eyalroz/

cuda-api-wrappers

Thin C++-flavored wrappers for the CUDA Runtime API

Owners:

$0.00

USD raised

Recent activities

eyalroz added the repository

about 4 years ago

Issues

Funded Unfunded Rewarded Closed

Have all launch config validation code in single place

Unfunded#548created byeyalroz

$0.00

Streamline `cuda-api-wrappers-config.cmake` + no need to use preprocessing for it

Unfunded#562created byeyalroz

$0.00

Unused / irrelevant functions in the launch config builder

Unfunded#561created byeyalroz

$0.00

Work around CMake restriction since 3.25 on setting properties of targets built outside of project

Unfunded#558created byeyalroz

$0.00

Fix the way we set WINDOWS_EXPORT_ALL_SYMBOLS

Unfunded#557created byeyalroz

$0.00

The CompileWithWarnings module is only relevant for the examples

Unfunded#556created byeyalroz

$0.00

Needless enabling of languages when not building example programs

Unfunded#555created byeyalroz

$0.00

Add port to VCPKG

Unfunded#554created byeclipse0922

$0.00

Shouldn't assume containers have cbegin()

Unfunded#553created byeyalroz

$0.00

vectorAddMMAP fails setting access mode to virtual memory mapping

Unfunded#552created byeyalroz

$0.00

std::is_trivially_copy_constructible requirement for kernel parameters still too strong

Unfunded#551created bypauleonix

$0.00

Block dimension validation is comparing in the wrong direction

Unfunded#550created byeyalroz

$0.00

Avoid using properties().max_shared_memory_per_block()

Unfunded#549created byeyalroz

$0.00

Must enable either C or C++ to use cuda-api-wrappers

Unfunded#563created byeyalroz

$0.00

In DEBUG mode, perform more launch config validation before launch

Unfunded#547created byeyalroz

$0.00

File `cuda/rtc.hpp` fails to compile.

Unfunded#546created byaryan-programmer

$0.00

Kernel launch errors from `cudaGetLastError` are not handled

Unfunded#545created byaryan-programmer

$0.00

Avoid trying to stick compilation logs of size N into buffers of size N-1 + Fix size-0 handling

Unfunded#544created byeyalroz

$0.00

Fall back to cudaGetErrorString() for Runtime-API-only errors

Unfunded#543created byeyalroz

$0.00

Robustify namespace use in our macros

Unfunded#542created byeyalroz

$0.00

Support for additional vendor GPUs

Unfunded#541created bydmcdougall

$0.00

Compilation logs are returned with the trailing nul as a a valid character

Unfunded#539created byeyalroz

$0.00

Scheduling a memory copy on stream 0x056b5efe0: context is destroyed

Unfunded#537created bythomas-bouvier

$0.00

Calling set_endpoint by mistake in a set_endpoint_untyped method of copy_parameters_t

Unfunded#535created byeyalroz

$0.00

some example program unique_ptr's need to be float[], not float

Unfunded#534created byeyalroz

$0.00

module proxy class doesn't need to hold its link options

Unfunded#576created byeyalroz

$0.00

Add detail_ function for allocating host memory in the current context

Unfunded#588created byeyalroz

$0.00

Missing memory::host::zero() variant taking a region

Unfunded#587created byeyalroz

$0.00

Factor the span class out of types.hpp

Unfunded#586created byeyalroz

$0.00

Have the global-device-memory versions of make_unique and make_unique_region be the default

Unfunded#585created byeyalroz

$0.00

Have all targets mark dependencies of CUDA library dependencies which NVIDIA ignores

Unfunded#583created byeyalroz

$0.00

Improve launch configuration validation

Unfunded#582created byeyalroz

$0.00

Robustify dimension comparison operators

Unfunded#581created byeyalroz

$0.00

Add standard-library-compatible allocators for different memory spaces

Unfunded#197created byeyalroz

$0.00

Support cudaMemRangeGetAttribute, cudaMemRangeGetAttributes

Unfunded#36created byeyalroz

$0.00

Support use_max_active_blocks_per_multiprocessor() on a config builder

Unfunded#580created byeyalroz

$0.00

config builder is missing a method for setting the device

Unfunded#579created byeyalroz

$0.00

Use the launch config builder in more examples

Unfunded#578created byeyalroz

$0.00

program_t::add_registered_globals should be able to take a container of any string-like type

Unfunded#529created byeyalroz

$0.00

Move apriori_compiled_kernel_t into the kernel namespace

Unfunded#575created byeyalroz

$0.00

Remove redundant inclusions of NVIDIA headers

Unfunded#573created byeyalroz

$0.00

Compatibility broken for CUDA 9.x

Unfunded#572created byeyalroz

$0.00

Add link options getter to the module class

Unfunded#571created byeyalroz

$0.00

Redundant definition of `cuda::module::handle_t`

Unfunded#570created byeyalroz

$0.00

Leftover warning regarding validate_all_dimensions_compatibility not being used

Unfunded#569created byeyalroz

$0.00

cuda::kernel_t::set_attribute is broken!

Unfunded#568created byeyalroz

$0.00

Want alias for runtime-and-driver

Unfunded#567created byeyalroz

$0.00

Want ability to set function attributes on function pointers and handles

Unfunded#566created byeyalroz

$0.00

Support CUlibrary's and CUkernel's - CUDA driver low-level "library management"

Unfunded#565created byeyalroz

$0.00

Full representation of launch configurations + launch-with-full-config support

Unfunded#564created byeyalroz

$0.00

Merge the `by_driver_api_module` and `by_runtime_api_module` example program directories

Unfunded#489created byeyalroz

$0.00

Add "unsupported device sync" error from NVRTC 12.1

Unfunded#501created byeyalroz

$0.00

Switch from NVVM to LTO-IR

Unfunded#500created byeyalroz

$0.00

Disable surface-related functionality with CUDA 12.x

Unfunded#499created byeyalroz

$0.00

Support num_blocks() as an alias for grid_size() in the launch config builder

Unfunded#498created byeyalroz

$0.00

Use relative include paths within the library

Unfunded#497created byeyalroz

$0.00

fatbin targets placed in "wrong" target folder

Unfunded#495created byeyalroz

$0.00

MSVC balks at set_access_mode() in vectorAddMMAP.cpp

Unfunded#494created byeyalroz

$0.00

launch config builder's overall_size() should take overall_dimension_t

Unfunded#493created byeyalroz

$0.00

Drop inclusion of cooperative_groups.h

Unfunded#492created byeyalroz

$0.00

Avoid MSVC warnings about standard-library facilities

Unfunded#491created byeyalroz

$0.00

Switch from CUDA::nvToolkitExt to CUDA::nvtx3 (and to CMake 3.25)

Unfunded#490created byeyalroz

$0.00

Move physical allocation namespace out of `virtual_` and up into `memory`

Unfunded#485created byeyalroz

$0.00

Get rid of dependencies on helper_cuda.hpp

Unfunded#502created byeyalroz

$0.00

cuda::memory::host::allocate should return a region_t

Unfunded#488created byeyalroz

$0.00

Better handling of negative device id's

Unfunded#487created byeyalroz

$0.00

Adopting code to `cuda::kernel::wrap` API changes

Unfunded#486created byfvanmaele

$0.00

Support setting kernel block cluster dimensions

Unfunded#484created byeyalroz

$0.00

Support converting regions into spans - at least explicitly

Unfunded#483created byeyalroz

$0.00

cuda::span improvements

Unfunded#482created byeyalroz

$0.00

stream_t::enqueue::free() is not const!

Unfunded#481created byeyalroz

$0.00

Try to build an executable for each C++ version after 11 supported by the compiler

Unfunded#480created byeyalroz

$0.00

The IPC pointer import mechanism is unsafe, inconsistent with rest of lib

Unfunded#479created byeyalroz

$0.00

By default, create CUDA C++ programs

Unfunded#477created byeyalroz

$0.00

Support capture-related API calls

Unfunded#222created byeyalroz

$0.00

Use explicit cast and correct types for write_single_value

Unfunded#516created byeyalroz

$0.00

mapped memory allocation currently insists on CU_MEMHOSTALLOC_DEVICEMAP

Unfunded#528created byeyalroz

$0.00

Add support for the read-only attribute for cuda::memory::host::register()

Unfunded#527created byeyalroz

$0.00

Use proper C++ random number generators in example programs

Unfunded#526created byeyalroz

$0.00

copy_parameters_t::set_single_context shouldn't take an endpoint_t

Unfunded#525created byeyalroz

$0.00

Spaces -> tabs fixes

Unfunded#524created byeyalroz

$0.00

Pass device_t's by const-ref more

Unfunded#523created byeyalroz

$0.00

Rename synch_ to sync_ on everything

Unfunded#522created byeyalroz

$0.00

Reduce boilerplate + avoid warning when overriding context for the current scope

Unfunded#521created byeyalroz

$0.00

Don't use a context-wrapper scoped-context-override when unnecessary

Unfunded#520created byeyalroz

$0.00

Support use of external memory resources

Unfunded#519created byeyalroz

$0.00

Avoid C-style casts

Unfunded#518created byeyalroz

$0.00

typed_set mistakenly accepts values of size 8

Unfunded#517created byeyalroz

$0.00

Factor the region class and related definitions out of `types.hpp`

Unfunded#589created byeyalroz

$0.00

Missing includes

Unfunded#515created bycodecircuit

$0.00

Definitions of cuda::memory::managed::allocate(device, num_bytes)

Unfunded#514created bycodecircuit

$0.00

Support creation of CUDA runtime error exceptions with fully-overriden what() messages

Unfunded#511created byeyalroz

$0.00

When NVRTC/PTX compiler return an invalid-option error - put the options in the error message

Unfunded#510created byeyalroz

$0.00

kernel compilation option --gpu-architecture can cause NVRTC_ERROR_INVALID_OPTION

Unfunded#509created byDanCher1403

$0.00

With CUDA >= 11.3, don't avoid creating a module just because CUBIN is unavailable

