All About NVIDIA GPUs PyTorch Developer podcast

Artwork

Tech PyTorch Edward Yang Team PyTorch Deep Learning Machine Learning

Player FM - Internet Radio Done Right

32 subscribers

Agregado hace cuatro años

Contenido proporcionado por PyTorch, Edward Yang, and Team PyTorch. Todo el contenido del podcast, incluidos episodios, gráficos y descripciones de podcast, lo carga y proporciona directamente PyTorch, Edward Yang, and Team PyTorch o su socio de plataforma de podcast. Si cree que alguien está utilizando su trabajo protegido por derechos de autor sin su permiso, puede seguir el proceso descrito aquí https://es.player.fm/legal.

<

<div class="span index">1</div> <span><a class="" data-remote="true" data-type="html" href="/series/state-secrets-inside-the-making-of-the-electric-state">State Secrets: Inside The Making Of The Electric State</a></span>

<div class="span index">1</div> <span><a class="" data-remote="true" data-type="html" href="/series/state-secrets-inside-the-making-of-the-electric-state">State Secrets: Inside The Making Of The Electric State</a></span> podcast artwork

<div class="span index">1</div> <span><a class="" data-remote="true" data-type="html" href="/series/state-secrets-inside-the-making-of-the-electric-state">State Secrets: Inside The Making Of The Electric State</a></span> podcast artwork

1
State Secrets: Inside The Making Of The Electric State

Cancelar la suscripción

hace 20 díashace 20d ago

Cancelar la suscripción

Semanalmente+

Step inside the world of The Electric State! Join host Francesca Amiker as she takes you behind the scenes of Anthony and Joe Russo’s epic new Netflix adventure. Over six in-depth episodes, explore how the filmmakers transformed Simon Stålenhag’s stunning graphic novel into a cinematic experience like no other. Exclusive interviews from the cast and crew—including Millie Bobby Brown, Chris Pratt, Stanley Tucci, and the Russo Brothers—will help break down the film’s jaw-dropping visuals, emotional core, groundbreaking technology AND the creative secrets that make The Electric State. Join us in the countdown to The Electric State premiering exclusively on Netflix on March 14th. State Secrets: Inside The Making of The Electric State coming March 7th.

PyTorch Developer Podcast « »
All about NVIDIA GPUs

hace 4 años 19:29

Compartir

MP3•Episodio en casa

Contenido proporcionado por PyTorch, Edward Yang, and Team PyTorch. Todo el contenido del podcast, incluidos episodios, gráficos y descripciones de podcast, lo carga y proporciona directamente PyTorch, Edward Yang, and Team PyTorch o su socio de plataforma de podcast. Si cree que alguien está utilizando su trabajo protegido por derechos de autor sin su permiso, puede seguir el proceso descrito aquí https://es.player.fm/legal.

PyTorch is in the business of shipping numerical software that can run fast on your CUDA-enabled NVIDIA GPU, but it turns out there is a lot of heterogeneity in NVIDIA’s physical GPU offering and when it comes to what is fast and what is slow, what specific GPU you have on hand matters quite a bit. Yet there are literally hundreds of distinct NVIDIA GPU models on the market, how do you make sense of the madness? Today, Natalia Gimelshein joins me to talk about everything that’s going on in the NVIDIA GPU market, and what, as a framework developer, you have to care about to make sense of it all.

Further reading.

NVIDIA microarchitectures on Wikipedia https://en.wikipedia.org/wiki/Category:Nvidia_microarchitectures
A slightly old post about matching SM to architecture https://arnon.dk/matching-sm-architectures-arch-and-gencode-for-various-nvidia-cards/

… continue reading

83 episodios

#Tech #PyTorch #Edward Yang #Team PyTorch #Deep Learning #Machine Learning

Artwork

All about NVIDIA GPUs

PyTorch Developer Podcast

32 subscribers

published hace 4 años

Compartir

MP3•Episodio en casa

Contenido proporcionado por PyTorch, Edward Yang, and Team PyTorch. Todo el contenido del podcast, incluidos episodios, gráficos y descripciones de podcast, lo carga y proporciona directamente PyTorch, Edward Yang, and Team PyTorch o su socio de plataforma de podcast. Si cree que alguien está utilizando su trabajo protegido por derechos de autor sin su permiso, puede seguir el proceso descrito aquí https://es.player.fm/legal.

PyTorch is in the business of shipping numerical software that can run fast on your CUDA-enabled NVIDIA GPU, but it turns out there is a lot of heterogeneity in NVIDIA’s physical GPU offering and when it comes to what is fast and what is slow, what specific GPU you have on hand matters quite a bit. Yet there are literally hundreds of distinct NVIDIA GPU models on the market, how do you make sense of the madness? Today, Natalia Gimelshein joins me to talk about everything that’s going on in the NVIDIA GPU market, and what, as a framework developer, you have to care about to make sense of it all.

