Grants will need a line for tokens

Predictions are of course dangerous. They’re also pretty much impossible and random. Who saw that location services on cell phones would create Uber and lead to disruption in the taxi industry? VisiCalc was supposed to kill accountants, but we have even more of them now, they just do different stuff. You should also beware anyone who predicts things like “here’s why I’ll be fine but everyone else is in trouble.” With that out of the way, here’s mine. ...

2026-10-30 · 2 min · Tyler Collins

Writing an AGENTS.md for a national cluster

Agents have started causing problems on our clusters. Heavy work kicked off on shared login nodes, Slurm getting polled in tight loops, that sort of thing. I’d been using a small CLAUDE.md for myself on Nibi for a while, and I was asked to turn it into something for everyone. The Alliance clusters get their environment from ComputeCanada/software-stack-config, so I opened PR #111 to add an agents/AGENTS.md there on request. Claude Code reads /etc/claude-code/CLAUDE.md system wide, and most other harnesses look for an AGENTS.md, so one file and a symlink covers a lot of ground. ...

2026-10-20 · 3 min · Tyler Collins

Your LLM Loop Is a Batch Job

Today I had to rate limit a user on SHARCNET’s internal LLM service for the first time. In the last 24 hours they had sent about 800,000 requests, and every single one was under 500 tokens. To be clear, they didn’t do anything wrong. The service is set up to let them do it. They had a pile of small work and a working endpoint, so they wrote some sort of loop. I probably would have done the same. But a loop that sends hundreds of thousands of tiny prompts is a batch job, and this was a really good reminder of why batch clusters are still super important. ...

2026-10-15 · 3 min · Tyler Collins

What Does a GPU-Hour Actually Measure?

You’ve probably seen a line like this in a paper or grant report: We used 10,000 GPU-hours. It looks like a useful answer. There’s a number and a unit. Very official. But what does a GPU-hour actually measure? On its own, pretty much nothing. I can’t compare that number with another project until I know what counted as a GPU, whether it was a whole device or a slice, and how the authors calculated the total. ...

2026-09-22 · 4 min · Tyler Collins

Writing More Code with AI Agents

I recently gave a SHARCNET General Interest Webinar called “Writing More Code with AI Agents.” More than 300 people registered. Attendance was excellent, there were lots of questions, and I’ve had a decent number of follow-up conversations over email. Pretty happy with how it turned out! The response also confirmed why I wanted to give the talk. People are constantly asking me agents and what they should be doing with them. They’re watching other researchers and developers move very quickly with these tools, and there’s a real fear of missing out. They want to try them, but they don’t necessarily know where to start or how much of the output they should trust. ...

2026-08-27 · 6 min · Tyler Collins

Serial Farms: Package Options and When to Switch to Farming

Submitting a few small jobs one at a time is fine. Submitting hundreds that way is annoying for you and hard on the scheduler. This webinar compared ways to group lots of serial tasks into manageable jobs. I covered when ordinary submissions stop making sense, how array indexes can drive repeated runs, and when it is time to use a serial farm. Abstract Small jobs are convenient to submit individually, but at scale they can overload a scheduler, inflate queue times, and ultimately reduce throughput. This webinar examines practical strategies for consolidating large numbers of short tasks, including job arrays, task-bundling techniques, and wrapper-based aggregation. We will discuss how these approaches differ in overhead, portability, scheduler behavior, and job-failure handling. The session will also provide guidance on recognizing when packaging options no longer yield sufficient throughput and when transitioning to a serial-farming model becomes advantageous. Serial farms can mitigate scheduler pressure, improve wait times on busy clusters, and offer more predictable performance. Examples will be provided throughout the webinar and shared on GitHub for future reference. ...

2025-12-03 · 1 min · Tyler Collins

Diagnosing Wasted Resources from User-Facing Portals

A slow job and a slow queue are different problems. This talk was about using the national cluster portals to figure out which one you’re dealing with. I covered job history, account priority, wait times, and resource-use reports. The examples looked at CPU, memory, and GPU requests that did not match the work being done. The point was to check the evidence before changing a job request. Abstract Researchers often leave resources on the table when specifying their job requirements on the national systems. This talk builds on previous sessions and uses the Digital Research Alliance of Canada’s User Facing Portals to explore what different types of jobs look like when they waste resources. Demonstrations will include interactive jobs, parallel jobs, GPU workflows, and more. With more accurate job specifications, researchers can expect shorter wait times and more throughput on any general-purpose system. ...

2024-08-14 · 1 min · Tyler Collins

Modern Approaches to Profiling in Python with Scalene

Before rewriting slow Python or moving it into Cython, it helps to find out what is actually slow. This Compute Ontario Colloquium was an introduction to Scalene. Scalene separates time spent in Python from time spent in native libraries, and it can also profile memory and GPU use. I ran it on the Alliance systems from a Jupyter notebook, then compared native Python, vectorized code, Cython, and just-in-time compilation. Abstract Python is a language developers choose to write in for convenience rather than speed. However, speed can be recovered by offloading calculations to libraries which leverage lower-level languages like NumPy, Cython, and more. Scalene is a high-performance CPU, GPU, and memory profiler which can illustrate where code should be passing calculations to other libraries for significant increases in speed. Scalene also includes support for Jupyter Notebooks, OpenAI suggestions for vectorizing code, as well as a significantly lower overhead and higher accuracy than other profilers. This talk will introduce the concepts required for understanding why external libraries are faster than native Python, interactions with approaches such as Cython and just-in-time compilers, as well as a live demonstration of Scalene on the Alliance systems inside of a Jupyter Notebook. Familiarity with Python, virtual environments, and Jupyter notebooks will be assumed. ...

2023-05-03 · 2 min · Tyler Collins

ViewClust: Early Days

In the early days of working for SHARCNET, my colleague and I decided to standardize how cluster metrics were computed across our internal data frames. As mentioned in a previous post, part of the solution was pandas. The second part was figuring out how to deploy the package for others to contribute to, as well as install on their own specific HPC clusters. Some quick searching revealed that PyPI and pip were the way to go. ...

2022-03-22 · 1 min · Tyler Collins

Conquering the Scheduler

Coming from a slightly different angle this time, I found that researchers were often isolating themselves to fewer resources on HPC systems by not investigating what the node feature mixture looked like. As such, this talk was created to help direct potentially abstract development efforts toward optimizing for the feature sets that are most available on an HPC cluster. Below is my abstract for the talk as well as the recording: ...

2022-03-22 · 1 min · Tyler Collins