emilio andere

cofounder/ceo @wafer_ai. math @uchicago

[risa @risashimmura]

we’re powering token pass on @vercel 🤝 @wafer_ai

available as an add-on to vercel pro. more models coming soon

https://t.co/8FapF5sULM

[emilio andere @gpuemi]

@vercel_dev awesome partnering here to power token pass!

[Vercel Developers @vercel_dev]

DeepSeek V4.1 Flash 1017 is now included in Vercel Token Pass.

• New weights from DeepSeek
• 1M context window
• No config change

Get $80 of tokens for $30/mo ↓https://t.co/VGjgv7AgJz

[steve @gpusteve]

excited to partner with @vercel to power Token Pass

for $30 per month, your team gets $80 in monthly credit for:

- GLM 5.3
- Kimi K3
- DeepSeek V4.1 Flash 1017

powered by @wafer_ai , with more models coming soon

available as an add-on to Vercel Pro.

links in thread 🧵

[emilio andere @gpuemi]

@wafer_ai @vercel always a pleasure working with the @vercel team

[wafer @wafer_ai]

excited to partner with @Vercel to power Token Pass!

for $30 per month, your team gets $80 in monthly credit for:

- GLM 5.3,
- Kimi K3
- DeepSeek V4.1 Flash 1017

powered by Wafer, with more models coming soon!

available as an add-on to Vercel Pro.

links in thread 🧵

[emilio andere @gpuemi]

Vercel Token Pass:
https://t.co/rCgRUTKP4L

[emilio andere @gpuemi]

excited to partner with @vercel to power Token Pass!

for $30 per month, your team gets $80 in monthly credit for:

- GLM 5.3
- Kimi K3
- DeepSeek V4.1 Flash 1017

powered by @wafer_ai, with more models coming soon!

links in thread 🧵

[emilio andere @gpuemi]

@rauchg Excited to support Token Pass 🫡

[Guillermo Rauch @rauchg]

DeepSeek v4.1 Flash (1017) is now included in ▲ Token Pass.

Fast, mind-bogglingly cheap, and served with ZDR (zero-data retention).

It's a phenomenal model to run subagents that execute plans conceived by larger and more expensive models. You can port TypeScript to Rust 10 times and not run out of credits 😂

[will o’brien @Willob]

Another takeaway after three years of building a company is that the returns to just "being a good person", and not acting transactionally, are higher than I expected.

There are so many transactional, boring, and at times immoral people in the tech industry that if you act with sustained warmth, authenticity, humor, and integrity, great gifts will be laid before you. White pill.

[georgy @georgezyst]

the only real moat is love of the game

[François Chollet @fchollet]

If you build an automation machine, the way to monetize it is to sell it to as many people as possible -- anyone who has tasks to automate.

But if what you build is an invention machine, then the best way to monetize it is to use it yourself.

[emilio andere @gpuemi]

@utsav_khairnar @wafer_ai I’ve written a lot on LinkedIn! will start posting here too

[emilio andere @gpuemi]

@wafer_ai's gpu performance engineering guide is getting some big updates!

what resources are still missing?

[wafer @wafer_ai]

we're doing a massive update on Wafer's popular open source guide to GPU performance engineering!

a few cool resources going into this update:
⠀
1. Stream-K: why a matmul can leave part of the GPU idle while the last blocks finish, and how dividing the multiply-accumulate work more evenly helps.
⠀
2. tcgen05 for dummies: gau-nernst builds a Blackwell matmul step by step. Tensor Memory, asynchronous copies and pipelining, with code you can follow as the kernel develops.
⠀
3. Thinking Machines' Defeating Nondeterminism: how changing the batch can change the order of floating-point additions, and why that can change inference results. then, how to make the kernels batch-invariant.
⠀
4. Triton's Linear Layouts: how tensor elements are arranged across threads and registers, and the data movement needed to change that arrangement.
⠀
the update also expands the quantization, speculative decoding and hardware references.
⠀
the guide starts with one inference request, then moves through GPU fundamentals, kernels, serving engines and distributed systems. start at the beginning if you're new; use the individual sections when you hit a specific problem.
⠀
if you know a primary source that explains one of these topics better, open a PR.

[emilio andere @gpuemi]

@nikunj @wafer_ai love to hear that Nikunj :)

[steve @gpusteve]

huge updates coming to the gpu perf engineering repo

what resources are we missing?

[wafer @wafer_ai]

we're doing a massive update on Wafer's popular open source guide to GPU performance engineering!

a few cool resources going into this update:
⠀
1. Stream-K: why a matmul can leave part of the GPU idle while the last blocks finish, and how dividing the multiply-accumulate work more evenly helps.
⠀
2. tcgen05 for dummies: gau-nernst builds a Blackwell matmul step by step. Tensor Memory, asynchronous copies and pipelining, with code you can follow as the kernel develops.
⠀
3. Thinking Machines' Defeating Nondeterminism: how changing the batch can change the order of floating-point additions, and why that can change inference results. then, how to make the kernels batch-invariant.
⠀
4. Triton's Linear Layouts: how tensor elements are arranged across threads and registers, and the data movement needed to change that arrangement.
⠀
the update also expands the quantization, speculative decoding and hardware references.
⠀
the guide starts with one inference request, then moves through GPU fundamentals, kernels, serving engines and distributed systems. start at the beginning if you're new; use the individual sections when you hit a specific problem.
⠀
if you know a primary source that explains one of these topics better, open a PR.

[emilio andere @gpuemi]

@chenzhangreal yes

[emilio andere @gpuemi]

you'll know more about blackwell tuning than 93.73% of people if you fully understand the Blackwell Tuning Guide

it covers register limits, thread-block clusters, distributed shared memory, shared-memory carveouts, L2 persistence, and NVLink peer access.

follow and save to keep up with the series. links in thread 🧵

[wafer @wafer_ai]

we launched the most comprehensive ai performance engineering repo in the world

we're posting every single resource

follow and save to keep up with the series. links in thread 🧵

part 14: Blackwell Tuning Guide

NVIDIA covers the execution limits, memory hierarchy, and communication mechanisms behind tuning AI workloads on Blackwell:

- register budgets, resident warps, and occupancy in fused AI kernels.

- shared-memory limits for attention and matmul tiles across Blackwell compute capabilities.

- thread-block clusters and distributed shared memory for exchanging tensor data across blocks.

- 8-block portable clusters, B200's optional 16-block clusters, and the parallelism tradeoff of larger clusters.

- L1/shared-memory carveouts and the balance between caching and explicit tensor staging.

- L2 cache persistence for inference kernels that reuse the same tensors.

- NVLink peer access and direct tensor transfers between GPUs in model-parallel inference.

the guide connects Blackwell's hardware limits to kernel design decisions like tile size, block cooperation, and tensor reuse, helping engineers choose what to test when tuning attention, matmul, and multi-GPU inference.