Merge remote-tracking branch 'origin/main' into cuda_update-gpu_test

Author: Alexey-Rivkin

.ci/jenkins/pipeline/proj-jjb.yaml

@@ -269,7 +269,7 @@
       - string:
           name: "NIXL_VERSION"
           default: "{jjb_branch}"
-          description: "NIXL version to use (tag like 0.7.1, branch name, or commit hash)"
+          description: "NIXL version to use (tag like 0.8.0, branch name, or commit hash)"
       - string:
           name: "UCX_VERSION"
           default: "v1.20.x"

Cargo.lock

@@ -20,7 +20,7 @@ members = [
 resolver = "3"
 
 [workspace.package]
-version = "0.7.1"
+version = "0.8.0"
 edition = "2021"
 description = "Low-level bindings to NIXL - NVIDIA Inference Xfer Library"
 authors = ["NIXL Developers <[email protected]>"]

Cargo.toml

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-project('nixlbench', 'CPP', version: '0.7.1',
+project('nixlbench', 'CPP', version: '0.8.0',
     default_options: ['buildtype=release',
                 'werror=true',
                 'cpp_std=c++17',

benchmark/nixlbench/meson.build

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-project('nixl', 'CPP', version: '0.7.1',
+project('nixl', 'CPP', version: '0.8.0',
     default_options: ['buildtype=release',
                 'werror=true',
                 'cpp_std=c++17',

examples/rust/Cargo.lock

@@ -19,7 +19,7 @@ build-backend = "mesonpy"
 
 [project]
 name = 'nixl-cu12'
-version = '0.7.1'
+version = '0.8.0'
 description = 'NIXL Python API'
 readme = 'README.md'
 license = {file = 'LICENSE'}

meson.build

@@ -1,4 +1,4 @@
-#NIXL Libfabric Plugin
+# NIXL Libfabric Plugin
 
 This plugin provides a high-performance RDMA backend for NIXL using the OpenFabrics Interfaces (OFI) Libfabric library.
 
@@ -7,43 +7,44 @@ This plugin provides a high-performance RDMA backend for NIXL using the OpenFabr
 The Libfabric plugin provides a high-performance RDMA communication backend with the following key capabilities:
 
 - **Multi-Rail RDMA**: Automatic discovery and utilization of multiple network devices for increased bandwidth
-- **GPU Direct Support**: Zero-copy transfers between GPU memory (VRAM) and remote systems with CUDA integration. And GDR support is currently mandated
+- **GPU Direct Support**: Zero-copy transfers between GPU memory (VRAM) and remote systems with CUDA integration. GDR (GPU Direct RDMA) support is currently required.
 - **Scalable Connection Management**: Efficient multi-agent connectivity with robust state tracking and automatic reconnection
 - **Asynchronous Processing**: Non-blocking RDMA operations with pre-allocated request pools and completion processing
 - **Thread-Safe Concurrency**: Background progress threads with lock-free data structures and configurable threading patterns
-
-EFA Specific **Topology-Aware Optimization**: Hardware-aware GPU-to-EFA and NUMA-to-EFA mapping using hwloc for optimal performance
+- **Topology-Aware Optimization**: Hardware-aware GPU-to-EFA and NUMA-to-EFA mapping using hwloc for optimal performance (EFA-specific)
 
 ## Dependencies
 
 ### Required Dependencies
 
 - **Libfabric**
-  - Many system will have installed libfabric already. If not, custom libfabric installation is available via https://ofiwg.github.io/libfabric/ - Minimum required version: v1.21.0
-  - For EFA enabled AWS instances, it is recommanded to install through AWS EFA installer: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/efa-start.html - Recommend to use the latest version
+  - Many systems will have libfabric already installed. If not, custom libfabric installation is available via https://ofiwg.github.io/libfabric/ - Minimum required version: `v1.21.0`
+  - For EFA enabled AWS instances, it is recommended to install through AWS EFA installer: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/efa-start.html - Recommend to use the latest version
 
 - **hwloc**
   - hwloc is used to understand the underlying architecture to optimize application performance. Suggested version: 2.10.0 or newer
 
 ### Network Hardware Requirements
 
-Validated compatiblity with:
+Validated compatibility with:
+
 - **AWS EFA** (Elastic Fabric Adapter)
 
 Any other Libfabric providers should also work but have not been validated in production environments. Community validation and feedback are highly appreciated!
 
 ## Build Instructions
 
 ```bash
-#Basic build setup with default options
+# Basic build setup with default options
 $ meson setup <name_of_build_dir>
 
-#Setup with custom options(example)
+# Setup with custom options (example)
 $ meson setup <name_of_build_dir> \
     -Dlibfabric_path=/path/to/libfabric
 
-#Build and install
-ninja && ninja install
+# Build and install
+$ cd build
+$ ninja && ninja install
 ```
 
 ## API Reference
@@ -62,23 +63,25 @@ ninja && ninja install
 ### Debug Information
 
 Enable debug logging by setting environment variables:
+
 ```bash
-#Libfabric debug logging
+# Libfabric debug logging
 export FI_LOG_LEVEL=debug
 export FI_LOG_PROV=efa  # or verbs, tcp, etc.
 
-#NIXL debug logging
+# NIXL debug logging
 export NIXL_LOG_LEVEL=debug
 ```
 
 ### Common Issues
 
 **No network devices detected:**
+
 ```bash
-#Check available fabric interfaces
+# Check available fabric interfaces
 fi_info -l
 
-#For checking specific devices(e.g.EFA as an example)
+# For checking specific devices (e.g. EFA as an example)
 fi_info -p efa
 ```