[
https://issues.apache.org/jira/browse/ARROW-2447?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16788780#comment-16788780
]
Pearu Peterson commented on ARROW-2447:
---------------------------------------
Re [~pitrou] comment: need a way to query device-specific buffer properties
(such as `cuda_buffer->context()`) ..
Currently, Arrow CUDA support uses primary context management which means that
to get the CUDA context, one only needs to know the device number (use
[cuDevicePrimaryCtxRetain|https://docs.nvidia.com/cuda/cuda-driver-api/group__CUDA__PRIMARY__CTX.html#group__CUDA__PRIMARY__CTX_1g9051f2d5c31501997a6cb0530290a300]).
The device number can be retrieved from the memory pointer (use
[cudaPointerGetAttributes|https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__UNIFIED.html#group__CUDART__UNIFIED_1gd89830e17d399c064a2f3c3fa8bb4390]).
So, it would be sufficient to know that the pointer is a CUDA device pointer
to establish its accessibility properties as well as context if needed.
> [C++] Create a device abstraction
> ---------------------------------
>
> Key: ARROW-2447
> URL: https://issues.apache.org/jira/browse/ARROW-2447
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++, GPU
> Affects Versions: 0.9.0
> Reporter: Antoine Pitrou
> Assignee: Pearu Peterson
> Priority: Major
> Fix For: 0.14.0
>
>
> Right now, a plain Buffer doesn't carry information about where it actually
> lies. That information also cannot be passed around, so you get APIs like
> {{PlasmaClient}} which take or return device number integers, and have
> implementations which hardcode operations on CUDA buffers. Also, unsuspecting
> receivers of a {{Buffer}} pointer may try to act on the underlying memory
> without knowing whether it's CPU-reachable or not.
> Here is a sketch for a proposed Device abstraction:
> {code}
> class Device {
> enum DeviceKind { KIND_CPU, KIND_CUDA };
> virtual DeviceKind kind() const;
> //MemoryPool* default_memory_pool() const;
> //std::shared_ptr<Buffer> Allocate(...);
> };
> class CpuDevice : public Device {};
> class CudaDevice : public Device {
> int device_num() const;
> };
> class Buffer {
> virtual DeviceKind device_kind() const;
> virtual std::shared_ptr<Device> device() const;
> virtual bool on_cpu() const {
> return true;
> }
> const uint8_t* cpu_data() const {
> return on_cpu() ? data() : nullptr;
> }
> uint8_t* cpu_mutable_data() {
> return on_cpu() ? mutable_data() : nullptr;
> }
> virtual CopyToCpu(std::shared_ptr<Buffer> dest) const;
> virtual CopyFromCpu(std::shared_ptr<Buffer> src);
> };
> class CudaBuffer : public Buffer {
> virtual bool on_cpu() const {
> return false;
> }
> };
> CopyBuffer(std::shared_ptr<Buffer> dest, const std::shared_ptr<Buffer> src);
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
