ILGPU JIT Compiler for high-performance .Net GPU programs
OTHER License
Bot releases are hidden (Show)
Published by m4rs-mt over 3 years ago
This new beta offers significant performance improvements to the generated kernel programs and includes a lot of amazing new features (get the Nuget package).
Please note that this version has some breaking changes compared to previous ILGPU versions.
Refer to the v1.0-beta1 summary for more information.
Published by m4rs-mt over 3 years ago
This new beta offers significant performance improvements to the generated kernel programs and includes a lot of amazing new features (get the Nuget package).
Please note that this version has some breaking changes compared to previous ILGPU versions.
Memory API
, involving ArrayView
and MemoryBuffer
types has been significantly improved to support explicit Stride
information (see below).IndexX
and LongIndexX
types have been renamed to IndexXD
and LongIndexXD
to have a unified programming experience with respect to memory buffers and array views (see below).Device API
has been redesigned to explicitly enable, filter and configure the available hardware accelerator devices (see below).Memory API
to support explicit stride information (#421, #475, #483).Device API
to enable, filter and configure the available hardware accelerator devices (#428).OpenCL 3.0
API (#464).ProfilingMarker
s (#482).Warp
/Group
/Multiprocessor
configurations (#402, #484).IRBuilder
(#477).OpenCL
kernels in the presence of constant switch conditions (#441).The new API distinguishes between a coherent, strongly typed ArrayView<T>
structure and its n-D versions ArrayViewXD<T, TStride>
, which carry dimension-dependent stride information (The actual logic for computing element addresses is moved from the IndexXD
types to the newly added StrideXD
types). This allows developers to explicitly specify a particular stride of a view, reinterpret
the data layout itself (by changing the stride), and perform compile-time optimizations based on explicitly typed stride information. Consequently, ILGPU's optimization pipeline is able to remove the overhead of these abstractions in most cases (except in rare use cases where strange-looking strides are used). It also makes all memory transfer-related operations explicit in terms of what memory layout the underlying data will have after an operation is performed.
In addition, it moves all copy
related methods to the ArrayView
instances instead of exposing them on the memory buffers. This realizes a "separation of concerns": One the one hand, a MemoryBuffer
holds a reference to the native memory area and controls its lifetime. On the other hand, ArrayView
structures manage the contents of these buffers and make them available to the actual GPU kernels.
Example:
// Simple 1D allocation of 1024 longs with TStride = Stride1D.Dense (all elements are accessed contiguously in memory)
var t = accl.Allocate1D<long>(1024);
// Advanced 1D allocation of 1024 longs with TStride = Stride1D.General(2) (each memory access will skip 2 elements)
// -> allocates 1024 * 2 longs to be able to access all of them
var t = accl.Allocate1D<long, Stride1D.General>(1024, new Stride1D.General(2));
// Simple 1D allocation of 1024 longs using the array provided
var data1 = new long[1024];
var t2 = accl.Allocate1D(data1);
// Simple 2D allocation of 1024 * 1024 longs using the array provided with TStride = Stride2D.DenseX
// (all elements in X dimension are accessed contiguously in memory)
// -> this will *not* transpose the input buffer as the memory layout will be identical on CPU and GPU
var data2 = new long[1024, 1024];
var t3 = accl.Allocate2DDenseX(data2);
// Simple 2D allocation of 1024 * 1024 longs using the array provided, with TStride = Stride2D.DenseY
// (all elements in Y dimension are accessed contiguously in memory)
// -> this *will* transpose the input buffer to match the desired data layout
var data3 = new long[1024, 1024];
var t4 = accl.Allocate2DDenseY(data3);
The major changes/features of the new Memory API are:
Index1
|Index2
|Index3
types have been renamed to Index1D
|Index2D
|Index3D
to match the naming scheme of ArrayViewXD
and MemoryBufferXD
types.LongIndex1
|LongIndex2
|LongIndex3
types have been renamed to LongIndex1D
|LongIndex2D
|LongIndex3D
to match the naming scheme of the ArrayViewXD
and MemoryBufferXD
types.MemoryBuffer
and ArrayView
instances:
ArrayView...
structures represent and manage the contents of buffers (or chunks of buffers).MemoryBuffer...
classes manage the lifetime of allocated memory chunks on a device.ILGPU.ArrayView
intrinsic structure implements the newly added IContiguousArrayView
interface that marks contiguous memory sections.ILGPU.Runtime.MemoryBuffer...
classes implement the newly added IContiguousArrayView
interface that marks contiguous memory sections.IContiguousArrayView
interface provide extension methods for initializing, copying from and to the memory region (not supported on accelerators).Stride
s. ILGPU contains built-in common strides for 1D, 2D and 3D views.
Stride1D.Dense
represents contiguous chunks of memory that pack elements side by side.Stride1D.General
represents strides that skip a certain number of elements.Stride2D.DenseX
represents 2D strides that pack elements side by side in dimension X (transfers from a to views with this stride involve transpose operations).Stride2D.DenseY
represents 2D strides that pack elements in the Y dimension side by side.Stride2D.General
represents strides that skip a certain number of elements in the X and Y dimensions.Stride3D.DenseXY
represents 3D strides that pack elements in the X,Y dimension side by side (transfers from a to views with this stride involve transposition operations).Stride3D.DenseZY
represents 3D strides that pack elements in the Z,Y dimension side by side.Stride3D.General
represents strides that omit a certain number of elements in the X, Y and Z dimensions.ArrayViewXD
types have been moved to the ILGPU.Runtime
namespace.ArrayViewXD
types do not implement IContiguousArrayView
, as they support arbitrary stride information.ArrayView1D<T, Stride1D.Dense>
specialization has an implicit conversion to ArrayView<T>
(and vice versa) for auxiliary purposes.CopyFromCPU
and CopyToCPU
methods are provided with additional hints as to whether they are transposing the input elements or keeping the original layout.GetAsXDArray(...)
always returns elements in .Net standard layout for 1D, 2D and 3D arrays (this may result in transposing the input elements of the buffer on the CPU).view.AsContiguous().GetAsArray()
to get the memory layout of the input buffer.The new Device API removes the enumeration ContextFlags
and implements the same functionality in an object oriented way using a Context.Builder
class. It offers a fluent-API like configuration interface which makes it easy to set up:
// Enables all supported accelerators (default CPU accelerator only) and puts the context
// into auto-assertion mode via "AutoAssertions()". In other words, if a debugger is attached,
// the `Context` instance will turn on all assertion checks. This behavior is identical
// to the current implementation via new Context();
using var context = Context.CreateDefault();
// Turns on O2 and enables all compatible Cuda devices.
using var context = Context.Create(builder =>
{
builder.Optimize(OptimizationLevel.O2).Cuda();
});
// Turns on all assertions, enables the IR verifier and enables all compatible OpenCL devices.
using var context = Context.Create(builder =>
{
builder.Assertions().Verify().OpenCL();
});
// Turns on kernel source-line annotations, fast math using 32-bit float and enables
// *all* (even incompatible) OpenCL devices.
using var context = Context.Create(builder =>
{
builder
.DebugSymbols(DebugSymbolsMode.KernelSourceAnnotations)
.Math(MathMode.Fast32BitOnly)
.OpenCL(device => true);
});
// Selects an OpenCL device with a warp size of at least 32:
using var context = Context.Create(builder =>
{
builder.OpenCL(device => device.WarpSize >= 32);
});
// Turns on all assertions in debug mode (same behavior like calling CreateDefault()):
using var context = Context.Create(builder =>
{
builder.AutoAssertions();
});
// Turns on debug optimizations (level O0) and all assertions if a debugger is attached:
using var context = Context.Create(builder =>
{
builder.AutoDebug();
});
// Turns on debug mode (optimization level P0, assertions and kernel debug information):
using var context = Context.Create(builder =>
{
builder.Debug();
});
// Disable caching, enable conservative inlining and inline mutable static field values:
using var context = Context.Create(builder =>
{
builder
.Caching(CachingMode.Disabled)
.Inlining(InliningMode.Conservative)
.StaticFields(StaticFieldMode.MutableStaticFields);
});
// Turn on *all* CPU accelerators that simulate different hardware platforms:
using var context = Context.Create(builder => builder.CPU());
// Turn on an AMD-based CPU accelerator:
using var context = Context.Create(builder => builder.CPU(CPUDeviceKind.AMD));
Note that by default all debug symbols are automatically turned off when a debugger is attached. If you want to turn on the debug information in all cases, call .builder.DebugSymbols(DebugSymbolsMode.Basic)
. At the same time, this PR introduces the notion of a Device
, which replaces the implementation of AcceleratorId
. This allows us to query detailed device information without explicitly instantiating an accelerator:
// Print all device information without instantiating a single accelerator
// (device context) instance.
using var context = Context.Create(...);
foreach (var device in context)
{
// Print detailed accelerator information
device.PrintInformation();
// ...
}
Note that we removed the ability to call the accelerator constructors (e.g. new CudaAccelerator(...)
) directly. Either use the CreateAccelerator
methods defined in the Device
classes or use one of the extension methods like CreateCudaAccelerator(...)
of the Context
class itself:
using var context = Context.Create(...);
foreach (var device in context)
{
// Instantiate an accelerator instance on this device
using Accelerator accel = device.CreateAccelerator();
// ...
}
// Instantiate the 2nd Cuda accelerator (NOTE that this is the *2nd* Cuda device
// and *not* the 2nd device of your machine).
using CudaAccelerator cudaDevice = context.CreateCudaAccelerator(1);
// Instantiate the 1st OpenCL accelerator (NOTE that this is the *1st* OpenCL device
// and *not* the 1st device of your machine).
using CLAccelerator clDevice = context.CreateOpenCLAccelerator(0);
Context
properties that expose types from other (ILGPU internal) namespaces that cannot/should not (?) be covered by the API/ABI guarantees we want to give, has been made internal
properties. To access these properties, use one of the available extension methods located in the corresponding namespaces:
using var context = ...
// OLD way
var internalIRContext = context.IRContext;
// NEW way:
// using namespace ILGPU.IR;
var internalIRContext = context.GetIRContext();
The new CPU runtime significantly improves the existing CPUAccelerator
runtime by adding support for user-defined warp
, group
and multiprocessor
configurations. It changes the internal functionality to simulate a single warp of at least 2 threads (which ensures that all shuffle-based/reduction-like algorithms can also be run on the CPU by default). At the same time, each virtual multiprocessor can only execute a single thread group at a time. Increasing the number of virtual multiprocessors allows the user to simulate multiple concurrent groups. Most use cases will not require more than a single multiprocessor in practice.
Note that all device-wide static Grid
/Group
/Atomic
/Warp
classes are fully supported to debug/simulate all ILGPU kernels on the CPU.
Note that a custom warp size must be a multiple of 2.
This PR adds a new set of static creation methods:
CreateDefaultSimulator(...)
which creates a CPUAccelerator
instance with 4 threads per warp, 4 warps per multiprocessor and a single multiprocessor (MaxGroupSize = 16
).CreateNvidiaSimulator(...)
which creates a CPUAccelerator
instance with 32 threads per warp, 32 warps per multiprocessor and a single multiprocessor (MaxGroupSize = 1024
).CreateAMDSimulator(...)
which creates a CPUAccelerator
instance with 32 threads per warp, 8 warps per multiprocessor and a single multiprocessor (MaxGroupSize = 256
).CreateLegacyAMDSimulator(...)
which creates a CPUAccelerator
instance with 64 threads per warp, 4 warps per multiprocessor and a single multiprocessor (MaxGroupSize = 256
).CreateIntelSimulator(...)
which creates a CPUAccelerator
instance with 16 threads per warp, 8 warps per multiprocessor and a single multiprocessor (MaxGroupSize = 128
).Furthermore, this PR adds support for advanced debugging features that enable a "sequential-like" execution mode. In this mode, each thread of a group will run sequentially one after another until it hits a synchronization barrier or exits the kernel function. This allows users to conveniently debug larger thread groups consisting of concurrent threads without switching to single-threaded execution. This behavior can be controlled via the newly added CPUAcceleratorMode
enum:
/// <summary>
/// The accelerator mode to be used with the <see cref="CPUAccelerator"/>.
/// </summary>
public enum CPUAcceleratorMode
{
/// <summary>
/// The automatic mode uses <see cref="Sequential"/> if a debugger is attached.
/// It uses <see cref="Parallel"/> if no debugger is attached to the
/// application.
/// </summary>
/// <remarks>
/// This is the default mode.
/// </remarks>
Auto = 0,
/// <summary>
/// If the CPU accelerator uses a simulated sequential execution mechanism. This
/// is particularly useful to simplify debugging. Note that different threads for
/// distinct multiprocessors may still run in parallel.
/// </summary>
Sequential = 1,
/// <summary>
/// A parallel execution mode that runs all execution threads in parallel. This
/// reduces processing time but makes it harder to use a debugger.