Unfunded#508created byeyalroz

$0.00

Run the simpleStreams example with all possible synch modes

Unfunded#507created byeyalroz

$0.00

GitHub Windows builds fail due to missing cudaProfiler.h

Unfunded#506created byeyalroz

$0.00

Re-establish example program build success on GitHub actions Windows runners

Unfunded#505created byeyalroz

$0.00

Build fails with cooperative_groups on GitHub Actions Windows runners with CUDA >= 11.7

Unfunded#504created byeyalroz

$0.00

Get rid of dependency on helper_string.h in example programs

Unfunded#503created byeyalroz

$0.00

Support querying the CUDA module loading mode

Unfunded#675created byeyalroz

$0.00

poor man's optional const&& constructor causes trouble in some cases

Unfunded#666created byeyalroz

$0.00

Support building (or rather exporting) targets which will statically link against CUDA libraries

Unfunded#699created byeyalroz

$0.00

Add "shortcut" function for getting the default device

Unfunded#698created byeyalroz

$0.00

When building example programs, make sure kernels are built for _all_ relevant architectures

Unfunded#694created byeyalroz

$0.00

Add an optional_ref class

Unfunded#691created byeyalroz

$0.00

Unify async and non-async operations using optional_ref<const stream_t> optional arguments

Unfunded#689created byeyalroz

$0.00

Missing support for asynchronous copying using CUDA_MEMCPY2D and CUDA_MEMCPY3D_PEER parameters

Unfunded#688created byeyalroz

$0.00

sync and async versions of copy_single disagree - pointers vs references

Unfunded#687created byeyalroz

$0.00

Support getting kernels from a library with a target context in which they should be available

Unfunded#686created byeyalroz

$0.00

About cuLibraryLoadFromFile

Unfunded#685created byCallmeZhangChenchen

$0.00

Change unique_span to take a deleter object taking a span, not a template param accepting taking pointer

Unfunded#678created byeyalroz

$0.00

optional::value_or() doesn't return a value

Unfunded#676created byeyalroz

$0.00

Add a matrixMulCUBLAS example program

Unfunded#680created byeyalroz

$0.00

Potential use-before-initialization in simpleIPC.cu

Unfunded#673created byeyalroz

$0.00

Rename macro to be less-clashy

Unfunded#671created byeyalroz

$0.00

Support Module function enumeration

Unfunded#669created byeyalroz

$0.00

Add square-brackets element acces into `grid::dimensions_t` and `grid::overall_dimensions_t`

Unfunded#668created byeyalroz

$0.00

Some methods in the dimensions classes are missing noexcept and/or constexpr

Unfunded#667created byeyalroz

$0.00

Address unique_span issues from cr.SX

Unfunded#665created byeyalroz

$0.00

unique_span deleted copy constructors should ignore source deleter type

Unfunded#662created bykdkavanagh

$0.00

unique_span assignement operator bug with C++20.

Unfunded#660created byraplonu

$0.00

The context_t::create_stream and context_t::create_event methods should be const

Unfunded#659created byeyalroz

$0.00

Offer stand-alone functions for stream capture begin and end

Unfunded#657created byeyalroz

$0.00

Default to capturing on CUDA streams using the global capture mode

Unfunded#656created byeyalroz

$0.00

(Deleted)

Unfunded#707created byeyalroz

$0.00

Should we support "CUDA in Graphics" context creation?

Unfunded#748created byeyalroz

$0.00

Fix incorrect values in max_in_flight_threads_per_processor() + comments

Unfunded#723created byeyalroz

$0.00

enqueueing a new event instead of the intended existing event during capture in simpleCudaGraphs

Unfunded#722created byeyalroz

$0.00

The max_in_flight_threads_on_device() property can be unsigned, doesn't need unsigned long long

Unfunded#721created byeyalroz

$0.00

Verify suspicious implicit numeric conversion in example programs

Unfunded#720created byeyalroz

$0.00

Remove redundant cast operators into regions from the unique_span class

Unfunded#719created byeyalroz

$0.00

Our exported CMake targets depend on CUDAToolkit targets that may be missing later on

Unfunded#717created byeyalroz

$0.00

unique_region_t move constructor and assigment operator doesn't compile

Unfunded#713created byralwing

$0.00

Update static tables with Blackwell architecture info

Unfunded#711created byeyalroz

$0.00

Fix C++20-implicit and pre-C++20 comparison operators for launch_configuration_t

Unfunded#710created byeyalroz

$0.00

Span-to-c-array copy function ignores stream of choice, acts synchronously

Unfunded#709created byeyalroz

$0.00

Infinite recursion in one variant of link_t::add_file()

Unfunded#708created byeyalroz

$0.00

Align the logic of context_t::create_module() with context::module::create()

Unfunded#654created byeyalroz

$0.00

context_t::flags() should not be virtual

Unfunded#706created byeyalroz

$0.00

Avoid accidental infinite recursion in library_t::get_managed()

