Trignometric functions on GPU

Specific Domains GPU
1

I tried executing a simple function involving sin and cos as below:

using CUDAdrv,CUDAnative,CuArrays
sine(x::Float32) = sin(x)
cosine(x::Float32) = cos(x)
CuArrays.@cufunc sine(x::Real) =  sin(x)
CuArrays.@cufunc cosine(x::Real) =  cos(x)

function update(cedg,b,s,c)
    index = (blockIdx().x - 1) * blockDim().x + threadIdx().x
    stride = blockDim().x * gridDim().x
    for i = index:stride:length(cedg)
      if(cedg[i]!=0)
        @inbounds s+=cedg[i]*sine(b[i])
        @inbounds c+=cedg[i]*cosine(b[i])
      end
    end
    return nothing
end

On trying to run it,

cedg=CuArrays.rand(10000)
b=CuArrays.rand(10000)
s=0f0
c=0f0
numblocks = ceil(Int, 10000/256)
@cuda threads=256 blocks=numblocks update!(cedg,b,s,c)

I get the following error

┌ Warning: calls to Base intrinsics might be GPU incompatible
│   exception = (CUDAnative.MethodSubstitutionWarning(sin(x::T) where T<:Union{Float32, Float64} in Base.Math at special/trig.jl:30, sin(x::Float32) in CUDAnative at /root/.julia/packages/CUDAnative/2WQzk/src/device/cuda/math.jl:13), Base.StackTraces.StackFrame[sin at trig.jl:30, update! at In[30]:2])
└ @ CUDAnative /root/.julia/packages/CUDAnative/2WQzk/src/compiler/irgen.jl:116
┌ Warning: calls to Base intrinsics might be GPU incompatible
│   exception = (CUDAnative.MethodSubstitutionWarning(cos(x::T) where T<:Union{Float32, Float64} in Base.Math at special/trig.jl:100, cos(x::Float32) in CUDAnative at /root/.julia/packages/CUDAnative/2WQzk/src/device/cuda/math.jl:6), Base.StackTraces.StackFrame[cos at trig.jl:100, update! at In[30]:2])
└ @ CUDAnative /root/.julia/packages/CUDAnative/2WQzk/src/compiler/irgen.jl:116
LLVM error: Cannot select: 0xc22bf70: i64,glue = sube Constant:i64<0>, 0xc22bea0, 0xc22bf08:1
  0x514c820: i64 = Constant<0>
  0xc22bea0: i64 = add 0x514c618, 0xc22be38
    0x514c618: i64 = add 0x514cb60, 0x514c958
      0x514cb60: i64 = mul 0xc22b000, 0x514c1a0
        0xc22b000: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %13
          0x514cf70: i64 = Register %13
        0x514c1a0: i64 = or 0x514c2d8, Constant:i64<4503599627370496>
          0x514c2d8: i64 = and 0x514c410, Constant:i64<4503599627370495>
            0x514c410: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %0
              0x514cd00: i64 = Register %0
            0x514c068: i64 = Constant<4503599627370495>
          0x514c8f0: i64 = Constant<4503599627370496>
      0x514c958: i64 = mulhu 0xc22b2d8, 0x514c1a0
        0xc22b2d8: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %14
          0x514cc30: i64 = Register %14
        0x514c1a0: i64 = or 0x514c2d8, Constant:i64<4503599627370496>
          0x514c2d8: i64 = and 0x514c410, Constant:i64<4503599627370495>
            0x514c410: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %0
              0x514cd00: i64 = Register %0
            0x514c068: i64 = Constant<4503599627370495>
          0x514c8f0: i64 = Constant<4503599627370496>
    0xc22be38: i64 = select 0xc22bdd0, Constant:i64<1>, 0x514c4e0
      0xc22bdd0: i1 = setcc 0xc22bc98, 0xc22bc30, setult:ch
        0xc22bc98: i64 = add 0xc22b1a0, 0xc22bc30
          0xc22b1a0: i64 = mul 0xc22b2d8, 0x514c1a0
            0xc22b2d8: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %14
              0x514cc30: i64 = Register %14
            0x514c1a0: i64 = or 0x514c2d8, Constant:i64<4503599627370496>
              0x514c2d8: i64 = and 0x514c410, Constant:i64<4503599627370495>
                0x514c410: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %0
                  0x514cd00: i64 = Register %0
                0x514c068: i64 = Constant<4503599627370495>
              0x514c8f0: i64 = Constant<4503599627370496>
          0xc22bc30: i64 = mulhu 0x514c7b8, 0x514c1a0
            0x514c7b8: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %15
              0x514c3a8: i64 = Register %15
            0x514c1a0: i64 = or 0x514c2d8, Constant:i64<4503599627370496>
              0x514c2d8: i64 = and 0x514c410, Constant:i64<4503599627370495>
                0x514c410: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %0
                  0x514cd00: i64 = Register %0
                0x514c068: i64 = Constant<4503599627370495>
              0x514c8f0: i64 = Constant<4503599627370496>
        0xc22bc30: i64 = mulhu 0x514c7b8, 0x514c1a0
          0x514c7b8: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %15
            0x514c3a8: i64 = Register %15
          0x514c1a0: i64 = or 0x514c2d8, Constant:i64<4503599627370496>
            0x514c2d8: i64 = and 0x514c410, Constant:i64<4503599627370495>
              0x514c410: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %0
                0x514cd00: i64 = Register %0
              0x514c068: i64 = Constant<4503599627370495>
            0x514c8f0: i64 = Constant<4503599627370496>
      0x514c478: i64 = Constant<1>
      0x514c4e0: i64 = zero_extend 0xc22bd00
        0xc22bd00: i1 = setcc 0xc22bc98, 0xc22b1a0, setult:ch
          0xc22bc98: i64 = add 0xc22b1a0, 0xc22bc30
            0xc22b1a0: i64 = mul 0xc22b2d8, 0x514c1a0
              0xc22b2d8: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %14
                0x514cc30: i64 = Register %14
              0x514c1a0: i64 = or 0x514c2d8, Constant:i64<4503599627370496>
                0x514c2d8: i64 = and 0x514c410, Constant:i64<4503599627370495>
                  0x514c410: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %0

