Pycuda中gpuarray.to_gpu能向共享GPU内存中申请空间吗

Question

GPU信息

---

运行如下代码 比较两个矩阵，取最大值为一个新矩阵

import pycuda.driver as drv
import pycuda.autoinit
import numpy
import numpy.linalg as la
from pycuda import driver, compiler, gpuarray, tools
from pycuda.compiler import SourceModule
import os

if (os.system("cl.exe")):
    os.environ['PATH'] += ';'+r"C:\\Program Files (x86)\\Microsoft Visual Studio 14.0\\VC\\bin\\amd64\\"
if (os.system("cl.exe")):
    raise RuntimeError("cl.exe still not found, path probably incorrect")


a = numpy.random.randint(0,3,(23600,26919)).astype(numpy.float32)
b = numpy.random.randint(1,4,(23600,26919)).astype(numpy.float32)

a_gpu = gpuarray.to_gpu(a)
b_gpu = gpuarray.to_gpu(b)

result_array_GPU = gpuarray.maximum(a_gpu,b_gpu)

print(a_gpu.get())
print(b_gpu.get())
print(result_array_GPU,type(result_array_GPU.get()))

---

出现错误 out of memory

---

查看GPU性能，发现_**共享GPU内存**_还有空间，请问能利用这一块内存吗？如何利用这一块的内存呢

---

网上看说可以去BOIS总结去设置大小，也有的说当专有GPU内存不够会向共享自动申请
具体我还要怎么操作呢，还请不吝赐教

Answer 1

你这个GPU不行，内存太小了，建议使用 RTX2080Super