A low-latency GPU memory copy library based on NVIDIA GPUDirect RDMA technology.
While GPUDirect RDMA is meant for direct access to GPU memory from third-party devices, it is possible to use these same APIs to create perfectly valid CPU mappings of the GPU memory.
The advantage of a CPU driven copy is the very small overhead involved. That might be useful when low latencies are required.
GDRCopy offers the infrastructure to create user-space mappings of GPU memory, which can then be manipulated as if it was plain host memory (caveats apply here).
A simple by-product of it is a copy library with the following characteristics:
-
very low overhead, as it is driven by the CPU. As a reference, currently a cudaMemcpy can incur in a 6-7us overhead.
-
An initial memory pinning phase is required, which is potentially expensive, 10us-1ms depending on the buffer size.
-
Fast H-D, because of write-combining. H-D bandwidth is 6-8GB/s on Ivy Bridge Xeon but it is subject to NUMA effects.
-
Slow D-H, because the GPU BAR, which backs the mappings, can't be prefetched and so burst reads transactions are not generated through PCIE
The library comes with a few tests like:
- sanity, which contains unit tests for the library and the driver.
- copybw, a minimal application which calculates the R/W bandwidth for a specific buffer size.
- copylat, a benchmark application which calculates the R/W copy latency for a range of buffer sizes.
GPUDirect RDMA requires NVIDIA Tesla or Quadro class GPUs based on Kepler, Pascal, Volta, or Turing, see GPUDirect RDMA. For more technical informations, please refer to the official GPUDirect RDMA design document.
The device driver requires GPU display driver >= 418.40 on ppc64le and >= 331.14 on other platforms. The library and tests require CUDA >= 6.0. Additionally, the sanity test requires check >= 0.9.8 and subunit.
DKMS is a prerequisite for installing GDRCopy kernel module package. On RHEL, however, users have an option to build kmod and install it instead of the DKMS package. See Build and installation section for more details.
# On RHEL
# dkms can be installed from epel-release. See https://fedoraproject.org/wiki/EPEL.
$ sudo yum install dkms check check-devel subunit subunit-devel
# On Debian
$ sudo apt install check libsubunit0 libsubunit-dev
CUDA and GPU display driver must be installed before building and/or installing GDRCopy. The installation instructions can be found in https://developer.nvidia.com/cuda-downloads.
GPU display driver header files are also required. They are installed as a part of the driver (or CUDA) installation with runfile. If you install the driver via package management, we suggest
- On RHEL,
sudo dnf module install nvidia-driver:latest-dkms
. - On Debian,
sudo apt install nvidia-dkms-<your-nvidia-driver-version>
.
The supported architectures are Linux x86_64, ppc64le, and arm64. The supported platforms are RHEL7, RHEL8, Ubuntu16_04, Ubuntu18_04, and Ubuntu20_04.
Root privileges are necessary to load/install the kernel-mode device driver.
We provide three ways for building and installing GDRCopy.
$ sudo yum groupinstall 'Development Tools'
$ sudo yum install dkms rpm-build make check check-devel subunit subunit-devel
$ cd packages
$ CUDA=<cuda-install-top-dir> ./build-rpm-packages.sh
$ sudo rpm -Uvh gdrcopy-kmod-<version>dkms.noarch.<platform>.rpm
$ sudo rpm -Uvh gdrcopy-<version>.<arch>.<platform>.rpm
$ sudo rpm -Uvh gdrcopy-devel-<version>.noarch.<platform>.rpm
DKMS package is the default kernel module package that build-rpm-packages.sh
generates. To create kmod package, -m
option must be passed to the script.
Unlike the DKMS package, the kmod package contains a prebuilt GDRCopy kernel
module which is specific to the NVIDIA driver version and the Linux kernel
version used to build it.