Unfunded#705created byeyalroz

$0.00

Compiler warnings coming from release 0.8.0, when using clang

Unfunded#704created bybgianfo

$0.00

swap method of unique_span doesn't swap the deleters

Unfunded#703created byeyalroz

$0.00

Use range-based copy, and fill - like in the C++ standard library

Unfunded#11created byeyalroz

$0.00

Fix typo in --no-source-include NVRTC option

Unfunded#702created byeyalroz

$0.00

Add support for libnvfatbin APIs

Unfunded#681created byeyalroz

$0.00

Respect lack of support for --version-ident with CUDA 12.2 and later

Unfunded#701created byeyalroz

$0.00

Unify the random values generator across vectorAdd* examples

Unfunded#683created byeyalroz

$0.00

Take more launch-config-related sizes as size_t

Unfunded#684created byeyalroz

$0.00

poor_mans_option::value_or() should be const

Unfunded#682created byeyalroz

$0.00

Make `cuda::memory::region_t` more constexpr-friendly

Unfunded#602created byeyalroz

$0.00

Two bugs with CUDA-12 related launch configuration fields

Unfunded#619created byeyalroz

$0.00

Use defaulted comparison operator for launch configurations

Unfunded#618created byeyalroz

$0.00

Replace memory::external::mapped_region_t with unique_region, or rebase the former on the latter

Unfunded#617created byeyalroz

$0.00

Reduce code repetition between context_t and primary_context_t

Unfunded#614created byeyalroz

$0.00

We don't need a context flags type, but do need a primary context flags type

Unfunded#612created byeyalroz

$0.00

Drop cuda::dynarray in favor of cuda::unique_span

Unfunded#610created byeyalroz

$0.00

"Robustify" the RTC buffer getters (log, ptx, lto_ir)

Unfunded#609created byeyalroz

$0.00

Split off a type_traits.hpp header from types.hpp

Unfunded#607created byeyalroz

$0.00

Support copying from standard-library containers with contiguous storage

Unfunded#606created byeyalroz

$0.00

Support for constructing const_region_t's from region_t&&

Unfunded#605created byeyalroz

$0.00

Make memory::region_t CUDA-independent

Unfunded#604created byeyalroz

$0.00

Use our unique_span's instead of the faux-dynarray class

Unfunded#603created byeyalroz

$0.00

context::current::scoped_override_t should be declared in current_context.hpp

Unfunded#620created byeyalroz

$0.00

Add an empty() method to span()

Unfunded#601created byeyalroz

$0.00

Stick to an underscore suffix for proxy class field name

Unfunded#600created byeyalroz

$0.00

An invalid file, `name_caching_program.hpp`, has snuck into our code

Unfunded#599created byeyalroz

$0.00

library kernel get_attribute and set_attribute have return issues

Unfunded#598created byeyalroz

$0.00

Refactoring of `region_t` introduced a sometimes-incorrect path

Unfunded#597created byeyalroz

$0.00

Checking against the optional rather than the value in launch_config_builder_t

Unfunded#596created byeyalroz

$0.00

Use region_t's for mapped memory

Unfunded#594created byeyalroz

$0.00

Fixes and additions to the copy parameters wrapper structure

Unfunded#593created byeyalroz

$0.00

cuda::memory::managed::region_t is not a useful definition, stick to cuda::memory::region_t

Unfunded#592created byeyalroz

$0.00

Some more work on copy functions

Unfunded#591created byeyalroz

$0.00

Avoid warnings about narrowing conversion and shadowing

Unfunded#590created byeyalroz

$0.00

Support new JIT compilation flags

Unfunded#635created byeyalroz

$0.00

Support stream-unaware launching in an arbitrary CUDA context

Unfunded#653created byeyalroz

$0.00

Let a unique_span have a default ctor

Unfunded#652created byeyalroz

$0.00

Can't access the default stream of a (non-primary) context

Unfunded#651created byeyalroz

$0.00

Consider using gsl::span's

Unfunded#37created byeyalroz

$0.00

cuCtxGetSharedMemConfig is deprecated as of CUDA 12.3, acknowledge this

Unfunded#649created byeyalroz

$0.00

Let a mapped pair of spans "decay" into a pair of regions

Unfunded#647created byeyalroz

$0.00

Managed memory deleter uses wrong free() function leaks PC reference

Unfunded#645created byeyalroz

$0.00

Move memory::pool::ipc::ptr_handle_t out of ipc.hpp

Unfunded#643created byeyalroz

$0.00

Rework the range_attribute_t and advice code

Unfunded#640created byeyalroz

$0.00

Simply access-permissions-related code

Unfunded#639created byeyalroz

$0.00

Drop the `native_word_t` type

Unfunded#638created byeyalroz

$0.00

Hide "flags" from the interface of devices and contexts

Unfunded#637created byeyalroz

$0.00

device::get() marked noexcept

Unfunded#475created byeyalroz

$0.00

Move `rtc::marshalled_options_t` and `rtc::marshal()` into the `detail_` subnamespace

Unfunded#632created byeyalroz

$0.00

Drop `examples/rtc_common.hpp`, use cuda::rtc::render to print compilation options

