pull in dependencies in more finely grained way (fixes #181). 2015-11-24 17:32:04 +08:00
whitequark 66b1388a63 transforms.artiq_ir_generator: never put TVars in dicts. 
A TVar looks just like whatever it points to, but it does not
compare equal, nor is its hash the same.
 2015-11-24 02:59:15 +08:00
whitequark fec5c2ebf0 transforms.interleaver: add a diagnostic for interleave inlining failure. 2015-11-24 02:57:03 +08:00
whitequark 8527e306c3 testbench.embedding: use dmgr to get core and export it. 2015-11-24 02:02:34 +08:00
whitequark 2bfc72fba9 testbench.embedding: fix ref_period mismatch. 2015-11-24 00:54:20 +08:00
whitequark d3f0059cab compiler.iodelay: correctly fold max(0, [0, ]...). 2015-11-24 00:46:55 +08:00
whitequark 9bc62fa3d2 transforms.iodelay_estimator: correctly handle functions with empty body. 2015-11-24 00:46:26 +08:00
whitequark 32fe4a8a0c transforms.llvm_ir_generator: don't assert on inlined functions. 2015-11-24 00:20:33 +08:00
whitequark abb36b42be compiler.iodelay: fold and eval SToMU to an int, not float. 2015-11-24 00:19:33 +08:00
whitequark 178ff74da2 transforms.interleaver: inline calls. 2015-11-24 00:02:07 +08:00
whitequark 2a82eb7219 compiler.ir: return dict from Delay.substs, not pair iterable. 2015-11-24 00:01:10 +08:00
whitequark 02f2763ea8 compiler.iodelay: always fully fold SToMU and MUToS. 2015-11-23 23:59:59 +08:00
whitequark f3da227e2d compiler.ir: change argument order for BasicBlock.insert. 2015-11-23 23:59:25 +08:00
whitequark f0fd6cd0ca compiler.algorithms.inline: implement. 2015-11-23 23:58:37 +08:00
whitequark a4525b21cf compiler.ir: print even blocks without predecessors. 2015-11-23 23:55:12 +08:00
whitequark d92b3434a0 compiler.ir: print basic blocks in reverse postorder for readability. 2015-11-23 21:44:38 +08:00
whitequark c73b2c1a78 compiler.ir: fix typo. 2015-11-23 21:21:01 +08:00
whitequark 0bf425eefa compiler.ir: maintain use lists while mutating instructions. 2015-11-23 19:18:58 +08:00
whitequark 03b4e4027c transforms.interleaver: fix IR type/value mismatch. 2015-11-23 18:53:42 +08:00
whitequark 73845279ae transforms.interleaver: determine when inlining is not necessary. 2015-11-23 18:08:33 +08:00
whitequark a01e328b4a transforms.interleaver: don't assume all delay expressions are folded. 2015-11-21 17:24:00 +08:00
whitequark 5cd12ffd28 compiler.iodelay: fold MUToS and SToMU. 2015-11-21 17:23:20 +08:00
whitequark 82b470891f transforms.interleaver: handle function calls (as atomic so far). 
This commit solves issue #2 described in 50e7b44; a function call
is now a valid decomposition for a delay instruction, and this
metadata is propagated when the interleaver converts delays.

However, the interleaver does not yet detect that a called function
is compound, i.e. it is not correct.
 2015-11-21 03:34:24 +08:00
whitequark 57dd163d37 transforms.artiq_ir_generator: fix decomposition of explicit delay_mu(). 2015-11-21 03:27:06 +08:00
whitequark cb3b811fd7 compiler: maintain both the IR and iodelay forms of delay expressions. 
After this commit, the delay instruction (again) does not generate
any LLVM IR: all heavy lifting is relegated to the delay and delay_mu
intrinsics. When the interleave transform needs to adjust the global
timeline, it synthesizes a delay_mu intrinsnic. This way,
the interleave transformation becomes composable, as the input and
the output IR invariants are the same.

Also, code generation is adjusted so that a basic block is split off
not only after a delay call, but also before one; otherwise, e.g.,
code immediately at the beginning of a `with parallel:` branch
would have no choice but to execute after another branch has already
advanced the timeline.

This takes care of issue #1 described in 50e7b44 and is a step
to solving issue #2.
 2015-11-21 03:22:47 +08:00
whitequark 50e7b44d04 compiler: actually implement interleaving correctly (calls are still broken). 
The previous implementation was completely wrong: it always advanced
the global timeline by the same amount as the non-interleaved basic
block did.

The new implementation only advances the global timeline by
the difference between its current time and the virtual time of
the branch, which requires it to adjust the delay instructions.

Previously, the delay expression was present in the IR twice: once
as the iodelay.Expr transformation-visible form, and once as regular
IR instructions, with the latter form being passed to the delay_mu
builtin and advancing the runtime timeline.

As a result of this change, this strategy is no longer valid:
we can meaningfully mutate the iodelay.Expr form but not the IR
instruction form. Thus, IR instructions are no longer generated for
delay expressions, and the LLVM lowering pass now has to lower
the iodelay.Expr objects as well.

This works OK for flat `with parallel:` expressions, but breaks down
outside of `with parallel:` or when calls are present. The reasons
it breaks down are as follows:

  * Outside of `with parallel:`, delay() and delay_mu() must accept
    any expression, but iodelay.Expr's are not nearly expressive
    enough. So, the IR instruction form must actually be kept as well.

  * A delay instruction is currently inserted after a call to
    a user-defined function; this delay instruction introduces
    a point where basic block reordering is possible as well as
    provides delay information. However, the callee knows nothing
    about the context in which it is called, which means that
    the runtime timeline is advanced twice. So, a new terminator
    instruction must be added that combines the properties of delay
    and call instructions (and another for delay and invoke as well).
 2015-11-21 00:02:47 +08:00
whitequark 73c358a59a Reformat. 2015-11-20 23:33:06 +08:00
whitequark cb9e7d15bf compiler.iodelay: subtraction certainly shouldn't be commutative. 2015-11-20 22:15:03 +08:00
whitequark cc623c13b4 Reformat. 2015-11-20 21:45:34 +08:00