mirror of
https://anongit.gentoo.org/git/repo/gentoo.git
synced 2025-12-13 13:41:05 +00:00
Closes: https://bugs.gentoo.org/957618 Signed-off-by: Paul Zander <negril.nx+gentoo@gmail.com> Part-of: https://github.com/gentoo/gentoo/pull/43702 Closes: https://github.com/gentoo/gentoo/pull/43702 Signed-off-by: Sam James <sam@gentoo.org>
115 lines
5.1 KiB
Diff
115 lines
5.1 KiB
Diff
From 5a1306ddccb709cc5c8c4bb503125ab49a6a95b3 Mon Sep 17 00:00:00 2001
|
|
From: Paul Zander <negril.nx+gentoo@gmail.com>
|
|
Date: Wed, 6 Aug 2025 14:27:28 +0200
|
|
Subject: [PATCH] cuda 13.0 support
|
|
|
|
Signed-off-by: Paul Zander <negril.nx+gentoo@gmail.com>
|
|
|
|
diff --git a/modules/core/src/cuda_info.cpp b/modules/core/src/cuda_info.cpp
|
|
index 2558ec8..35b76e7 100644
|
|
--- a/modules/core/src/cuda_info.cpp
|
|
+++ b/modules/core/src/cuda_info.cpp
|
|
@@ -424,7 +424,9 @@ int cv::cuda::DeviceInfo::clockRate() const
|
|
#ifndef HAVE_CUDA
|
|
throw_no_cuda();
|
|
#else
|
|
- return deviceProps().get(device_id_)->clockRate;
|
|
+ int i32value;
|
|
+ cudaSafeCall( cudaDeviceGetAttribute(&i32value, cudaDevAttrClockRate, device_id_) );
|
|
+ return i32value;
|
|
#endif
|
|
}
|
|
|
|
@@ -487,7 +489,9 @@ bool cv::cuda::DeviceInfo::kernelExecTimeoutEnabled() const
|
|
#ifndef HAVE_CUDA
|
|
throw_no_cuda();
|
|
#else
|
|
- return deviceProps().get(device_id_)->kernelExecTimeoutEnabled != 0;
|
|
+ int i32value;
|
|
+ cudaSafeCall( cudaDeviceGetAttribute(&i32value, cudaDevAttrKernelExecTimeout, device_id_) );
|
|
+ return (i32value != 0);
|
|
#endif
|
|
}
|
|
|
|
@@ -522,7 +526,9 @@ DeviceInfo::ComputeMode cv::cuda::DeviceInfo::computeMode() const
|
|
ComputeModeExclusiveProcess
|
|
};
|
|
|
|
- return tbl[deviceProps().get(device_id_)->computeMode];
|
|
+ int i32value;
|
|
+ cudaSafeCall( cudaDeviceGetAttribute(&i32value, cudaDevAttrComputeMode, device_id_) );
|
|
+ return tbl[i32value];
|
|
#endif
|
|
}
|
|
|
|
@@ -554,7 +560,10 @@ int cv::cuda::DeviceInfo::maxTexture1DLinear() const
|
|
#ifndef HAVE_CUDA
|
|
throw_no_cuda();
|
|
#else
|
|
- return deviceProps().get(device_id_)->maxTexture1DLinear;
|
|
+ std::size_t i32value;
|
|
+ cudaChannelFormatDesc format;
|
|
+ cudaSafeCall( cudaDeviceGetTexture1DLinearMaxWidth(&i32value, &format, device_id_) );
|
|
+ return i32value;
|
|
#endif
|
|
}
|
|
|
|
@@ -793,7 +802,9 @@ int cv::cuda::DeviceInfo::memoryClockRate() const
|
|
#ifndef HAVE_CUDA
|
|
throw_no_cuda();
|
|
#else
|
|
- return deviceProps().get(device_id_)->memoryClockRate;
|
|
+ int i32value;
|
|
+ cudaSafeCall( cudaDeviceGetAttribute(&i32value, cudaDevAttrClockRate, device_id_) );
|
|
+ return i32value;
|
|
#endif
|
|
}
|
|
|
|
