config_schema.yml

# Copyright 2019-2021 ETH Zurich and the DaCe authors. All rights reserved.
# Schema file for DaCe Preferences

# Metadata fields for elements:
#   type: any python type (dict, list, int, bool, float, str)
#   title: short name to show in GUI
#   description: tooltip to show in GUI
#   required: required sub-fields (for dict fields)
#   default: default value. Can be platform-specific (see below)
#   default_<platformname>: default value for platform <platformname> (overrides default)
#   template_vars: template variables to include when processing (str fields only)

# Top-level element is a dictionary (record)
type: dict
title: General
description: DaCe Preferences
required:
    #############################################
    # Categories
    optimizer:
        type: dict
        title: Optimizer
        description: Preferences of the SDFG Optimizer
        required:
            autospecialize:
                type: bool
                default: false
                title: Auto-specialize symbols
                description: >
                    Automatically specialize every SDFG to the symbol values
                    at call-time. Requires all symbols to be set.

            autooptimize:
                type: bool
                default: false
                title: Run auto-optimization heuristics
                description: >
                    Automatically runs the set of optimizing transformation
                    heuristics on any program called via the Python frontend.

            autotile_size:
                type: int
                default: 128
                title: Default tile size in auto-optimization
                description: >
                    Sets the default tile size for the optimization heuristics.

            autotile_partial_parallelism:
                type: bool
                default: true
                title: Prefer partial parallelism over write-conflict tiling
                description: >
                    If true, sets the auto-optimizer to prefer extracting map
                    parallel dimensions over tiling for atomic write-conflict
                    resolution edges. This may be slower in case of small
                    parallel dimensions vs. conflicted dimensions. This
                    preference only applies to symbolic ranges or ranges over
                    the autotile_size parameter.

            transform_on_call:
                type: bool
                default: false
                title: Transform SDFGs on Invocation
                description: >
                    Automatically calls sdfg.optimize() every time a DaCe
                    program or SDFG is invoked.

            interface:
                type: str
                default: dace.transformation.optimizer.SDFGOptimizer
                title: SDFG Optimizer
                description: >
                    SDFG optimization class to import and call on
                    sdfg.optimize(). Defaults to the transformation CLI,
                    empty string or an invalid class name skips optimization.

            visualize_sdfv:
                type: bool
                default: false
                title: Visualize SDFG
                description: Open a SDFV in browser every transformation.

            save_intermediate:
                type: bool
                default: false
                title: Save intermediate SDFGs
                description: Save SDFG files after every transformation.

            automatic_simplification:
                type: bool
                default: true
                title: Automatic SDFG simplification
                description: >
                    Automatically performs SDFG simplification on programs.

            detect_control_flow:
                type: bool
                default: true
                title: Detect control flow from state transitions
                description: >
                    Attempts to infer control flow constructs "if",
                    "for" and "while" from state transitions, allowing
                    code generators to generate appropriate code.

            symbolic_positive:
                type: bool
                default: true
                title: Treat all symbolic expressions as positive
                description: >
                    Every expression in which a symbolic value appears
                    is treated as strictly positive. This is necessary
                    for certain Range evaluations using Subgraph Fusion.

            match_exception:
                type: bool
                default: false
                title: Treat exceptions in "can_be_applied" as errors
                description: >
                    When an exception is raised in a transformation "can_be_applied"
                    function, if True the exception is raised further. Otherwise
                    the exception is printed as a warning.
    compiler:
        type: dict
        title: Compiler
        description: Preferences of the compiler
        required:
            use_cache:
                type: bool
                default: false
                title: Use cache
                description: >
                    If enabled, does not recompile code generated from SDFGs
                    if shared library (.so/.dll) file is present.

            library_prefix:
                type: str
                default: ""
                default_Linux: "lib"
                default_Darwin: "lib"
                title: Library prefix
                description: Filename prefix for shared libraries.

            library_extension:
                type: str
                default: so
                default_Linux: so
                default_Windows: dll
                default_Darwin: dylib
                title: Library extension
                description: File extension of shared libraries.

            indentation_spaces:
                type: int
                default: 4
                title: Indentation width
                description: >
                    Number of spaces used when indenting generated code.

            build_type:
                type: str
                default: RelWithDebInfo
                title: Build configuration
                description: >
                    Configuration type for CMake build (can be Debug, Release,
                    RelWithDebInfo, or MinSizeRel).

