You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We need some overview documents to help ghidra users get a quick estimate of what a piece of extension code is doing.
What patterns are generated when?
local optimization of similar scalar reads and writes, especially in initialization code
vectorization of simple loops
processing of vector instrinsic code
vectorization of complex loops, especially iterating transforms over arrays of structures
Group vector instructions by complexity
vector math and logical operations are easy
vector gather, scatter, slideup, slidedown, segmented, strided, and indexed ops are more confusing
Gather concepts and glossary items
element width, multipliers, segments, group, and all of the acronyms used in their definitions
what rules does a compiler follow when deciding how to vectorize scalar code - exceptions, alignment, vector register pressure, memory latency, and branch prediction? How are these options passed to the compiler and to the runtime execution environment?
Modelling
extract feature elements from binary code to help categorize semantics - how many reads and writes per loop, what are the likely type conversions present, is this an example of loop unrolling?
what should the Ghidra decompiler window show when vector instructions are found?
should Ghidra track the vector context registers?
is there a vector or functional notation that captures vectorized code better than C?
The text was updated successfully, but these errors were encountered:
We need some overview documents to help ghidra users get a quick estimate of what a piece of extension code is doing.
What patterns are generated when?
Group vector instructions by complexity
Gather concepts and glossary items
Modelling
The text was updated successfully, but these errors were encountered: