backend.ai-kernel-binary 26.9.4__tar.gz → 26.9.6__tar.gz

{backend_ai_kernel_binary-26.9.4/backend.ai_kernel_binary.egg-info → backend_ai_kernel_binary-26.9.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: backend.ai-kernel-binary
-Version: 26.9.4
+Version: 26.9.6
 Summary: Backend.AI Kernel Runner Prebuilt Binaries Package
 Home-page: https://github.com/lablup/backend.ai
 Author: Lablup Inc. and contributors

backend_ai_kernel_binary-26.9.6/ai/backend/runner/VERSION

@@ -0,0 +1 @@
+26.9.6

{backend_ai_kernel_binary-26.9.4 → backend_ai_kernel_binary-26.9.6}/ai/backend/runner/all-smi.1

@@ -1,4 +1,4 @@
-.TH ALL-SMI 1 "October 11, 2026" "all-smi 0.27.1" "User Commands"
+.TH ALL-SMI 1 "October 20, 2026" "all-smi 0.28.1" "User Commands"
 .SH NAME
 all-smi \- Command-line utility for monitoring GPU hardware
 .SH SYNOPSIS
@@ -497,6 +497,9 @@ Native support for M1/M2/M3/M4 series with Metal API integration, ANE power moni
 .B NVIDIA Jetson
 Tegra-specific metrics with DLA (Deep Learning Accelerator) support
 .TP
+.B NVIDIA RTX Spark N1X
+Windows on ARM64 SoC with a Blackwell GPU; the NPU is shown as a DLA gauge on the GPU row and exported as all_smi_dla_* metrics
+.TP
 .B Google Cloud TPUs
 Support for TPU v2-v7/Ironwood architectures, HBM memory tracking, and integration with libtpu/JAX workflows
 .TP
@@ -507,7 +510,7 @@ Support for Wormhole and Blackhole architectures with multiple temperature senso
 Support for Trainium and Inferentia devices: one row per NeuronCore, device HBM and PCI topology from neuron-ls, per-core memory from the neuron driver sysfs tree, and best-effort per-core utilization from neuron-monitor while a Neuron runtime is attached
 .TP
 .B Rebellions NPUs
-Support for ATOM, ATOM+, and ATOM Max variants with PCIe Gen5 x16 interfaces. An ATOM Max card has four dies that share one power reading; it is counted once, on the lowest\-numbered die of the card, and the other dies show the card value in parentheses and export no power series
+Support for the ATOM line (ATOM, ATOM+, ATOM Max) and its successor REBEL line (REBEL100) with PCIe Gen5 x16 interfaces. An ATOM Max card has four dies that share one power reading; it is counted once, on the lowest\-numbered die of the card, and the other dies show the card value in parentheses and export no power series
 .TP
 .B Furiosa NPUs
 Support for RNGD architecture with 8-core NPUs and PE utilization tracking

{backend_ai_kernel_binary-26.9.4 → backend_ai_kernel_binary-26.9.6/backend.ai_kernel_binary.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: backend.ai-kernel-binary
-Version: 26.9.4
+Version: 26.9.6
 Summary: Backend.AI Kernel Runner Prebuilt Binaries Package
 Home-page: https://github.com/lablup/backend.ai
 Author: Lablup Inc. and contributors

{backend_ai_kernel_binary-26.9.4 → backend_ai_kernel_binary-26.9.6}/setup.py

@@ -84,7 +84,7 @@ setup(**{
     },
     'python_requires': '>=3.13,<3.14',
     'url': 'https://github.com/lablup/backend.ai',
-    'version': """26.9.4
+    'version': """26.9.6
 """,
     'zip_safe': False,
 })

backend_ai_kernel_binary-26.9.4/ai/backend/runner/VERSION

@@ -1 +0,0 @@
-26.9.4