Unfunded#631created byeyalroz

$0.00

Move program_base_t into the detail namespace

Unfunded#630created byeyalroz

$0.00

Add a device ID field to the texture_view object

Unfunded#629created byeyalroz

$0.00

Correct signatures of context::current::peer_to_peer functions

Unfunded#628created byeyalroz

$0.00

Drop context-specifying functions for allocation host memory

Unfunded#627created byeyalroz

$0.00

Use optional's instead of a special bespoke "unused" value in pci_location_t's

Unfunded#626created byeyalroz

$0.00

Drop specify_default_load_caching_mode from link::options_t

Unfunded#625created byeyalroz

$0.00

Collect the log-related link options into a sub-structure of `link::options_t`

Unfunded#624created byeyalroz

$0.00

link::marshalled_options_t and link::option_t should be in namespace detail_

Unfunded#622created byeyalroz

$0.00

Improve naming of fallback_strategy_t and its values

Unfunded#621created byeyalroz

$0.00

vectorAdd doesn't compile with C++17 and GCC 12 on my machine

Unfunded#323created byQuarticCat

$0.00

Redundant 0x in error messages

Unfunded#335created byeyalroz

$0.00

Reduce the number of redundant informative API calls

Unfunded#334created byeyalroz

$0.00

Don't treat freeing in a destroyed context as an error

Unfunded#333created byeyalroz

$0.00

Freeing a pointer from a destroyed context should not be an error

Unfunded#332created byeyalroz

$0.00

module::load_from_file() creates a module in the current context

Unfunded#331created byeyalroz

$0.00

launch config builder trying to use maximum linear block without device or kernel

Unfunded#330created byeyalroz

$0.00

marshalled_options.hpp errors with C++17

Unfunded#329created byeyalroz

$0.00

Reduce gratuitous API calls in current_device::detail::set_()

Unfunded#328created byeyalroz

$0.00

`make_unique` without device behaves strangely in the example `vectorAdd`

Unfunded#327created byQuarticCat

$0.00

Drop redundant inclusions and make include order more "challenging" in vectorAdd examples

Unfunded#326created byeyalroz

$0.00

nvrtc/compilation_options.hpp needs to know about device_t

Unfunded#325created byeyalroz

$0.00

marshalled_options.hpp needs cuda::span, but doesn't see it

Unfunded#324created byeyalroz

$0.00

unique_ptrs throws an instance of 'cuda::runtime_error' in version v0.5.1

Unfunded#336created byDiamonDinoia

$0.00

std::trivially-copyable requirement for kernel parameters too strong

Unfunded#321created byeyalroz

$0.00

module::create missing when only including runtime_api.hpp

Unfunded#320created byeyalroz

$0.00

Add an example which only includes the runtime API header collection

Unfunded#319created byeyalroz

$0.00

Support memory pools (introduced in CUDA 11.2)

Unfunded#249created byeyalroz

$0.00

Support making grids from OpenCL-style "global+local work sizes"

Unfunded#214created byeyalroz

$0.00

Leaking a current context and a primary context reference unit in scoped_current_device_fallback_t

Unfunded#317created byeyalroz

$0.00

Cannot rely on scoped_current_device_fallback_t

Unfunded#316created byeyalroz

$0.00

Launching on the current device instead of the selected device

Unfunded#315created byeyalroz

$0.00

Potential failure to set context when recording an event

Unfunded#314created byeyalroz

$0.00

examples fail on kepler GPU

Unfunded#313created byrainli323

$0.00

Make launch config construction more convenient with a builder class

Unfunded#311created byeyalroz

$0.00

std:: -> ::std::

Unfunded#347created byeyalroz

$0.00

Missing implementation of context_t::create_event()

Unfunded#360created byeyalroz

$0.00

Allow enqueueing kernel launches using type-erased marshalled arguments

Unfunded#359created byeyalroz

$0.00

link_t missing a device_id_ member

Unfunded#358created byeyalroz

$0.00

assignment operators not updated for primary context refcount unit holding

Unfunded#357created byeyalroz

$0.00

proxy classes must be move-assignable

Unfunded#356created byeyalroz

$0.00

Support taking the language dialect from an std::string

Unfunded#355created byeyalroz

$0.00

cuda::rtc::program::create() is too clunky

Unfunded#354created byeyalroz

$0.00

missing initializer for member ‘cuda::device::pci_location_t::function’

Unfunded#353created byDiamonDinoia

$0.00

Simplify the option marshalling code

Unfunded#352created byeyalroz

$0.00

memMapIPCDrv example does not compile with MSVC C++17

Unfunded#351created byQ-Minh

$0.00

Prefix CMake options with CUDA_API_WRAPPERS

Unfunded#349created byQ-Minh

$0.00

MSVC has problems with append_to_current since size_t isn't unsigned long

Unfunded#348created byeyalroz

$0.00

Switch the NVTX wrapper to use driver-API-style calls

Unfunded#310created byeyalroz

$0.00

GitHub actions script for MS Windows fails to install CUDA

Unfunded#346created byeyalroz

$0.00

Try setting up Appveyor-based CI environment

Unfunded#79created byeyalroz

$0.00

vectorAdd_profiled fails to compile with VS2019

Unfunded#345created byQ-Minh

$0.00

vectorAddMMAP target does not compile due to virtual memory feature set_access_mode

Unfunded#344created byQ-Minh

$0.00

std::experimental::filesystem is deprecated for recent MSVC versions causing compilation failures

Unfunded#343created byQ-Minh

$0.00

mkstemp unsupported on Windows

Unfunded#342created byeyalroz

$0.00

Two overloads with unsigned and size_t clash on Windows

Unfunded#341created byeyalroz

$0.00

Mark shared-mem size-determiner function with CUDA_CB

Unfunded#340created byeyalroz

$0.00

Avoid MSVC compiler warnings

Unfunded#339created byeyalroz

$0.00

Example programs fail compilation due to HANDLE being undefined

Unfunded#338created byQ-Minh

$0.00

Library does not apply MSVC __cplusplus workaround consistently

Unfunded#337created byQ-Minh

$0.00

Is it acceptable to launch with kernel_t's based on the arguments' types?

