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.

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

pyxrt not compatible with numpy v2.0.0 #8258

Open
andrew-bolin opened this issue Jun 28, 2024 · 0 comments
Open

pyxrt not compatible with numpy v2.0.0 #8258

andrew-bolin opened this issue Jun 28, 2024 · 0 comments

Comments

@andrew-bolin
Copy link

andrew-bolin commented Jun 28, 2024

Using pyxrt with numpy v2.0.0 results in all bytes being read as 1.

Minimal reproduction script below.
Uses xclbin file available here: https://gitlab.com/ska-telescope/low-cbf/ska-low-cbf-fw-cnic/-/packages/26156556
but I think any xclbin file or different Alveo card is likely to produce the same result.

import pyxrt
device = pyxrt.device("0000:d2:00.1")
xclbin = pyxrt.xclbin("./cnic.xclbin")
uuid = device.load_xclbin(xclbin)
kernel_names = [kernel.get_name() for kernel in xclbin.get_kernels()]
cu_name = kernel_names[0]
kernel = pyxrt.kernel(device, uuid, cu_name, pyxrt.kernel.cu_access_mode(0))

mem_group_ids = []
num_args = 0
for n in range(100):
    try:
        gid = kernel.group_id(n)
        num_args += 1
        if gid not in (-1, 0xFFFF):
            mem_group_ids.append(gid)
            print(f"Kernel group: {n}, memory bank {gid}")
    except IndexError as e:
        print(f"_load_firmware; n:{n}, {e}")
        break

buf = pyxrt.bo(device, 128<<10, pyxrt.bo.flags.normal, mem_group_ids.pop(0))

task = kernel(0, 0x8000*4, buf, 64)

task.wait()
buf.sync(pyxrt.xclBOSyncDirection.XCL_BO_SYNC_BO_FROM_DEVICE, 64, 0)
result = buf.read(64, 0)
print(result)


# Wrong result w/ numpy v2.0.0
# array([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
#       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
#       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
#      dtype=int8)


# Correct result w/ numpy v1.23.0
# array([  1, 112, 106,  -7,  18,  25,   9,  35,   1,   0,   0,   0,   0,
#         0,   0,   0, 102, 102, 102, 102,   0,   0,   0,   0,   1,   0,
#         0,   0,   8,   0,   0,   0,   0,   0,   0,   0,  67,  73,  78,
#        67,  18,  39, -49,  99,  78,  73,  65,  77,  64,   2,   0,   0,
#         1, 117, -57,  57,  62,   2,   0,   0,   0,   0,   0,   0],
#      dtype=int8)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant