pfalcon / graph-llvm-ir
Licence: other
Visualization of LLVM IR
Stars: ✭ 43
Programming Languages
python
139335 projects - #7 most used programming language
Labels
Projects that are alternatives of or similar to graph-llvm-ir
nolimix86
LLVM-based x86 emulator with support for unlimited virtual registers, used before the register allocation pass
Stars: ✭ 19 (-55.81%)
Mutual labels: llvm
js-ziju
Compile javascript to LLVM IR, x86 assembly and self interpreting
Stars: ✭ 112 (+160.47%)
Mutual labels: llvm
llvm-kaleidoscope
LLVM Tutorial: Kaleidoscope (Implementing a Language with LLVM)
Stars: ✭ 124 (+188.37%)
Mutual labels: llvm
llvm-statepoint-utils
Runtime support for LLVM's GC Statepoints
Stars: ✭ 35 (-18.6%)
Mutual labels: llvm
neeilang
Fast, type-safe, object-oriented language by yours truly
Stars: ✭ 42 (-2.33%)
Mutual labels: llvm
lleaves
Compiler for LightGBM gradient-boosted trees, based on LLVM. Speeds up prediction by ≥10x.
Stars: ✭ 132 (+206.98%)
Mutual labels: llvm
llvm-project-prepo
Fork of LLVM with modifications to support a program repository
Stars: ✭ 27 (-37.21%)
Mutual labels: llvm
cosmix
A Compiler-based System for Secure Memory Instrumentation and Execution in Enclaves
Stars: ✭ 22 (-48.84%)
Mutual labels: llvm
(Useful) combinations of rendering options: --control (default): Renders both explicit control flow present in LLVM IR (sequential between statements in basic blocks, jumps between basic blocks), and dataflow dependencies. Control flow has higher weight (that means that control flow edges tend to be more straight). --dag-control: Ignore explicit flow control present in LLVM IR and instead compute order of evaluation of independent (i.e. disconnected) dataflow DAGs within basic block. Root node of a DAG consider to be an instruction of type void. (The idea is that void instruction is executed solely for side effect, and then it must be last instruction in evaluation of some DAG, i.e. its root. This is clearly a heuristic, which needs to be tested on various inputs yet.) --block For both options above, you can add --block to clusterize instructions of the same basic block together within a rectangle block. This seems like natural way to do it, but leaves questions open where to put leaves of dataflow graphs (i.e. variables, constants, etc.) So far, these are rendered as DAG structure also, which means they are not part of any basic block cluster. But rendering them in such way leads to edges going from different basic blocks to the same leaf nodes, leading to a mess in the graph. Possible other options: duplicate leaf nodes; don't render at all (can be kinda assumed). --block-edges This makes control edges between basic blocks actually go between basic blocks, not specific instructions in them. This may be useful for some kinds of presentations. This also removes extra nodes to represent labels. Results of the latter changes are mixed though, it leads to not ideal placing of leaf non-cluster nodes and thus deformed graphs.
Note that the project description data, including the texts, logos, images, and/or trademarks,
for each open source project belongs to its rightful owner.
If you wish to add or remove any projects, please contact us at [email protected].