/// </summary>
Parallel = 2,
}
By default, all CPUAccelerator
instances use the automatic mode (CPUAcceleratorMode.Auto
) that switches to a sequential execution model as soon as a debugger is attached to the application.
Note that threads in the scope of multiple multiprocessors may still run in parallel.
Special thanks to @MoFtZ, @Joey9801, @jgiannuzzi and @GPSnoopy for their contributions to this release in form of code, feedback, ideas and proposals. Furthermore, we would like to thank the entire ILGPU community (especially @MPSQUARK, @Nnelg, @Ruberik, @Yey007, @faruknane, @mikhail-khalizev, @nullandkale and @yuryGotham) for providing feedback, submitting issues and feature requests.
Published by m4rs-mt over 3 years ago
The new stable version contains several bug fixes and improves the code quality of the generated kernel programs (get the Nuget package).
It is strongly recommended to upgrade to this version as soon as possible to avoid known bugs and some CPU-buffer deallocation issues.
Special thanks to @MoFtZ, @marcin-krystianc and @jgiannuzzi for their contributions to this release and to the entire ILGPU community for providing feedback, submitting issues and feature requests.
Published by m4rs-mt over 3 years ago
The new stable version offers significant performance improvements of the generated kernel programs and contains critical resource deallocation fixes (get the Nuget package).
It is strongly recommended to upgrade to this version as soon as possible to avoid resource and GC related deallocation issues.
ExchangeBuffer
class has been changed to avoid exposing internal memory buffers. If you previously relied on the immediate inheritance from ExchangeBufferBase
on MemoryBuffer
, you have to adapt your program to use the intermediate base class MemoryBuffer<T, TIndex>
instead (see diff).MemoryBufferXD
classes have been removed to avoid ownership related GC-free issues (see diff).We have decided to remove dangerous properties from several memory buffer classes. The use of these properties can lead to program crashes, since buffers could be disposed asynchronously in the background by the GC without further notice.
O2
to O1
(release mode) to improve performance in release builds using an additional of stable optimization passes (#344).Cuda
backend to O1
pipeline to generate vectorized IO operations in release builds (#350).sizeof
IL instruction (#380).PrintInformation
method to Accelerator
instances to print detailed accelerator information (#389).ArrayView
accesses on GPU devices (Use flag ContextFlags.EnableAsserations
or attach a debugger to your application to enable assertion checks. Make sure to use the portable
debug information format for detailed source location information) (#375).CPU
, Cuda
and OpenCL
accelerators (#342).AlignTo
alignment methods to explicitly align ArrayView
instances to a particular alignment in bytes (#316).LocalMemory
class (#316).PopCount
, CLZ
and CTZ
operations (#324).MemSet
functions to all memory buffers (#338).O2
pipeline (#328).LongGlobalIndex
helper to simplify correct computations using 64-bit integers (#337).CLPlatformVersion
and fixed OpenCL 1.2 compatibility issues (#335).SharedMemory
in implicitly grouped kernels (#354).CudaAccelerator
and CLAccelerator
instances to run on non-native OS .NET versions (#396).LoopUnrolling
(#373).printf
formats for int64
and uintX
types (#391).DebugArrayView
implementations (#345).RuntimeSystem
to avoid concurrency/reflection-API issues (#393).Special thanks to @MoFtZ, @Ruberik and @jgiannuzzi for their contributions to this release and to the entire ILGPU community for providing feedback, submitting issues and feature requests.
Published by m4rs-mt over 3 years ago
This new beta version offers important bug fixes and performance improvements of the generated kernel programs and a set of new features (get the Nuget package).
sizeof
IL instruction (#380).PrintInformation
method to Accelerator
instances to print detailed accelerator information (#389).ArrayView
accesses on GPU devices (Use flag ContextFlags.EnableAsserations
or attach a debugger to your application to enable assertion checks. Make sure to use the portable
debug information format for detailed source location information) (#375).SharedMemory
in implicitly grouped kernels (#354).CudaAccelerator
and CLAccelerator
instances to run on non-native OS .NET versions (#396).LoopUnrolling
(#373).printf
formats for int64
and uintX
types (#391).Major internal changes:
RuntimeSystem
to avoid concurrency/reflection-API issues (#393).Special thanks to @MoFtZ, @Ruberik for their contributions to this release and to the entire ILGPU community for providing feedback, submitting issues and feature requests.
Published by m4rs-mt almost 4 years ago
This new stable version offers significant performance and code quality improvements of the generated kernel programs.
