Skip to main content
Scour
Browse
Getting Started
Login
Sign Up
You are offline. Trying to reconnect...
Copied to clipboard
Unable to share or copy to clipboard
Prefill Decoding
⏱️ Prefill Decoding
Specific
prefill phase, decode phase, chunked prefill, time-to-first-token
Filter Results
Timeframe
Fresh
Past Hour
Today
This Week
This Month
Feeds to Scour
Subscribed
All
Scoured
179
posts in
9.6
ms
RKSC: Reasoning-Aware
KV
Cache
Sharing and Confident Early Exit for Multi-Step LLM Inference
💾
KV Cache
Content type:
Academic
arxiv.org
·
19h
19 hours ago
Actions for RKSC: Reasoning-Aware KV Cache Sharing and Confident Early Exit for Multi-Step LLM Inference
Less-relevant results
A system programmer’s guide to LLM inference
💰
Inference Cost
Content type:
Blog
blog.xiangpeng.systems
·
2d
2 days ago
·
Hacker News
Actions for A system programmer’s guide to LLM inference
2x GH200 for LLM inference, Part 2:
vLLM
, DeepSeek V4 Flash, and MTP
🧠
Inference Engineering
Content type:
Blog
dnhkng.github.io
·
3d
3 days ago
Actions for 2x GH200 for LLM inference, Part 2: vLLM, DeepSeek V4 Flash, and MTP
LangChain vs LlamaIndex 2026: Response
Time
on 10 RAG Tasks
⚙️
ML Compilers
Content type:
Blog
Content type:
Discussion
tildalice.io
·
8h
8 hours ago
Actions for LangChain vs LlamaIndex 2026: Response Time on 10 RAG Tasks
Two Leaps to 1000
Tokens/s
on a 1T-Parameter Model: On Inference Systems, Execution Boundaries, and Co-Design
🧠
Inference Engineering
Content type:
Blog
tilert.ai
·
2d
2 days ago
·
Hacker News
Actions for Two Leaps to 1000 Tokens/s on a 1T-Parameter Model: On Inference Systems, Execution Boundaries, and Co-Design
MLPerf and the rise of
latency-aware
LLM benchmarking
🧠
Inference Engineering
edn.com
·
5d
5 days ago
Actions for MLPerf and the rise of latency-aware LLM benchmarking
KaiFelixBennett/gemma4-turboquant-rdna4: Run Gemma-4-31B at full 256K context on a $1,400 AMD RDNA4 GPU (gfx1201): TurboQuant
KV
cache
+ HIP-graph-safe Flash-Attention for llama.cpp, fully measured on real hardware.
⚡
FlashAttention
Content type:
Code
github.com
·
7h
7 hours ago
·
Hacker News
Actions for KaiFelixBennett/gemma4-turboquant-rdna4: Run Gemma-4-31B at full 256K context on a $1,400 AMD RDNA4 GPU (gfx1201): TurboQuant KV cache + HIP-graph-safe Flash-Attention for llama.cpp, fully measured on real hardware.
The economics of speculative
decoding
🚀
Speculative Decoding
Content type:
Blog
fergusfinn.com
·
2d
2 days ago
·
Hacker News
Actions for The economics of speculative decoding
Running LLM Inference on Kubernetes: What It Actually Takes
🧠
Inference Engineering
Content type:
Blog
fairwinds.com
·
5d
5 days ago
Actions for Running LLM Inference on Kubernetes: What It Actually Takes
From GPU to
Token
: The 8-Layer Observability Stack for AI Infrastructure
💰
Inference Cost
Content type:
Blog
jimmysong.io
·
1d
1 day ago
Actions for From GPU to Token: The 8-Layer Observability Stack for AI Infrastructure
ReasonAlloc: Hierarchical
Decoding-Time
KV
Cache Budget Allocation for Reasoning Models
🧠
Inference Engineering
Content type:
Academic
arxiv.org
·
19h
19 hours ago
Actions for ReasonAlloc: Hierarchical Decoding-Time KV Cache Budget Allocation for Reasoning Models
Massive AI Storage Demand Creates a New Memory Wall
🧠
Inference Engineering
Content type:
News
eetimes.com
·
9h
9 hours ago
Actions for Massive AI Storage Demand Creates a New Memory Wall
The Death of the Four Golden Signals: Designing Telemetry for Non-Deterministic Infrastructure
🛡️
SRE
devops.com
·
5d
5 days ago
Actions for The Death of the Four Golden Signals: Designing Telemetry for Non-Deterministic Infrastructure
AMD's Lemonade SDK For Local AI Adds NVIDIA CUDA Support
🎮
GPU Computing
phoronix.com
·
7h
7 hours ago
Actions for AMD's Lemonade SDK For Local AI Adds NVIDIA CUDA Support
Which is faster: Gemini 3.5 Flash or Kimi K2.6 on Cerebras
🔢
FP8 Training
Content type:
Blog
cerebras.ai
·
5d
5 days ago
Actions for Which is faster: Gemini 3.5 Flash or Kimi K2.6 on Cerebras
Benchmarking dots.tts on Strix Halo
🎮
GPU Computing
sleepingrobots.com
·
2d
2 days ago
Actions for Benchmarking dots.tts on Strix Halo
What Breaks When Multi-Agent Systems Scale
🛡️
SRE
digitalocean.com
·
16h
16 hours ago
Actions for What Breaks When Multi-Agent Systems Scale
Breaking the Ice: Analyzing Cold Start
Latency
in
vLLM
💾
KV Cache
Content type:
Academic
arxiv.org
·
2d
2 days ago
·
Hacker News
Actions for Breaking the Ice: Analyzing Cold Start Latency in vLLM
Nvidia DGX Spark GB10 – AI Models and Guide with
vLLM
and Autonomous Script
💾
KV Cache
Content type:
Code
github.com
·
5d
5 days ago
·
Hacker News
Actions for Nvidia DGX Spark GB10 – AI Models and Guide with vLLM and Autonomous Script
Cohere open-sources a coding agent that runs on a single H100
🎮
GPU Computing
venturebeat.com
·
1d
1 day ago
Actions for Cohere open-sources a coding agent that runs on a single H100
« Page 1
·
Page 3 »
Log in to enable infinite scrolling
Keyboard Shortcuts
Navigation
Next / previous item
j
/
k
Open post
o
or
Enter
Preview post
v
Post Actions
Love post
a
Like post
l
Dislike post
d
Undo reaction
u
Save / unsave
s
Recommendations
Add interest / feed
Enter
Not interested
x
Go to
Home
g
h
Interests
g
i
Feeds
g
f
Likes
g
l
History
g
y
Changelog
g
c
Settings
g
s
Browse
g
b
Search
/
Pagination
Next page
n
Previous page
p
General
Show this help
?
Submit feedback
!
Close modal / unfocus
Esc
Press
?
anytime to show this help