Remove device "gpu number" specifications from model benchmarks. (ire…

…e-org#18315)

We'll be configuring which GPU is visible at the runner level, not the
test source code level. This way, we'll be able to have multiple runners
on the same physical machine (with multiple GPUs).

(As discussed previously here:
iree-org#17842 (comment))

ci-exactly: build_packages, regression_test

.github/workflows/pkgci_regression_test.yml

@@ -342,7 +342,6 @@ jobs:
             --goldensize-rocm-unet-bytes 2280000  \
             --goldensize-rocm-clip-bytes 860000 \
             --goldensize-rocm-vae-bytes 840000 \
-            --gpu-number 6 \
             --rocm-chip gfx90a \
             --log-cli-level=info \
             --retries 7
@@ -364,7 +363,6 @@ jobs:
             --goldensize-rocm-unet-bytes 2270000 \
             --goldensize-rocm-clip-bytes 860000  \
             --goldensize-rocm-vae-bytes 840000 \
-            --gpu-number 0 \
             --rocm-chip gfx942 \
             --log-cli-level=info \
             --retries 7

experimental/benchmarks/sdxl/benchmark_sdxl_rocm.py

@@ -46,7 +46,7 @@ def run_iree_command(args: Sequence[str] = ()):
     return 1, proc.stdout
 
 
-def run_sdxl_rocm_benchmark(rocm_chip, gpu_number):
+def run_sdxl_rocm_benchmark(rocm_chip):
     exec_args = [
         "iree-compile",
         f"{benchmark_dir}/sdxl_pipeline_bench_f16.mlir",
@@ -67,7 +67,7 @@ def run_sdxl_rocm_benchmark(rocm_chip, gpu_number):
         return 1, stdout
     exec_args = [
         "iree-benchmark-module",
-        f"--device=hip://{gpu_number}",
+        f"--device=hip",
         "--device_allocator=caching",
         f"--module={prompt_encoder_dir}/model.rocm_{rocm_chip}.vmfb",
         f"--parameters=model={prompt_encoder_dir}/real_weights.irpa",
@@ -90,10 +90,10 @@ def run_sdxl_rocm_benchmark(rocm_chip, gpu_number):
     return run_iree_command(exec_args)
 
 
-def run_sdxl_unet_rocm_benchmark(gpu_number, rocm_chip):
+def run_sdxl_unet_rocm_benchmark(rocm_chip):
     exec_args = [
         "iree-benchmark-module",
-        f"--device=hip://{gpu_number}",
+        f"--device=hip",
         "--device_allocator=caching",
         f"--module={scheduled_unet_dir}/model.rocm_{rocm_chip}.vmfb",
         f"--parameters=model={scheduled_unet_dir}/real_weights.irpa",
@@ -111,10 +111,10 @@ def run_sdxl_unet_rocm_benchmark(gpu_number, rocm_chip):
     return run_iree_command(exec_args)
 
 
-def run_sdxl_prompt_encoder_rocm_benchmark(gpu_number, rocm_chip):
+def run_sdxl_prompt_encoder_rocm_benchmark(rocm_chip):
     exec_args = [
         "iree-benchmark-module",
-        f"--device=hip://{gpu_number}",
+        f"--device=hip",
         "--device_allocator=caching",
         f"--module={prompt_encoder_dir}/model.rocm_{rocm_chip}.vmfb",
         f"--parameters=model={prompt_encoder_dir}/real_weights.irpa",
@@ -130,10 +130,10 @@ def run_sdxl_prompt_encoder_rocm_benchmark(gpu_number, rocm_chip):
     return run_iree_command(exec_args)
 
 
-def run_sdxl_vae_decode_rocm_benchmark(gpu_number, rocm_chip):
+def run_sdxl_vae_decode_rocm_benchmark(rocm_chip):
     exec_args = [
         "iree-benchmark-module",
-        f"--device=hip://{gpu_number}",
+        f"--device=hip",
         "--device_allocator=caching",
         f"--module={vae_decode_dir}/model.rocm_{rocm_chip}.vmfb",
         f"--parameters=model={vae_decode_dir}/real_weights.irpa",
@@ -194,7 +194,6 @@ def test_sdxl_rocm_benchmark(
     goldentime_rocm_unet,
     goldentime_rocm_clip,
     goldentime_rocm_vae,
-    gpu_number,
     rocm_chip,
     goldendispatch_rocm_unet,
     goldendispatch_rocm_clip,
@@ -204,7 +203,7 @@ def test_sdxl_rocm_benchmark(
     goldensize_rocm_vae,
 ):
     # e2e benchmark
-    ret_value, output = run_sdxl_rocm_benchmark(rocm_chip, gpu_number)
+    ret_value, output = run_sdxl_rocm_benchmark(rocm_chip)
     benchmark_e2e_mean_time = job_summary_process(ret_value, output)
     mean_line = (
         f"E2E Benchmark Time: {str(benchmark_e2e_mean_time)} ms"
@@ -213,7 +212,7 @@ def test_sdxl_rocm_benchmark(
     logging.getLogger().info(mean_line)
 
     # unet benchmark
-    ret_value, output = run_sdxl_unet_rocm_benchmark(gpu_number, rocm_chip)
+    ret_value, output = run_sdxl_unet_rocm_benchmark(rocm_chip)
     benchmark_unet_mean_time = job_summary_process(ret_value, output)
     mean_line = (
         f"Scheduled Unet Benchmark Time: {str(benchmark_unet_mean_time)} ms"
@@ -242,7 +241,7 @@ def test_sdxl_rocm_benchmark(
     logging.getLogger().info(compilation_line)
 
     # prompt encoder benchmark
-    ret_value, output = run_sdxl_prompt_encoder_rocm_benchmark(gpu_number, rocm_chip)
+    ret_value, output = run_sdxl_prompt_encoder_rocm_benchmark(rocm_chip)
     benchmark_clip_mean_time = job_summary_process(ret_value, output)
     mean_line = (
         f"Prompt Encoder Benchmark Time: {str(benchmark_clip_mean_time)} ms"
@@ -271,7 +270,7 @@ def test_sdxl_rocm_benchmark(
     logging.getLogger().info(compilation_line)
 
     # vae decode benchmark
-    ret_value, output = run_sdxl_vae_decode_rocm_benchmark(gpu_number, rocm_chip)
+    ret_value, output = run_sdxl_vae_decode_rocm_benchmark(rocm_chip)
     benchmark_vae_mean_time = job_summary_process(ret_value, output)
     mean_line = (
         f"VAE Decode Benchmark Time: {str(benchmark_vae_mean_time)} ms"

experimental/benchmarks/sdxl/conftest.py

@@ -68,13 +68,6 @@ def pytest_addoption(parser):
         type=int,
         help="Golden vmfb size to test benchmark",
     )
-    parser.addoption(
-        "--gpu-number",
-        action="store",
-        default=0,
-        type=int,
-        help="IREE GPU device number to test on",
-    )
     parser.addoption(
         "--rocm-chip",
         action="store",
@@ -137,8 +130,3 @@ def goldensize_rocm_vae(request):
 @pytest.fixture
 def rocm_chip(request):
     return request.config.getoption("--rocm-chip")
-
-
-@pytest.fixture
-def gpu_number(request):
-    return request.config.getoption("--gpu-number")