                  0x514c068: i64 = Constant<4503599627370495>
                0x514c8f0: i64 = Constant<4503599627370496>
            0xc22bc30: i64 = mulhu 0x514c7b8, 0x514c1a0
              0x514c7b8: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %15
                0x514c3a8: i64 = Register %15
              0x514c1a0: i64 = or 0x514c2d8, Constant:i64<4503599627370496>
                0x514c2d8: i64 = and 0x514c410, Constant:i64<4503599627370495>
                  0x514c410: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %0

                  0x514c068: i64 = Constant<4503599627370495>
                0x514c8f0: i64 = Constant<4503599627370496>
          0xc22b1a0: i64 = mul 0xc22b2d8, 0x514c1a0
            0xc22b2d8: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %14
              0x514cc30: i64 = Register %14
            0x514c1a0: i64 = or 0x514c2d8, Constant:i64<4503599627370496>
              0x514c2d8: i64 = and 0x514c410, Constant:i64<4503599627370495>
                0x514c410: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %0
                  0x514cd00: i64 = Register %0
                0x514c068: i64 = Constant<4503599627370495>
              0x514c8f0: i64 = Constant<4503599627370496>
  0xc22bf08: i64,glue = subc Constant:i64<0>, 0xc22bc98
    0x514c820: i64 = Constant<0>
    0xc22bc98: i64 = add 0xc22b1a0, 0xc22bc30
      0xc22b1a0: i64 = mul 0xc22b2d8, 0x514c1a0
        0xc22b2d8: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %14
          0x514cc30: i64 = Register %14
        0x514c1a0: i64 = or 0x514c2d8, Constant:i64<4503599627370496>
          0x514c2d8: i64 = and 0x514c410, Constant:i64<4503599627370495>
            0x514c410: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %0
              0x514cd00: i64 = Register %0
            0x514c068: i64 = Constant<4503599627370495>
          0x514c8f0: i64 = Constant<4503599627370496>
      0xc22bc30: i64 = mulhu 0x514c7b8, 0x514c1a0
        0x514c7b8: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %15
          0x514c3a8: i64 = Register %15
        0x514c1a0: i64 = or 0x514c2d8, Constant:i64<4503599627370496>
          0x514c2d8: i64 = and 0x514c410, Constant:i64<4503599627370495>
            0x514c410: i64,ch = CopyFromReg 0x9fa5040, Register:i64 %0
              0x514cd00: i64 = Register %0
            0x514c068: i64 = Constant<4503599627370495>
          0x514c8f0: i64 = Constant<4503599627370496>
In function: julia_paynehanek_19687

Stacktrace:
 [1] handle_error(::Cstring) at /root/.julia/packages/LLVM/ICZSf/src/core/context.jl:103
 [2] macro expansion at /root/.julia/packages/LLVM/ICZSf/src/base.jl:18 [inlined]
 [3] LLVMTargetMachineEmitToMemoryBuffer at /root/.julia/packages/LLVM/ICZSf/lib/6.0/libLLVM_h.jl:2726 [inlined]
 [4] emit(::LLVM.TargetMachine, ::LLVM.Module, ::LLVM.API.LLVMCodeGenFileType) at /root/.julia/packages/LLVM/ICZSf/src/targetmachine.jl:42
 [5] mcgen(::CUDAnative.CompilerJob, ::LLVM.Module, ::LLVM.Function) at /root/.julia/packages/CUDAnative/2WQzk/src/compiler/mcgen.jl:87
 [6] macro expansion at /root/.julia/packages/TimerOutputs/7Id5J/src/TimerOutput.jl:214 [inlined]
 [7] macro expansion at /root/.julia/packages/CUDAnative/2WQzk/src/compiler/driver.jl:204 [inlined]
 [8] macro expansion at /root/.julia/packages/TimerOutputs/7Id5J/src/TimerOutput.jl:214 [inlined]
 [9] #codegen#152(::Bool, ::Bool, ::Bool, ::Bool, ::Bool, ::typeof(CUDAnative.codegen), ::Symbol, ::CUDAnative.CompilerJob) at /root/.julia/packages/CUDAnative/2WQzk/src/compiler/driver.jl:201
 [10] #codegen at ./none:0 [inlined]
 [11] #compile#151(::Bool, ::Bool, ::Bool, ::Bool, ::Bool, ::typeof(CUDAnative.compile), ::Symbol, ::CUDAnative.CompilerJob) at /root/.julia/packages/CUDAnative/2WQzk/src/compiler/driver.jl:47
 [12] #compile at ./none:0 [inlined]
 [13] #compile#150 at /root/.julia/packages/CUDAnative/2WQzk/src/compiler/driver.jl:28 [inlined]
 [14] #compile at ./none:0 [inlined] (repeats 2 times)
 [15] macro expansion at /root/.julia/packages/CUDAnative/2WQzk/src/execution.jl:403 [inlined]
 [16] #cufunction#194(::Nothing, ::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}, ::typeof(cufunction), ::typeof(update!), ::Type{Tuple{CuDeviceArray{Float32,1,CUDAnative.AS.Global},CuDeviceArray{Float32,1,CUDAnative.AS.Global},Float32,Float32}}) at /root/.julia/packages/CUDAnative/2WQzk/src/execution.jl:368
 [17] cufunction(::Function, ::Type) at /root/.julia/packages/CUDAnative/2WQzk/src/execution.jl:368
 [18] top-level scope at /root/.julia/packages/CUDAnative/2WQzk/src/execution.jl:176
 [19] top-level scope at gcutils.jl:91
 [20] top-level scope at /root/.julia/packages/CUDAnative/2WQzk/src/execution.jl:173
 [21] top-level scope at In[60]:1

How do I fix this?

