Skip to content

Commit

Permalink
Merge pull request #34 from LuxDL/ap/multigpu
Browse files Browse the repository at this point in the history 
MultiGPU Setup
  • Loading branch information
@avik-pal
avik-pal authored Feb 27, 2024
2 parents 6137fb7 + 8a9985c commit ce8b898
Show file tree
Hide file tree
Showing 8 changed files with 257 additions and 74 deletions.
5 changes: 3 additions & 2 deletions Project.toml
View file
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
name = "LuxDeviceUtils"
uuid = "34f89e08-e1d5-43b4-8944-0b49ac560553"
authors = ["Avik Pal <[email protected]> and contributors"]
version = "0.1.15"
version = "0.1.16"

[deps]
Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
Expand All @@ -11,7 +11,6 @@ LuxCore = "bb33d45b-7691-41d6-9220-0943567d0623"
PrecompileTools = "aea7be01-6a6a-4083-8856-8a6e6704d82a"
Preferences = "21216c6a-2e73-6563-6e65-726566657250"
Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"

[weakdeps]
FillArrays = "1a297f60-69ca-5386-bcde-b61e274b549b"
Expand All @@ -20,6 +19,7 @@ LuxAMDGPU = "83120cb1-ca15-4f04-bf3b-6967d2e6b60b"
LuxCUDA = "d0bbae9a-e099-4d5b-a835-1c6931763bda"
Metal = "dde4c033-4e86-420c-a63e-0dd931031962"
RecursiveArrayTools = "731186ca-8d62-57ce-b412-fbd966d074cd"
SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"

[extensions]
Expand All @@ -29,6 +29,7 @@ LuxDeviceUtilsLuxAMDGPUExt = "LuxAMDGPU"
LuxDeviceUtilsLuxCUDAExt = "LuxCUDA"
LuxDeviceUtilsMetalGPUArraysExt = ["GPUArrays", "Metal"]
LuxDeviceUtilsRecursiveArrayToolsExt = "RecursiveArrayTools"
LuxDeviceUtilsSparseArraysExt = "SparseArrays"
LuxDeviceUtilsZygoteExt = "Zygote"

[compat]
Expand Down
43 changes: 39 additions & 4 deletions ext/LuxDeviceUtilsLuxAMDGPUExt.jl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,19 +5,54 @@ import Adapt: adapt_storage, adapt

__init__() = reset_gpu_device!()

LuxDeviceUtils.__is_loaded(::LuxAMDGPUDevice) = true
LuxDeviceUtils.__is_functional(::LuxAMDGPUDevice) = LuxAMDGPU.functional()
LuxDeviceUtils.__is_loaded(::Union{LuxAMDGPUDevice, <:Type{LuxAMDGPUDevice}}) = true
function LuxDeviceUtils.__is_functional(::Union{LuxAMDGPUDevice, <:Type{LuxAMDGPUDevice}})
return LuxAMDGPU.functional()
end

function LuxDeviceUtils._with_device(::Type{LuxAMDGPUDevice}, ::Nothing)
return LuxAMDGPUDevice(nothing)
end
function LuxDeviceUtils._with_device(::Type{LuxAMDGPUDevice}, id::Int)
id > length(AMDGPU.devices()) &&
throw(ArgumentError("id = $id > length(AMDGPU.devices()) = $(length(AMDGPU.devices()))"))
old_dev = AMDGPU.device()
AMDGPU.device!(AMDGPU.devices()[id])
device = LuxAMDGPUDevice(AMDGPU.device())
AMDGPU.device!(old_dev)
return device
end

LuxDeviceUtils._get_device_id(dev::LuxAMDGPUDevice) = AMDGPU.device_id(dev.device)

# Default RNG
LuxDeviceUtils.default_device_rng(::LuxAMDGPUDevice) = AMDGPU.rocrand_rng()

# Query Device from Array
LuxDeviceUtils.get_device(::AMDGPU.AnyROCArray) = LuxAMDGPUDevice()
LuxDeviceUtils.get_device(x::AMDGPU.AnyROCArray) = LuxAMDGPUDevice(AMDGPU.device(x))

# Device Transfer
## To GPU
adapt_storage(::LuxAMDGPUAdaptor, x) = roc(x)
adapt_storage(::LuxAMDGPUAdaptor{Nothing}, x) = roc(x)
function adapt_storage(to::LuxAMDGPUAdaptor, x)
old_dev = AMDGPU.device() # remember the current device
if !(x isa AMDGPU.AnyROCArray)
AMDGPU.device!(to.device)
x_new = roc(x)
AMDGPU.device!(old_dev)
return x_new
elseif AMDGPU.device_id(AMDGPU.device(x)) == AMDGPU.device_id(to.device)
return x
else
AMDGPU.device!(to.device)
x_new = copy(x)
AMDGPU.device!(old_dev)
return x_new
end
end
adapt_storage(::LuxAMDGPUAdaptor{Nothing}, rng::AbstractRNG) = rng
adapt_storage(::LuxAMDGPUAdaptor, rng::AbstractRNG) = rng
adapt_storage(::LuxAMDGPUAdaptor{Nothing}, rng::Random.TaskLocalRNG) = AMDGPU.rocrand_rng()
adapt_storage(::LuxAMDGPUAdaptor, rng::Random.TaskLocalRNG) = AMDGPU.rocrand_rng()

adapt_storage(::LuxCPUAdaptor, rng::AMDGPU.rocRAND.RNG) = Random.default_rng()
Expand Down
43 changes: 39 additions & 4 deletions ext/LuxDeviceUtilsLuxCUDAExt.jl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,19 +5,54 @@ import Adapt: adapt_storage, adapt

