Skip to content

Simplify copy kernel #28352

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

Jump to bottom
Merged
zasdfgbnm merged 1 commit into from
Oct 22, 2019
Merged

Simplify copy kernel #28352

zasdfgbnm merged 1 commit into from
Oct 22, 2019

Conversation

@zasdfgbnm
Copy link
Collaborator

@zasdfgbnm zasdfgbnm commented Oct 21, 2019 •
edited
Loading

Stack from ghstack:

Using the new type promotion and dynamic casting added to
TensorIterator, the copy kernels could be greatly simplified.

Script:

import torch
import timeit
import pandas
import itertools
from tqdm import tqdm
import math
print(torch.__version__)
print()

_10M = 10 * 1024 ** 2

d = {}

for from_, to in tqdm(itertools.product(torch.testing.get_all_dtypes(),
repeat=2)):
    if from_ not in d:
        d[from_] = {}
    a = torch.zeros(_10M, dtype=from_)
    min_ = math.inf
    for i in range(100):
        start = timeit.default_timer()
        a.to(to)
        end = timeit.default_timer()
        elapsed = end - start
        if elapsed < min_:
            min_ = elapsed
    d[from_][to] = int(elapsed * 1000 * 1000)

pandas.DataFrame(d)

Before:
image

After:
image

This was referenced Oct 21, 2019
Merged
Merged
zasdfgbnm added a commit that referenced this pull request Oct 21, 2019
@zasdfgbnm
Simplify copy kernel
1c1c778 
Using the new type promotion and dynamic casting added to
`TensorIterator`, the copy kernels could be greatly simplified.

**Script:**
```python
import torch
import timeit
import pandas
import itertools
from tqdm import tqdm
import math
print(torch.__version__)
print()

_10M = 10 * 1024 ** 2

d = {}

for from_, to in tqdm(itertools.product(torch.testing.get_all_dtypes(),
repeat=2)):
    if from_ not in d:
        d[from_] = {}
    a = torch.zeros(_10M, dtype=from_)
    min_ = math.inf
    for i in range(100):
        start = timeit.default_timer()
        a.to(to)
        end = timeit.default_timer()
        elapsed = end - start
        if elapsed < min_:
            min_ = elapsed
    d[from_][to] = int(elapsed * 1000 * 1000)

pandas.DataFrame(d)
```

**Before:**
![image](https://user-images.githubusercontent.com/1032377/67171274-2e93d000-f36b-11e9-8fa0-91edd7dbc8ec.png)

**After:**
![image](https://user-images.githubusercontent.com/1032377/67171200-d361dd80-f36a-11e9-9b22-66292e395a09.png)

ghstack-source-id: 3e43f97
Pull Request resolved: #28352
@zasdfgbnm
Simplify copy kernel
d3cbbca 
Using the new type promotion and dynamic casting added to
`TensorIterator`, the copy kernels could be greatly simplified.

**Script:**
```python
import torch
import timeit
import pandas
import itertools
from tqdm import tqdm
import math
print(torch.__version__)
print()

_10M = 10 * 1024 ** 2

d = {}

for from_, to in tqdm(itertools.product(torch.testing.get_all_dtypes(),
repeat=2)):
    if from_ not in d:
        d[from_] = {}
    a = torch.zeros(_10M, dtype=from_)
    min_ = math.inf
    for i in range(100):
        start = timeit.default_timer()
        a.to(to)
        end = timeit.default_timer()
        elapsed = end - start
        if elapsed < min_:
            min_ = elapsed
    d[from_][to] = int(elapsed * 1000 * 1000)

pandas.DataFrame(d)
```

**Before:**
![image](https://user-images.githubusercontent.com/1032377/67171274-2e93d000-f36b-11e9-8fa0-91edd7dbc8ec.png)

**After:**
![image](https://user-images.githubusercontent.com/1032377/67171200-d361dd80-f36a-11e9-9b22-66292e395a09.png)

[ghstack-poisoned]
@zasdfgbnm zasdfgbnm requested a review from colesbury October 21, 2019 03:48
@zasdfgbnm zasdfgbnm added better-engineering Relatively self-contained tasks for better engineering contributors module: operators labels Oct 21, 2019
@zasdfgbnm
Copy link
Collaborator Author

zasdfgbnm commented Oct 21, 2019

Sorry there is a bug in my benchmark code: I am not reporting the minimum time but reporting the last result instead. Here is the fixed script and result:

script:

import torch
import timeit
import pandas
import itertools
from tqdm.notebook import tqdm
import math
print(torch.__version__)
print()

_10M = 10 * 1024 ** 2

d = {}

for from_, to in tqdm(itertools.product(torch.testing.get_all_dtypes(), repeat=2)):
    if from_ not in d:
        d[from_] = {}
    a = torch.zeros(_10M, dtype=from_)
    min_ = math.inf
    for i in range(100):
        start = timeit.default_timer()
        a.to(to)
        end = timeit.default_timer()
        elapsed = end - start
        if elapsed < min_:
            min_ = elapsed
    d[from_][to] = int(min_ * 1000 * 1000)
    
pandas.DataFrame(d)

before:
image

after:
image

@zasdfgbnm zasdfgbnm merged commit d3cbbca into gh/zasdfgbnm/10/base Oct 22, 2019
zasdfgbnm added a commit that referenced this pull request Oct 22, 2019
@zasdfgbnm
Simplify copy kernel
0e17040 
Using the new type promotion and dynamic casting added to
`TensorIterator`, the copy kernels could be greatly simplified.