            allow_shadowing:
                type: bool
                default: true
                title: Allow variable shadowing
                description: >
                   Allowing shadowing of variables in the code (reduces
                   exceptions to warnings when shadowing is encountered).

            codegen_lineinfo:
                type: bool
                default: false
                title: Annotate code generator lines
                description: >
                    Keep a source mapping between generated code and the file/line
                    of the code generator that generated it. Used for debugging
                    code generation.

            default_data_types:
                type : str
                default: Python
                title : Default data types
                description: >
                    Specify the default data types to use in generating code.
                    If "Python", Python's semantics will be followed (i.e., `float`  and `int`
                    are represented using 64 bits). If the property is set to "C", C's semantcs will be
                    used (`float` and `int` are represented using 32bits).

            unique_functions:
                type: str
                default: hash
                title: Generate unique functions
                description: >
                    Determine if and how to generate the code for equivalent NestedSDFGs:
                    "hash": hashing is used to determine if multiple NestedSDFGs with equivalent contents exist.
                    If this is the case, the code is generated only once.
                    "unique_name": the unique_name property of SDFG is used to determine if two NestedSDFGs are equal,
                    generating the code only once.  This gives more control to the programmer, that can explicitly
                    decide what NestedSDFG code can be replicated and what not.
                    "none": a separate function is code generated for each NestedSDFG

            allow_view_arguments:
                type: bool
                default: false
                title: Allow numpy views as arguments
                description: >
                    If true, allows users to call DaCe programs with NumPy views
                    (for example, "A[:, 1]" or "w.T"). As this can create pointer
                    aliasing issues with two arrays pointing to the same memory,
                    or analyzability issue with strides and alignment, this option
                    is disabled by default.

            inline_sdfgs:
                type: bool
                default: false
                title: Inline all nested SDFGs
                description: >
                    If set to true, inlines all nested SDFGs upon code generation by default.

            max_stack_array_size:
                type: int
                default: 65536
                title: Max stack-allocated array size (bytes)
                description: >
                    All stack allocated arrays (i.e. StorageType.Register) with
                    size larger than this will be allocated on the heap.

            #############################################
            # CPU compiler
            cpu:
                type: dict
                title: CPU
                description: CPU compiler preferences
                required:
                    executable:
                        type: str
                        default: ''
                        title: Compiler executable override
                        description: File path or name of compiler executable

                    args:
                        type: str
                        title: Arguments
                        description: Compiler argument flags
                        default: '-std=c++14 -fPIC -Wall -Wextra -O3 -march=native -ffast-math -Wno-unused-parameter -Wno-unused-label'
                        default_Windows: '/O2 /fp:fast /arch:AVX2 /D_USRDLL /D_WINDLL /D__restrict__=__restrict'

                    libs:
                        type: str
                        title: Additional libraries
                        description: Additional linked libraries required by target
                        default: ''

                    openmp_sections:
                        type: bool
                        default: true
                        title: Use OpenMP sections
                        description: >
                            If set to true, multiple connected components will
                            generate "#pragma omp parallel sections" code around
                            them.

            #############################################
            # GPU (CUDA/HIP) compiler
            cuda:
                type: dict
                title: GPU
                description: GPU (CUDA/HIP) compiler preferences
                required:
                    backend:
                        type: str
                        default: 'cuda'
                        title: Compilation backend
                        description: >
                            Backend to compile for ('cuda' for NVIDIA or 'hip' for AMD).

                    path:
                        type: str
                        default: ''
                        title: CUDA/HIP path override
                        description: Path to CUDA toolkit or ROCm/HIP root directory

                    args:
                        type: str
                        title: nvcc Arguments
                        description: Compiler argument flags for CUDA
                        default: '-std=c++14 -Xcompiler -fPIC -O3 -Xcompiler -march=native --use_fast_math -Xcompiler -Wno-unused-parameter'
                        default_Windows: '-std=c++14 -O3 --use_fast_math'

                    hip_args:
                        type: str
                        title: hipcc Arguments
                        description: Compiler argument flags for HIP
                        default: '-std=c++17 -fPIC -O3 -ffast-math -Wno-unused-parameter'

                    cuda_arch:
                        type: str
                        title: Additional CUDA architectures
                        description: >
                            Additional CUDA architectures (separated by commas)
                            to compile GPU code for, excluding the current
                            architecture on the compiling machine.
                        default: '35'

                    hip_arch:
                        type: str
                        title: Additional HIP architectures
                        description: >
                            Additional HIP architectures (separated by commas)
                            to compile GPU code for, excluding the current
                            architecture on the compiling machine.
                        default: '906'

