Re: [Beignet] How to allocate SVM
Hi, Why only CL_DEVICE_SVM_COARSE_GRAIN_BUFFER is supported? Capabilities flag seem to indicate that the Intel device is capable for fine grain memory. At the moment, Is it possible to get support for fine grain memory features on Intel platform, perhaps by switching to different OpenCL SDK? I am running Ubuntu 17.10 and soon going to upgrade to 18.04 LTS. Thanks 2018-02-24 3:23 GMT+02:00 Yang, Rong R : > Beignet only support CL_DEVICE_SVM_COARSE_GRAIN_BUFFER svm. > > > > Thanks, > > Yang Rong > > > > *From:* Beignet [mailto:beignet-boun...@lists.freedesktop.org] *On Behalf > Of *Sami Pietilä > *Sent:* Thursday, February 22, 2018 4:47 PM > *To:* beignet@lists.freedesktop.org > *Subject:* [Beignet] How to allocate SVM > > > > Hi, > > > > I have Skylake (and Broadwell) CPU and I am running beignet 1.3 platform. I > have checked that device capabilities has CL_DEVICE_SVM_FINE_GRAIN_BUFFER > flag. > > > > Following code gives CL_INVALID_VALUE. Even removing CL_MEM_SVM_ATOMICS > does not help. What could be wrong with the code? Additionally, how to the > same SVM reservation by OpenCL C++ bindings (cl2.hpp)? > > > > size_t SVM_size = 42; > > > > void* p = clSVMAlloc ( > > context(), > > CL_MEM_READ_WRITE | CL_MEM_SVM_FINE_GRAIN_BUFFER | CL_MEM_SVM_ATOMICS, > > SVM_size, > > 0); > > > > cl_int err = clEnqueueSVMMap(queue(), CL_TRUE, CL_MAP_READ | CL_MAP_WRITE, > p, SVM_size, 0, NULL, NULL); > > > > Thanks > ___ Beignet mailing list Beignet@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/beignet
Re: [Beignet] How to allocate SVM
Beignet only support CL_DEVICE_SVM_COARSE_GRAIN_BUFFER svm. Thanks, Yang Rong From: Beignet [mailto:beignet-boun...@lists.freedesktop.org] On Behalf Of Sami Pietilä Sent: Thursday, February 22, 2018 4:47 PM To: beignet@lists.freedesktop.org Subject: [Beignet] How to allocate SVM Hi, I have Skylake (and Broadwell) CPU and I am running beignet 1.3 platform. I have checked that device capabilities has CL_DEVICE_SVM_FINE_GRAIN_BUFFER flag. Following code gives CL_INVALID_VALUE. Even removing CL_MEM_SVM_ATOMICS does not help. What could be wrong with the code? Additionally, how to the same SVM reservation by OpenCL C++ bindings (cl2.hpp)? size_t SVM_size = 42; void* p = clSVMAlloc ( context(), CL_MEM_READ_WRITE | CL_MEM_SVM_FINE_GRAIN_BUFFER | CL_MEM_SVM_ATOMICS, SVM_size, 0); cl_int err = clEnqueueSVMMap(queue(), CL_TRUE, CL_MAP_READ | CL_MAP_WRITE, p, SVM_size, 0, NULL, NULL); Thanks ___ Beignet mailing list Beignet@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/beignet
[Beignet] How to allocate SVM
Hi, I have Skylake (and Broadwell) CPU and I am running beignet 1.3 platform. I have checked that device capabilities has CL_DEVICE_SVM_FINE_GRAIN_BUFFER flag. Following code gives CL_INVALID_VALUE. Even removing CL_MEM_SVM_ATOMICS does not help. What could be wrong with the code? Additionally, how to the same SVM reservation by OpenCL C++ bindings (cl2.hpp)? size_t SVM_size = 42; void* p = clSVMAlloc ( context(), CL_MEM_READ_WRITE | CL_MEM_SVM_FINE_GRAIN_BUFFER | CL_MEM_SVM_ATOMICS, SVM_size, 0); cl_int err = clEnqueueSVMMap(queue(), CL_TRUE, CL_MAP_READ | CL_MAP_WRITE, p, SVM_size, 0, NULL, NULL); Thanks ___ Beignet mailing list Beignet@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/beignet