@@ -924,6 +935,15 @@ void cv::cuda::printCudaDeviceInfo(int device)
|
|
cudaDeviceProp prop;
|
|
cudaSafeCall( cudaGetDeviceProperties(&prop, dev) );
|
|
|
|
+ int clockRate;
|
|
+ cudaSafeCall( cudaDeviceGetAttribute(&clockRate, cudaDevAttrClockRate, dev) );
|
|
+
|
|
+ int KernelExecTimeout;
|
|
+ cudaSafeCall( cudaDeviceGetAttribute(&KernelExecTimeout, cudaDevAttrKernelExecTimeout, dev) );
|
|
+
|
|
+ int ComputeMode;
|
|
+ cudaSafeCall( cudaDeviceGetAttribute(&ComputeMode, cudaDevAttrComputeMode, dev) );
|
|
+
|
|
printf("\nDevice %d: \"%s\"\n", dev, prop.name);
|
|
printf(" CUDA Driver Version / Runtime Version %d.%d / %d.%d\n", driverVersion/1000, driverVersion%100, runtimeVersion/1000, runtimeVersion%100);
|
|
printf(" CUDA Capability Major/Minor version number: %d.%d\n", prop.major, prop.minor);
|
|
@@ -933,7 +953,7 @@ void cv::cuda::printCudaDeviceInfo(int device)
|
|
if (cores > 0)
|
|
printf(" (%2d) Multiprocessors x (%2d) CUDA Cores/MP: %d CUDA Cores\n", prop.multiProcessorCount, cores, cores * prop.multiProcessorCount);
|
|
|
|
- printf(" GPU Clock Speed: %.2f GHz\n", prop.clockRate * 1e-6f);
|
|
+ printf(" GPU Clock Speed: %.2f GHz\n", clockRate * 1e-6f);
|
|
|
|
printf(" Max Texture Dimension Size (x,y,z) 1D=(%d), 2D=(%d,%d), 3D=(%d,%d,%d)\n",
|
|
prop.maxTexture1D, prop.maxTexture2D[0], prop.maxTexture2D[1],
|
|
@@ -952,8 +972,8 @@ void cv::cuda::printCudaDeviceInfo(int device)
|
|
printf(" Maximum memory pitch: %u bytes\n", (int)prop.memPitch);
|
|
printf(" Texture alignment: %u bytes\n", (int)prop.textureAlignment);
|
|
|
|
- printf(" Concurrent copy and execution: %s with %d copy engine(s)\n", (prop.deviceOverlap ? "Yes" : "No"), prop.asyncEngineCount);
|
|
- printf(" Run time limit on kernels: %s\n", prop.kernelExecTimeoutEnabled ? "Yes" : "No");
|
|
+ printf(" Concurrent copy and execution: %s with %d copy engine(s)\n", (prop.asyncEngineCount ? "Yes" : "No"), prop.asyncEngineCount);
|
|
+ printf(" Run time limit on kernels: %s\n", (KernelExecTimeout != 0) ? "Yes" : "No");
|
|
printf(" Integrated GPU sharing Host Memory: %s\n", prop.integrated ? "Yes" : "No");
|
|
printf(" Support host page-locked memory mapping: %s\n", prop.canMapHostMemory ? "Yes" : "No");
|
|
|
|
@@ -964,7 +984,7 @@ void cv::cuda::printCudaDeviceInfo(int device)
|
|
printf(" Device supports Unified Addressing (UVA): %s\n", prop.unifiedAddressing ? "Yes" : "No");
|
|
printf(" Device PCI Bus ID / PCI location ID: %d / %d\n", prop.pciBusID, prop.pciDeviceID );
|
|
printf(" Compute Mode:\n");
|
|
- printf(" %s \n", computeMode[prop.computeMode]);
|
|
+ printf(" %s \n", computeMode[ComputeMode]);
|
|
}
|
|
|
|
printf("\n");
|
|
--
|
|
2.50.1
|
|
|