                    default_block_size:
                        type: str
                        title: Default thread-block size
                        description: >
                            Default thread-block size for GPU kernels when
                            explicit GPU block maps are not defined.
                            Can be set to 'max' to maximize occupancy.
                        default: '32,1,1'

                    dynamic_map_block_size:
                        type: str
                        title: Thread-Block size for GPU_ThreadBlock_Dynamic
                        description: >
                            Thread-Block size for maps using GPU_ThreadBlock_Dynamic
                            scheduler. Can be set to 'max' to maximize occupancy.
                        default: '128,1,1'

                    dynamic_map_fine_grained:
                        type: bool
                        title: Enable fine grained load balancing for GPU_ThreadBlock_Dynamic
                        description: >
                            If true the scheduler will dynamically redistribute the
                            combined work of all threads in the warp equally across the
                            warp (fine grained). Otherwise, each warp works sequentially
                            only on its tasks (potential load imbalance).
                        default: true

                    persistent_map_SM_fraction:
                        type: float
                        title: Fraction of SMs to use for persistent GPU map
                        description: >
                            Sets the fraction of the number of SMs of the Device
                            that the GPU_Persistent map can use. Together with
                            persistent_map_occupancy this specifies the grid
                            size of the kernel being launched.
                            0.0 < persistent_map_SM_fraction <= 1.0
                            The fraction will be rounded up to the next integer
                            number of SMs. The max value of SMs that can/will
                            be used is equal to cudaDevAttrMultiProcessorCount.
                        default: 1.0

                    persistent_map_occupancy:
                        type: int
                        title: Number of blocks to launch per SM used
                        description: >
                            Sets the number of thread block to be launched per
                            SM being used. Essentially this is a simple
                            multiplier to persistent_map_SM_fraction.
                            It is up to the user to check if the resulting
                            number of thread blocks can run efficiently on the
                            GPU.
                        default: 2

                    max_concurrent_streams:
                        type: int
                        title: Concurrent execution streams
                        description: >
                            Maximum number of concurrent CUDA/HIP streams to
                            generate. Special values: -1 only uses the
                            default stream, 0 uses infinite concurrent streams.
                        default: 0

                    syncdebug:
                        type: bool
                        title: Synchronous Debugging
                        description: >
                            Enables Synchronous Debugging mode, where each library call
                            is followed by full-device synchronization and error checking.
                        default: false

                    libs:
                        type: str
                        title: Additional libraries
                        description: Additional linked libraries required by target
                        default: ''

            #############################################
            # General FPGA flags
            fpga:
                type: dict
                title: FPGA
                description: "Common preferences for FPGA compilation."
                required:

                    autobuild_bitstreams:
                        type: bool
                        default: true
                        title: Automatically build bitstreams
                        description: >
                            If set to true, CMake will automatically build missing
                            bitstreams when running an FPGA program. This can take a
                            very long time, and users might want to do this manually.
                            If set to false, the program will optimistically assume
                            that the bitstream is present in the build directory, and
                            will crash if this is not the case.

                    minimum_fifo_depth:
                        type: int
                        default: ''
                        title: Minimum depth of FIFOs
                        description: Sets the minimum depth of any generated FIFO.

                    vendor:
                        type: str
                        default: xilinx
                        title: FPGA vendor
                        description: >
                            Target Xilinx ("xilinx") or Intel ("intel_fpga") FPGAs when
                            generating code.

            #############################################
            # FPGA (Xilinx) compiler flags
            xilinx:
                type: dict
                title: Xilinx
                description: FPGA (Xilinx) compiler preferences
                required:

                    mode:
                        type: str
                        default: simulation
                        title: Compilation mode
                        description: Target of FPGA kernel build (simulation/software_emulation/hardware_emulation/hardware)

                    path:
                        type: str
                        default: ''
                        title: Vitis installation override
                        description: >
                            Path to specific Vitis/SDx/SDAccel installation to
                            use instead of just searching PATH and environment
                            variables.

                    platform:
                        type: str
                        default: xilinx_u250_xdma_201830_2
                        title: Target platform for Xilinx
                        description: Platform name of Vitis/SDx/SDAccel target.

                    frequency:
                        type: str
                        default: ''
                        title: Target frequency for Xilinx kernels
                        description: >
                            Target frequency, in MHz, when compiling kernels
                            for Xilinx. Will not necessarily be achieved in
                            practice. To enable multiple clocks, enter values
                            in the format "clock_id:frequency", with frequency
                            being specified in MHz separated by an escaped bar,
                            all enclosed in quotes. E.g. "0:250\|1:500".