Unfunded#236created byeyalroz

$0.00

Don't launch wrapped kernels on devices other than the wrapper's

Unfunded#288created byeyalroz

$0.00

Direct component access, or access via other wrappers?

Unfunded#287created byeyalroz

$0.00

Don't use cooperative groups with CUDA 8...

Unfunded#286created byeyalroz

$0.00

Change kernel_t construction pattern to fit other wrapper classes

Unfunded#285created byeyalroz

$0.00

Properly define a grid-and-block dims structure

Unfunded#284created byeyalroz

$0.00

Factor out device, event, stream etc. identification code

Unfunded#283created byeyalroz

$0.00

Add methods for getting individual kernel attributes - to the runtime-API-only branch

Unfunded#282created byeyalroz

$0.00

Rename "stream ids" and "event ids" into stream and event handles

Unfunded#280created byeyalroz

$0.00

Consider dropping device_t::create_event and device_t::create_stream

Unfunded#277created byeyalroz

$0.00

Support pitched/padded arrays in 2D, 3D copies

Unfunded#275created byeyalroz

$0.00

Missing some cudaSetDevice() calls in stream_t methods

Unfunded#273created byeyalroz

$0.00

Support getting and setting stream synchronization policy

Unfunded#272created byeyalroz

$0.00

Use wrap() for device_t's

Unfunded#289created byeyalroz

$0.00

Add support for run-time compilation using the NVRTC library

Unfunded#228created byeyalroz

$0.00

Possible Improper Use of Lock Guard

Unfunded#271created byDeveloperPaul123

$0.00

Build Issues with Examples

Unfunded#270created byDeveloperPaul123

$0.00

Rename resource_limit_t -> limit_t

Unfunded#269created byeyalroz

$0.00

Export dependency to CUDAToolkit

Unfunded#268created bycodecircuit

$0.00

Correct link to `CUDA::nvToolsExt` of target `nvtx`

Unfunded#267created bycodecircuit

$0.00

Unfunded#266created bymorsingher

$0.00

Support for NVRTC diag-suppress/error/warn

Unfunded#265created byeyalroz

$0.00

Use a named handle type for CUDA arrays

Unfunded#264created byeyalroz

$0.00

Support the NVRTC LTO flag

Unfunded#263created byeyalroz

$0.00

Support new CUDA 11.* flags in NVRTC

Unfunded#262created byeyalroz

$0.00

Move wrap() methods for significant wrapper classes out of `detail_::`

Unfunded#298created bycodecircuit

$0.00

Add support for naming streams, events and devices in NVTX

Unfunded#309created byeyalroz

$0.00

In NVTX, be willing to accept both narrow and wide character strings

Unfunded#308created byeyalroz

$0.00

Make the nvtx library header-only

Unfunded#307created byeyalroz

$0.00

Use driver API error enum values whenever applicable

Unfunded#306created byeyalroz

$0.00

Add missing named errors

Unfunded#305created byeyalroz

$0.00

Restore CUDA 9.x compatibility

Unfunded#304created byeyalroz

$0.00

Use CUDA_VERSION instead of CUDART_VERSION

Unfunded#303created byeyalroz

$0.00

Add functionality from the CUDA driver, not just the runtime API

Unfunded#9created byeyalroz

$0.00

Add support for CUDA versions before 10.x to the driver-wrappers branch

Unfunded#302created byeyalroz

$0.00

Can't assign launch configurations

Unfunded#301created byeyalroz

$0.00

Checking CUDA_VERSION instead of CUDART_VERSION

Unfunded#300created byeyalroz

$0.00

Bug in 0.4.4: Use of old `stream.id()` instead of `stream.handle()` in `cuda::copy_attributes(const stream_t&, const stream_t&)`

Unfunded#299created bypauleonix

$0.00

If the language dialect is "set" to an empty/null string - unset it

Unfunded#361created byeyalroz

$0.00

Any plan to support vcpkg?

Unfunded#297created byeladmaimoni

$0.00

Windows x64-to-x32 cross-build fails to match the stream_launched_host_function_adapter template