**Script:**
```python
import torch
import timeit
import pandas
import itertools
from tqdm import tqdm
import math
print(torch.__version__)
print()

_10M = 10 * 1024 ** 2

d = {}

for from_, to in tqdm(itertools.product(torch.testing.get_all_dtypes(),
repeat=2)):
    if from_ not in d:
        d[from_] = {}
    a = torch.zeros(_10M, dtype=from_)
    min_ = math.inf
    for i in range(100):
        start = timeit.default_timer()
        a.to(to)
        end = timeit.default_timer()
        elapsed = end - start
        if elapsed < min_:
            min_ = elapsed
    d[from_][to] = int(elapsed * 1000 * 1000)

pandas.DataFrame(d)
```

**Before:**
![image](https://user-images.githubusercontent.com/1032377/67171274-2e93d000-f36b-11e9-8fa0-91edd7dbc8ec.png)

**After:**
![image](https://user-images.githubusercontent.com/1032377/67171200-d361dd80-f36a-11e9-9b22-66292e395a09.png)

ghstack-source-id: d7fc960
Pull Request resolved: #28352
@zasdfgbnm zasdfgbnm mentioned this pull request Oct 22, 2019
Closed
zasdfgbnm added a commit that referenced this pull request Oct 22, 2019
@zasdfgbnm
Update on "Simplify copy kernel"
0cf634c 
Using the new type promotion and dynamic casting added to
`TensorIterator`, the copy kernels could be greatly simplified.

For benchmark, see #28352 (comment)

[ghstack-poisoned]
zasdfgbnm added a commit that referenced this pull request Oct 23, 2019
@zasdfgbnm
Update on "Simplify copy kernel"
e84ac97 
Using the new type promotion and dynamic casting added to
`TensorIterator`, the copy kernels could be greatly simplified.

For benchmark, see #28352 (comment)

[ghstack-poisoned]
zasdfgbnm added a commit that referenced this pull request Oct 24, 2019
@zasdfgbnm
Update on "Simplify copy kernel"
5fc3a07 
Using the new type promotion and dynamic casting added to
`TensorIterator`, the copy kernels could be greatly simplified.

For benchmark, see #28352 (comment)

[ghstack-poisoned]
zasdfgbnm added a commit that referenced this pull request Oct 24, 2019
@zasdfgbnm
Update on "Simplify copy kernel"
1a1c3f8 
Using the new type promotion and dynamic casting added to
`TensorIterator`, the copy kernels could be greatly simplified.

For benchmark, see #28352 (comment)

[ghstack-poisoned]
zasdfgbnm added a commit that referenced this pull request Oct 25, 2019
@zasdfgbnm
Update on "Simplify copy kernel"
44b398e 
Using the new type promotion and dynamic casting added to
`TensorIterator`, the copy kernels could be greatly simplified.

For benchmark, see #28352 (comment)

[ghstack-poisoned]
zasdfgbnm added a commit that referenced this pull request Oct 25, 2019
@zasdfgbnm
Update on "Simplify copy kernel"
dd612d0 
Using the new type promotion and dynamic casting added to
`TensorIterator`, the copy kernels could be greatly simplified.

For benchmark, see #28352 (comment)

[ghstack-poisoned]
zasdfgbnm added a commit that referenced this pull request Oct 26, 2019
@zasdfgbnm
Update on "Simplify copy kernel"
25e7b33 
Using the new type promotion and dynamic casting added to
`TensorIterator`, the copy kernels could be greatly simplified.

For benchmark, see #28352 (comment)

[ghstack-poisoned]
zasdfgbnm added a commit that referenced this pull request Oct 26, 2019
@zasdfgbnm
Update on "Simplify copy kernel"
27ecdf9 
Using the new type promotion and dynamic casting added to
`TensorIterator`, the copy kernels could be greatly simplified.

For benchmark, see #28352 (comment)

[ghstack-poisoned]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@colesbury colesbury Awaiting requested review from colesbury

@VitalyFedyunin VitalyFedyunin Awaiting requested review from VitalyFedyunin

@ngimel ngimel Awaiting requested review from ngimel

Assignees

No one assigned

Labels

better-engineering Relatively self-contained tasks for better engineering contributors

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@zasdfgbnm