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

Jump to bottom

feat: User Selectable GPUs and GPU-based Model Recommendations #1730

Merged
merged 9 commits into from
Feb 6, 2024
Merged

feat: User Selectable GPUs and GPU-based Model Recommendations #1730

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
9 commits
Select commit Hold shift click to select a range
f07f5e5
Add a field to settings.json to help manage selected GPUs
hientominh Jan 19, 2024
8625e8c
Setting get list GPU from file
hientominh Jan 23, 2024
a8a9e6b
Fix eslint
hientominh Jan 23, 2024
82bb48e
Add gpu monitor to bottom bar
hientominh Jan 24, 2024
27363de
Setting get list GPU from file
hientominh Jan 23, 2024
156055f
Add recommend models in hub base on GPU vRam in GPU mode
hientominh Jan 25, 2024
b4dc2ce
Increase interval time to get System Resource from 0.5s to 2s
hientominh Jan 25, 2024
6d61ba9
Merge dev to feature/user-select-gpus
hientominh Feb 6, 2024
3f0b62c
Fix eslint and correct setUseGpuIfPossible function call
hientominh Feb 6, 2024
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
Next Next commit
Add a field to settings.json to help manage selected GPUs
  • Loading branch information
@hientominh @hiro-v
hientominh authored and hiro-v committed Feb 5, 2024
commit f07f5e5b56016f2cb3094de6910019ce747d5deb
12 changes: 6 additions & 6 deletions extensions/inference-nitro-extension/src/node/execute.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,11 26,11 @@ export const executableNitroFile = (): NitroExecutableOptions => {
binaryFolder = path.join(binaryFolder, "win-cpu");
} else {
if (nvidiaInfo["cuda"].version === "12") {
binaryFolder = path.join(binaryFolder, "win-cuda-12-0");
} else {
binaryFolder = path.join(binaryFolder, "win-cuda-11-7");
} else {
binaryFolder = path.join(binaryFolder, "win-cuda-12-0");
}
cudaVisibleDevices = nvidiaInfo["gpu_highest_vram"];
cudaVisibleDevices = nvidiaInfo["gpus_in_use"];
}
binaryName = "nitro.exe";
} else if (process.platform === "darwin") {
Expand All @@ -51,11 51,11 @@ export const executableNitroFile = (): NitroExecutableOptions => {
binaryFolder = path.join(binaryFolder, "linux-cpu");
} else {
if (nvidiaInfo["cuda"].version === "12") {
binaryFolder = path.join(binaryFolder, "linux-cuda-12-0");
} else {
binaryFolder = path.join(binaryFolder, "linux-cuda-11-7");
} else {
binaryFolder = path.join(binaryFolder, "linux-cuda-12-0");
}
cudaVisibleDevices = nvidiaInfo["gpu_highest_vram"];
cudaVisibleDevices = nvidiaInfo["gpus_in_use"];
}
}
return {
Expand Down
57 changes: 27 additions & 30 deletions extensions/inference-nitro-extension/src/node/nvidia.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 19,8 @@ const DEFALT_SETTINGS = {
},
gpus: [],
gpu_highest_vram: "",
gpus_in_use: "",
is_initial: true,
};

/**
Expand Down Expand Up @@ -48,11 50,15 @@ export interface NitroProcessInfo {
*/
export async function updateNvidiaInfo() {
if (process.platform !== "darwin") {
await Promise.all([
updateNvidiaDriverInfo(),
updateCudaExistence(),
updateGpuInfo(),
]);
let data;
try {
data = JSON.parse(readFileSync(NVIDIA_INFO_FILE, "utf-8"));
} catch (error) {
data = DEFALT_SETTINGS;
writeFileSync(NVIDIA_INFO_FILE, JSON.stringify(data, null, 2));
}
updateNvidiaDriverInfo();
updateGpuInfo();
}
}

Expand All @@ -73,12 79,7 @@ export async function updateNvidiaDriverInfo(): Promise<void> {
exec(
"nvidia-smi --query-gpu=driver_version --format=csv,noheader",
(error, stdout) => {
let data;
try {
data = JSON.parse(readFileSync(NVIDIA_INFO_FILE, "utf-8"));
} catch (error) {
data = DEFALT_SETTINGS;
}
let data = JSON.parse(readFileSync(NVIDIA_INFO_FILE, "utf-8"));

if (!error) {
const firstLine = stdout.split("\n")[0].trim();
Expand Down Expand Up @@ -107,7 108,7 @@ export function checkFileExistenceInPaths(
/**
* Validate cuda for linux and windows
*/
export function updateCudaExistence() {
export function updateCudaExistence(data: Record<string, any> = DEFALT_SETTINGS): Record<string, any> {
let filesCuda12: string[];
let filesCuda11: string[];
let paths: string[];
Expand Down Expand Up @@ -141,19 142,14 @@ export function updateCudaExistence() {
cudaVersion = "12";
}

let data;
try {
data = JSON.parse(readFileSync(NVIDIA_INFO_FILE, "utf-8"));
} catch (error) {
data = DEFALT_SETTINGS;
}

data["cuda"].exist = cudaExists;
data["cuda"].version = cudaVersion;
if (cudaExists) {
console.log(data["is_initial"], data["gpus_in_use"]);
if (cudaExists && data["is_initial"] && data["gpus_in_use"] !== "") {
data.run_mode = "gpu";
}
writeFileSync(NVIDIA_INFO_FILE, JSON.stringify(data, null, 2));
data.is_initial = false;
return data;
}

/**
Expand All @@ -163,12 159,7 @@ export async function updateGpuInfo(): Promise<void> {
exec(
"nvidia-smi --query-gpu=index,memory.total --format=csv,noheader,nounits",
(error, stdout) => {
let data;
try {
data = JSON.parse(readFileSync(NVIDIA_INFO_FILE, "utf-8"));
} catch (error) {
data = DEFALT_SETTINGS;
}
let data = JSON.parse(readFileSync(NVIDIA_INFO_FILE, "utf-8"));

if (!error) {
// Get GPU info and gpu has higher memory first
Expand All @@ -187,12 178,18 @@ export async function updateGpuInfo(): Promise<void> {
return { id, vram };
});

data["gpus"] = gpus;
data["gpu_highest_vram"] = highestVramId;
data.gpus = gpus;
data.gpu_highest_vram = highestVramId;
} else {
data["gpus"] = [];
data.gpus = [];
data.gpu_highest_vram = "";
}

if (!data["gpus_in_use"] || data["gpus_in_use"] === "") {
data.gpus_in_use = data["gpu_highest_vram"];
}

data = updateCudaExistence(data);
writeFileSync(NVIDIA_INFO_FILE, JSON.stringify(data, null, 2));
Promise.resolve();
}
Expand Down