                    enable_debugging:
                        type: bool
                        default: false
                        title: Enable debugging for hardware kernels
                        description: >
                            Injects debugging cores on the interfaces of the
                            kernel, allowing fine-grained debugging of hardware
                            runs at the cost of additional resources. This is
                            always enabled for emulation runs.

                    host_flags:
                        type: str
                        title: Host arguments
                        description: Extra host compiler argument flags
                        default: "-Wno-unknown-pragmas -Wno-unused-label"

                    synthesis_flags:
                        type: str
                        title: Synthesis arguments
                        description: High-level synthesis C++ flags
                        default: "-std=c++11"

                    build_flags:
                        type: str
                        title: Arguments
                        description: Kernel build C++ flags
                        default: ""

                    decouple_array_interfaces:
                        type: bool
                        default: false
                        title: Decouple array memory interfaces
                        description: >
                            If an array is both read and written, this option decouples
                            its accesses, by creatin a memory interface for reading and one
                            for writing.
                            Note that this may hide potential Read-After-Write or
                            Write-After-Read dependencies.


            #############################################
            # Intel FPGA compiler flags
            intel_fpga:
                type: dict
                title: Intel FPGA
                description: Intel FPGA compiler preferences.
                required:

                    mode:
                        type: str
                        default: emulator
                        title: Compilation mode
                        description: >
                          Target of FPGA kernel build
                          (emulator/simulator/hardware).

                    path:
                        type: str
                        default: ''
                        title: Intel FPGA OpenCL SDK installation override
                        description: >
                            Path to specific Intel FPGA OpenCL SDK installation
                            to use instead of just searching PATH and
                            environment variables.

                    board:
                        type: str
                        default: a10gx
                        title: Target FPGA board

                    enable_debugging:
                        type: bool
                        default: false
                        title: Enable debugging for hardware kernels
                        description: Injects debugging cores where available.

                    host_flags:
                        type: str
                        title: Host arguments
                        description: Extra host compiler argument flags
                        default: "-Wno-unknown-pragmas"

                    kernel_flags:
                        type: str
                        title: Kernel flags
                        description: High-level synthesis C++ flags
                        default: "-fp-relaxed -cl-no-signed-zeros -cl-fast-relaxed-math -cl-single-precision-constant -no-interleaving=default"

            #############################################
            # RTL (SystemVerilog) compiler
            rtl:
                type: dict
                title: RTL
                description: RTL (SystemVerilog) compiler preferences
                required:
                    verbose:
                        type: bool
                        default: false
                        title: Verbose Build & Execution Output
                        description: Output full build and execution (incl internal state) log.
                    verilator_flags:
                        type: str
                        default: ''
                        title: Additional Verilator Arguments
                        description: Additional arguments feed to verilator.
                    verilator_lint_warnings:
                        type: bool
                        default: true
                        title: Verilator Lint Warnings
                        description: Enable/Disable detailed SV lint checker output.
                    verilator_enable_debug:
                        type: bool
                        default: false
                        title: Verilator Enable Debug
                        description: Enable/disable verbose internal state debug output.

            #############################################
            # MPI compiler
            mpi:
                type: dict
                title: MPI
                description: MPI compiler preferences
                required:
                    executable:
                        type: str
                        default: ''
                        title: Compiler executable override
                        description: File path or name of compiler executable

            #############################################
            # Linker
            linker:
                type: dict
                title: Linker
                description: Linker preferences
                required:
                    executable:
                        type: str
                        default: ''
                        title: Linker executable override
                        description: File path or name of linker executable

                    args:
                        type: str
                        title: Arguments
                        description: Linker argument flags
                        # Tell linker to use rpath instead of runpath. Intel
                        # FPGA programs fail to find certain libraries at
                        # runtime with runpath.
                        default: '-Wl,--disable-new-dtags'
                        default_Darwin: ''
                        default_Windows: ''


    execution:
        type: dict
        title: Execution
        description: Binary execution preferences
        required:
            general:
                type: dict
                title: General
                description: General execution preferences
                required:
                    host:
                        type: str
                        default: localhost
                        title: Host
                        description: Hostname to use for execution

                    workdir:
                        type: str
                        default: '/tmp/'
                        title: Working directory
                        description: Working directory on the remote host

                    check_args:
                        type: bool
                        default: true
                        title: Check arguments
                        description: >
                            Do strict verification that arguments passed when
                            calling a DaCe program match the expected dtypes.

