v
@vsoch.bsky.social
I’m the Vanessasaurus! 🥑
https://vsoch.github.io
https://vsoch.github.io
We used state machines to orchestrate a complex #HPC workload in #Kubernetes, and performed a Usability study across #Azure, #GoogleCloud, and #AWS with over 28K CPU and 256 GPU that resulted in over 26K datasets!
dl.acm.org/doi/10.1145/...
dl.acm.org/doi/10.1145/...
Will be presented at #SC25.
dl.acm.org/doi/10.1145/...
dl.acm.org/doi/10.1145/...
Will be presented at #SC25.
November 7, 2025 at 7:22 PM
We used state machines to orchestrate a complex #HPC workload in #Kubernetes, and performed a Usability study across #Azure, #GoogleCloud, and #AWS with over 28K CPU and 256 GPU that resulted in over 26K datasets!
dl.acm.org/doi/10.1145/...
dl.acm.org/doi/10.1145/...
Will be presented at #SC25.
dl.acm.org/doi/10.1145/...
dl.acm.org/doi/10.1145/...
Will be presented at #SC25.
Belated Halloween Dance! 🎃
This is the first time that I've ever attempted this dance - I've seen it enough times that I got the general gist! But let's be real, it's terrible. 😄 I guess I don't care, because I had fun. Happy belated Halloween!
This is the first time that I've ever attempted this dance - I've seen it enough times that I got the general gist! But let's be real, it's terrible. 😄 I guess I don't care, because I had fun. Happy belated Halloween!
November 2, 2025 at 2:57 AM
Belated Halloween Dance! 🎃
This is the first time that I've ever attempted this dance - I've seen it enough times that I got the general gist! But let's be real, it's terrible. 😄 I guess I don't care, because I had fun. Happy belated Halloween!
This is the first time that I've ever attempted this dance - I've seen it enough times that I got the general gist! But let's be real, it's terrible. 😄 I guess I don't care, because I had fun. Happy belated Halloween!
This is a progressive tempo run, where you start easily and go up to a comfortably fast pace. There is something to say about strength (of heart and muscle) per unit of body weight! I can't win otherwise. 😆
October 28, 2025 at 2:40 AM
This is a progressive tempo run, where you start easily and go up to a comfortably fast pace. There is something to say about strength (of heart and muscle) per unit of body weight! I can't win otherwise. 😆
And so it begins! Is lifting the bar hard for me? Yes. Is my shirt inside out? Also yes. 😂 Am I going to be consistent, and steadfast in my ability to manifest what I want in life? HELL yes. 👏
October 15, 2025 at 10:45 PM
And so it begins! Is lifting the bar hard for me? Yes. Is my shirt inside out? Also yes. 😂 Am I going to be consistent, and steadfast in my ability to manifest what I want in life? HELL yes. 👏
Thanks Brian! If it's OK, I'd rather not make a chat or forum account, so I'll post here. You can see the failure here, and a picture below:
github.com/flux-framewo...
The base image is regularly rebuilt from this Dockerfile github.com/flux-framewo... and CI fails for it too. Thanks for the help!
github.com/flux-framewo...
The base image is regularly rebuilt from this Dockerfile github.com/flux-framewo... and CI fails for it too. Thanks for the help!
October 9, 2025 at 3:18 AM
Thanks Brian! If it's OK, I'd rather not make a chat or forum account, so I'll post here. You can see the failure here, and a picture below:
github.com/flux-framewo...
The base image is regularly rebuilt from this Dockerfile github.com/flux-framewo... and CI fails for it too. Thanks for the help!
github.com/flux-framewo...
The base image is regularly rebuilt from this Dockerfile github.com/flux-framewo... and CI fails for it too. Thanks for the help!
When things are hard, we remember that we are responsible for ourselves, and the outcomes of our choices. Today I chose strength, and to prove to myself that I am still the strongest person that I know. With time (and protein!) you will see it too. I embrace darkness, but I choose to be a light.
October 6, 2025 at 5:00 AM
When things are hard, we remember that we are responsible for ourselves, and the outcomes of our choices. Today I chose strength, and to prove to myself that I am still the strongest person that I know. With time (and protein!) you will see it too. I embrace darkness, but I choose to be a light.
