Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

scx_utils: cleanup gpu topology part #1211

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

scx_utils: cleanup gpu topology part #1211

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
79 changes: 38 additions & 41 deletions rust/scx_utils/src/gpu.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,51 +29,48 @@ pub fn create_gpus() -> BTreeMap<usize, Vec<Gpu>> {
let Ok(nvml) = Nvml::init_with_flags(InitFlags::NO_GPUS) else {
return BTreeMap::new();
};
match nvml.device_count() {
Ok(nvidia_gpu_count) => {
for i in 0..nvidia_gpu_count {
let Ok(nvidia_gpu) = nvml.device_by_index(i) else {
continue;
};
let graphics_boost_clock = nvidia_gpu
.max_customer_boost_clock(Clock::Graphics)
.unwrap_or(0);
let sm_boost_clock = nvidia_gpu.max_customer_boost_clock(Clock::SM).unwrap_or(0);
let Ok(memory_info) = nvidia_gpu.memory_info() else {
continue;
};
let Ok(pci_info) = nvidia_gpu.pci_info() else {
continue;
};
let Ok(index) = nvidia_gpu.index() else {
continue;
};
if let Ok(nvidia_gpu_count) = nvml.device_count() {
for i in 0..nvidia_gpu_count {
let Ok(nvidia_gpu) = nvml.device_by_index(i) else {
continue;
};
let graphics_boost_clock = nvidia_gpu
.max_customer_boost_clock(Clock::Graphics)
.unwrap_or(0);
let sm_boost_clock = nvidia_gpu.max_customer_boost_clock(Clock::SM).unwrap_or(0);
let Ok(memory_info) = nvidia_gpu.memory_info() else {
continue;
};
let Ok(pci_info) = nvidia_gpu.pci_info() else {
continue;
};
let Ok(index) = nvidia_gpu.index() else {
continue;
};

// The NVML library doesn't return a PCIe bus ID compatible with sysfs. It includes
// uppercase bus ID values and an extra four leading 0s.
let bus_id = pci_info.bus_id.to_lowercase();
let fixed_bus_id = bus_id.strip_prefix("0000").unwrap_or("");
let numa_path = format!("/sys/bus/pci/devices/{}/numa_node", fixed_bus_id);
let numa_node = read_file_usize(&Path::new(&numa_path)).unwrap_or(0);
// The NVML library doesn't return a PCIe bus ID compatible with sysfs. It includes
// uppercase bus ID values and an extra four leading 0s.
let bus_id = pci_info.bus_id.to_lowercase();
let fixed_bus_id = bus_id.strip_prefix("0000").unwrap_or("");
let numa_path = format!("/sys/bus/pci/devices/{}/numa_node", fixed_bus_id);
let numa_node = read_file_usize(&Path::new(&numa_path)).unwrap_or(0);

let gpu = Gpu {
index: GpuIndex::Nvidia { nvml_id: index },
node_id: numa_node as usize,
max_graphics_clock: graphics_boost_clock as usize,
max_sm_clock: sm_boost_clock as usize,
memory: memory_info.total,
};
if !gpus.contains_key(&numa_node) {
gpus.insert(numa_node, vec![gpu]);
continue;
}
if let Some(gpus) = gpus.get_mut(&numa_node) {
gpus.push(gpu);
}
let gpu = Gpu {
index: GpuIndex::Nvidia { nvml_id: index },
node_id: numa_node as usize,
max_graphics_clock: graphics_boost_clock as usize,
max_sm_clock: sm_boost_clock as usize,
memory: memory_info.total,
};
if !gpus.contains_key(&numa_node) {
gpus.insert(numa_node, vec![gpu]);
continue;
}
if let Some(gpus) = gpus.get_mut(&numa_node) {
gpus.push(gpu);
}
}
_ => {}
};
}

gpus
}
26 changes: 10 additions & 16 deletions rust/scx_utils/src/topology.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -270,9 +270,9 @@ impl Topology {
#[cfg(feature = "gpu-topology")]
pub fn gpus(&self) -> BTreeMap<GpuIndex, &Gpu> {
let mut gpus = BTreeMap::new();
for (_, node) in &self.nodes {
for node in self.nodes.values() {
for (idx, gpu) in &node.gpus {
gpus.insert(idx.clone(), gpu);
gpus.insert(*idx, gpu);
}
}
gpus
Expand Down Expand Up @@ -582,14 +582,11 @@ fn create_default_node(
#[cfg(feature = "gpu-topology")]
{
let system_gpus = create_gpus();
match system_gpus.get(&0) {
Some(gpus) => {
for gpu in gpus {
node.gpus.insert(gpu.index, gpu.clone());
}
if let Some(gpus) = system_gpus.get(&0) {
for gpu in gpus {
node.gpus.insert(gpu.index, gpu.clone());
}
_ => {}
};
}
}

if !Path::new("/sys/devices/system/cpu").exists() {
Expand Down Expand Up @@ -647,14 +644,11 @@ fn create_numa_nodes(

#[cfg(feature = "gpu-topology")]
{
match system_gpus.get(&node_id) {
Some(gpus) => {
for gpu in gpus {
node.gpus.insert(gpu.index, gpu.clone());
}
if let Some(gpus) = system_gpus.get(&node_id) {
for gpu in gpus {
node.gpus.insert(gpu.index, gpu.clone());
}
_ => {}
};
}
}

let cpu_pattern = numa_path.join("cpu[0-9]*");
Expand Down