Further reading.

NVIDIA microarchitectures on Wikipedia https://en.wikipedia.org/wiki/Category:Nvidia_microarchitectures
A slightly old post about matching SM to architecture https://arnon.dk/matching-sm-architectures-arch-and-gencode-for-various-nvidia-cards/

… continue reading

83 episodios

#Tech #PyTorch #Edward Yang #Team PyTorch #Deep Learning #Machine Learning

Todos os episódios

×

Artwork

1
Compiler collectives 16:33

hace 36 weeks16:33

Reproducir más Tarde

Reproducir más Tarde

16:33

Compiler collectives are a PT2 feature where by compiler instances across multiple ranks use NCCL collectives to communicate information to other instances. This is used to ensure we consistently decide if inputs or static or dynamic across all ranks. See also PR at https://github.com/pytorch/pytorch/pull/130935…

Artwork

1
TORCH_TRACE and tlparse 15:28

hace 50 weeks15:28

Reproducir más Tarde

Reproducir más Tarde

15:28

TORCH_TRACE and tlparse are a structured log and log parser for PyTorch 2. It gives useful information about what code was compiled and what the intermediate build products look like.

Artwork

1
Higher order operators 17:10

hace 51 weeks17:10

Reproducir más Tarde

Reproducir más Tarde

17:10

Higher order operators are a special form of operators in torch.ops which have relaxed input argument requirements: in particular, they can accept any form of argument, including Python callables. Their name is based off of their most common use case, which is to represent higher order functions like control flow operators. However, they are also used to implement other variants of basic operators and can also be used to smuggle in Python data that is quite unusual. They are implemented using a Python dispatcher.…

Artwork

1
Inductor - Post-grad FX passes 24:07

hace 52 weeks24:07

Reproducir más Tarde

Reproducir más Tarde

24:07

The post-grad FX passes in Inductor run after AOTAutograd has functionalized and normalized the input program into separate forward/backward graphs. As such, they generally can assume that the graph in question is functionalized, except for some mutations to inputs at the end of the graph. At the end of post-grad passes, there are special passes that reintroduce mutation into the graph before going into the rest of Inductor lowering which is generally aware of passes. The post-grad FX passes are varied but are typically domain specific passes making local changes to specific parts of the graph.…

Artwork

1
CUDA graph trees 20:50

hace 1 year20:50

Reproducir más Tarde

Reproducir más Tarde

20:50

CUDA graph trees are the internal implementation of CUDA graphs used in PT2 when you say mode="reduce-overhead". Their primary innovation is that they allow the reuse of memory across multiple CUDA graphs, as long as they form a tree structure of potential paths you can go down with the CUDA graph. This greatly reduced the memory usage of CUDA graphs in PT2. There are some operational implications to using CUDA graphs which are described in the podcast.…

Artwork

1
Min-cut partitioner 15:56

hace 1 year15:56

Reproducir más Tarde

Reproducir más Tarde

15:56

The min-cut partitioner makes decisions about what to save for backwards when splitting the forward and backwards graph from the joint graph traced by AOTAutograd. Crucially, it doesn't actually do a "split"; instead, it is deciding how much of the joint graph should be used for backwards. I also talk about the backward retracing problem.…

Artwork

1
AOTInductor 17:30

hace 1 year17:30

Reproducir más Tarde

Reproducir más Tarde

17:30

AOTInductor is a feature in PyTorch that lets you export an inference model into a self-contained dynamic library, which can subsequently be loaded and used to run optimized inference. It is aimed primarily at CUDA and CPU inference applications, for situations when your model export once to be exported once while your runtime may still get continuous updates. One of the big underlying organizing principles is a limited ABI which does not include libtorch, which allows these libraries to stay stable over updates to the runtime. There are many export-like use cases you might be interested in using AOTInductor for, and some of the pieces should be useful, but AOTInductor does not necessarily solve them.…

Artwork

1
Tensor subclasses and PT2 13:25

hace 1 year13:25

Reproducir más Tarde

Reproducir más Tarde

13:25

Tensor subclasses allow you to add extend PyTorch with new types of tensors without having to write any C++. They have been used to implement DTensor, FP8, Nested Jagged Tensor and Complex Tensor. Recent work by Brian Hirsh means that we can compile tensor subclasses in PT2, eliminating their overhead. The basic mechanism by which this compilation works is a desugaring process in AOTAutograd. There are some complications involving views, dynamic shapes and tangent metadata mismatch.…

Artwork

1
Compiled autograd 18:07

hace 1 year18:07

Reproducir más Tarde

Reproducir más Tarde

18:07

Compiled autograd is an extension to PT2 that permits compiling the entirety of a backward() call in PyTorch. This allows us to fuse accumulate grad nodes as well as trace through arbitrarily complicated Python backward hooks. Compiled autograd is an important part of our plans for compiled DDP/FSDP as well as for whole-graph compilation.…

Artwork

1
PT2 extension points 15:54

hace 1 year15:54

Reproducir más Tarde

Reproducir más Tarde

15:54

We discuss some extension points for customizing PT2 behavior across Dynamo, AOTAutograd and Inductor.

