Add PCIe transfer benchmark (H2D/D2H) with Smart Chunking

Author: linamy85

Ironwood/configs/host_device/host_device_smart_chunking_single_chip.yaml

@@ -0,0 +1,11 @@
+benchmarks:
+- benchmark_name: host_device_smart_chunking
+  num_runs: 20
+  benchmark_sweep_params:
+  # Single Chip (1 Chip, 2 Devices)
+  - {
+      num_devices: 2,
+      data_size_mb_list: [1, 16, 128, 256, 512, 1024, 2048, 4096, 8192, 16384, 32768]
+    }
+  csv_path: "../microbenchmarks/host_device/single_chip_smart"
+  trace_dir: "../microbenchmarks/host_device/single_chip_smart/trace"

Ironwood/guides/host_device/host_device.md

@@ -2,7 +2,7 @@
 
 This guide provides instructions for running Host Device (Host-to-Device and Device-to-Host) microbenchmarks on tpu7x-2x2x1 Google Kubernetes Engine (GKE) clusters. It covers creating a node pool, running the benchmarks, and viewing the output.
 
-> [!WARNING]
+> [!NOTE]
 > This benchmark is currently a Work In Progress (WIP). Expected bandwidth numbers are not yet finalized.
 
 ## Create Node Pools

Ironwood/scripts/run_host_device_benchmark.sh

@@ -27,13 +27,8 @@ while [[ "$#" -gt 0 ]]; do
 done
 
 echo "--- Starting Host-Device Transfer Benchmark (H2D/D2H) ---"
-echo "********************************************************"
-echo "WARNING: This benchmark is currently a WORK IN PROGRESS"
-echo "********************************************************"
-echo ""
-echo "Configuration:"
-echo "    Interleaved: $INTERLEAVED"
-echo ""
+echo "Note: This benchmark is work in progress"
+echo "Interleaved: $INTERLEAVED"
 
 if [ -n "$SPECIFIC_CONFIG" ]; then
     CONFIGS=("$SPECIFIC_CONFIG")