more compute intensive hello world example #1872
Conversation
jlewitt1
force-pushed
the
compute-intensive-example
branch
2 times, most recently
from
defddbc to
3e4a3d3
Compare
jlewitt1
changed the title
jlewitt1
force-pushed
the
compute-intensive-example
branch
2 times, most recently
from
f23eb9f to
67badfa
Compare
README.md
Outdated
| for i in range(num_iterations): | ||
| a = np.random.rand(matrix_size, matrix_size) | ||
| b = np.random.rand(matrix_size, matrix_size) | ||
|
|
||
| product = a @ b | ||
| iteration_sum = np.sum(product) | ||
| total_sum += iteration_sum | ||
|
|
||
| print(f"Iteration {i+1}: sum={iteration_sum:.3f}") | ||
| time.sleep(1) |
There was a problem hiding this comment.
down for a slightly more involved hello world that numpy to showcase the compute/image, but I do think the content here in the for loop is not super necessary and the focus should be on the kt portion. thoughts on cutting it down more simply? e.g.
for i in range(num_iterations):
total_sum += np.rand()
print(f"Iteration {i+1}: sum={iteration_sum:.3f}")
time.sleep(1)
There was a problem hiding this comment.
i agree the code isn't so meaningful, the only benefit is that it allows us to stream utilization metrics back.. so i'm a little torn
carolineechen
left a comment
carolineechen
left a comment
There was a problem hiding this comment.
left a comment on potentially simplifying the numpy function. but don't feel super strongly if you think leaving it as is showcases the use case better
jlewitt1
force-pushed
the
compute-intensive-example
branch
from
67badfa to
615f6ae
Compare
| def simulate_workload(num_iterations: int = 5, vector_size: int = 10000): | ||
| import numpy as np | ||
|
|
||
| return sum(np.dot(np.random.rand(vector_size), np.random.rand(vector_size)) for _ in range(num_iterations)) |
There was a problem hiding this comment.
yup, can we also pass in new (kw)args to the remote_fn call below to also show that the function takes in args just as the original python function does?
No description provided.