Unfunded#296created byeyalroz

$0.00

Build fails in MS Visual Studio due to narrowing conversion

Unfunded#295created byvic4key

$0.00

Support CUDA execution graphs

Unfunded#175created byneoblizz

$0.00

Make "cooperativity" be part of the launch configuration

Unfunded#258created byeyalroz

$0.00

Drop per-architecture properties which are actually per-compute-capability

Unfunded#293created byeyalroz

$0.00

Missing device properties values for Ampere GPUs

Unfunded#292created byeyalroz

$0.00

Add a `unique_region_t` class or some equivalent mechanism

Unfunded#291created byeyalroz

$0.00

symbol handling is broken

Unfunded#290created byeyalroz

$0.00

Support memory functions for copying between peer devices

Unfunded#225created byeyalroz

$0.00

Properly diagnose library and driver mismatches

Unfunded#254created byeyalroz

$0.00

Robustify error handling

Unfunded#426created byeyalroz

$0.00

Avoid move'ing compilation_options_t unless necessary

Unfunded#439created byeyalroz

$0.00

bitwise-AND-ing instead of bitwise-OR-ing in make_cuda_host_alloc_flags

Unfunded#438created byeyalroz

$0.00

Avoid half-precision includes and functionality when CUDA_NO_HALF is defined

Unfunded#437created byeyalroz

$0.00

Update static tables with Lovelace, Hopper architecture info

Unfunded#435created byeyalroz

$0.00

(deleted)

Unfunded#434created byeyalroz

$0.00

Pass program name by value instead of const-ref

Unfunded#433created byeyalroz

$0.00

More non-::-qualified namespace std

Unfunded#432created byeyalroz

$0.00

Drop unnecessary nvtx dependencies

Unfunded#431created byeyalroz

$0.00

Pass kernels by reference rather than by value to kernel launch functions

Unfunded#430created byeyalroz

$0.00

Some array_t copy() code tweaks

Unfunded#429created byeyalroz

$0.00

Size checks in one of the array copy function is reversed

Unfunded#428created byeyalroz

$0.00

Drop copy constructors creating non-owning variants (stream_t, event_t)

Unfunded#427created byeyalroz

$0.00

Should wrapper classes throw when failing on destruction?

Unfunded#440created byeyalroz

$0.00

Build error strings lazily

Unfunded#425created byeyalroz

$0.00

Pass link options by const-ref, not by value

Unfunded#424created byeyalroz

$0.00

Implement missing module getters: texture and surface reference

Unfunded#423created byeyalroz

$0.00

Some context override gadgets are named "set_device_..."

Unfunded#420created byeyalroz

$0.00

Move multi-wrapper-dependent functionality out of module.hpp

Unfunded#419created byeyalroz

$0.00

Multi GPU support is broken from v0.5.4 to the current development branch

Unfunded#418created byDiamonDinoia

$0.00

fix up relations of memory.hpp and multi_wrapper_impls/memory.hpp

Unfunded#417created byeyalroz

$0.00

std::experimental headers for optional and any mistakenly included within namespace cuda

Unfunded#416created byeyalroz

$0.00

optional fixes vis-a-vis later C++ standard versions

Unfunded#415created byeyalroz

$0.00

Apply the rule-of-five to the scoped context ensurer

Unfunded#414created byeyalroz

$0.00

Fix up dependencies and includes

Unfunded#413created byeyalroz

$0.00

cuda::kernel::get() should not take an arbitrary context

Unfunded#459created byeyalroz

$0.00

devices begin() can throw - don't have it as noexcept

Unfunded#474created byeyalroz

$0.00

Avoid the fancy-shmancy heap allocation for stream callbacks

Unfunded#473created byeyalroz

$0.00

stream_t::enqueue::memset() should support regions

Unfunded#472created byeyalroz

$0.00

Make more comparison operators constexpr and noexcept

Unfunded#471created byeyalroz

$0.00

Need a convenient function for getting a kernel from a __global__ function

Unfunded#469created byeyalroz

$0.00

Have memory::type_of treat non-CUDA-allocated memory as a non-error

Unfunded#468created byeyalroz

$0.00

Array-to-raw-mem copy function must either get a context or determine it

Unfunded#467created byeyalroz

$0.00

Make copy_parameters_t user-facing and beef it up

Unfunded#466created byeyalroz

$0.00

Unable to build examples with instructions in README.md

Unfunded#464created byqy3u

$0.00

Allow obtaining a pointer's device and context without wrapping it

Unfunded#463created byeyalroz

$0.00

Support obtaining a pointer's device in CUDA 9.1 and earlier

Unfunded#462created byeyalroz

$0.00

Support colcon build

Unfunded#460created bywep21

$0.00

bandwidthtest doesn't include <iostream>

Unfunded#412created byeyalroz

$0.00

Support execution graph capture via stream_t's

Unfunded#458created byeyalroz

$0.00

Missing definition for allocate_managed() in src/cuda/api/context.hpp

Unfunded#455created bykingcrimsontianyu

$0.00

Set flags for the flush-remote-writes stream operation

Unfunded#453created byeyalroz

$0.00

Support the memory barrier "stream memory operation"

Unfunded#452created byeyalroz

$0.00

Stream callbacks should use CUDA_CB, not CUDART_CB

Unfunded#451created byeyalroz

$0.00

Memory leak: Allocation without release in cuda::memory::virtual::set_access_mode

Unfunded#450created byeyalroz

$0.00

`apriori_compiled_kernel_t::get_attribute` should be marked inline

Unfunded#449created byjhol

$0.00

cuda::profiling::mark::range_start and range_end call create_attributions the wrong way

Unfunded#448created byeyalroz

$0.00

Member initialisation order error.