$ sudo apt install build-essential devscripts debhelper check libsubunit-dev fakeroot pkg-config dkms
$ cd packages
$ CUDA=<cuda-install-top-dir> ./build-deb-packages.sh
$ sudo dpkg -i gdrdrv-dkms_<version>_<arch>.<platform>.deb
$ sudo dpkg -i libgdrapi_<version>_<arch>.<platform>.deb
$ sudo dpkg -i gdrcopy-tests_<version>_<arch>.<platform>.deb
$ sudo dpkg -i gdrcopy_<version>_<arch>.<platform>.deb
$ make prefix=<install-to-this-location> CUDA=<cuda-install-top-dir> all install
$ sudo ./insmod.sh
If libcheck
is installed in a non-standard path and therefore is not
picked by pkg-config
, you can set the PKG_CONFIG_PATH
environment
variable to the directory which contains the check.pc
file and pass it
down to make:
$ PKG_CONFIG_PATH=/check_install_path/lib/pkgconfig/ make <...>
Execute provided tests:
$ sanity
Running suite(s): Sanity
100%: Checks: 27, Failures: 0, Errors: 0
$ copybw
GPU id:0; name: Tesla V100-SXM2-32GB; Bus id: 0000:06:00
GPU id:1; name: Tesla V100-SXM2-32GB; Bus id: 0000:07:00
GPU id:2; name: Tesla V100-SXM2-32GB; Bus id: 0000:0a:00
GPU id:3; name: Tesla V100-SXM2-32GB; Bus id: 0000:0b:00
GPU id:4; name: Tesla V100-SXM2-32GB; Bus id: 0000:85:00
GPU id:5; name: Tesla V100-SXM2-32GB; Bus id: 0000:86:00
GPU id:6; name: Tesla V100-SXM2-32GB; Bus id: 0000:89:00
GPU id:7; name: Tesla V100-SXM2-32GB; Bus id: 0000:8a:00
selecting device 0
testing size: 131072
rounded size: 131072
gpu alloc fn: cuMemAlloc
device ptr: 7f1153a00000
map_d_ptr: 0x7f1172257000
info.va: 7f1153a00000
info.mapped_size: 131072
info.page_size: 65536
info.mapped: 1
info.wc_mapping: 1
page offset: 0
user-space pointer:0x7f1172257000
writing test, size=131072 offset=0 num_iters=10000
write BW: 9638.54MB/s
reading test, size=131072 offset=0 num_iters=100
read BW: 530.135MB/s
unmapping buffer
unpinning buffer
closing gdrdrv
$ copylat
GPU id:0; name: Tesla V100-SXM2-32GB; Bus id: 0000:06:00
GPU id:1; name: Tesla V100-SXM2-32GB; Bus id: 0000:07:00
GPU id:2; name: Tesla V100-SXM2-32GB; Bus id: 0000:0a:00
GPU id:3; name: Tesla V100-SXM2-32GB; Bus id: 0000:0b:00
GPU id:4; name: Tesla V100-SXM2-32GB; Bus id: 0000:85:00
GPU id:5; name: Tesla V100-SXM2-32GB; Bus id: 0000:86:00
GPU id:6; name: Tesla V100-SXM2-32GB; Bus id: 0000:89:00
GPU id:7; name: Tesla V100-SXM2-32GB; Bus id: 0000:8a:00
selecting device 0
device ptr: 0x7fa2c6000000
allocated size: 16777216
gpu alloc fn: cuMemAlloc
map_d_ptr: 0x7fa2f9af9000
info.va: 7fa2c6000000
info.mapped_size: 16777216
info.page_size: 65536
info.mapped: 1
info.wc_mapping: 1
page offset: 0
user-space pointer: 0x7fa2f9af9000
gdr_copy_to_mapping num iters for each size: 10000
WARNING: Measuring the API invocation overhead as observed by the CPU. Data
might not be ordered all the way to the GPU internal visibility.
