gaspersic/triton
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[DOC] Fixed formatting issues in tutorial

Browse Source
This commit is contained in:
Philippe Tillet
2020-02-10 03:27:36 -05:00
committed by Philippe Tillet
parent 5778eea546
commit 97293440fc
2 changed files with 10 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
docs/tutorials/matrix-multiplication.rst
View File

@@ -35,7 +35,7 @@ Matrix multiplications of the form `C = A x B` can be implemented in Triton-C fa
TYPE a[TM, TK] = *pa; //(9) TYPE a[TM, TK] = *pa; //(9)
TYPE b[TK, TN] = *pb; //(10) TYPE b[TK, TN] = *pb; //(10)
// matrix-multiply accumulate // matrix-multiply accumulate
c += a @ b; //(11) c += dot(a, b); //(11)
// increment pointers // increment pointers
pa = pa + TK * 1; //(12) pa = pa + TK * 1; //(12)
pb = pb + TK * ldb; //(13) pb = pb + TK * ldb; //(13)
@@ -88,7 +88,7 @@ The purpose of pre-fetching is to overlap the update of the accumulator `c` with
TYPE a[TM, TK] = *pa; //(9) TYPE a[TM, TK] = *pa; //(9)
TYPE b[TK, TN] = *pb; //(10) TYPE b[TK, TN] = *pb; //(10)
for(int k = K; k > 0; k-= TK){ for(int k = K; k > 0; k-= TK){
c += a @ b; c += dot(a, b);
pa = pa + TK * 1; pa = pa + TK * 1;
pb = pb + TK * ldb; pb = pb + TK * ldb;
// don't prefetch last iteration // don't prefetch last iteration
@@ -144,7 +144,7 @@ It is common for optimized matrix-multiplication implementations (e.g., BLAS) to
TYPE b[SHAPE_B] = (*pb); TYPE b[SHAPE_B] = (*pb);
// reduction loop // reduction loop
for(int k = K; k > 0; k-= TK){ for(int k = K; k > 0; k-= TK){
c += USE_A @ USE_B; c += dot(USE_A, USE_B);
pa = pa + TK * STRIDE_AK; pa = pa + TK * STRIDE_AK;
pb = pb + TK * STRIDE_BK; pb = pb + TK * STRIDE_BK;
a = *pa; a = *pa;

4
docs/tutorials/matrix-transposition.rst
View File

@@ -53,7 +53,7 @@ which will be used in statements (5) and (6) to construct tiles of pointers
- Statements (5) constructs the following array of pointers `px` using numpy-style broadcasting semantics: - Statements (5) constructs the following array of pointers `px` using numpy-style broadcasting semantics:
.. code-block:: C ::
│ X + (pidm*TM + 0) + (pidn*TN + 0)*ldx, ..., ..., X + (pidm*TM + 0) + (pidn*TN + TN - 1)*ldx) │ │ X + (pidm*TM + 0) + (pidn*TN + 0)*ldx, ..., ..., X + (pidm*TM + 0) + (pidn*TN + TN - 1)*ldx) │
│ ⋮ ⋮ │ │ ⋮ ⋮ │
@@ -63,7 +63,7 @@ which will be used in statements (5) and (6) to construct tiles of pointers
- Statement (6) constructs the following array of pointers `py` using numpy-style broadcasting semantics: - Statement (6) constructs the following array of pointers `py` using numpy-style broadcasting semantics:
.. code-block:: C ::
│ Y + (pidn*TN + 0) + (pidm*TM + 0)*ldy, ..., ..., Y + (pidn*TN + 0) + (pidm*TM + TM - 1)*ldy) │ │ Y + (pidn*TN + 0) + (pidm*TM + 0)*ldy, ..., ..., Y + (pidn*TN + 0) + (pidm*TM + TM - 1)*ldy) │
│ ⋮ ⋮ │ │ ⋮ ⋮ │