首页下载资源操作系统弹道导弹的点点滴滴的点点滴滴的点点滴滴的

ZIP弹道导弹的点点滴滴的点点滴滴的点点滴滴的

weixin_4401362643.28MB需要积分:1

资源文件列表:

tvmriscv截图.zip 大约有21个文件
  1. tvmriscv截图/
  2. tvmriscv截图/2024-09-23 10-56-15屏幕截图.png 19.65KB
  3. tvmriscv截图/2024-09-23 10-55-38屏幕截图.png 16.84KB
  4. tvmriscv截图/2024-09-14 08-47-00屏幕截图.png 118.49KB
  5. tvmriscv截图/2024-09-14 11-14-03屏幕截图.png 48.58KB
  6. tvmriscv截图/2024-09-14 09-36-09屏幕截图.png 47.7KB
  7. tvmriscv截图/2024-09-14 15-40-07屏幕截图.png 93.38KB
  8. tvmriscv截图/riscv.png 59.63KB
  9. tvmriscv截图/2024-09-14 08-46-13屏幕截图.png 27.54KB
  10. tvmriscv截图/2024-09-14 08-47-56屏幕截图.png 34.21KB
  11. tvmriscv截图/2024-09-14 08-52-15屏幕截图.png 111.95KB
  12. tvmriscv截图/2024-09-23 11-24-34屏幕截图.png 74.06KB
  13. tvmriscv截图/2024-09-23 11-24-15屏幕截图.png 36.32KB
  14. tvmriscv截图/2024-09-14 08-46-13屏幕截图(复件).png 27.54KB
  15. tvmriscv截图/2024-09-14 15-22-37屏幕截图.png 35.43KB
  16. tvmriscv截图/2024-09-14 15-20-00屏幕截图.png 102.42KB
  17. tvmriscv截图/2024-09-14 15-21-19屏幕截图.png 53.2KB
  18. tvmriscv截图/riscv-tvm.py 2.02KB
  19. tvmriscv截图/cpp_deploy_pack 801.1KB
  20. tvmriscv截图/libtvm_runtime_pack.o 4.11MB
  21. tvmriscv截图/test_resnet18.so 46.47MB

资源介绍:

弹道导弹的点点滴滴的点点滴滴的点点滴滴的
import numpy as np from tvm import te import onnx import numpy as np import tvm import tvm.relay as relay from tvm import rpc from tvm.contrib import utils x = np.zeros((1,3, 640, 640)) from tvm.contrib import graph_executor # target = tvm.target.Target('llvm -mtriple=riscv64-unknown-linux-gnu -mcpu=generic-rv64 -mfloat-abi=hard') #target = "llvm -mtriple=riscv64-unknown-linux-gnu -mcpu=generic-rv64" -mcpu=sifive-u54 -mcpu=sifive-u54 -device=arm_cpu target = 'llvm -mtriple=riscv64-unknown-linux-gnu -mcpu=generic-rv64 -device=arm_cpu' input_name = "input.1" shape_dict = {input_name: x.shape} #导入onnx模型 onnx_model = onnx.load('resnet18.onnx') mod, params = relay.frontend.from_onnx(onnx_model, shape_dict) # 这里利用TVM构建出优化后模型的信息 # with relay.build_config(opt_level=3): # lib = relay.build_module.build(sym, target, params=params) with tvm.transform.PassContext(opt_level=3): graph, lib, params = relay.build(mod, target=target, params=params) lib.save("model.ll") #这个relay.build_module.build与relay.build一样; dtype = 'float32' from tvm.contrib import graph_runtime # 下面的函数导出我们需要的动态链接库,地址可自定义 print("Output model files") libpath = "model1.so" #with tvm.transform.PassContext(opt_level=3): # graph, lib, params = relay.build(mod, target=target, params=params) #lib.save("model.ll") from tvm.contrib import cc ##################### Test code start ############################## # import subprocess # def c_compile(file_name, objects, **kwargs): # cmd = "/usr/bin/riscv64-linux-gnu-gcc" # options = ["-march=rv64g", "-mabi=lp64", "-o", file_name] + objects # subprocess.check_call(cmd + " " + " ".join(options), shell=True) # lib.export_library(libpath, fcompile=c_compile) ##################### Test code end ############################## lib.export_library(libpath, cc="/usr/bin/riscv64-linux-gnu-gcc") #lib.export_library("model.so", cc="riscv64-unknown-linux-gnu-g++", options=["-march=rv64gc", "-mabi=lp64d", "-static-libstdc++"])
100+评论
captcha