Test Size(B) Avg.Time(us)
gdr_copy_to_mapping 1 0.0889
gdr_copy_to_mapping 2 0.0884
gdr_copy_to_mapping 4 0.0884
gdr_copy_to_mapping 8 0.0884
gdr_copy_to_mapping 16 0.0905
gdr_copy_to_mapping 32 0.0902
gdr_copy_to_mapping 64 0.0902
gdr_copy_to_mapping 128 0.0952
gdr_copy_to_mapping 256 0.0983
gdr_copy_to_mapping 512 0.1176
gdr_copy_to_mapping 1024 0.1825
gdr_copy_to_mapping 2048 0.2549
gdr_copy_to_mapping 4096 0.4366
gdr_copy_to_mapping 8192 0.8141
gdr_copy_to_mapping 16384 1.6155
gdr_copy_to_mapping 32768 3.2284
gdr_copy_to_mapping 65536 6.4906
gdr_copy_to_mapping 131072 12.9761
gdr_copy_to_mapping 262144 25.9459
gdr_copy_to_mapping 524288 51.9100
gdr_copy_to_mapping 1048576 103.8028
gdr_copy_to_mapping 2097152 207.5990
gdr_copy_to_mapping 4194304 415.2856
gdr_copy_to_mapping 8388608 830.6355
gdr_copy_to_mapping 16777216 1661.3285
gdr_copy_from_mapping num iters for each size: 100
Test Size(B) Avg.Time(us)
gdr_copy_from_mapping 1 0.9069
gdr_copy_from_mapping 2 1.7170
gdr_copy_from_mapping 4 1.7169
gdr_copy_from_mapping 8 1.7164
gdr_copy_from_mapping 16 0.8601
gdr_copy_from_mapping 32 1.7024
gdr_copy_from_mapping 64 3.1016
gdr_copy_from_mapping 128 3.4944
gdr_copy_from_mapping 256 3.6400
gdr_copy_from_mapping 512 2.4394
gdr_copy_from_mapping 1024 2.8022
gdr_copy_from_mapping 2048 4.6615
gdr_copy_from_mapping 4096 7.9783
gdr_copy_from_mapping 8192 14.9209
gdr_copy_from_mapping 16384 28.9571
gdr_copy_from_mapping 32768 56.9373
gdr_copy_from_mapping 65536 114.1008
gdr_copy_from_mapping 131072 234.9382
gdr_copy_from_mapping 262144 496.4011
gdr_copy_from_mapping 524288 985.5196
gdr_copy_from_mapping 1048576 1970.7057
gdr_copy_from_mapping 2097152 3942.5611
gdr_copy_from_mapping 4194304 7888.9468
gdr_copy_from_mapping 8388608 18361.5673
gdr_copy_from_mapping 16777216 36758.8342
unmapping buffer
unpinning buffer
closing gdrdrv
$ apiperf -s 8
GPU id:0; name: Tesla V100-SXM2-32GB; Bus id: 0000:06:00
GPU id:1; name: Tesla V100-SXM2-32GB; Bus id: 0000:07:00
GPU id:2; name: Tesla V100-SXM2-32GB; Bus id: 0000:0a:00
GPU id:3; name: Tesla V100-SXM2-32GB; Bus id: 0000:0b:00
GPU id:4; name: Tesla V100-SXM2-32GB; Bus id: 0000:85:00
GPU id:5; name: Tesla V100-SXM2-32GB; Bus id: 0000:86:00
GPU id:6; name: Tesla V100-SXM2-32GB; Bus id: 0000:89:00
GPU id:7; name: Tesla V100-SXM2-32GB; Bus id: 0000:8a:00
selecting device 0
device ptr: 0x7f1563a00000
allocated size: 65536
Size(B) pin.Time(us) map.Time(us) get_info.Time(us) unmap.Time(us)
unpin.Time(us)
65536 1346.034060 3.603800 0.340270 4.700930 676.612800
Histogram of gdr_pin_buffer latency for 65536 bytes
[1303.852000 - 2607.704000] 93
[2607.704000 - 3911.556000] 0
[3911.556000 - 5215.408000] 0
[5215.408000 - 6519.260000] 0
[6519.260000 - 7823.112000] 0
[7823.112000 - 9126.964000] 0
[9126.964000 - 10430.816000] 0
[10430.816000 - 11734.668000] 0
[11734.668000 - 13038.520000] 0
[13038.520000 - 14342.372000] 2
closing gdrdrv
Depending on the platform architecture, like where the GPU are placed in the PCIe topology, performance may suffer if the processor which is driving the copy is not the one which is hosting the GPU, for example in a multi-socket server.