2

Hello,

this is what I’ve found on this issue (see the bottom of the page):

Many mathematical functions are provided by the libdevice library, and are wrapped by CUDAnative.jl. These functions implement interfaces that are similar to existing functions in Base , albeit often with support for fewer types.

and

in kernel code, call CUDAnative.sin instead of plain sin

1 Like
3

I tried what you suggested,

function update(cedg,b,s,c)
    index = (blockIdx().x - 1) * blockDim().x + threadIdx().x
    stride = blockDim().x * gridDim().x
    for i = index:stride:length(cedg)
      if(cedg[i]!=0)
        @inbounds s+=cedg[i]*CUDAnative.sin(b[i])
        @inbounds c+=cedg[i]*CUDAnative.cos(b[i])
      end
    end
    return nothing
end

I still get the following error:

Warning: calls to Base intrinsics might be GPU incompatible
│   exception = (CUDAnative.MethodSubstitutionWarning(sin(x::T) where T<:Union{Float32, Float64} in Base.Math at special/trig.jl:30, sin(x::Float32) in CUDAnative at /root/.julia/packages/CUDAnative/2WQzk/src/device/cuda/math.jl:13), Base.StackTraces.StackFrame[sin at trig.jl:30, update! at In[30]:2])
└ @ CUDAnative /root/.julia/packages/CUDAnative/2WQzk/src/compiler/irgen.jl:116
┌ Warning: calls to Base intrinsics might be GPU incompatible
│   exception = (CUDAnative.MethodSubstitutionWarning(cos(x::T) where T<:Union{Float32, Float64} in Base.Math at special/trig.jl:100, cos(x::Float32) in CUDAnative at /root/.julia/packages/CUDAnative/2WQzk/src/device/cuda/math.jl:6), Base.StackTraces.StackFrame[cos at trig.jl:100, update! at In[30]:2])
└ @ CUDAnative /root/.julia/packages/CUDAnative/2WQzk/src/compiler/irgen.jl:116
LLVM error: Cannot select: 0xe206f70: i64,glue = sube Constant:i64<0>, 0xe206ea0, 0xe206f08:1
  0xc228820: i64 = Constant<0>
  0xe206ea0: i64 = add 0xc228618, 0xe206e38
    0xc228618: i64 = add 0xc228b60, 0xc228958
      0xc228b60: i64 = mul 0xe206000, 0xc2281a0
        0xe206000: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %13
          0xc228f70: i64 = Register %13
        0xc2281a0: i64 = or 0xc2282d8, Constant:i64<4503599627370496>
          0xc2282d8: i64 = and 0xc228410, Constant:i64<4503599627370495>
            0xc228410: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %0
              0xc228d00: i64 = Register %0
            0xc228068: i64 = Constant<4503599627370495>
          0xc2288f0: i64 = Constant<4503599627370496>
      0xc228958: i64 = mulhu 0xe2062d8, 0xc2281a0
        0xe2062d8: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %14
          0xc228c30: i64 = Register %14
        0xc2281a0: i64 = or 0xc2282d8, Constant:i64<4503599627370496>
          0xc2282d8: i64 = and 0xc228410, Constant:i64<4503599627370495>
            0xc228410: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %0
              0xc228d00: i64 = Register %0
            0xc228068: i64 = Constant<4503599627370495>
          0xc2288f0: i64 = Constant<4503599627370496>
    0xe206e38: i64 = select 0xe206dd0, Constant:i64<1>, 0xc2284e0
      0xe206dd0: i1 = setcc 0xe206c98, 0xe206c30, setult:ch
        0xe206c98: i64 = add 0xe2061a0, 0xe206c30
          0xe2061a0: i64 = mul 0xe2062d8, 0xc2281a0
            0xe2062d8: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %14
              0xc228c30: i64 = Register %14
            0xc2281a0: i64 = or 0xc2282d8, Constant:i64<4503599627370496>
              0xc2282d8: i64 = and 0xc228410, Constant:i64<4503599627370495>
                0xc228410: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %0
                  0xc228d00: i64 = Register %0
                0xc228068: i64 = Constant<4503599627370495>
              0xc2288f0: i64 = Constant<4503599627370496>
          0xe206c30: i64 = mulhu 0xc2287b8, 0xc2281a0
            0xc2287b8: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %15
              0xc2283a8: i64 = Register %15
            0xc2281a0: i64 = or 0xc2282d8, Constant:i64<4503599627370496>
              0xc2282d8: i64 = and 0xc228410, Constant:i64<4503599627370495>
                0xc228410: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %0
                  0xc228d00: i64 = Register %0
                0xc228068: i64 = Constant<4503599627370495>
              0xc2288f0: i64 = Constant<4503599627370496>
        0xe206c30: i64 = mulhu 0xc2287b8, 0xc2281a0
          0xc2287b8: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %15
            0xc2283a8: i64 = Register %15
          0xc2281a0: i64 = or 0xc2282d8, Constant:i64<4503599627370496>
            0xc2282d8: i64 = and 0xc228410, Constant:i64<4503599627370495>
              0xc228410: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %0
                0xc228d00: i64 = Register %0
              0xc228068: i64 = Constant<4503599627370495>
            0xc2288f0: i64 = Constant<4503599627370496>
      0xc228478: i64 = Constant<1>
      0xc2284e0: i64 = zero_extend 0xe206d00
        0xe206d00: i1 = setcc 0xe206c98, 0xe2061a0, setult:ch
          0xe206c98: i64 = add 0xe2061a0, 0xe206c30
            0xe2061a0: i64 = mul 0xe2062d8, 0xc2281a0
              0xe2062d8: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %14
                0xc228c30: i64 = Register %14
              0xc2281a0: i64 = or 0xc2282d8, Constant:i64<4503599627370496>
                0xc2282d8: i64 = and 0xc228410, Constant:i64<4503599627370495>
                  0xc228410: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %0

