release: prepare 0.40.1 (hotfix: #968)

CHANGELOG.md

@@ -2,6 +2,38 @@
 
 ## [Unreleased]
 
+## [0.40.1] - 2026-08-12
+
+Headline: **correctness hotfix — silently wrong output, not a crash.** `DefaultCpuOps.transpose()` for packed quantized weights (Q4_0/Q5_0/Q5_1/Q8_0/Q4_K/Q5_K/Q6_K) performed a shape-only relabel instead of a real block-grid byte permutation whenever a row spanned more than one quant block (`blocksPerInputDim > 1` — true of virtually every real model). `ops.matmul(x, ops.transpose(W))` fed the packed-quant kernels bytes in the wrong order across all three kernel tiers — scalar, Panama-vector, and native (FFM/JNI) — silently producing wrong numbers, sometimes all-zero output, with no exception raised. Upgrading is strongly recommended for anyone using packed-quantized weights with `ops.transpose()`.
+
+### Fixed
+
+- **Packed-quant `transpose()` silently corrupted matmul output**
+  ([#968](https://github.com/SKaiNET-developers/SKaiNET/issues/968),
+  [#969](https://github.com/SKaiNET-developers/SKaiNET/pull/969))
+  — `DefaultCpuOps.transpose()` swapped the tensor's shape metadata without
+  physically reordering the underlying `packedData` bytes, on the assumption
+  that the packed-quant matmul kernels index those bytes block-major
+  regardless of layout. That assumption only holds when there is a single
+  quant block per row (`blocksPerInputDim == 1`); for any wider row the
+  canonical (row-major) and kernel-native block orderings are literal
+  transposes of the `(outputDim, blocksPerInputDim)` block grid and do not
+  coincide, so a freshly-loaded weight run through `ops.transpose()` fed the
+  scalar, Panama-vector, and native (FFM/JNI) kernel tiers alike bytes in the
+  wrong order — for all seven packed formats (Q4_0/Q5_0/Q5_1/Q8_0/Q4_K/Q5_K/Q6_K).
+  `transpose()` now performs a real `O(bytes)` block-grid permutation
+  (`transposePackedBlocks`); a misaligned packed tensor (`inputDim` not a
+  multiple of the format's block size) now throws `IllegalArgumentException`
+  instead of silently truncating a partial trailing block.
+  `DefaultCpuOpsJvm`'s separate, independently-buggy shape-swap-only
+  interception for `Q4_KTensorData` is removed, falling through to the
+  shared corrected implementation. Caught by a new ground-truth regression
+  test (`NativeLazyTransposeGroundTruthReproTest`) that dequants each packed
+  format's canonical and kernel-native byte layouts independently and checks
+  both the classic (`transpose` + matmul) and pre-transposed paths against
+  that ground truth, per format — the kind of test the original "same bytes,
+  new shape" optimization lacked.
+
 ## [0.40.0] - 2026-08-11
 