In the example below, GPU ID 0 is hosted by CPU socket 0. By explicitly playing with the OS process and memory affinity, it is possible to run the test onto the optimal processor:
$ numactl -N 0 -l copybw -d 0 -s $((64 * 1024)) -o $((0 * 1024)) -c $((64 * 1024))
GPU id:0; name: Tesla V100-SXM2-32GB; Bus id: 0000:06:00
GPU id:1; name: Tesla V100-SXM2-32GB; Bus id: 0000:07:00
GPU id:2; name: Tesla V100-SXM2-32GB; Bus id: 0000:0a:00
GPU id:3; name: Tesla V100-SXM2-32GB; Bus id: 0000:0b:00
GPU id:4; name: Tesla V100-SXM2-32GB; Bus id: 0000:85:00
GPU id:5; name: Tesla V100-SXM2-32GB; Bus id: 0000:86:00
GPU id:6; name: Tesla V100-SXM2-32GB; Bus id: 0000:89:00
GPU id:7; name: Tesla V100-SXM2-32GB; Bus id: 0000:8a:00
selecting device 0
testing size: 65536
rounded size: 65536
gpu alloc fn: cuMemAlloc
device ptr: 7f5817a00000
map_d_ptr: 0x7f583b186000
info.va: 7f5817a00000
info.mapped_size: 65536
info.page_size: 65536
info.mapped: 1
info.wc_mapping: 1
page offset: 0
user-space pointer:0x7f583b186000
writing test, size=65536 offset=0 num_iters=1000
write BW: 9768.3MB/s
reading test, size=65536 offset=0 num_iters=1000
read BW: 548.423MB/s
unmapping buffer
unpinning buffer
closing gdrdrv
or on the other socket:
$ numactl -N 1 -l copybw -d 0 -s $((64 * 1024)) -o $((0 * 1024)) -c $((64 * 1024))
GPU id:0; name: Tesla V100-SXM2-32GB; Bus id: 0000:06:00
GPU id:1; name: Tesla V100-SXM2-32GB; Bus id: 0000:07:00
GPU id:2; name: Tesla V100-SXM2-32GB; Bus id: 0000:0a:00
GPU id:3; name: Tesla V100-SXM2-32GB; Bus id: 0000:0b:00
GPU id:4; name: Tesla V100-SXM2-32GB; Bus id: 0000:85:00
GPU id:5; name: Tesla V100-SXM2-32GB; Bus id: 0000:86:00
GPU id:6; name: Tesla V100-SXM2-32GB; Bus id: 0000:89:00
GPU id:7; name: Tesla V100-SXM2-32GB; Bus id: 0000:8a:00
selecting device 0
testing size: 65536
rounded size: 65536
gpu alloc fn: cuMemAlloc
device ptr: 7fbb63a00000
map_d_ptr: 0x7fbb82ab0000
info.va: 7fbb63a00000
info.mapped_size: 65536
info.page_size: 65536
info.mapped: 1
info.wc_mapping: 1
page offset: 0
user-space pointer:0x7fbb82ab0000
writing test, size=65536 offset=0 num_iters=1000
write BW: 9224.36MB/s
reading test, size=65536 offset=0 num_iters=1000
read BW: 521.262MB/s
unmapping buffer
unpinning buffer
closing gdrdrv
GDRCopy works with regular CUDA device memory only, as returned by cudaMalloc. In particular, it does not work with CUDA managed memory.
gdr_pin_buffer()
accepts any addresses returned by cudaMalloc and its family.
In contrast, gdr_map()
requires that the pinned address is aligned to the GPU page.
Neither CUDA Runtime nor Driver APIs guarantees that GPU memory allocation
functions return aligned addresses. Users are responsible for proper alignment
of addresses passed to the library.
Two cudaMalloc'd memory regions may be contiguous. Users may call
gdr_pin_buffer
and gdr_map
with address and size that extend across these
two regions. This use case is not well-supported in GDRCopy. On rare occassions,
users may experience 1.) an error in gdr_map
, or 2.) low copy performance
because gdr_map
cannot provide write-combined mapping.
In some GPU driver versions, pinning the same GPU address multiple times consumes additional BAR1 space. This is because the space is not properly reused. If you encounter this issue, we suggest that you try the latest version of NVIDIA GPU driver.
On POWER9 where CPU and GPU are connected via NVLink, CUDA9.2 and GPU Driver v396.37 are the minimum requirements in order to achieve the full performance. GDRCopy works with ealier CUDA and GPU driver versions but the achievable bandwidth is substantially lower.
For reporting issues you may be having using any of NVIDIA software or reporting suspected bugs we would recommend you use the bug filing system which is available to NVIDIA registered developers on the developer site.
If you are not a member you can sign up.
Once a member you can submit issues using this form. Be sure to select GPUDirect in the "Relevant Area" field.
You can later track their progress using the My Bugs link on the left of this view.
If you find this software useful in your work, please cite: R. Shi et al., "Designing efficient small message transfer mechanism for inter-node MPI communication on InfiniBand GPU clusters," 2014 21st International Conference on High Performance Computing (HiPC), Dona Paula, 2014, pp. 1-10, doi: 10.1109/HiPC.2014.7116873.