Hacker News: boywitharupee

New comment by boywitharupee in "NanoClaw moved from Apple Containers to Docker"

boywitharupee — Sun, 22 Feb 2026 20:06:20 +0000

they have a watchdog loop, it runs periodically

New comment by boywitharupee in "CUDA Tile Open Sourced"

boywitharupee — Thu, 25 Dec 2025 21:03:53 +0000

shouldn't the title be "CUDA Tile IR Open Sourced"?

New comment by boywitharupee in "Faster, easier 2D vector rendering [video]"

boywitharupee — Fri, 13 Jun 2025 03:52:55 +0000

is there a document or reference implementation that describes the full algorithm? tiling, sorting, merging, and strip conversion.

New comment by boywitharupee in "Is `(expession, lvalue) = rvalue` a valid assignment in C or C++?"

boywitharupee — Fri, 15 Nov 2024 11:58:47 +0000

> In C++, it's an rvalue reference , which can be effectively thought of as an lvalue

hmm...this doesn't sound quite right? the comma operator's result in C++ is not an rvalue reference - it takes on exactly the value category of its right operand (which in this case is an lvalue)

New comment by boywitharupee in "ThunderKittens: Simple, fast, and adorable AI kernels"

boywitharupee — Thu, 31 Oct 2024 00:33:17 +0000

so, these are hand optimized primitives for specific model of nvidia gpus? do you still have to make launch/scheduling decisions to maximize occupancy? how does this approach scale to other target devices with specialized instruction sets and different architecture?

New comment by boywitharupee in "Tracy: A real time, nanosecond resolution frame profiler"

boywitharupee — Tue, 24 Sep 2024 11:54:50 +0000

can someone explain how is profiling tools like this written for GPU applications? wouldn't you need access to internal runtime api?

for ex. Apple wraps Metal buffers as "Debug" buffers to record allocations/deallocations.

New comment by boywitharupee in "AlphaProteo generates novel proteins for biology and health research"

boywitharupee — Thu, 05 Sep 2024 19:44:05 +0000

what kind of model architecture was used for this? is it safe to assume they used a transformer model or a variant of it?

New comment by boywitharupee in "Fitting an elephant with four non-zero parameters"

boywitharupee — Sun, 14 Jul 2024 17:16:42 +0000

what's the purpose of this? is it one of those 'fun' problems to solve?

New comment by boywitharupee in "GraphRAG is now on GitHub"

boywitharupee — Tue, 02 Jul 2024 18:35:31 +0000

how different is this compared to Facebook's open-source tool Faiss[1]?

[1] https://github.com/facebookresearch/faiss/

New comment by boywitharupee in "Nvidia Warp: A Python framework for high performance GPU simulation and graphics"

boywitharupee — Fri, 14 Jun 2024 19:24:25 +0000

In a similar fashion, you'll see that JAX has frontend code being open-sourced, while device-related code is distributed as binaries. For example, if you're on Google's TPU, you'll see libtpu.so, and on macOS, you'll see pjrt_plugin_metal_1.x.dylib.

The main optimizations (scheduler, vectorizer, etc.) are hidden behind these shared libraries. If open-sourced, they might reveal hints about proprietary algorithms and provide clues to various hardware components, which could potentially be exploited.

New comment by boywitharupee in "Copy-and-Patch: Fast compilation for high-level languages and bytecode (2020)"

boywitharupee — Wed, 05 Jun 2024 00:36:21 +0000

or have the right entitlements?

https://developer.apple.com/documentation/bundleresources/en...

New comment by boywitharupee in "Copy-and-Patch: Fast compilation for high-level languages and bytecode (2020)"

boywitharupee — Sun, 02 Jun 2024 18:28:53 +0000

> At runtime, C&P generates executable code by copying the object code and patching the holes with runtime known values.

how would this work on OSs under hardened runtime rules?

New comment by boywitharupee in "Show HN: Get paid to do your own ML research"

boywitharupee — Tue, 28 May 2024 03:05:26 +0000

will you also provide compute resources?

New comment by boywitharupee in "Show HN: Boldly go where Gradient Descent has never gone before with DiscoGrad"

boywitharupee — Sun, 26 May 2024 18:16:41 +0000

how would you compare this to the polytope model? https://en.wikipedia.org/wiki/Polytope_model

New comment by boywitharupee in "Implementation of Google's Griffin Architecture – RNN LLM"

boywitharupee — Thu, 11 Apr 2024 10:44:31 +0000

and is Griffin a state space model?

New comment by boywitharupee in "ClangQL: A tool to run SQL-like query on C/C++ Code"

boywitharupee — Mon, 08 Apr 2024 16:11:23 +0000

i wonder if we can train a foundational model on this data which will eventually allow to semantically search the codebase?

New comment by boywitharupee in "CityGaussian: Real-time high-quality large-scale scene rendering with Gaussians"

boywitharupee — Tue, 02 Apr 2024 19:28:07 +0000

what's the memory and compute requirements for this?

New comment by boywitharupee in "LLaMA now goes faster on CPUs"

boywitharupee — Mon, 01 Apr 2024 14:20:16 +0000

but which model to tokenize with? is there a leaderboard for models that are good for RAG?

New comment by boywitharupee in "Tiny bitfield based text renderer (2022)"

boywitharupee — Sat, 16 Mar 2024 04:24:16 +0000

is this known as a procedural generation?

New comment by boywitharupee in "Building Meta's GenAI infrastructure"

boywitharupee — Wed, 13 Mar 2024 01:16:59 +0000

care to explain why attention has precision issues with fp8?