profile
viewpoint
If you are wondering where the data of this site comes from, please visit https://api.github.com/users/dignifiedquire/events. GitMemory does not store any data, but only uses NGINX to cache data for a period of time. The idea behind GitMemory is simply to give users a better reading experience.
Friedel Ziegelmayer dignifiedquire Germany

async-rs/async-std 2996

Async version of the Rust standard library

deltachat/deltachat-core-rust 359

Delta Chat Rust Core library, used by Android/iOS/desktop apps and bindings

deltachat/deltachat-core 309

Delta.Chat C-Library with e2e chat-over-email functionality & Python bindings

deltachat/deltachat-ios 198

Email-based instant messaging for iOS.

async-rs/async-tls 130

A TLS implementation over AsyncRead and AsyncWrite

browserify/browserify-zlib 47

Full zlib module for browserify

dignifiedquire/async-tar 46

A tar archive reading/writing library for async Rust.

async-rs/async-global-executor 36

A global executor built on top of async-executor and smol

daviddias/js-ethereum-libp2p 22

⛓ Process the ethereum blockchain in the Browser ou Node.js with ♦libp2p 🎉

issue commentfilecoin-project/bellperson

do windowpost err

This means your machine is likely running out of memory (RAM)

duzy555666

comment created time in 6 hours

Pull request review commentfilecoin-project/bellperson

perf: use multiple GPUs for FFTs

 fn best_fft<E: Engine + gpu::GpuEngine, T: Group<E::Fr>>(     } } +fn best_fft3<E: Engine + gpu::GpuEngine, T: Group<E::Fr>>(+    worker: &Worker,+    kern: &mut Option<gpu::LockedFFTKernel<E>>,+    a0: &mut [T],+    a1: &mut [T],+    a2: &mut [T],

I am not a fan of puting mutable references in lists (there is no good reason, just a personal preference). This could probably be abstraced further, and just made to work with n elements instead of fixed 3.

dignifiedquire

comment created time in a day

PullRequestReviewEvent

PR closed supranational/blst

feat: export fp12_raise_to_z

I need this to implement a subgroup check in fp12

+8 -0

3 comments

2 changed files

dignifiedquire

pr closed time in a day

pull request commentsupranational/blst

feat: export fp12_raise_to_z

Closing as the core method now exists https://github.com/supranational/blst/commit/d01bb838805b6837083c3bd395174ec9b6dbecfa

Thank you @dot-asm

dignifiedquire

comment created time in a day

pull request commentfilecoin-project/rust-fil-proofs

fix: upgrade to latest versions

the required tests changed with the blst/pairing change from the earlier PR

vmx

comment created time in a day

issue commentfilecoin-project/bellperson

Test Failures on GTX3070

we just merged cuda support, could you try running with that, and see if the error persists? the other reason for sudden failures I have seen are incompatible nvidia drivers. did you update them recently?

meawoppl

comment created time in a day

PullRequestReviewEvent
PullRequestReviewEvent

push eventfilecoin-project/bellperson

dignifiedquire

commit sha b4b936bf698bfd727fb2a715298c02051ded4741

perf: use multiple GPUs for FFTs

view details

dignifiedquire

commit sha 76f20d919ae63399089436ae8402a25dd87749cf

test: add fft3 test

view details

push time in 2 days

push eventfilecoin-project/bellperson

dignifiedquire

commit sha 2a3ee5b878ea6f92a3901f20371b2300eeb1d612

test: add fft3 test

view details

push time in 2 days

create barnchfilecoin-project/bellperson

branch : perf-fft-parallel

created branch time in 2 days

push eventfilecoin-project/bellperson

Friedel Ziegelmayer

commit sha f0f8fe9c51f871e178fe4d462c82b06d052ad78e

fix: restructure use of rayon to avoid deadlocks - perf: Uses rayon in the places where the work is cpu bound only - fix: Do not use custom rayon thread pools, as multiple of these can result in cross pool deadlocking. - fix: Proxy `BELLMAN_NUM_CPUS` to `RAYON_NUM_THREADS` for now, and deprecate it. - refactor: Use `RAYON_NUM_THREADS` as the baseline for determining the number of threads to to use - perf: Use `in_place_scope` instead of `scope` to avoid spawning unnecessary threads. - ci: Add CI tests to ensure the currently deadlocking test for `groth16_bench` is checked Co-authored-by: nemo <nemo@protocol.ai>

view details

nemo

commit sha c19cc076653598c31c980f85b4b2c0b8d3a26655

fix: update incorrectly named test (#225)

view details

nemo

commit sha 4e8d2eb28072ef71117dcc89466bcb3faef05369

chore(bellperson): release 0.16.3

view details

DrPeterVanNostrand

commit sha d1b99396c7a2e25a60444cdc1e7c0ba4fd2a1094

feat: switch to upstream versions of group, ff and pairing Breaking Change: Removes the `pairing` backend, reducing to only the `blst` backend. Co-authored-by: dignifiedquire <me@dignifiedquire.com> Co-authored-by: nemo <nemo@protocol.ai>

view details

Volker Mische

commit sha 584f05283515e20bc31d993d5ce5d6bbb6b60e9d

feat: add CUDA support A new environment variable `BELLMAN_GPU_FRAMEWORK` is introduced to select CUDA or OpenCL if Bellperson is compiled with support for both. Currently only the multiexp calculation is supported by CUDA, the FFT calculation still runs on OpenCL. BREAKING CHANGE: `gpu` feature is replaced with `cuda` and `opencl` feature. The `gpu` feature is now split into two pieces, `opencl` and `cuda`, you can compile the library with either or both.