It handles logical operators within groups (each associated with one cpuset mask) so the user can break down shape by task or other (second picture).
More to come! Right now we are using it for testing. 🤓
More to come! Right now we are using it for testing. 🤓
October 4, 2025 at 11:56 PM
It handles logical operators within groups (each associated with one cpuset mask) so the user can break down shape by task or other (second picture).
More to come! Right now we are using it for testing. 🤓
More to come! Right now we are using it for testing. 🤓
It's about time we think more about our job shape. This is something I'm really excited about for #FluxFramework! Have you ever thought about the implications of your flag choices for binding? I made a little tool today (WIP) that can detect (and then predict) topology (binding) down to PU.
October 4, 2025 at 11:56 PM
It's about time we think more about our job shape. This is something I'm really excited about for #FluxFramework! Have you ever thought about the implications of your flag choices for binding? I made a little tool today (WIP) that can detect (and then predict) topology (binding) down to PU.
Developer tip - taking the time to write comments for the next day. If you are a gerbil, a monster, a three toed sloth, or a goblin that will die alone, make sure to be honest with yourself about that. 🧌
September 21, 2025 at 6:46 AM
Developer tip - taking the time to write comments for the next day. If you are a gerbil, a monster, a three toed sloth, or a goblin that will die alone, make sure to be honest with yourself about that. 🧌
What did I do for the holiday weekend? 🤔 Biked around for 15 miles looking for a track. I didn't find one, so opted for paved paths instead. I didn't hit the pace I wanted, but in retrospect, this was a good effort! I ran this at noon on Saturday, when it was in the 90s. 😎
September 2, 2025 at 12:16 AM
What did I do for the holiday weekend? 🤔 Biked around for 15 miles looking for a track. I didn't find one, so opted for paved paths instead. I didn't hit the pace I wanted, but in retrospect, this was a good effort! I ran this at noon on Saturday, when it was in the 90s. 😎
We are immensely grateful to the larger community, our collaborators, and tutorial participants! Folks came with a huge set of questions that you'll hear in the Q&A of our tutorial video. Thank you for the questions and discussion - keep it coming! 🚗
We hope to see you at #SC25. 👀
We hope to see you at #SC25. 👀
August 21, 2025 at 5:20 AM
We are immensely grateful to the larger community, our collaborators, and tutorial participants! Folks came with a huge set of questions that you'll hear in the Q&A of our tutorial video. Thank you for the questions and discussion - keep it coming! 🚗
We hope to see you at #SC25. 👀
We hope to see you at #SC25. 👀
At the end we jumped right into the deep end (1:16:00) and walked through Chapter 4, demonstrating running LAMMPS on bare-metal and in the Flux Operator in User-space Kubernetes. The notebook also has MuMMI component examples, along showcasing the the #Kubeflow Trainer.
August 21, 2025 at 5:20 AM
At the end we jumped right into the deep end (1:16:00) and walked through Chapter 4, demonstrating running LAMMPS on bare-metal and in the Flux Operator in User-space Kubernetes. The notebook also has MuMMI component examples, along showcasing the the #Kubeflow Trainer.
The tutorial portion this year was so good! 🥳 We developed a custom EC2 Spawner, meaning our participants got a web interface notebook running on their private hpc7g instance on #AWS. The instance came with Flux and User-space Kubernetes! This is a new Chapter 4 on Converged Computing. 😎
August 21, 2025 at 5:20 AM
The tutorial portion this year was so good! 🥳 We developed a custom EC2 Spawner, meaning our participants got a web interface notebook running on their private hpc7g instance on #AWS. The instance came with Flux and User-space Kubernetes! This is a new Chapter 4 on Converged Computing. 😎
The third section of the talk is near and dear to me - Converged Computing (25:52). Portability is a new metric of performance. Flux is a portability layer for #HPC and cloud. Learn about the Flux Operator, User-space #Kubernetes, and recent work to automate experiments with helm, and #eBPF.
August 21, 2025 at 5:20 AM
The third section of the talk is near and dear to me - Converged Computing (25:52). Portability is a new metric of performance. Flux is a portability layer for #HPC and cloud. Learn about the Flux Operator, User-space #Kubernetes, and recent work to automate experiments with helm, and #eBPF.