 Headline: **big models fit on real devices, and SKaiNET reaches iOS/macOS

README.md

@@ -53,7 +53,7 @@ Add the core dependencies (Gradle Kotlin DSL):
 ```kotlin
 dependencies {
     // Recommended: import the umbrella BOM and drop versions on the engine modules.
-    implementation(platform("sk.ainet:skainet-bom:0.40.0"))
+    implementation(platform("sk.ainet:skainet-bom:0.40.1"))
 
     implementation("sk.ainet.core:skainet-lang-core")
     implementation("sk.ainet.core:skainet-backend-cpu")
@@ -297,7 +297,11 @@ val withoutLabel = dataPipeline<RawDataset>()
 
 ---
 
-## What's New in 0.40.0
+## What's New in 0.40.1
+
+- **Correctness hotfix: packed-quant `transpose()` was silently wrong, not crashing.** `ops.matmul(x, ops.transpose(W))` on a packed-quantized weight (Q4_0/Q5_0/Q5_1/Q8_0/Q4_K/Q5_K/Q6_K) with more than one quant block per row produced silently incorrect output — sometimes all-zero — across the scalar, Panama-vector, *and* native (FFM/JNI) kernel tiers, with no exception raised. `transpose()` now performs a real block-grid byte permutation instead of a shape-only relabel; a misaligned packed tensor now throws instead of silently truncating. Closes [#968](https://github.com/SKaiNET-developers/SKaiNET/issues/968). **Upgrading is strongly recommended** for anyone calling `ops.transpose()` on packed-quantized weights.
+
+### Previously, in 0.40.0
 
 - **Android models grow past the ART heap cap.** Off-heap/mmap tensor storage shares the JVM's memory-mapped weight loading with Android — dense F32 tensors serve as zero-heap mapped views, and weight bytes live in OS-paged file-backed pages instead of the managed heap. A 640 MB dense model now loads with **1.4 MB** of heap allocation.
 - **GGUF `DEQUANTIZE_TO_FP32` no longer over-allocates.** A 1.1B Q4_K_M GGUF transiently needed >12 GB heap against a ~4.4 GB dense-FP32 floor. Three compounding allocation sources in the loader and K-quant kernels are fixed, bringing peak live allocation to ~1.05x of the dense FP32 size.
@@ -354,6 +358,10 @@ We love contributions! Whether it's a new operator, documentation, or a bug fix:
 
 Browse the full codebase documentation on [DeepWiki](https://deepwiki.com/SKaiNET-developers/SKaiNET).
 
+### Contributors (0.40.1)
+
+- **Michal Harakal** ([@michalharakal](https://github.com/michalharakal)) — packed-quant `transpose()` block-grid correctness fix, all three kernel tiers (#968, #969)
+
 ### Contributors (0.40.0)
 
 - **Michal Harakal** ([@michalharakal](https://github.com/michalharakal)) — off-heap/mmap tensor storage on Android (#921), GGUF `DEQUANTIZE_TO_FP32` over-allocation fix (#782), native Q5_0/Q5_1 packed matmul kernels (#708), Apple arm64 runtime FEAT_DotProd dispatch (#958), Apple iOS/macOS Kotlin/Native kernel targets (#959)

docs/antora.yml

@@ -15,7 +15,7 @@ asciidoc:
     framework_name: SKaiNET
     # Current SKaiNET release — bump once per release; referenced as
     # {skainet_version} in dependency snippets (blocks need subs="attributes+").
-    skainet_version: 0.40.0
+    skainet_version: 0.40.1
     ksp_version: 2.2.21-2.0.5
     dokka_version: 2.1.0
     asciidoctorj_version: 3.0.0

docs/modules/ROOT/pages/reference/kernel-support-matrix.adoc

@@ -1,7 +1,7 @@
 = Kernel × platform support matrix
 :description: Which compute-kernel provider serves each weight format on each KMP target.
 
-Generated from `kernel-support.json` (version `0.40.0`) by `KernelSupportMatrixTest` — registry introspection of the registered `KernelProvider` implementations. Do not edit by hand; run `./gradlew generateKernelMatrix` to refresh.
+Generated from `kernel-support.json` (version `0.40.1`) by `KernelSupportMatrixTest` — registry introspection of the registered `KernelProvider` implementations. Do not edit by hand; run `./gradlew generateKernelMatrix` to refresh.
 
 Each cell is the best (highest-priority) provider that serves `Float32 × format` `matmul` on that platform: *native-ffm* (100) → *panama-vector* (50) → *scalar* (0). An empty cell (`—`) means no provider carries a kernel there (the format is dequant-to-FP32 only).
 

gradle.properties

@@ -1,5 +1,5 @@
 GROUP=sk.ainet.core
-VERSION_NAME=0.40.0
+VERSION_NAME=0.40.1
 POM_DESCRIPTION=SKaiNET
 
 POM_URL=https://github.com/SKaiNET-developers/skainet/