rvc/configs/config.py

import torch
import json
import os

version_config_list = [
    "v1/32000.json",
    "v1/40000.json",
    "v1/48000.json",
    "v2/48000.json",
    "v2/32000.json",
]


def singleton_variable(func):
    def wrapper(*args, **kwargs):
        if not wrapper.instance:
            wrapper.instance = func(*args, **kwargs)
        return wrapper.instance

    wrapper.instance = None
    return wrapper


@singleton_variable
class Config:
    def __init__(self):
        self.device = "cuda:0"
        self.is_half = True
        self.use_jit = False
        self.n_cpu = 0
        self.gpu_name = None
        self.json_config = self.load_config_json()
        self.gpu_mem = None
        self.instead = ""
        self.x_pad, self.x_query, self.x_center, self.x_max = self.device_config()

    @staticmethod
    def load_config_json() -> dict:
        d = {}
        for config_file in version_config_list:
            with open(f"rvc/configs/{config_file}", "r") as f:
                d[config_file] = json.load(f)
        return d

    @staticmethod
    def has_mps() -> bool:
        if not torch.backends.mps.is_available():
            return False
        try:
            torch.zeros(1).to(torch.device("mps"))
            return True
        except Exception:
            return False

    @staticmethod
    def has_xpu() -> bool:
        if hasattr(torch, "xpu") and torch.xpu.is_available():
            return True
        else:
            return False

    def use_fp32_config(self):
        print(
            f"Using FP32 config instead of FP16 due to GPU compatibility ({self.gpu_name})"
        )
        for config_file in version_config_list:
            self.json_config[config_file]["train"]["fp16_run"] = False
            with open(f"rvc/configs/{config_file}", "r") as f:
                strr = f.read().replace("true", "false")
            with open(f"rvc/configs/{config_file}", "w") as f:
                f.write(strr)
        with open("rvc/train/preprocess/preprocess.py", "r") as f:
            strr = f.read().replace("3.7", "3.0")
        with open("rvc/train/preprocess/preprocess.py", "w") as f:
            f.write(strr)

    def device_config(self) -> tuple:
        if torch.cuda.is_available():
            if self.has_xpu():
                self.device = self.instead = "xpu:0"
                self.is_half = True
            i_device = int(self.device.split(":")[-1])
            self.gpu_name = torch.cuda.get_device_name(i_device)
            if (
                ("16" in self.gpu_name and "V100" not in self.gpu_name.upper())
                or "P40" in self.gpu_name.upper()
                or "P10" in self.gpu_name.upper()
                or "1060" in self.gpu_name
                or "1070" in self.gpu_name
                or "1080" in self.gpu_name
            ):
                self.is_half = False
                self.use_fp32_config()
            self.gpu_mem = int(
                torch.cuda.get_device_properties(i_device).total_memory
                / 1024
                / 1024
                / 1024
                + 0.4
            )
            if self.gpu_mem <= 4:
                with open("rvc/train/preprocess/preprocess.py", "r") as f:
                    strr = f.read().replace("3.7", "3.0")
                with open("rvc/train/preprocess/preprocess.py", "w") as f:
                    f.write(strr)
        elif self.has_mps():
            print("No supported Nvidia GPU found")
            self.device = self.instead = "mps"
            self.is_half = False
            self.use_fp32_config()
        else:
            print("No supported Nvidia GPU found")
            self.device = self.instead = "cpu"
            self.is_half = False
            self.use_fp32_config()

        if self.n_cpu == 0:
            self.n_cpu = os.cpu_count()

        if self.is_half:
            x_pad = 3
            x_query = 10
            x_center = 60
            x_max = 65
        else:
            x_pad = 1
            x_query = 6
            x_center = 38
            x_max = 41

        if self.gpu_mem is not None and self.gpu_mem <= 4:
            x_pad = 1
            x_query = 5
            x_center = 30
            x_max = 32

        return x_pad, x_query, x_center, x_max


def max_vram_gpu(gpu):
    if torch.cuda.is_available():
        gpu_properties = torch.cuda.get_device_properties(gpu)
        total_memory_gb = round(gpu_properties.total_memory / 1024 / 1024 / 1024)
        return total_memory_gb
    else:
        return "0"


def get_gpu_info():
    ngpu = torch.cuda.device_count()
    gpu_infos = []
    if torch.cuda.is_available() or ngpu != 0:
        for i in range(ngpu):
            gpu_name = torch.cuda.get_device_name(i)
            mem = int(
                torch.cuda.get_device_properties(i).total_memory / 1024 / 1024 / 1024
                + 0.4
            )
            gpu_infos.append("%s: %s %s GB" % (i, gpu_name, mem))
    if len(gpu_infos) > 0:
        gpu_info = "\n".join(gpu_infos)
    else:
        gpu_info = "Unfortunately, there is no compatible GPU available to support your training."
    return gpu_info