Skip to content

lessons: actually use allocated buffers in Step12 lesson #91

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

lessons: actually use allocated buffers in Step12 lesson #91

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 4 additions & 4 deletions lessons/15_Step_12.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -221,7 +221,7 @@
"outputs": [],
"source": [
"def pressure_poisson_periodic(p, dx, dy):\n",
" pn = numpy.empty_like(p)\n",
" pn[:] = numpy.empty_like(p)\n",
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

pn doesn’t exist yet, so pn[:] = ... will raise an error

even if it did exist, you’d just be copying junk from a freshly allocated array into pn—no gain here

" \n",
" for q in range(nit):\n",
" pn = p.copy()\n",
Expand Down Expand Up @@ -313,10 +313,10 @@
"stepcount = 0\n",
"\n",
"while udiff > .001:\n",
" un = u.copy()\n",
" vn = v.copy()\n",
" un[:] = u.copy()\n",
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

you allocate a temporary copy and then copy it again into un/vn. Double work.

Would suggest this instead (fastest without allocation):

un[...] = u          # or: numpy.copyto(un, u)
vn[...] = v          # or: numpy.copyto(vn, v)

" vn[:] = v.copy()\n",
"\n",
" b = build_up_b(rho, dt, dx, dy, u, v)\n",
" b[:] = build_up_b(rho, dt, dx, dy, u, v)\n",
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

build_up_b still allocates a new array internally and returns it; then you copy that result into b. That’s an extra full-array copy every iteration

to improve the performance here you would need to refactor build_up_b to write in-place into a provided buffer, something like:

def build_up_b_out(b, rho, dt, dx, dy, u, v):
    b[1:-1, 1:-1] = ...
    b[1:-1, -1]   = ...
    b[1:-1, 0]    = ...

" p = pressure_poisson_periodic(p, dx, dy)\n",
"\n",
" u[1:-1, 1:-1] = (un[1:-1, 1:-1] -\n",
Expand Down