search for: ub3

...seful if someone can mention these examples using existing dialects in MLIR and also how the different transformations / lowerings are planned. B. In latest RFC(3), it is mentioned that the initial OpenMP dialect version will be as follows, omp.parallel { omp.do { fir.do %i = 0 to %ub3 : !fir.integer { ... } } } and then after the "LLVM conversion" it is converted as follows: omp.parallel { %ub3 = omp.do %i = 0 to %ub3 : !llvm.integer { ... } } a. Is it the same omp.do operation which now contains the bounds and indu...

...op operations as well. Example, affine.for and loop.for would have to be lowered to omp.do in different ways. >From slide 20 of fosdem presentation you mentioned, the LLVM + OpenMP dialect representation is as follows: ------------------------------ Mlir.region(…) { omp.parallel { %ub3 = … omp.do %i = 0 to %ub3 : !integer { … } } } ------------------------------- Currently, the LLVM Dialect doesn’t contain a high level loop operation. It is all based on CFG implementation. Will omp.do follow the same structure (SizedRegion<1>) as loop.for? OR there...

...gt; different ways. >> >> From slide 20 of fosdem presentation you mentioned, the LLVM + OpenMP >> dialect representation is as follows: >> >> ------------------------------ >> >> Mlir.region(…) { >> >> omp.parallel { >> >> %ub3 = … >> >> omp.do %i = 0 to %ub3 : !integer { >> >> … >> >> } >> >> } >> >> } >> >> ------------------------------- >> >> Currently, the LLVM Dialect doesn’t contain a high level loop operation. &gt...

...on you mentioned, the LLVM + OpenMP >>>> dialect representation is as follows: >>>> >>>> ------------------------------ >>>> >>>> Mlir.region(…) { >>>> >>>> omp.parallel { >>>> >>>> %ub3 = … >>>> >>>> omp.do %i = 0 to %ub3 : !integer { >>>> >>>> … >>>> >>>> } >>>> >>>> } >>>> >>>> } >>>> >>>> ------------------------------...

...ferent > > transformations / lowerings are planned. > > > > B. In latest RFC(3), it is mentioned that the initial OpenMP dialect > > version will be as follows, > > > > omp.parallel { > > > > omp.do { > > > > fir.do %i = 0 to %ub3 : !fir.integer { > > > > ... > > > > } > > > > } > > > > } > > > > and then after the "LLVM conversion" it is converted as follows: > > > > omp.parallel { > > > > %ub3 = > &gt...

...Fm)`s5m&>LPA2{ zvxLGxg(T^}i)W<4(PO4GdG9X*UY%EIZ~(BV8%WT|uW0|>Wiklk@L^Nr4@URLh4gwX zJyEvMT4H3vmB0tDSAT|_6r*}f3_Qe+jqBJvwkh%ZSEDPHEutbxA}LYZ*gUQo{~S8a zkvaoWnaTStR3_#)Y7=GK?Tc?<rp?CELtYt`etIwQ*5MBv8~{G?TEWSqhlyJJ7NVv^ z?>^4WUgX6e&hS#--ps$$o%gN|#i~^Q@4QfHQKP{DK&MoZ;^N7<kk(ub3P&UF-Efp7 z;q`|-|FbezIYz()g+eh^6h-rhhzM!+$M4fL_>e%+OHe4USFU!7f8?VnUl9{ipH=I& z5#%+A-l~s4s{Y=CsvgpFvcc5WNv|wHNp>nKm3&Uc#bs{}q3^B7q49$$X&Tu$)8dnT z9G}~V6H(g<Yg+g~#}{gh_;~LdXhj1g1Y#Pr(D$Qx#zT^B<i^B+e`%WW7rlP{0k1Dd zp>l12Rh@b$)rRC}rjd8~2<hrN{N^~BLmqt@baOf-1@gDdx}9F5-R2f...