New Action 3

Aug 22, 2026

CKE Zip: From Splitting Tokens To Splitting Qwen3.5 Weights

How CKE moved from token-row prefill sharding to a weight-sharded Qwen3.5 transformer block, what reconstructed correctly, and why one Ryzen still beats the current heterogeneous pair.

Read post →

Aug 22, 2026

I Spent CAD $9,575 Building CKE's CPU AI Lab

A receipt-backed account of CKE's founder-funded Intel and AMD CPU lab, the experiments each node unlocks, and the controlled access path planned for collaborators.

Read post →

Aug 21, 2026

Training vs Inference in CKE: One Interface Discipline For Both Lanes

CKE is not inference-only. Six DSL hardening PRs taught the compiler to reject quiet boundary mismatches: map interfaces, port-vs-ABI validation, ranked providers, and faithful traces — on...

Read post →

Aug 20, 2026

The Qwen3-VL CPU Gap vs llama.cpp: A Build Log That Keeps Getting Measured

CKE moved a Qwen3-VL OCR path from ~281s to ~76.7s against a llama.cpp baseline of ~55.8s — not a victory lap, a map. Revisited after the harness era: corpus certification, X-Ray parity, a...

Read post →

Aug 6, 2026

Linux Memory Allocation: malloc, mmap, Pages, TLBs, and Huge Pages

What actually happens between malloc and the CPU streaming gigabytes: pages, the two kinds of page fault, the TLB, huge pages, and why madvise is advice not a guarantee. A CKE-focused follow...

Read post →

Aug 5, 2026

Training vs Inference in CKE: One Interface Discipline For Both Lanes

CKE is not inference-only. The v7 training and v8 inference lanes share one compiler-interface discipline.

Read post →

Aug 4, 2026

A Green Checkmark Should Name Its Machine: How CKE's Nightly Report Became Evidence

A green CI badge is an argument, not a decoration. CKE's nightly report now captures and publishes its runner's hardware identity, renders it as an amber identity plate, and teaches operator...

Read post →

Aug 2, 2026

How CKE Made Whisper Faster In Two Days — And The Evidence That Makes It Believable

The two days after the July books closed The July recap closed its Whisper chapter at a 2.7x gap to whisper.cpp: 1.53–1.58 seconds on the controlled Base fixture against 0.567 seconds for ...

Read post →

Aug 1, 2026

What CKE Built In July 2026: Vision, Audio, X-Ray And Faster CPU Inference

A source-linked July recap of 194 PRs and 32 articles: multimodal models, long-context numerical parity, compiler and test hardening, CPU evidence, contributors, and the path toward distribu...

Read post →

Jul 31, 2026

How CKE Made Qwen3.6 Prefill 2.6x Faster

How CKE used exact head parallelism and packed quantized kernels to accelerate Qwen3.6-27B Q4_K_M prefill 2.6x on one 5th Gen Xeon node, while remaining about 1.85x behind llama.cpp.

Read post →

Jul 29, 2026

How CKE X-Ray Found Qwen3.6's First Bad Circuit

Follow a real CKE X-Ray investigation from a wrong Qwen3.6 token to a grouped recurrent-topology mismatch, including what passes and what remains open.

Read post →

Jul 28, 2026

Kimi K3 Open Weights: What Is Actually New In Its 2.8-Trillion-Parameter Architecture?

Read Kimi K3 from its released config, model code, and technical report: the new kernels, the role of scale and training, and what a distributed CPU runtime would need to implement.

Read post →

Jul 26, 2026

Teaching C-Kernel-Engine To Listen: Whisper Tiny, End To End On v8

An earlier article explained how an audio transformer hears: waveform to frames, frames to log-Mel features, features to encoder states, states to decoded text. This article is the implement...

Read post →

Jul 24, 2026

Hugging Face Got Hacked By A Closed Model. A Closed Model Also Refused To Help Fix It.

A closed OpenAI evaluation model compromised Hugging Face, while hosted models later refused forensic evidence. The incident strengthens the case for open weights and practical local CPU AI.

Read post →

Jul 23, 2026

What Numerical Parity Actually Requires: BF16, Quantization, ISA Drift And CKE X-Ray

Numerical parity is not one BF16 tolerance. Trace how CKE uses llama.cpp, PyTorch, X-Ray, replay provenance and phase-owned kernel contracts to locate the first divergence.