Unfunded#443created byj-horner-c4x

$0.00

Lingering `link::input_type_t` in `cuda/api/link.hpp`

Unfunded#442created byjhol

$0.00

kernel_t::context() should use wrap() to be noexcept

Unfunded#441created byeyalroz

$0.00

In debug mode, perform some sanity checks before enqueuing a kernel launch

Unfunded#373created byeyalroz

$0.00

Add support for the PTX compilation API

Unfunded#385created byeyalroz

$0.00

Make nvrtc depend on runtime-and-driver

Unfunded#384created byeyalroz

$0.00

region_pair has no type information

Unfunded#383created byEmmanuelMess

$0.00

Make the library usable with FetchContent

Unfunded#382created byeyalroz

$0.00

Provide instalation guide for installing with CMakeLists

Unfunded#381created byEmmanuelMess

$0.00

no problem

Unfunded#380created byjiudianren

$0.00

Let the user supply a string buffer for the compilation log

Unfunded#379created byeyalroz

$0.00

Compilation log vector<char> contains trailing '\0'

Unfunded#378created byeyalroz

$0.00

When rendering compilation options to a string, we get an extra space

Unfunded#377created byeyalroz

$0.00

When rendering compilation options to a string, we get an extra space

Unfunded#376created byeyalroz

$0.00

Support additional arbitrary NVRTC compilation options

Unfunded#375created byeyalroz

$0.00

More launch_config_builder tweaks

Unfunded#374created byeyalroz

$0.00

Add support for nvrtcGetSupportedArchs()

Unfunded#386created byeyalroz

$0.00

Launch config build dimensions finalization broken when specifying block and grid dims

Unfunded#372created byeyalroz

$0.00

Replace constexpr const with static constexpr const wherever relevant

Unfunded#371created byeyalroz

$0.00

cuda::context::current::exists() should not throw before initialziation

Unfunded#370created byeyalroz

$0.00

Launch config builder logic for detecting missing dimension spec wrong

Unfunded#369created byeyalroz

$0.00

Primary context activation error in p2pBandwidthLatencyTest

Unfunded#368created byeyalroz

$0.00

Drop redundant scoped context setting when enqueuing a kernel launch

Unfunded#367created byeyalroz

$0.00

Spruce up the FOR_THIS_SCOPE macros

Unfunded#366created byeyalroz

$0.00

Add module_t::get_kernel() taking a string

Unfunded#365created byeyalroz

$0.00

Make `marshal()` and `render()` independent functions

Unfunded#364created byeyalroz

$0.00

Need ability to render compilation_options_t into a buffer/string

Unfunded#363created byeyalroz

$0.00

Add a `clear_language_dialect()` method to compilation options class

Unfunded#362created byeyalroz

$0.00

Add the bandwidthtest CUDA sample as an example program

Unfunded#398created byeyalroz

$0.00

Occurrences of non-::-qualified std namespace have snuck in again

Unfunded#411created byeyalroz

$0.00

v0.5.5 does not compile

Unfunded#410created byDiamonDinoia

$0.00

Move pci_id_impl.hpp into the detail folder

Unfunded#409created byeyalroz

$0.00

constness mismatch of declaration and definition of event::create with contexts

Unfunded#408created byeyalroz

$0.00

Broken stream_t lvalue assignment operator

Unfunded#407created byeyalroz

$0.00

If we have a stream_t copy ctor, we should have an assignment operator, and vice-versa

Unfunded#406created byeyalroz

$0.00

overall_dimensions_t should be constructible from a dim3 struct

Unfunded#404created byeyalroz

$0.00

Accept vectors of std::strings for setting NVRTC definition terms and values

Unfunded#403created byeyalroz

$0.00

cuda::stream_t::enqueue_t methods are non-const

Unfunded#401created byeyalroz

$0.00

Allow using cuda::wait(my_event)

Unfunded#400created byeyalroz

$0.00

Support taking an event pair for time_elapsed_between

Unfunded#399created byeyalroz

$0.00

Support getting and setting stream attributes

Unfunded#223created byeyalroz

$0.00

Some NVRTC compilation options are not (directly) supported

Unfunded#397created byeyalroz

$0.00

scoped_existence_ensurer_t fails if the driver has not been initialized

Unfunded#396created byeyalroz

$0.00

Starting CUDA profiling: initialization error

Unfunded#395created byalsam

$0.00

Don't initialize driver when getting a device name

Unfunded#394created byeyalroz

$0.00

NVTX implicitly depends on libdl (at least on Linux)

Unfunded#393created byeyalroz

$0.00

NVRTC, NVTX wrappers can't be used in different translation units

Unfunded#392created byeyalroz

$0.00

Regression: Fix for #304 broke CUDA 10.0 compatibility

Unfunded#391created byeyalroz

$0.00

Avoid memory leak in device::detail_::get_name()

Unfunded#390created byeyalroz

$0.00

Support asynchronous memory allocation

Unfunded#248created byeyalroz

$0.00

A builder-class for NVRTC programs

Unfunded#388created byeyalroz

$0.00

nvrtc.h included in wrong file

Unfunded#387created byeyalroz

$0.00

Recent activities

eyalroz added the repository

about 4 years ago

cuda-api-wrappers

$0.00

Issues

Company

Community

Support

Connect