                  0xc228068: i64 = Constant<4503599627370495>
                0xc2288f0: i64 = Constant<4503599627370496>
            0xe206c30: i64 = mulhu 0xc2287b8, 0xc2281a0
              0xc2287b8: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %15
                0xc2283a8: i64 = Register %15
              0xc2281a0: i64 = or 0xc2282d8, Constant:i64<4503599627370496>
                0xc2282d8: i64 = and 0xc228410, Constant:i64<4503599627370495>
                  0xc228410: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %0

                  0xc228068: i64 = Constant<4503599627370495>
                0xc2288f0: i64 = Constant<4503599627370496>
          0xe2061a0: i64 = mul 0xe2062d8, 0xc2281a0
            0xe2062d8: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %14
              0xc228c30: i64 = Register %14
            0xc2281a0: i64 = or 0xc2282d8, Constant:i64<4503599627370496>
              0xc2282d8: i64 = and 0xc228410, Constant:i64<4503599627370495>
                0xc228410: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %0
                  0xc228d00: i64 = Register %0
                0xc228068: i64 = Constant<4503599627370495>
              0xc2288f0: i64 = Constant<4503599627370496>
  0xe206f08: i64,glue = subc Constant:i64<0>, 0xe206c98
    0xc228820: i64 = Constant<0>
    0xe206c98: i64 = add 0xe2061a0, 0xe206c30
      0xe2061a0: i64 = mul 0xe2062d8, 0xc2281a0
        0xe2062d8: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %14
          0xc228c30: i64 = Register %14
        0xc2281a0: i64 = or 0xc2282d8, Constant:i64<4503599627370496>
          0xc2282d8: i64 = and 0xc228410, Constant:i64<4503599627370495>
            0xc228410: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %0
              0xc228d00: i64 = Register %0
            0xc228068: i64 = Constant<4503599627370495>
          0xc2288f0: i64 = Constant<4503599627370496>
      0xe206c30: i64 = mulhu 0xc2287b8, 0xc2281a0
        0xc2287b8: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %15
          0xc2283a8: i64 = Register %15
        0xc2281a0: i64 = or 0xc2282d8, Constant:i64<4503599627370496>
          0xc2282d8: i64 = and 0xc228410, Constant:i64<4503599627370495>
            0xc228410: i64,ch = CopyFromReg 0x2d09c40, Register:i64 %0
              0xc228d00: i64 = Register %0
            0xc228068: i64 = Constant<4503599627370495>
          0xc2288f0: i64 = Constant<4503599627370496>
In function: julia_paynehanek_19754