We transition into Flux Projects, specifically flux-core and flux-sched, and updates since our presentation one year ago. Our Flux team is amazingly productive! 💪
August 21, 2025 at 5:20 AM
We transition into Flux Projects, specifically flux-core and flux-sched, and updates since our presentation one year ago. Our Flux team is amazingly productive! 💪
If you've never heard of Flux or are unfamiliar with workload managers, you'll want to watch from the beginning (1:20). In addition to basic definitions, we talk about the challenges Flux was designed to solve, and the features that make it special (10:35).
August 21, 2025 at 5:20 AM
If you've never heard of Flux or are unfamiliar with workload managers, you'll want to watch from the beginning (1:20). In addition to basic definitions, we talk about the challenges Flux was designed to solve, and the features that make it special (10:35).
This was 20 years ago. When we are young we often can't anticipate oncoming darkness. But we also can't anticipate our own resilience. What we come to realize with experience is that we are always in a mixture of light and shadow. It is a choice to not just see, but try to be a source of light. 🕯️
August 6, 2025 at 5:13 AM
This was 20 years ago. When we are young we often can't anticipate oncoming darkness. But we also can't anticipate our own resilience. What we come to realize with experience is that we are always in a mixture of light and shadow. It is a choice to not just see, but try to be a source of light. 🕯️
Our #SC25 tutorial is up! 🎉
sc25.conference-program.com/presentation...
I'm already excited! We will be co-presenting with #AWS and teaching you how to run #HPC workloads using #Kubernetes with the Flux Operator. And a taste of MuMMI, a workload with Ai/ML components. Hope to see you there! 🥳
sc25.conference-program.com/presentation...
I'm already excited! We will be co-presenting with #AWS and teaching you how to run #HPC workloads using #Kubernetes with the Flux Operator. And a taste of MuMMI, a workload with Ai/ML components. Hope to see you there! 🥳
August 4, 2025 at 3:58 AM
Our #SC25 tutorial is up! 🎉
sc25.conference-program.com/presentation...
I'm already excited! We will be co-presenting with #AWS and teaching you how to run #HPC workloads using #Kubernetes with the Flux Operator. And a taste of MuMMI, a workload with Ai/ML components. Hope to see you there! 🥳
sc25.conference-program.com/presentation...
I'm already excited! We will be co-presenting with #AWS and teaching you how to run #HPC workloads using #Kubernetes with the Flux Operator. And a taste of MuMMI, a workload with Ai/ML components. Hope to see you there! 🥳
For the first time - user-space Kubernetes running under a Flux allocation on a production cluster. This is OSU and LAMMPS. This has been months of work and persistence. We got this working on an old kernel, and hugely strict security policy. Experiments and more details coming soon! 🥳
June 27, 2025 at 1:20 AM
For the first time - user-space Kubernetes running under a Flux allocation on a production cluster. This is OSU and LAMMPS. This has been months of work and persistence. We got this working on an old kernel, and hugely strict security policy. Experiments and more details coming soon! 🥳
For most that missed the #ISC25 Flux Framework Tutorial, we just posted our slides online:
github.com/flux-framewo...
Thank you to those that attended, and see you next time! 👋
github.com/flux-framewo...
Thank you to those that attended, and see you next time! 👋
June 18, 2025 at 1:29 AM
For most that missed the #ISC25 Flux Framework Tutorial, we just posted our slides online:
github.com/flux-framewo...
Thank you to those that attended, and see you next time! 👋
github.com/flux-framewo...
Thank you to those that attended, and see you next time! 👋
For the last taste of current work, we talk about running user-space Kubernetes alongside Flux, a project we call "The Bare Metal Bros." Although slirp4netns adds network overhead, when we use bypass mechanisms (Infiniband and EFA) we get close to equivalent performance.
June 13, 2025 at 6:30 PM
For the last taste of current work, we talk about running user-space Kubernetes alongside Flux, a project we call "The Bare Metal Bros." Although slirp4netns adds network overhead, when we use bypass mechanisms (Infiniband and EFA) we get close to equivalent performance.