Artwork

1
Inductor - Define-by-run IR 12:06

hace 1 year12:06

Reproducir más Tarde

Reproducir más Tarde

12:06

Define-by-run IR is how Inductor defines the internal compute of a pointwise/reduction operation. It is characterized by a function that calls a number of functions in the 'ops' namespace, where these ops can be overridden by different handlers depending on what kind of semantic analysis you need to do. The ops Inductor supports include regular arithmetic operators, but also memory load/store, indirect indexing, masking and collective operations like reductions.…

Artwork

1
Unsigned integers 13:07

hace 1 year13:07

Reproducir más Tarde

Reproducir más Tarde

13:07

Traditionally, unsigned integer support in PyTorch was not great; we only support uint8. Recently, we added support for uint16, uint32 and uint64. Bare bones functionality works, but I'm entreating the community to help us build out the rest. In particular, for most operations, we plan to use PT2 to build anything else. But if you have an eager kernel you really need, send us a PR and we'll put it in. While most of the implementation was straightforward, there are some weirdnesses related to type promotion inconsistencies with numpy and dealing with the upper range of uint64. There is also upcoming support for sub-byte dtypes uint1-7, and these will exclusively be implemented via PT2.…

Artwork

1
Inductor - IR 18:00

hace 1 year18:00

Reproducir más Tarde

Reproducir más Tarde

18:00

Inductor IR is an intermediate representation that lives between ATen FX graphs and the final Triton code generated by Inductor. It was designed to faithfully represent PyTorch semantics and accordingly models views, mutation and striding. When you write a lowering from ATen operators to Inductor IR, you get a TensorBox for each Tensor argument which contains a reference to the underlying IR (via StorageBox, and then a Buffer/ComputedBuffer) that says how the Tensor was computed. The inner computation is represented via define-by-run, which allows for compact definition of IR representation, while still allowing you to extract an FX graph out if you desire. Scheduling then takes buffers of inductor IR and decides what can be fused. Inductor IR may have too many nodes, this would be a good thing to refactor in the future.…

Artwork

1
Dynamo - VariableTracker 15:55

hace 1 year15:55

Reproducir más Tarde

Reproducir más Tarde

15:55

I talk about VariableTracker in Dynamo. VariableTracker is Dynamo's representation of the Python. I talk about some recent changes, namely eager guards and mutable VT. I also tell you how to find the functionality you care about in VariableTracker ( https://docs.google.com/document/d/1XDPNK3iNNShg07jRXDOrMk2V_i66u1hEbPltcsxE-3E/edit#heading=h.i6v7gqw5byv6 ).…

Artwork

1
Unbacked SymInts 21:31

hace 2 years21:31

Reproducir más Tarde

Reproducir más Tarde

21:31

This podcast goes over the basics of unbacked SymInts. You might want to listen to this one before listening to https://pytorch-dev-podcast.simplecast.com/episodes/zero-one-specialization Some questions we answer (h/t from Gregory Chanan): - Are unbacked symints only for export? Because otherwise I could just break / wait for the actual size. But maybe I can save some retracing / graph breaks perf if I have them too? So the correct statement is "primarily" for export? - Why am I looking into the broadcasting code at all? Naively, I would expect the export graph to be just a list of ATen ops strung together. Why do I recurse that far down? Why can't I annotate DONT_TRACE_ME_BRO? - How does 0/1 specialization fit into this? I understand we may want to 0/1 specialize in a dynamic shape regime in "eager" mode (is there a better term?), but that doesn't seem to matter for export? - So far we've mainly been talking about how to handle our own library code. There is a worry about pushing complicated constraints downstream, similar to torchscript. What constraints does this actually push?…

Bienvenido a Player FM!

Player FM está escaneando la web en busca de podcasts de alta calidad para que los disfrutes en este momento. Es la mejor aplicación de podcast y funciona en Android, iPhone y la web. Regístrate para sincronizar suscripciones a través de dispositivos.

Escucha más de 500 temas

Escucha este programa mientras exploras