__init__() = reset_gpu_device!()

LuxDeviceUtils.__is_loaded(::LuxCUDADevice) = true
LuxDeviceUtils.__is_functional(::LuxCUDADevice) = LuxCUDA.functional()
LuxDeviceUtils.__is_loaded(::Union{LuxCUDADevice, Type{<:LuxCUDADevice}}) = true
function LuxDeviceUtils.__is_functional(::Union{LuxCUDADevice, Type{<:LuxCUDADevice}})
return LuxCUDA.functional()
end

function LuxDeviceUtils._with_device(::Type{LuxCUDADevice}, ::Nothing)
return LuxCUDADevice(nothing)
end
function LuxDeviceUtils._with_device(::Type{LuxCUDADevice}, id::Int)
id > length(CUDA.devices()) &&
throw(ArgumentError("id = $id > length(CUDA.devices()) = $(length(CUDA.devices()))"))
old_dev = CUDA.device()
CUDA.device!(id - 1)
device = LuxCUDADevice(CUDA.device())
CUDA.device!(old_dev)
return device
end

LuxDeviceUtils._get_device_id(dev::LuxCUDADevice) = CUDA.deviceid(dev.device) + 1

# Default RNG
LuxDeviceUtils.default_device_rng(::LuxCUDADevice) = CUDA.default_rng()

# Query Device from Array
LuxDeviceUtils.get_device(::CUDA.AnyCuArray) = LuxCUDADevice()
LuxDeviceUtils.get_device(x::CUDA.AnyCuArray) = LuxCUDADevice(CUDA.device(x))

# Device Transfer
## To GPU
adapt_storage(::LuxCUDAAdaptor, x) = cu(x)
adapt_storage(::LuxCUDAAdaptor{Nothing}, x) = cu(x)
function adapt_storage(to::LuxCUDAAdaptor, x)
old_dev = CUDA.device() # remember the current device
if !(x isa CUDA.AnyCuArray)
CUDA.device!(to.device)
x_new = cu(x)
CUDA.device!(old_dev)
return x_new
elseif CUDA.device(x).handle == to.device.handle
return x
else
CUDA.device!(to.device)
x_new = copy(x)
CUDA.device!(old_dev)
return x_new
end
end
adapt_storage(::LuxCUDAAdaptor{Nothing}, rng::AbstractRNG) = rng
adapt_storage(::LuxCUDAAdaptor, rng::AbstractRNG) = rng
adapt_storage(::LuxCUDAAdaptor{Nothing}, rng::Random.TaskLocalRNG) = CUDA.default_rng()
adapt_storage(::LuxCUDAAdaptor, rng::Random.TaskLocalRNG) = CUDA.default_rng()

adapt_storage(::LuxCPUAdaptor, rng::CUDA.RNG) = Random.default_rng()
Expand Down
6 changes: 4 additions & 2 deletions ext/LuxDeviceUtilsMetalGPUArraysExt.jl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,8 +5,10 @@ import Adapt: adapt_storage, adapt

__init__() = reset_gpu_device!()

LuxDeviceUtils.__is_loaded(::LuxMetalDevice) = true
LuxDeviceUtils.__is_functional(::LuxMetalDevice) = Metal.functional()
LuxDeviceUtils.__is_loaded(::Union{LuxMetalDevice, Type{<:LuxMetalDevice}}) = true
function LuxDeviceUtils.__is_functional(::Union{LuxMetalDevice, Type{<:LuxMetalDevice}})
return Metal.functional()
end

# Default RNG
LuxDeviceUtils.default_device_rng(::LuxMetalDevice) = GPUArrays.default_rng(MtlArray)
Expand Down
9 changes: 9 additions & 0 deletions ext/LuxDeviceUtilsSparseArraysExt.jl
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,9 @@
module LuxDeviceUtilsSparseArraysExt

import Adapt: adapt_storage
import LuxDeviceUtils: LuxCPUAdaptor
import SparseArrays: AbstractSparseArray

adapt_storage(::LuxCPUAdaptor, x::AbstractSparseArray) = x

end
Loading

2 comments on commit ce8b898

@avik-pal
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@JuliaRegistrator register

@JuliaRegistrator
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Registration pull request created: JuliaRegistries/General/101784

Tip: Release Notes

Did you know you can add release notes too? Just add markdown formatted text underneath the comment after the text
"Release notes:" and it will be added to the registry PR, and if TagBot is installed it will also be added to the
release that TagBot creates. i.e.

@JuliaRegistrator register

Release notes:

## Breaking changes

- blah

To add them here just re-invoke and the PR will be updated.

Tagging

After the above pull request is merged, it is recommended that a tag is created on this repository for the registered package version.

This will be done automatically if the Julia TagBot GitHub Action is installed, or can be done manually through the github interface, or via:

git tag -a v0.1.16 -m "<description of version>" ce8b89863a3952d99ff2b09c4fb1cdea77406daa
git push origin v0.1.16

Please sign in to comment.