
Growing @AMD, engineer 🧠, AI harbinger, film fanatic, retro gamer, fmr @IBM, @NCState CS alum, Gen X, @theAvettBros reject circus, NC native, RT ≠ endorsement
Wake Forest, NC 🇺🇸
Joined March 2007
- Tweets109K
- Following7.9K
- Followers8.2K
- Likes67K
Pinned Tweet
Replying to @netflix
The Descent 3. An all female astronaut crew crashed and they are trapped in a collapsed cave system in a crater on the dark side of the moon. Tensions are high. Secrets are revealed. And they are not alone.
Ryan A Boyles retweeted
I can finally talk about Mass Effect Universe.
Since Bioware closed I can now reveal that I was part of the Mass Effect Council, saw what BioWare was building, and provided feedback during development.
• Set 600+ years after the Reaper War
• You play as a customizable N7 agent
• Instead of being a walking gun rack, you're a cross between John Wick and James Bond
• Shepard's legacy preserved, regardless of your ending
• A galaxy divided into three major factions
• Humanity, Krogan, Batarians, and more allied together
• Council races forming their own alliance
• The Geth emerging as a third superpower
• A galactic Cold War instead of another Reaper style invasion with a shadowy adversary
• Geth-built megastructures
• Connections between the Milky Way and Andromeda
• Experimental technology implanted in you as a unique point of difference
• Hacking and manipulating the environment
• See through walls, identify systems, remotely trigger hazards and more
• Handcrafted missions like ME1-3
• But with Hitman-style freedom in how you complete them
• Brutal close-range executions
• Squadmates who can clash with each other
• Romance wasn't being sanitised
• Liara returns - now an incredibly powerful Asari Matriarch
From what I saw, BioWare was making Mass Effect for us, the fans, not for people who don't exist.
It looked incredible.
And now we may never get to play it.
Find out more here: youtu.be/ZOWrTNazZxM
#SaveMassEffect
Ryan A Boyles retweeted
In their first podcast interview, fired OpenAI safety researchers @balesni and @tomekkorbak give us their account of the events leading up to their dismissal. It's a wild story: youtube.com/watch?v=tjAg_KxA…
Ryan A Boyles retweeted
Space Bunny is here! 🐰 MiniMax M3.1 Flash hits >1K tok/s, bs1 decode with SGLang, production-ready from Day 0. This integration is a joint effort by @MiniMax_AI, @NVIDIAAI, @AMD, and @radixark, and you can also run RL with Miles!
-Broad hardware support: NVIDIA Blackwell and Hopper (GB300/B300/B200/H200), plus AMD Instinct MI350X and MI355X.
-Agentic serving at scale: ~100K total tok/s per GPU on the AgentX coding-agent trace with 96 concurrent sessions.
-Speculative decoding: Supports MiniMax's DSpark draft and a DFlash2 draft trained by @radixark; supports HiCache and multimodal input.
-Ultra-low-latency inference: MegaKernels deliver 2.15× faster single-request decode.
MiniMax M3.1 Flash weights are now public. 428B parameters total, 23B activated per token.
Agentic performance: A generational leap in agentic tasks, coding, and cybersecurity — 76.2 on BankerToolBench, 73.7% on DeepSWE v1.1, 61.7 on OSWorld V2, 29.6 on Agents' Last Exam, and 82.2 on CyberGym.
Efficient by design: MiniMax Sparse Attention and DSpark speculative decoding improve inference efficiency.
Hugging Face:
huggingface.co/MiniMaxAI/Min…
ModelScope:
modelscope.cn/models/MiniMax…
API: platform.minimax.io
M Plan: minimax.io/m-planMiniMax
MCode for desktop: agent.minimax.io/download
Ryan A Boyles retweeted
FlyDSL, an open source Python kernel DSL, now plugs into PyTorch on @AMD MI350-series GPUs via torch.compile and eager mode.
Up to 1.68x geomean scaled GEMM kernel speedups and up to 3.01x end-to-end #vLLM serving speedups.
Read more: bit.ly/3W1zwUv
Live research with local models using the new Perplexity Portable Computer.
We are building a local-first, personalized AI news priority feed running on the Perplexity PPLX 27B.
Building new possibilities on AMD Ryzen AI Max+ architecture. bit.ly/4khjBvc
Ryan A Boyles retweeted
My interview with Sam Altman (@sama):
0:00 Intro
0:52 When Will AI Discover New Treatments?
3:31 Why AI Hasn’t Killed Coding Jobs
11:47 Personal AI Assistants
15:39 Why Are We Busier Than Ever?
20:10 Why the AI Consciousness Debate Worries Sam
26:09 ChatGPT Is Too Agreeable
30:03 OpenAI’s Bet on Chips and Cheaper Compute
34:20 Why AI Needs a New Kind of Computer
37:41 Sam on Pressure: “Pick the Big Thing”
42:18 Why “AI Will Cure Cancer” Isn’t Enough
Ryan A Boyles retweeted
I wrote a eulogy of sorts for BioWare. It's called, "BioWare was the greatest RPG developer ever."
KOTOR, Mass Effect, Baldur's Gate, Jade Empire, and the list just keeps going. Wow. Check it out if you're interested: ign.com/articles/bioware-was…
Ryan A Boyles retweeted
Bioware has been a massive part of CDPR history. We befriended them when localizing and selling Baldur’s Gate games in Poland, Czech, and Slovakia. And later they were great help when lending us their engine for Witcher 1, sharing their wisdom and even giving us a spot at their booth at E3. I had a pleasure to visit them a few times in Edmonton and also host them in our Warsaw office. Very sad for the people affected, but also sad to see a legendary brand go down. Not a great day in the games industry.
Ryan A Boyles retweeted
Another @AllThingsOpen in the books. @jhibbets conducting one of his dozens of interviews on site.
Ryan A Boyles retweeted
We made a show! youtube.com/watch?v=OKDai2_a…
The CEO of @G_Bionics, @Daniele_Pucci_, shares his perspective on the future of robotics in a conversation with Salil Raje, SVP and GM of the AMD Embedded business, and Kirk Saban, CVP of Embedded Products, Software and Solutions. Together, they discuss the evolving role of physical AI and the technologies needed to support it.
Watch here to learn more: youtube.com/watch?v=tFWlucki…
let's go
Our new show with NPR, Machine Gods, is finally here! Episode 1 is live wherever you get your podcasts.
The official podcast of the singularity starts today!
youtube.com/watch?v=OKDai2_a…
Ryan A Boyles retweeted
It's such a tough choice in local AI right now for which model to main. I am conflicted between GLM 5.3 Flash, Deepseek V4.1 1017 and Qwen 3.8 Flash Next. Which one do you prefer and why?
22%GLM 5.3 Flash
48%Deepseek v4.1 1017
30%Qwen 3.8 Flash Next
1,879 votes • Final resultsRead full blog: lmsys.org/blog/2026-10-18-sg…
Upstream > downstream 👐
The best optimizations are the ones developers don't have to think about. ⚡️
Cool to see FlyDSL helping accelerate PyTorch workloads on AMD GPUs. 😎
Another reminder that open ecosystems compound. 🚀
FlyDSL, an open source Python kernel DSL, now plugs into PyTorch on @AMD MI350-series GPUs via torch.compile and eager mode.
Up to 1.68x geomean scaled GEMM kernel speedups and up to 3.01x end-to-end #vLLM serving speedups.
Read more: bit.ly/3W1zwUv




