Stacktrace:
 [1] handle_error(::Cstring) at /root/.julia/packages/LLVM/ICZSf/src/core/context.jl:103
 [2] macro expansion at /root/.julia/packages/LLVM/ICZSf/src/base.jl:18 [inlined]
 [3] LLVMTargetMachineEmitToMemoryBuffer at /root/.julia/packages/LLVM/ICZSf/lib/6.0/libLLVM_h.jl:2726 [inlined]
 [4] emit(::LLVM.TargetMachine, ::LLVM.Module, ::LLVM.API.LLVMCodeGenFileType) at /root/.julia/packages/LLVM/ICZSf/src/targetmachine.jl:42
 [5] mcgen(::CUDAnative.CompilerJob, ::LLVM.Module, ::LLVM.Function) at /root/.julia/packages/CUDAnative/2WQzk/src/compiler/mcgen.jl:87
 [6] macro expansion at /root/.julia/packages/TimerOutputs/7Id5J/src/TimerOutput.jl:214 [inlined]
 [7] macro expansion at /root/.julia/packages/CUDAnative/2WQzk/src/compiler/driver.jl:204 [inlined]
 [8] macro expansion at /root/.julia/packages/TimerOutputs/7Id5J/src/TimerOutput.jl:214 [inlined]
 [9] #codegen#152(::Bool, ::Bool, ::Bool, ::Bool, ::Bool, ::typeof(CUDAnative.codegen), ::Symbol, ::CUDAnative.CompilerJob) at /root/.julia/packages/CUDAnative/2WQzk/src/compiler/driver.jl:201
 [10] #codegen at ./none:0 [inlined]
 [11] #compile#151(::Bool, ::Bool, ::Bool, ::Bool, ::Bool, ::typeof(CUDAnative.compile), ::Symbol, ::CUDAnative.CompilerJob) at /root/.julia/packages/CUDAnative/2WQzk/src/compiler/driver.jl:47
 [12] #compile at ./none:0 [inlined]
 [13] #compile#150 at /root/.julia/packages/CUDAnative/2WQzk/src/compiler/driver.jl:28 [inlined]
 [14] #compile at ./none:0 [inlined] (repeats 2 times)
 [15] macro expansion at /root/.julia/packages/CUDAnative/2WQzk/src/execution.jl:403 [inlined]
 [16] #cufunction#194(::Nothing, ::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}, ::typeof(cufunction), ::typeof(update!), ::Type{Tuple{CuDeviceArray{Float32,1,CUDAnative.AS.Global},CuDeviceArray{Float32,1,CUDAnative.AS.Global},Float32,Float32}}) at /root/.julia/packages/CUDAnative/2WQzk/src/execution.jl:368
 [17] cufunction(::Function, ::Type) at /root/.julia/packages/CUDAnative/2WQzk/src/execution.jl:368
 [18] top-level scope at /root/.julia/packages/CUDAnative/2WQzk/src/execution.jl:176
 [19] top-level scope at gcutils.jl:91
 [20] top-level scope at /root/.julia/packages/CUDAnative/2WQzk/src/execution.jl:173
 [21] top-level scope at In[72]:1
4

Those error messages seem to be identical. Did you update the function definition?

5

You were right. For some reason, the redefinition of the function was not getting affected. And strangely now, both the versions of the function work! I have another question though, I initialise s and c outside of the function, but on executing the function their values remain unaffected. Why is this so?

1 Like