chenlin/Mujoco_WASM
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Allow dot products with sparse vectors to specify that one vector uses uncompressed memory.

Browse Source 
PiperOrigin-RevId: 561951337
Change-Id: I1310d7c09d9a85ad9b43877155844a9d0ce6edab
This commit is contained in:
Yuval Tassa
2023-09-01 07:39:27 -07:00
committed by Copybara-Service
parent 70c5fa50f2
commit 29aa5e4a41
7 changed files with 143 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/engine/engine_util_solve.c
+1 -1
View File
@@ -235,7 +235,7 @@ void mju_cholSolveSparse(mjtNum* res, const mjtNum* mat, const mjtNum* vec, int
// x(i) -= sum_j L(i,j)*x(j), j=0:i-1
if (nnz > 1) {
res[i] -= mju_dotSparse(mat+adr, res, nnz-1, colind+adr);
res[i] -= mju_dotSparse(mat+adr, res, nnz-1, colind+adr, /*flg_unc1=*/0);
// modulo AVX, the above line does
// for (int j=0; j<nnz-1; j++)
// res[i] -= mat[adr+j]*res[colind[adr+j]];