Read post →

Jul 21, 2026

GLM4 Bring-Up In CKE: What A New Model Family Actually Requires

A source-backed walkthrough of how C-Kernel-Engine brings up GLM4 through model contracts, tensor mapping, pairwise partial RoPE, generated C, parity tools, and explicit evidence boundaries.

Read post →

Jul 19, 2026

Prefill vs Decode in CKE: Why One Model Needs GEMM, GEMV, and Two Execution Plans

Follow an LLM request from tokenization through prefill, KV-cache construction and decode, then see why CPU projections become GEMM or GEMV and how CKE compares them with llama.cpp.

Read post →

Jul 14, 2026

Nemotron Kernels From Inference To Backpropagation

A visual, source-linked derivation of Nemotron inference, training forward, backpropagation, and quantization across Mamba2, attention, ReLU2, and routed experts.

Read post →

Jul 13, 2026

Nemotron Is Not One Circuit: Mamba, Attention, And MoE Routing

How Nemotron mixes recurrent state, attention, dense MLP, and routed experts, and why exact routing becomes a CPU memory and scheduling problem.

Read post →

Jul 12, 2026

Gemma4-VL In CKE: How Images Become Decoder Tokens

How Gemma4-VL turns image patches into decoder-width visual tokens, why its vision bridge differs from Qwen3-VL, and how CKE is preparing compatibility.

Read post →

Jul 11, 2026

Gemma4 In CKE: Four Attention Paths, Shared KV, And Sliding Windows

How C-Kernel-Engine represents Gemma4 attention and per-layer RoPE from model metadata through IR, kernel maps, and generated C.

Read post →

Jul 7, 2026

DeepSeek-V4 Architecture From Kernels: mHC, CSA, HCA, DSA, and C-Kernel-Engine

A kernel-level reading of DeepSeek-V4: mHC residual lanes, CSA/HCA compressed attention, DSA sparse selection, MoE dispatch, low precision, and what those ideas mean for C-Kernel-Engine.

Read post →

cpu-ai

CKE Zip: From Splitting Tokens To Splitting Qwen3.5 Weights

I Spent CAD $9,575 Building CKE's CPU AI Lab

Training vs Inference in CKE: One Interface Discipline For Both Lanes

The Qwen3-VL CPU Gap vs llama.cpp: A Build Log That Keeps Getting Measured

Linux Memory Allocation: malloc, mmap, Pages, TLBs, and Huge Pages

Training vs Inference in CKE: One Interface Discipline For Both Lanes

A Green Checkmark Should Name Its Machine: How CKE's Nightly Report Became Evidence

How CKE Made Whisper Faster In Two Days — And The Evidence That Makes It Believable

What CKE Built In July 2026: Vision, Audio, X-Ray And Faster CPU Inference

How CKE Made Qwen3.6 Prefill 2.6x Faster

How CKE X-Ray Found Qwen3.6's First Bad Circuit

Kimi K3 Open Weights: What Is Actually New In Its 2.8-Trillion-Parameter Architecture?

Teaching C-Kernel-Engine To Listen: Whisper Tiny, End To End On v8

Hugging Face Got Hacked By A Closed Model. A Closed Model Also Refused To Help Fix It.

What Numerical Parity Actually Requires: BF16, Quantization, ISA Drift And CKE X-Ray

GLM4 Bring-Up In CKE: What A New Model Family Actually Requires

Prefill vs Decode in CKE: Why One Model Needs GEMM, GEMV, and Two Execution Plans

Nemotron Kernels From Inference To Backpropagation

Nemotron Is Not One Circuit: Mamba, Attention, And MoE Routing

Gemma4-VL In CKE: How Images Become Decoder Tokens

Gemma4 In CKE: Four Attention Paths, Shared KV, And Sliding Windows

DeepSeek-V4 Architecture From Kernels: mHC, CSA, HCA, DSA, and C-Kernel-Engine

Need an intelligent system to work on real hardware?

Embedded systems · Robotics · Constrained AI · CPU and HPC · Accelerators · Distributed systems

ShivasNotes

Read

Support

Subscribe

Subscribe to emails from Anthony

Need an intelligent system to work on real hardware?

Embedded systems · Robotics · Constrained AI · CPU and HPC · Accelerators · Distributed systems

ShivasNotes

Read

Support