view details

Volker Mische

commit sha de72a1550cb7df3324e98fe42f8e6e869f842432

test: test BELLMAN_GPU_FRAMEWORK environment variable Refactor the multicore environment variable test infrastructure into its own module, so that it can also be used in other places.

view details

Friedel Ziegelmayer

commit sha 8c9528ddac45ffbf06c0005c5ea1906085e3b2f4

Merge pull request #218 from filecoin-project/cuda

view details

dignifiedquire

commit sha 8e641b04880cd8c6048ffc856d591d919997477a

optimize LinearCombination

view details

dignifiedquire

commit sha 6c71e3bd34324c703dd4411fdad4a7450dbeac7d

perf(lc): optimize sequential inserts

view details

dignifiedquire

commit sha ebeb9c502a9ae03a124c32e207937675ab5bb091

perf(gadgets): use optimized LC creation in Num

view details

push time in 2 days

push eventfilecoin-project/bellperson

Volker Mische

commit sha 584f05283515e20bc31d993d5ce5d6bbb6b60e9d

feat: add CUDA support A new environment variable `BELLMAN_GPU_FRAMEWORK` is introduced to select CUDA or OpenCL if Bellperson is compiled with support for both. Currently only the multiexp calculation is supported by CUDA, the FFT calculation still runs on OpenCL. BREAKING CHANGE: `gpu` feature is replaced with `cuda` and `opencl` feature. The `gpu` feature is now split into two pieces, `opencl` and `cuda`, you can compile the library with either or both.

view details

Volker Mische

commit sha de72a1550cb7df3324e98fe42f8e6e869f842432

test: test BELLMAN_GPU_FRAMEWORK environment variable Refactor the multicore environment variable test infrastructure into its own module, so that it can also be used in other places.

view details

Friedel Ziegelmayer

commit sha 8c9528ddac45ffbf06c0005c5ea1906085e3b2f4

Merge pull request #218 from filecoin-project/cuda

view details

dignifiedquire

commit sha bc8bb99b16428d5faf2a086d5d173405181cf203

refactor(domain) : remove `Group` abstraction This required us to use `unsafe` casting, even though we don't actually make use of the abstraction anywhere. With it gone, so is the `unsafe` casting.

view details

push time in 2 days

push eventfilecoin-project/bellperson

Volker Mische

commit sha 584f05283515e20bc31d993d5ce5d6bbb6b60e9d

feat: add CUDA support A new environment variable `BELLMAN_GPU_FRAMEWORK` is introduced to select CUDA or OpenCL if Bellperson is compiled with support for both. Currently only the multiexp calculation is supported by CUDA, the FFT calculation still runs on OpenCL. BREAKING CHANGE: `gpu` feature is replaced with `cuda` and `opencl` feature. The `gpu` feature is now split into two pieces, `opencl` and `cuda`, you can compile the library with either or both.

view details

Volker Mische

commit sha de72a1550cb7df3324e98fe42f8e6e869f842432

test: test BELLMAN_GPU_FRAMEWORK environment variable Refactor the multicore environment variable test infrastructure into its own module, so that it can also be used in other places.

view details

Friedel Ziegelmayer

commit sha 8c9528ddac45ffbf06c0005c5ea1906085e3b2f4

Merge pull request #218 from filecoin-project/cuda

view details

dignifiedquire

commit sha fab8f4f19af558b9087c9bf188302949933c47bc

perf(multiexp): optimize generation of exponents for GPU multiexp

view details

push time in 2 days

push eventfilecoin-project/bellperson

Volker Mische

commit sha 584f05283515e20bc31d993d5ce5d6bbb6b60e9d

feat: add CUDA support A new environment variable `BELLMAN_GPU_FRAMEWORK` is introduced to select CUDA or OpenCL if Bellperson is compiled with support for both. Currently only the multiexp calculation is supported by CUDA, the FFT calculation still runs on OpenCL. BREAKING CHANGE: `gpu` feature is replaced with `cuda` and `opencl` feature. The `gpu` feature is now split into two pieces, `opencl` and `cuda`, you can compile the library with either or both.

view details

Volker Mische

commit sha de72a1550cb7df3324e98fe42f8e6e869f842432

test: test BELLMAN_GPU_FRAMEWORK environment variable Refactor the multicore environment variable test infrastructure into its own module, so that it can also be used in other places.

