Comments (10)
wsmoses
commented on June 26, 2024
1
Yeah that post sparse todense is the one I remember working with you on during benchmarking, so that looks right.
from enzyme.
alecjacobson
commented on June 26, 2024
And if it's changed to N then it seems the i*3+4 etc. need a %(N*3) to wrap around the circle of springs.
from enzyme.
wsmoses
commented on June 26, 2024
Yeah I think I agree here. Fortunately this is just a test for correct
compilation and not a benchmark.
…On Sun, May 19, 2024 at 11:00 AM Alec Jacobson ***@***.***> wrote:
https://github.com/EnzymeAD/Enzyme/blob/2188ad8f1401886eca1e2045abad6b60424fc2c5/enzyme/test/Integration/Sparse/ringspring3Drestlengthone.cpp#L24
N seems to be the number of 3D nodes. So that pos and x are arrays of
length 3*N.
This N/3 seems to mean that only first ⅓ of the springs are considered.
—
Reply to this email directly, view it on GitHub
<#1886>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAJTUXHRZ74TX2H6742GI3DZDDEBRAVCNFSM6AAAAABH6PENKGVHI2DSMVQWIX3LMV43ASLTON2WKOZSGMYDINRXGQ3DKNI>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
from enzyme.
wsmoses
commented on June 26, 2024
I'm not sure about the % . I need to double check what this test is validating for, but the wrap around in the actual benchmarks is done via the todense to create the actual double pointers (and accessing them will implicitly wrap around)
from enzyme.
wsmoses
commented on June 26, 2024
Yeah okay I think this is only a partial test here
The commented out line
// input = __enzyme_todense((void*)mod_load, (void*)never_store, input, N);
is an example of what would be needed to ensure that the wrap around
happens internally if one doesn’t do so in the inner function.
@martinjm97 this is the actual benchmark code here right?
https://github.com/martinjm97/sparse_deriv_enzyme/blob/main/rest_length_1/enzyme_dense_hess.cpp
Or at least it does have the proper sizing and inner modulus you’re looking
for here.
We for sure should just add the full benchmark as another integration test
though
…On Sun, May 19, 2024 at 11:03 AM William Moses ***@***.***> wrote:
Yeah I think I agree here. Fortunately this is just a test for correct
compilation and not a benchmark.
On Sun, May 19, 2024 at 11:00 AM Alec Jacobson ***@***.***>
wrote:
>
> https://github.com/EnzymeAD/Enzyme/blob/2188ad8f1401886eca1e2045abad6b60424fc2c5/enzyme/test/Integration/Sparse/ringspring3Drestlengthone.cpp#L24
>
> N seems to be the number of 3D nodes. So that pos and x are arrays of
> length 3*N.
>
> This N/3 seems to mean that only first ⅓ of the springs are considered.
>
> —
> Reply to this email directly, view it on GitHub
> <#1886>, or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AAJTUXHRZ74TX2H6742GI3DZDDEBRAVCNFSM6AAAAABH6PENKGVHI2DSMVQWIX3LMV43ASLTON2WKOZSGMYDINRXGQ3DKNI>
> .
> You are receiving this because you are subscribed to this thread.Message
> ID: ***@***.***>
>
from enzyme.
alecjacobson
commented on June 26, 2024
Thanks. @martinjm97 do you mind adding me as a collaborator on that repo so I can copy the benchmark for our comparison?
from enzyme.
martinjm97
commented on June 26, 2024
Sorry, I just saw this. I've added you to the repo.
from enzyme.
martinjm97
commented on June 26, 2024
The benchmarking script is here:
The code that we run is in the test is here:
https://github.com/dromaiidae/sparse_deriv/blob/dc973f23d672232089c30efe9bccabf48ebab765/benchmark/SiggraphPerformance/ring_spring_rest_length_one_timing.py#L56
from enzyme.
wsmoses
commented on June 26, 2024
I don't think I have access to that repo, @deomaiidae if you can add me
from enzyme.
martinjm97
commented on June 26, 2024
@alecjacobson the code has a bug as you pointed out. Moreover, the benchmarking script calls the file at this location. However, when I ran the benchmarking script, I ran a version of the code that fixed this bug (and also printed out how much time the benchmark took to run). I've made a PR of the code I ran: #1889.
from enzyme.
Related Issues (20)
- Enzyme: Cannot cast __enzyme_autodiff primal argument 16 HOT 7
- enzyme_dupped parameter doesn't return gradient
- New C++ interface with lambda HOT 2
- Can't compile eigensumsqdyn-notmp.cpp with Eigen 3.4.0
- Injected headers for c++ break tooling
- Incorrect derivative result when nested void functions and recursive nature functions are used. HOT 6
- abort cmake when -DLLVM_DIR is an invalid path.
- check-enzyme-integration tests failures HOT 3
- Branch mismatcharg fails to compile HOT 6
- Building Enzyme CMake - Undefined symbol: main HOT 3
- Unnecessary caching for recursive functions
- Bug in Enzyme gsl branch HOT 1
- compilation slowdown associated with PreserveNVVMNewPM HOT 1
- incorrect derivative of function that returns struct HOT 2
- C++ interface templates appear to be broken HOT 7
- `std::vector.push_back()` causes segementation fault in Enzyme HOT 2
- EnzymeCreateForwardDiff missing from CApi.h HOT 1
- Clarify usage through linking HOT 8
- Failing unittest Enzyme/ReverseMode/gsl_sf_legendre_array_e.ll
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from enzyme.