Subscribe
Sign in
Home
Notes
Start Here
Archive
About
Latest
Top
Discussions
Inside Attention-FFN disaggregation: What Groq LP40 will look like
How AFD works, from ByteDance's MegaScale-Infer paper to NVIDIA's Groq LPUs, and why the LP40 roadmap makes the acquisition hard to explain.
17 hrs ago
•
Subbu
8
2
June 2026
The Psychology of Hype Cycles
A priest and the dot-com bubble, FOMO and the rational mind, Why we do this every time
Jun 14
•
Subbu
45
4
8
May 2026
DRAM was the worst business in chips, now Micron is worth $1 trillion: 6 cycles of boom and bust and what could come next
How the cycle that built and buried memory companies broke, and the mistake Micron can't afford to repeat
May 31
•
Subbu
12
5
1
The Uncomfortable Truth Behind Deploying the Latest NVIDIA GPUs: MFU, Silent Data Corruption, and the Real Moat for CSPs and Hyperscalers
Lessons from DeepSeek, ByteDance, Meta, Google and the long deployment gap between a Jensen keynote and production reality
May 10
•
Subbu
18
1
1
March 2026
Why speculative decoding wants two kinds of silicon: NVIDIA, Groq, d-Matrix, Gimlet Labs, NVLink Fusion
Disaggregated speculative decoding and why the new class of accelerators are a better fit than GPUs
Mar 14
•
Subbu
36
8
Arista, AMD Pensando, and the Making of a Billionaire CEO
How Cisco and its culture of spin-ins led to the creation of Pensando and Silicon Valley's richest Indian-American CEO
Mar 1
•
Subbu
31
9
February 2026
How Mellanox set the stage for NVIDIA’s AI dominance and signaled the Storage Supercycle
Deep dive of the MLNX+NVIDIA partnership, GPUDirect RDMA, GPUDirect Storage, Bluefield DPU, Spectrum-X, and what could be coming next
Feb 6
•
Subbu
18
3
January 2026
Analysis of NVIDIA’s Bluefield-4 DPU and KV-Cache Context Memory Storage Platform (CES 2026): Architecture, Strategy, Dynamo, WEKA…
At CES 2026, NVIDIA announced the Context Memory Storage Platform, a new appliance designed to expand KV cache capacity beyond the GPU rack.
Jan 11
•
Subbu
21
3
5
December 2025
A Deep Dive into NVIDIA Rubin CPX: History, Architecture, Splitwise/DistServe, Inference Economics, and Limitations
A first principles analysis of why NVIDIA decided to make this new class of accelerator
Dec 27, 2025
•
Subbu
76
3
26
October 2025
Analysis of NVIDIA DGX Spark's GB10 SoC
Architecture, methodology, and memory interface of the GB10 SoC
Oct 21, 2025
•
Subbu
16
2
5 different ways DRAM and Compute are integrated
Examples of how GDDR, DDR, LPDDR, HBM and 3D DRAM are integrated with compute
Oct 5, 2025
•
Subbu
29
6
August 2025
Fundamental guide to understanding DRAM performance and timing parameters
How DRAM timing parameters affect HBM, LPDDR, GDDR bandwidth. How to get the most performance out of your memory.
Aug 25, 2025
•
Subbu
28
3
This site requires JavaScript to run correctly. Please
turn on JavaScript
or unblock scripts