LoginSchedule demo
cedana / blog · field reports & engineering deep-dives

Writing from the
cedana team.

Engineering notes, benchmarks, and field reports covering how we build the automation layer for AI factories.
01Featured
Blueprint schematic of an eight-GPU node accumulating session state, with a state gauge rising toward the VRAM capacity lineBlueprint schematic of an eight-GPU node accumulating session state, with a state gauge rising toward the VRAM capacity line
FeaturedEngineering

The Era of Stateful Inference: how to improve cost per token.

We are entering the stateful inference era, driven by frontier models with longer context windows, longer in-flight sessions, and single instances spanning 8, 16, or more GPUs.

Jul 14, 2026· Cedana Editorial
read →
02Editor’s picks · 04
Blueprint fleet of GPU cells with only a third lit, next to a utilization gauge stuck at a dashed 30 percent ceiling and a second gauge breaking through itBlueprint fleet of GPU cells with only a third lit, next to a utilization gauge stuck at a dashed 30 percent ceiling and a second gauge breaking through it
Engineering

The Utilization Ceiling: Why AI and HPC Schedulers Hit 30% and How to Fix This

Nov 12, 2025
Blueprint timeline of two jobs crossing a 24-hour wall-time boundary: one is terminated and restarts from zero, the other checkpoints and continues into the next allocationBlueprint timeline of two jobs crossing a 24-hour wall-time boundary: one is terminated and restarts from zero, the other checkpoints and continues into the next allocation
Engineering

The Wall-Time Limit forces an expensive tradeoff in HPC.

Aug 12, 2026
Blueprint roofline plot: performance rises along the memory-bandwidth slope to a ridge point, then flattens at the compute ceilingBlueprint roofline plot: performance rises along the memory-bandwidth slope to a ridge point, then flattens at the compute ceiling
Engineering

Roofline Analysis and the Inference Value Chain

Jul 29, 2026
Blueprint of live migration: a dashed reclaimed spot instance, an accent state block in flight, and a healthy instance holding the restored stateBlueprint of live migration: a dashed reclaimed spot instance, an accent state block in flight, and a healthy instance holding the restored state
How-To

Using Cedana to Live-Migrate Stateful Workloads Between Spot Instances

Nov 2, 2024
03All posts · 08

insights from the team.

Field reports, engineering deep-dives, and benchmarks from the Cedana team.

Blueprint timeline of two jobs crossing a 24-hour wall-time boundary: one is terminated and restarts from zero, the other checkpoints and continues into the next allocationBlueprint timeline of two jobs crossing a 24-hour wall-time boundary: one is terminated and restarts from zero, the other checkpoints and continues into the next allocation
Engineering

The Wall-Time Limit forces an expensive tradeoff in HPC.

Jobs reach their wall-time limits and lose hours or days of in-memory progress. However, the limit itself is not the problem.

Aug 12, 2026 · Cedana Editorial
Blueprint roofline plot: performance rises along the memory-bandwidth slope to a ridge point, then flattens at the compute ceilingBlueprint roofline plot: performance rises along the memory-bandwidth slope to a ridge point, then flattens at the compute ceiling
Engineering

Roofline Analysis and the Inference Value Chain

Once you stop renting intelligence, you own performance.

Jul 29, 2026 · Cedana Editorial
Blueprint schematic of an eight-GPU node accumulating session state, with a state gauge rising toward the VRAM capacity lineBlueprint schematic of an eight-GPU node accumulating session state, with a state gauge rising toward the VRAM capacity line
Engineering

The Era of Stateful Inference: how to improve cost per token.

We are entering the stateful inference era, driven by frontier models with longer context windows, longer in-flight sessions, and single instances spanning 8, 16, or more GPUs.

Jul 14, 2026 · Cedana Editorial
Blueprint fleet of GPU cells with only a third lit, next to a utilization gauge stuck at a dashed 30 percent ceiling and a second gauge breaking through itBlueprint fleet of GPU cells with only a third lit, next to a utilization gauge stuck at a dashed 30 percent ceiling and a second gauge breaking through it
Engineering

The Utilization Ceiling: Why AI and HPC Schedulers Hit 30% and How to Fix This

GPU utilization across AI and HPC workloads is fundamentally capped at 30% because schedulers cannot migrate running jobs. Cedana's CPU and GPU migration capability surpasses this limitation, unlocking near-full utilization.

Nov 12, 2025 · Cedana Editorial
Blueprint reinforcement-learning loop: an agent block and an environment block joined by circular arrows, with reward pips along the return pathBlueprint reinforcement-learning loop: an agent block and an environment block joined by circular arrows, with reward pips along the return path
Engineering

Exploring new frontiers of Reinforcement Learning with Cedana

Cedana <3 RL !

Jul 1, 2025 · Niranjan Ravichandra