search for: isl_map

Hi, thanks to Tobias for doing most of the work to get the vector code generation working with ISL's code generation. Attached are two patches to port the vector code generation part of the testsuite from Cloog/CodeGen to Isl/CodeGen. Tobi, do you want to commit these two patches separately, or you want me to combine them, as the second patch is needed to make the testsuite pass? Also, let

...efficients a1, b1 > ... to check if they follow the linearity. I perform dynamic > instrumentation to compute the coefficients. Mh. I believe we should distinguish the data for Polly and for your calculations. I assumed we would use affine linear relations in the access functions (Actually isl_maps like {[i,j] -> List[10i + 30j + 10]) to define accesses such that Polly can use this access functions to calculate dependences and reschedule the code accordingly. The possibly non-affine accesses would then be hidden behind the virtual access. > However, for applying the transformation...

This is exactly want I need to achieve with Polly actually. I think a good idea would be to define intrinsics / metadata, as you mentioned, to notify Polly that even though it cannot analyse these accesses, to ignore them and perform the code transformations. We can go even further and maybe describe these accesses with some parametric linear functions. For instance: while (cond1){

..., Control, Flow, Anti, Output - Which instances depend on each other. This can be: - No instance information. All instances of S must be executed before any instance of T. - A vector of dependence distances. Basically MemoryDepChecker for each surrounding loop. - An polyhedral relational map (isl_map) describing which instances of T depend on which instances of S. The dependency type "register" means that T used an llvm::Value defined by S. This can also be modeled using flow and anti-dependencies but as shown before this can be destructive to transformation opportunitied. Transform...

** *Hi everyone,As you may know, stock LLVM does not provide the kind of advanced loop transformations necessary to provide good performance on many applications. LLVM's Polly project provides many of the required capabilities, including loop transformations such as fission, fusion, skewing, blocking/tiling, and interchange, all powered by state-of-the-art dependence analysis. Polly also