-
Notifications
You must be signed in to change notification settings - Fork 201
Comparing changes
Open a pull request
base repository: m-labs/artiq
base: 321ba57e84dd
head repository: m-labs/artiq
compare: d6510083b747
Commits on Apr 14, 2016
-
Configuration menu - View commit details
-
Copy full SHA for 00facbb - Browse repository at this point
Copy the full SHA 00facbbView commit details -
transforms.llvm_ir_generator: extract class function attributes.
This should give LLVM more visibility.
Configuration menu - View commit details
-
Copy full SHA for f72e050 - Browse repository at this point
Copy the full SHA f72e050View commit details -
Configuration menu - View commit details
-
Copy full SHA for e534941 - Browse repository at this point
Copy the full SHA e534941View commit details -
Configuration menu - View commit details
-
Copy full SHA for 8fa4281 - Browse repository at this point
Copy the full SHA 8fa4281View commit details -
Configuration menu - View commit details
-
Copy full SHA for ccb1d54 - Browse repository at this point
Copy the full SHA ccb1d54View commit details -
Configuration menu - View commit details
-
Copy full SHA for f2c92ff - Browse repository at this point
Copy the full SHA f2c92ffView commit details -
llvm_ir_generator: generate code more amenable to LLVM's GlobalOpt.
This exposes almost all embedded methods to inlining, with massive gains.
Configuration menu - View commit details
-
Copy full SHA for 92f3dc7 - Browse repository at this point
Copy the full SHA 92f3dc7View commit details -
Configuration menu - View commit details
-
Copy full SHA for fcf2a73 - Browse repository at this point
Copy the full SHA fcf2a73View commit details -
compiler: extract runtime checks into separate cold functions.
This reduces register pressure as well as function size, which favorably affects the inliner.
Configuration menu - View commit details
-
Copy full SHA for 3fa5762 - Browse repository at this point
Copy the full SHA 3fa5762View commit details -
Configuration menu - View commit details
-
Copy full SHA for 019f528 - Browse repository at this point
Copy the full SHA 019f528View commit details -
This doesn't do much, only frees some registers.
Configuration menu - View commit details
-
Copy full SHA for 718d411 - Browse repository at this point
Copy the full SHA 718d411View commit details -
Configuration menu - View commit details
-
Copy full SHA for 348e058 - Browse repository at this point
Copy the full SHA 348e058View commit details -
Configuration menu - View commit details
-
Copy full SHA for 665e59e - Browse repository at this point
Copy the full SHA 665e59eView commit details -
Configuration menu - View commit details
-
Copy full SHA for b5fd257 - Browse repository at this point
Copy the full SHA b5fd257View commit details -
Configuration menu - View commit details
-
Copy full SHA for 081edb2 - Browse repository at this point
Copy the full SHA 081edb2View commit details -
compiler: mark loads of kernel constant attributes as load invariant.
Also, enable LICM, since it can take advantage of this.
Configuration menu - View commit details
-
Copy full SHA for 0e0f81b - Browse repository at this point
Copy the full SHA 0e0f81bView commit details -
compiler: mark FFI functions as ModRef=Ref using TBAA metadata.
Fascinatingly, the fact that you can mark call instructions with !tbaa metadata is completely undocumented. Regardless, it is true: a !tbaa metadata for an "immutable" type will cause AliasAnalysis::getModRefBehavior to return OnlyReadsMemory for that call site. Don't bother marking loads with TBAA yet since we already place !load.invariant on them (which is as good as the TBAA "immutable" flag) and after that we're limited by lack of !nonnull anyway. Also, add TBAA analysis passes in our pipeline to actually engage it.
Configuration menu - View commit details
-
Copy full SHA for c6b2165 - Browse repository at this point
Copy the full SHA c6b2165View commit details -
compiler: allow flagging syscalls, providing information to optimizer.
This also fixes a crash in test_cache introduced in 1d8b0d4.
Configuration menu - View commit details
-
Copy full SHA for f7d4a37 - Browse repository at this point
Copy the full SHA f7d4a37View commit details -
Configuration menu - View commit details
-
Copy full SHA for 5fafcc1 - Browse repository at this point
Copy the full SHA 5fafcc1View commit details -
Configuration menu - View commit details
-
Copy full SHA for ee7e648 - Browse repository at this point
Copy the full SHA ee7e648View commit details -
Configuration menu - View commit details
-
Copy full SHA for dcb0ffd - Browse repository at this point
Copy the full SHA dcb0ffdView commit details -
Configuration menu - View commit details
-
Copy full SHA for 812e79b - Browse repository at this point
Copy the full SHA 812e79bView commit details -
Configuration menu - View commit details
-
Copy full SHA for f7603dc - Browse repository at this point
Copy the full SHA f7603dcView commit details -
test: relax lit/embedding/syscall_flags.
We currently have broken debug info. In either case, debug info is irrelevant to this test.
Configuration menu - View commit details
-
Copy full SHA for 50ae176 - Browse repository at this point
Copy the full SHA 50ae176View commit details -
Configuration menu - View commit details
-
Copy full SHA for e416246 - Browse repository at this point
Copy the full SHA e416246View commit details -
llvm_ir_generator: do not use 'coldcc' calling convention.
First, this calling convention doesn't actually exist in OR1K and trying to use it in Asserts build causes an UNREACHABLE. Second, I tried to introduce it and it does not appear to produce any measurable benefit: not only OR1K has a ton of CSRs but also it is quite hard, if not realistically impossible, to produce the kind of register pressure that would be relieved by sparing a few more CSRs for our exception raising function calls, since temporaries don't have to be preserved before a noreturn call and spilling over ten registers across an exceptional edge is not something that the code we care about would do. Third, it produces measurable drawbacks: it inflates code size of check:* functions by adding spills. Of course, this could be alleviated by making __artiq_raise coldcc as well, but what's the point anyway?
Configuration menu - View commit details
-
Copy full SHA for f59fd8f - Browse repository at this point
Copy the full SHA f59fd8fView commit details -
Configuration menu - View commit details
-
Copy full SHA for 932e680 - Browse repository at this point
Copy the full SHA 932e680View commit details -
Configuration menu - View commit details
-
Copy full SHA for 03b6555 - Browse repository at this point
Copy the full SHA 03b6555View commit details -
compiler: mark loaded pointers as !dereferenceable.
Also, lower the bound for test_pulse_rate_dds, since we generate better code for it now.
Configuration menu - View commit details
-
Copy full SHA for 1120c26 - Browse repository at this point
Copy the full SHA 1120c26View commit details -
compiler: add analysis passes from TargetMachine.
This doesn't have any effect right now, but is the right thing to do.
Configuration menu - View commit details
-
Copy full SHA for c89c27e - Browse repository at this point
Copy the full SHA c89c27eView commit details -
Configuration menu - View commit details
-
Copy full SHA for f26990a - Browse repository at this point
Copy the full SHA f26990aView commit details -
Configuration menu - View commit details
-
Copy full SHA for 66bbee5 - Browse repository at this point
Copy the full SHA 66bbee5View commit details -
coredevice: format backtrace RA as +0xN, not 0xN.
The absolute address is somewhere in the 0x4000000 range; the one that is displayed is an offset from the shared object base.
Configuration menu - View commit details
-
Copy full SHA for 7c520aa - Browse repository at this point
Copy the full SHA 7c520aaView commit details -
Configuration menu - View commit details
-
Copy full SHA for f958cba - Browse repository at this point
Copy the full SHA f958cbaView commit details -
Configuration menu - View commit details
-
Copy full SHA for 702e959 - Browse repository at this point
Copy the full SHA 702e959View commit details -
Configuration menu - View commit details
-
Copy full SHA for a2f6e81 - Browse repository at this point
Copy the full SHA a2f6e81View commit details -
Configuration menu - View commit details
-
Copy full SHA for 89326fb - Browse repository at this point
Copy the full SHA 89326fbView commit details -
conda: update llvmlite-artiq dependency.
Build 22 includes debug information support.
Configuration menu - View commit details
-
Copy full SHA for 31b5154 - Browse repository at this point
Copy the full SHA 31b5154View commit details -
Configuration menu - View commit details
-
Copy full SHA for 75252ca - Browse repository at this point
Copy the full SHA 75252caView commit details -
llvm_ir_generator: change !{→unconditionally_}dereferenceable.
Since LLVM 3.8, !dereferenceable is weaker, so we introduce !unconditionally_dereferenceable (http://reviews.llvm.org/D18738) to regain its functionality.
Configuration menu - View commit details
-
Copy full SHA for d4f1614 - Browse repository at this point
Copy the full SHA d4f1614View commit details -
llvm_ir_generator: add fast-math flags to fcmp.
This is allowed in 3.8.
Configuration menu - View commit details
-
Copy full SHA for c6f946a - Browse repository at this point
Copy the full SHA c6f946aView commit details -
conda: update llvmlite-artiq dependency.
Build 24 includes addc optimizations.
Configuration menu - View commit details
-
Copy full SHA for 3445462 - Browse repository at this point
Copy the full SHA 3445462View commit details -
Configuration menu - View commit details
-
Copy full SHA for e6666ce - Browse repository at this point
Copy the full SHA e6666ceView commit details -
Configuration menu - View commit details
-
Copy full SHA for 2248a2e - Browse repository at this point
Copy the full SHA 2248a2eView commit details -
Configuration menu - View commit details
-
Copy full SHA for 904379d - Browse repository at this point
Copy the full SHA 904379dView commit details -
Configuration menu - View commit details
-
Copy full SHA for d651008 - Browse repository at this point
Copy the full SHA d651008View commit details
There are no files selected for viewing