Видео с ютуба Skildcore

How 59 Million Deletes Broke GitHub: The MySQL Migration Outage

How 59 Million Deletes Broke GitHub: The MySQL Migration Outage
Stop Summarizing Agent History: Give the Agent the KV Delete Button

Stop Summarizing Agent History: Give the Agent the KV Delete Button
Did the API Actually Run That Model? Inside FlashProve zk-SNARKs

Did the API Actually Run That Model? Inside FlashProve zk-SNARKs
SGLang vs llama.cpp: The Slot Table That Unlocked 205 tok/s MoE Decode

SGLang vs llama.cpp: The Slot Table That Unlocked 205 tok/s MoE Decode
The LRU Cache That Doubled Local MoE Speeds: llama.cpp Expert Offloading

The LRU Cache That Doubled Local MoE Speeds: llama.cpp Expert Offloading
The SSM Layer Nobody Made Fast (Until Now): ReplaySSM Explained

The SSM Layer Nobody Made Fast (Until Now): ReplaySSM Explained
Why $400,000/Day Vanishes in AI Clusters: The Scheduler Placement Tax

Why $400,000/Day Vanishes in AI Clusters: The Scheduler Placement Tax
Bolt-On vs Native Drafters: Medusa, EAGLE, and MiniMax DSpark Architecture

Bolt-On vs Native Drafters: Medusa, EAGLE, and MiniMax DSpark Architecture
700 tok/s on a 2.8 Trillion Model: The 92-Layer TPU Megakernel

700 tok/s on a 2.8 Trillion Model: The 92-Layer TPU Megakernel
1.9x Higher MoE Throughput: The Overlap Math Behind Micro-Batch Ping-Pong

1.9x Higher MoE Throughput: The Overlap Math Behind Micro-Batch Ping-Pong
Stop Evicting Tokens: How InfiniGen Predicts Attention Before It Happens

Stop Evicting Tokens: How InfiniGen Predicts Attention Before It Happens
Why NVIDIA Locked Your RTX 3090 to Half Speed (And How to Bypass It)

Why NVIDIA Locked Your RTX 3090 to Half Speed (And How to Bypass It)
Why PagedAttention Slows Down Your LLM: Virtual Memory to the Rescue

Why PagedAttention Slows Down Your LLM: Virtual Memory to the Rescue
How vLLM's NIXL Push Mode Cuts Disaggregated TTFT by 3× (RDMA Write Deep Dive)

How vLLM's NIXL Push Mode Cuts Disaggregated TTFT by 3× (RDMA Write Deep Dive)
Mooncake Architecture Deep Dive | Disaggregated KV Cache & LLM Serving

Mooncake Architecture Deep Dive | Disaggregated KV Cache & LLM Serving
How SGLang's Unified Radix Cache Solves the Hybrid Model KV-Cache Crisis

How SGLang's Unified Radix Cache Solves the Hybrid Model KV-Cache Crisis
I Tested NVFP4 vs FP8 KV Cache on Blackwell (One Refused to Boot)

I Tested NVFP4 vs FP8 KV Cache on Blackwell (One Refused to Boot)
Your NVMe SSD is Dying in 5 Months: The Cost of KV-Cache Offload

Your NVMe SSD is Dying in 5 Months: The Cost of KV-Cache Offload
Why Your Local LLM Is 9x Slower on First Launch (The Page Cache Fix)

Why Your Local LLM Is 9x Slower on First Launch (The Page Cache Fix)