view details

Friedel Ziegelmayer

commit sha 8c9528ddac45ffbf06c0005c5ea1906085e3b2f4

Merge pull request #218 from filecoin-project/cuda

view details

dignifiedquire

commit sha 262a1f9b40eae422c22d9b78620005b460c03591

prover: improve param loading

view details

dignifiedquire

commit sha 9b606e3e0456c78fbc43bb3294cdfc4e74d39c1d

prover: parallelize param loading

view details

dignifiedquire

commit sha d80ccc1cab32cb150edf1822635ef6691d48d238

fixup

view details

push time in 2 days

delete branch filecoin-project/neptune

delete branch : cuda

delete time in 2 days

push eventfilecoin-project/neptune

Volker Mische

commit sha 5bf574ec78ac92a432a9bba646052d2c6520da17

feat: add support for CUDA Neptune now also runs on CUDA. Closes #104.

view details

Volker Mische

commit sha 9eb0a4fb0cc806d026407ab9693fd97ec19123bd

fix: compile gbench with correct arities

view details

Friedel Ziegelmayer

commit sha 9d31500282855173c64cbd5608d20ded814839b4

Merge pull request #109 from filecoin-project/cuda

view details

push time in 2 days

PR merged filecoin-project/neptune

feat: add support for CUDA

Neptune now also runs on CUDA. The OpenCL version is a bit slower than previous versions, due to the removal of 64-bit limbs from the kernel.

Closed #104.

+717 -334

1 comment

19 changed files

vmx

pr closed time in 2 days

issue closedfilecoin-project/neptune

Add CUDA support

Neptune should also support to run on CUDA.

closed time in 2 days

vmx

delete branch filecoin-project/bellperson

delete branch : cuda

delete time in 2 days

push eventfilecoin-project/bellperson

Volker Mische

commit sha 584f05283515e20bc31d993d5ce5d6bbb6b60e9d

feat: add CUDA support A new environment variable `BELLMAN_GPU_FRAMEWORK` is introduced to select CUDA or OpenCL if Bellperson is compiled with support for both. Currently only the multiexp calculation is supported by CUDA, the FFT calculation still runs on OpenCL. BREAKING CHANGE: `gpu` feature is replaced with `cuda` and `opencl` feature. The `gpu` feature is now split into two pieces, `opencl` and `cuda`, you can compile the library with either or both.

view details

Volker Mische

commit sha de72a1550cb7df3324e98fe42f8e6e869f842432

test: test BELLMAN_GPU_FRAMEWORK environment variable Refactor the multicore environment variable test infrastructure into its own module, so that it can also be used in other places.

view details

Friedel Ziegelmayer

commit sha 8c9528ddac45ffbf06c0005c5ea1906085e3b2f4

Merge pull request #218 from filecoin-project/cuda

view details

push time in 2 days

PR merged filecoin-project/bellperson

feat: Add CUDA support

A new environment variable BELLMAN_GPU_FRAMEWORK is introduced to select CUDA or OpenCL if Bellperson is compiled with support for both.

Currently only the multiexp calculation is supported by CUDA, the FFT calculation still runs on OpenCL.

BREAKING CHANGE: gpu feature is replaced with cuda and opencl feature.

The gpu feature is now split into two pieces, opencl and cuda, you can compile the library with either or both.

+656 -368

1 comment

23 changed files

vmx

pr closed time in 2 days

issue commentRustCrypto/crypto-bigint

Feature wishlist tracking ticket

I would suggest having the different algorithms behind specific methods, such that downstream implementors can choose the algorithm, eg fn sqrt_tonelli_shank().

tarcieri

comment created time in 2 days

PullRequestReviewEvent
PullRequestReviewEvent
PullRequestReviewEvent

Pull request review commentfilecoin-project/bellperson

feat: Add CUDA support

+use std::env;++#[cfg(feature = "opencl")]+use ec_gpu_gen::Limb64;+use log::info;+#[cfg(feature = "cuda")]+use rust_gpu_tools::cuda;+#[cfg(feature = "opencl")]+use rust_gpu_tools::opencl;+use rust_gpu_tools::{Device, Framework, GPUError as GpuToolsError, Program};++#[cfg(not(all(feature = "cuda", feature = "opencl")))]+use crate::gpu::error::GPUError;+use crate::gpu::error::GPUResult;+#[cfg(feature = "opencl")]+use crate::gpu::sources;+use crate::gpu::GpuEngine;+use pairing::Engine;++/// Returns the program for the preferred [`rust_gpu_tools::device::Framework`].+///+/// If the device supports CUDA, then CUDA is used, else OpenCL. You can force a selection with+/// the environment variable `BELLMAN_GPU_FRAMEWORK`, which can be set either to `cuda` or `opencl`.+pub fn program<E>(device: &Device) -> GPUResult<Program>+where+    E: Engine + GpuEngine,+{+    let framework = match env::var("BELLMAN_GPU_FRAMEWORK") {

can you add a test for this? I have added a helper in multicore for testing env variable based behaviour

vmx

comment created time in 2 days

PullRequestReviewEvent