                    execcmd:
                        type: str
                        title: Command
                        description: >
                            Command to use to execute ${command} on ${host}
                        default: 'ssh ${host} ${command}'
                        template_vars:
                            - host
                            - command

                    copycmd_r2l:
                        type: str
                        default: 'scp ${host}:${srcfile} ${dstfile}'
                        title: "Remote->Local copy command"
                        description: >
                            Command to use to copy ${srcfile} on ${host} to
                            the local ${dstfile}.
                        template_vars:
                            - host
                            - srcfile
                            - dstfile

                    copycmd_l2r:
                        type: str
                        default: "scp ${srcfile} ${host}:${dstfile}"
                        title: "Local->Remote copy command"
                        description: >
                            Command to use to copy the local ${srcfile} to the
                            remote ${dstfile}.
                        template_vars:
                            - host
                            - srcfile
                            - dstfile

                    repetitions:
                        type: int
                        default: 5
                        title: "Repetitions per Run"
                        description: >
                            Number of repetitions to run for each click of the
                            Run button (median value will be reported in the
                            performance chart).
            mpi:
                type: dict
                title: MPI
                description: MPI execution preferences
                required:
                    mpiexec:
                        type: str
                        default: 'mpirun -n ${num_procs} ${command}'
                        title: mpirun command
                        description: >
                            Command to use to execute MPI job ${command} with
                            ${num_procs} processes.
                        template_vars:
                            - num_procs
                            - command

                    num_procs:
                        type: int
                        default: 4
                        title: Number of processes
                        description: Number of MPI processes to use

    instrumentation:
        type: dict
        title: Instrumentation
        description: Instrumentation preferences
        required:
            report_each_invocation:
                type: bool
                title: Save report for each invocation
                default: true
                description: >
                    Save an instrumentation report file for each invocation of
                    the SDFG, rather than one report that spans from SDFG
                    initialization to finalization.

            papi:
                type: dict
                title: PAPI
                description: PAPI configuration
                required:
                    default_counters:
                        type: str
                        title: Default PAPI counters
                        default: "['PAPI_TOT_INS', 'PAPI_TOT_CYC', 'PAPI_L2_TCM', 'PAPI_L3_TCM']"
                        description: >
                            Sets the default PAPI counter list, formatted as
                            a Python list of strings.
                    overhead_compensation:
                        type: bool
                        title: Compensate Overhead
                        default: true
                        description: >
                            Subtracts the minimum measured overhead from every measurement.
                    vectorization_analysis:
                        type: bool
                        title: Enable vectorization check
                        default: false
                        description: >
                            Enables analysis of gcc vectorization information. Only gcc/g++ is supported.

            print_fpga_runtime:
                type: bool
                default: false
                title: Print FPGA runtime
                description: Prints the runtime of instrumented FPGA kernel states to standard output.

    #############################################
    # Python frontend settings

    frontend:
        type: dict
        title: Frontend
        description: Python frontend preferences
        required:
            cache_size:
                type: int
                title: Program cache size
                default: 32
                description: >
                    The number of compiled programs to cache (based on argument
                    types, closure constants, and closure array types) to avoid
                    reparsing/compiling when calling a @dace.program or method.

            implicit_recursion_depth:
                type: int
                title: Auto-parsing recursion depth
                default: 64
                description: >
                    The maximum call-stack depth allowed when automatically
                    parsing called dace functions or methods.

            raise_nested_parsing_errors:
                type: bool
                title: Raise nested parsing errors
                default: false
                description: >
                    Raise all errors out of nested function parsing contexts
                    instead of trying to create a callback implicitly.

            verbose_errors:
                type: bool
                title: Show preprocessed AST on parsing errors
                default: false
                description: >
                    Prints out the preprocessed unparsed AST in case of a parsing
                    error.

            preprocessing_passes:
                type: int
                title: Number of preprocessing passes on Python code
                default: 5
                description: >
                    Number of times to run the Python preprocessing passes (e.g., constant
                    folding) on the input code. Set to zero to disable preprocessing
                    optimizations, set to -1 to run until the code has not changed.

            dont_fuse_callbacks:
                type: bool
                title: Do not fuse callbacks
                default: false
                description: >
                    Stricter mode of operation where callbacks into Python don't participate
                    in state fusion transformations.

            unroll_threshold:
                type: int
                title: Automatic unroll loop size threshold
                default: -1
                description: >
                    Threshold for automatic loop unrolling of any generator (e.g., including ``range``) with a
                    compile-time size. A value of -1 (default) means not to unroll any loop automatically, a value of
                    0 means unrolling every loop, and a value above zero sets a size threshold beyond which a
                    constant-sized loop will not be automatically unrolled.



