gitadmin/foxhunt
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactor(cuda): eliminate all CPU fallbacks — CUDA mandatory across ML stack

Browse Source 
- Remove ALL #[cfg(feature = "cuda")] guards (~400+ occurrences)
- Remove ALL #[cfg_attr(not(feature = "cuda"), ignore)] test annotations (~250)
- Make cuda default feature in 9 ML crates (ml, ml-core, ml-dqn, ml-ppo, etc.)
- Convert nvrtc JIT compilation to precompiled nvcc (searchsorted, prefix_sum)
- Move compile_ptx_for_device() to ml-core for shared access
- Delete dead CPU code: multi_step.rs, self_supervised_pretraining.rs,
  training_guard_gpu_tests.rs, CPU PER buffer paths, CPU Q-diagnostics
- Replace unwrap_or(Device::Cpu) with hard errors everywhere
- Remove dead is_cuda() else branches in DQN/PPO/hyperopt trainers
- Change config defaults from "cpu" to "cuda" (rainbow, tlob, pipeline)
- Port IQL value network to GPU kernel (5 CUDA entry points)
- Port HER goal relabeling to GPU kernel (warp-per-sample)
- Wire DSR GPU-to-CPU sync in training loop
- cfg!(feature = "cuda") → true in inference_validator

Zero warnings, zero errors across entire workspace.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
jgrusewski
2026-03-16 21:01:28 +01:00
parent 979f135271
commit 450c23a6d0
221 changed files with 3879 additions and 17772 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
testing/integration/gpu/cuda_initialization_test.rs
View File

@@ -71,32 +71,29 @@ mod tests {
info!("📊 Testing CUDA device properties");
#[cfg(feature = "cuda")]
{
match cudarc::driver::CudaDevice::new(0) {
Ok(device) => {
info!("✅ CUDA device initialized successfully");
// Test device properties
let device_name = device.name().unwrap_or("Unknown".to_string());
let total_memory = device.total_memory().unwrap_or(0);
info!("Device name: {}", device_name);
info!("Total memory: {} bytes ({:.2} GB)",
total_memory, total_memory as f64 / (1024.0 * 1024.0 * 1024.0));
// Basic validation
assert!(!device_name.is_empty(), "Device should have a name");
assert!(total_memory > 0, "Device should have memory");
info!("✅ CUDA device properties validation passed");
}
Err(e) => {
warn!("⚠️ CUDA device initialization failed: {}", e);
}
match cudarc::driver::CudaDevice::new(0) {
Ok(device) => {
info!("✅ CUDA device initialized successfully");
// Test device properties
let device_name = device.name().unwrap_or("Unknown".to_string());
let total_memory = device.total_memory().unwrap_or(0);
info!("Device name: {}", device_name);
info!("Total memory: {} bytes ({:.2} GB)",
total_memory, total_memory as f64 / (1024.0 * 1024.0 * 1024.0));
// Basic validation
assert!(!device_name.is_empty(), "Device should have a name");
assert!(total_memory > 0, "Device should have memory");
info!("✅ CUDA device properties validation passed");
}
Err(e) => {
warn!("⚠️ CUDA device initialization failed: {}", e);
}
}
}
#[test]

5
testing/integration/gpu/cuda_kernel_test.rs
View File

@@ -5,14 +5,12 @@
use super::utils::*;
use log::{info, warn};
#[cfg(feature = "cuda")]
use cudarc::driver::{CudaDevice, DevicePtr, LaunchAsync, LaunchConfig};
#[cfg(test)]
mod tests {
use super::*;
#[cfg(feature = "cuda")]
#[test]
fn test_cuda_kernel_manager_creation() {
init_gpu_test_env();
@@ -49,7 +47,6 @@ mod tests {
}
}
#[cfg(feature = "cuda")]
#[test]
fn test_cuda_memory_operations() {
init_gpu_test_env();
@@ -121,7 +118,6 @@ mod tests {
}
}
#[cfg(feature = "cuda")]
#[test]
fn test_cuda_kernel_compilation() {
init_gpu_test_env();
@@ -270,7 +266,6 @@ mod tests {
}
}
#[cfg(feature = "cuda")]
#[test]
fn test_cuda_stream_operations() {
init_gpu_test_env();

1
testing/integration/gpu/gpu_memory_management_test.rs
View File

@@ -219,7 +219,6 @@ mod tests {
info!("✅ GPU memory pool behavior test completed");
}
#[cfg(feature = "cuda")]
#[test]
fn test_cuda_memory_info() {
init_gpu_test_env();

55
testing/integration/gpu/production_gpu_integration_test.rs
View File

@@ -368,38 +368,35 @@ mod tests {
let device = get_test_device();
// Test GPU monitoring
#[cfg(feature = "cuda")]
{
match cudarc::driver::CudaDevice::new(0) {
Ok(cuda_device) => {
if let Ok(total_memory) = cuda_device.total_memory() {
info!("GPU monitoring data:");
info!(" Total GPU memory: {} bytes ({:.2} GB)",
total_memory, total_memory as f64 / (1024.0 * 1024.0 * 1024.0));
// Test memory usage monitoring during allocation
let start_time = Instant::now();
let tensor_result = Tensor::zeros((2000, 2000), DType::F32, &device);
let allocation_time = start_time.elapsed();
match tensor_result {
Ok(_tensor) => {
let estimated_usage = 2000 * 2000 * 4; // 4 bytes per f32
info!(" Allocation time: {:?}", allocation_time);
info!(" Estimated memory usage: {} bytes ({:.2} MB)",
estimated_usage, estimated_usage as f64 / (1024.0 * 1024.0));
info!(" ✅ GPU monitoring data collected successfully");
}
Err(e) => {
warn!(" ❌ GPU allocation failed during monitoring: {}", e);
}
match cudarc::driver::CudaDevice::new(0) {
Ok(cuda_device) => {
if let Ok(total_memory) = cuda_device.total_memory() {
info!("GPU monitoring data:");
info!(" Total GPU memory: {} bytes ({:.2} GB)",
total_memory, total_memory as f64 / (1024.0 * 1024.0 * 1024.0));
// Test memory usage monitoring during allocation
let start_time = Instant::now();
let tensor_result = Tensor::zeros((2000, 2000), DType::F32, &device);
let allocation_time = start_time.elapsed();
match tensor_result {
Ok(_tensor) => {
let estimated_usage = 2000 * 2000 * 4; // 4 bytes per f32
info!(" Allocation time: {:?}", allocation_time);
info!(" Estimated memory usage: {} bytes ({:.2} MB)",
estimated_usage, estimated_usage as f64 / (1024.0 * 1024.0));
info!(" ✅ GPU monitoring data collected successfully");
}
Err(e) => {
warn!(" ❌ GPU allocation failed during monitoring: {}", e);
}
}
}
Err(e) => {
warn!("❌ Failed to initialize CUDA device for monitoring: {}", e);
}
}
Err(e) => {
warn!("❌ Failed to initialize CUDA device for monitoring: {}", e);
}
}