Special thanks to @MoFtZ, @Yey007 and @jgiannuzzi for contributing to this release.
Published by m4rs-mt almost 4 years ago
Special thanks to @MoFtZ, @Yey007 and @jgiannuzzi for contributing to this release.
Published by m4rs-mt almost 4 years ago
The new stable version offers significant performance and code quality improvements of the generated kernel programs.
ContextFlags.EnableVerifier
(#121).Published by m4rs-mt almost 4 years ago
The new beta version offers significant performance improvements of the generated kernel programs.
ContextFlags.EnableVerifier
(#121).Published by m4rs-mt almost 4 years ago
The new stable version offers significant performance and code quality improvements of the generated kernel programs.
GetSubView
in the context of generic and multidimensional array views (#19).Special thanks to @MoFtZ for contributing to this release.
Published by m4rs-mt almost 4 years ago
OpenCL
code generation issues (#85, #88, #91, #92)Special thanks to @MoFtZ for contributing to this release.
Published by m4rs-mt almost 4 years ago
PTX
and OpenCL
code by enabling more aggressive optimizations and clever code generation (#70).enum
-value interop (#66).PTXBackend
to support all API changes and to fix several critical code-generation issues. This also includes emission of PTX instructions that mimic the Cuda
compiler.OpenCL
backend to support all API changes and to fix several critical code-generation issues (#72, #73, #74, #78).IR-rewriter
API to perform more advanced IR transformations.rewriter API
.Special thanks to @MoFtZ for contributing to this release.
Published by m4rs-mt almost 4 years ago
CPU
& Cuda
backends).KernelConfig
structure to specify launch dimensions for explicitly grouped kernels.KernelConfig
structure instead of GroupedIndex
types.Grid
and Group
properties.Index1
structure to avoid name clashes with new System.Index
structure.Index2
and Index3
types.EntryPointDescription
structure to specify an entry point and its index type.RuntimeKernelConfig
structure to combine static and dynamic information about a particular kernel launch.GroupedIndex
types.PTXInstructions
to support bool-based IOs in PTXBackend
(#68).ExchangeBuffer
to use new page-locked memory allocation (if available).CudaAPI
to supported paged-lock host-memory allocation functions.GetSubView
in the context of generic and multidimensional array views (#19).AtomicCAS
operations on AMD hardware (#67).Published by m4rs-mt almost 4 years ago
Cuda
and CPU
-based array views.CPU
and GPU
memory.CLAccelerator
.CL
accelerators.AcceleratorId
classes.OpenCL
code generator for float values that are assign integers values.OpenCL
backend.ABI
thread safe to support concurrent queries of size/alignment information.Published by m4rs-mt almost 4 years ago
OpenCL
-compatible GPUs (beta)CUDA
driver version detection.CPUAccelerator
.Utility.Select
method that can be used to create highly-efficient select instructions in favor of if branches.XMath
functions to the ILGPU.Algorihtms
library. Use the new IntrinsicMath
class for math functions that are supported on all platforms.AcceleratorId
functionality.CudaMemoryBuffer
to support MemSetToZero
using alternate streams.Accelerator
to resolve a launch extent with the maximum number of groups.PTXBackend
.Special thanks to @MoFtZ for contributing to this release.
Published by m4rs-mt almost 4 years ago
Greatly improved ILGPU version that included significant performance and code quality improvements.
PTXBackend
.PTXBackend
.PTXBackend
.SequencePoint
.PTXBackend
.Special thanks to @MoFtZ and @mikhail-khalizev for contributing to this release.
Published by m4rs-mt almost 4 years ago
Improved version of v0.5
that contains bug fixes and performance improvements and features based on community feedback.
DebuggerDisplay
attributes on array views.MemoryBuffer.CopyTo
.PTXBackend
.null
values in PTXBackend
.Published by m4rs-mt almost 4 years ago
Note that this has been the first standalone compiler version without any native library dependencies.
Published by m4rs-mt almost 4 years ago
Note that this version is based on the LLVM compiler framework and contains native dependencies. Please use ILGPU >= v0.5
to use the platform independent ILGPU compiler version.
*** This release has been the last LLVM-based compiler version ***
DLLLoader
and PTXBackend
.Published by m4rs-mt almost 4 years ago
Note that this version is based on the LLVM compiler framework and contains native dependencies. Please use ILGPU >= v0.5
to use the platform independent ILGPU compiler version.