    #############################################
    # General settings

    debugprint:
        type: bool
        default: false
        title: Debug printing
        description: Enable verbose printouts.

    progress:
        type: bool
        default: true
        title: Progress reports
        description: Enable progress report printouts.

    cache:
        type: str
        default: name
        title: Naming of cache entry
        description: >
            Determine the name of the generated dacecache folder.
            "name" uses the name of the SDFG directly, causing it to be
            overridden by other programs using the same SDFG name.
            "hash" uses a mangled name based on the hash of the SDFG, such that
            any change to the SDFG will generate a different cache folder.
            "unique" uses a name based on the currently running Python process
            at code generation time, such that no caching or clashes can happen
            between different processes or subsequent invocations of Python.
            "single" uses a single cache folder for all SDFGs, saving space and
            potentially build time, but disallows executing SDFGs in parallel
            and caching of more than one simultaneous SDFG.

    store_history:
        type: bool
        default: true
        title: Store SDFG transformation history
        description: Store the history of transformations on the SDFG file.

    default_build_folder:
        type: str
        default: .dacecache
        title: Default SDFG build folder
        description: >
            Default folder in which compiled DaCe programs and SDFGs are stored.
            Can either be a relative path (by default) or absolute.

    profiling:
        type: bool
        default: false
        title: Profiling
        description: Enable profiling support.

    profiling_status:
        type: bool
        default: true
        title: Status bar for profiling
        description: >
            Enable tqdm status bar while profiling. If tqdm is not installed
            a warning will appear. To disable this feature (and the warning) set
            this option to false.

    treps:
        type: int
        default: 100
        title: Profiling Repetitions
        description: Number of times to run program for profiling.

    #############################################
    # Experimental features

    experimental:
        type: dict
        title: Experimental
        description: Experimental features
        required:
            validate_undefs:
                type: bool
                default: false
                title: Undefined Symbol Check
                description: >
                    Check for undefined symbols in memlets during SDFG validation.

    #############################################
    # Features for unit testing

    testing:
        type: dict
        title: Testing
        description: Unit testing settings
        required:
            serialization:
                type: bool
                default: false
                title: Test Serialization on validation
                description: >
                    Before generating code, verify that a serialization/deserialization loop
                    generates the same SDFG.

            deserialize_exception:
                type: bool
                default: false
                title: Treat exceptions in deserialization as errors
                description: >
                    When an exception is raised in a deserialization process (e.g., due to missing library node),
                    by default a warning is issued. If this setting is True, the exception will be raised as-is.

    #############################################
    # DaCe library settings

    library:
        type: dict
        title: Library
        description: Settings for handling the use of DaCe libraries.
        required:
            blas:
                type: dict
                title: BLAS
                description: Built-in BLAS DaCe library.
                required:
                    default_implementation:
                        type: str
                        default: pure
                        description: Default implementation BLAS library nodes.
                    override:
                        type: bool
                        default: false
                        description: >
                            Force the default implementation, even if an
                            implementation has been explicitly set on a node.
                    fpga:
                        type: dict
                        title: FPGA
                        description: FPGA-specific BLAS options.
                        required:
                            default_stream_depth:
                                type: int
                                default: 32
                                title: Default FPGA stream depth
                                description: >
                                    Default FPGA stream depth used in the BLAS
                                    library nodes and the corresponding
                                    streaming transformations
            lapack:
                type: dict
                title: LAPACK
                description: Built-in LAPACK DaCe library.
                required:
                    default_implementation:
                        type: str
                        default: OpenBLAS
                        description: Default implementation of LAPACK library nodes.
                    override:
                        type: bool
                        default: false
                        description: >
                            Force the default implementation, even if an
                            implementation has been explicitly set on a node.
            linalg:
                type: dict
                title: linalg
                description: Built-in NumPy linalg DaCe library.
                required:
                    default_implementation:
                        type: str
                        default: OpenBLAS
                        description: Default implementation of linalg library nodes.
                    override:
                        type: bool
                        default: false
                        description: >
                            Force the default implementation, even if an
                            implementation has been explicitly set on a node.