add static compilation example

C_FHist/C_FHist.jl

@@ -0,0 +1,14 @@
+module C_FHist
+
+using FHist: Hist1D, _fast_bincounts_1d!
+
+Base.@ccallable function hist1d(input::Ptr{Cdouble}, Ninput::Clong, bincounts::Ptr{Cdouble}, Nbincounts::Clong, start::Cdouble, step::Cdouble, stop::Cdouble)::Cvoid
+    np_input = unsafe_wrap(Array{Float64}, input, Ninput, own=false)
+    np_bincounts = unsafe_wrap(Array{Float64}, bincounts, Nbincounts, own=false)
+    binedges = start:step:stop
+    h = Hist1D(; bincounts=np_bincounts, binedges)
+    _fast_bincounts_1d!(h, np_input, binedges)
+    return nothing
+end
+
+end

C_FHist/README.md

@@ -0,0 +1,96 @@
+## Setup Julia nightly:
+Follow instruction at https://julialang.org/install/
+```
+curl -fsSL https://install.julialang.org | sh
+juliaup add 1.12
+```
+
+## How to test
+
+```bash
+>julia +1.12 --project=.. -e "using Pkg; Pkg.update()"
+# the the -beta4+0 is the latest version at the time of writing, change it to the latest version
+# the x64.linux.gnu need to be changed to your platform
+>julia +1.12 --project=.. ~/.julia/juliaup/julia-1.12.0-beta4+0.x64.linux.gnu/share/julia/juliac.jl --output-lib libfhistjl.so --compile-ccallable --experimental --trim C_FHist.jl
+
+>pixi ls -x
+Package  Version  Build            Size     Kind   Source
+numpy    2.3.1    py313h41a2e72_0  6.3 MiB  conda  numpy
+
+> pixi run python test.py
+```
+
+## Results on macOS ARM (Mac Mini M4)
+```
+> julia +1.12 -e "using InteractiveUtils; versioninfo()"
+Julia Version 1.12.0-beta4
+Commit 600ac61d3d2 (2025-06-05 07:03 UTC)
+Build Info:
+  Official https://julialang.org release
+Platform Info:
+  OS: macOS (arm64-apple-darwin24.0.0)
+  CPU: 10 × Apple M4
+  WORD_SIZE: 64
+  LLVM: libLLVM-18.1.7 (ORCJIT, apple-m1)
+  GC: Built with stock GC
+
+> pixi run python test.py
+=====================================
+Input size: 1000
+All close: True
+Numpy    time (ms): 0.021641701459884644
+FHist.jl time (ms): 0.003958120942115784
+=====================================
+Input size: 10000
+All close: True
+Numpy    time (ms): 0.05824156105518341
+FHist.jl time (ms): 0.011725164949893951
+=====================================
+Input size: 100000
+All close: True
+Numpy    time (ms): 0.438716821372509
+FHist.jl time (ms): 0.0859750434756279
+=====================================
+Input size: 1000000
+All close: True
+Numpy    time (ms): 3.94724179059267
+FHist.jl time (ms): 0.829133577644825
+```
+
+## Results on Linux x86_64 (Ryzen 9 3900X)
+
+```
+> julia +1.12 -e "using InteractiveUtils; versioninfo()"
+Julia Version 1.12.0-beta4
+Commit 600ac61d3d2 (2025-06-05 07:03 UTC)
+Build Info:
+  Official https://julialang.org release
+Platform Info:
+  OS: Linux (x86_64-linux-gnu)
+  CPU: 24 × AMD Ryzen 9 3900X 12-Core Processor
+  WORD_SIZE: 64
+  LLVM: libLLVM-18.1.7 (ORCJIT, znver2)
+  GC: Built with stock GC
+
+> pixi run python test.py
+=====================================
+Input size: 1000
+All close: True
+Numpy    time (ms): 0.05090880440548062
+FHist.jl time (ms): 0.007097609341144562
+=====================================
+Input size: 10000
+All close: True
+Numpy    time (ms): 0.19860140746459365
+FHist.jl time (ms): 0.020549201872199774
+=====================================
+Input size: 100000
+All close: True
+Numpy    time (ms): 1.3177349930629134
+FHist.jl time (ms): 0.19442759221419692
+=====================================
+Input size: 1000000
+All close: True
+Numpy    time (ms): 6.48591200588271
+FHist.jl time (ms): 1.8390380078926682
+```

C_FHist/benchmarks.md

@@ -0,0 +1,33 @@
+## On a AMD Zen2 system:
+
+### JULIA_CPU_TARGET='znver2;generic' julia +1.12
+```
+Input size: 1000000
+All close: True
+Numpy    time (μs): 6.558584200683981
+FHist.jl time (μs): 1.7967560008401051
+```
+
+### JULIA_CPU_TARGET='generic;znver2' julia +1.12
+```
+Input size: 1000000
+All close: True
+Numpy    time (μs): 6.538515799911693
+FHist.jl time (μs): 2.6081674004672095
+```
+
+### JULIA_CPU_TARGET='znver2;generic' julia +nightly
+```
+Input size: 1000000
+All close: True
+Numpy    time (μs): 6.641940600820817
+FHist.jl time (μs): 1.5721895993920043
+```
+
+### JULIA_CPU_TARGET='generic;znver2' julia +nightly
+```
+Input size: 1000000
+All close: True
+Numpy    time (μs): 6.617425798322074
+FHist.jl time (μs): 2.40858779870905
+```

C_FHist/jlsum.jl

@@ -0,0 +1,4 @@
+Base.@ccallable function jlsum(p::Ptr{Cdouble}, N::Clong)::Cdouble
+    np_input = unsafe_wrap(Array{Float64}, p, N, own=false)
+    return sum(np_input)
+end

C_FHist/jlsum.py

@@ -0,0 +1,26 @@
+import ctypes
+
+lib = ctypes.CDLL('./libjlsum.so')
+lib.jlsum.argtypes = [
+        ctypes.POINTER(ctypes.c_double), ctypes.c_long,
+        ]
+lib.jlsum.restype = ctypes.c_double
+def jlsum(a):
+    res = lib.jlsum(
+            a.ctypes.data_as(ctypes.POINTER(ctypes.c_double)), 
+            ctypes.c_long(len(a)), 
+            )
+    return res
+
+
+import numpy as np
+import timeit
+for N_input in [10**4, 10**5, 10**6]:
+    input_data = np.random.rand(N_input)
+    print("=====================================")
+    print(f"Input size: {N_input}")
+    print("np.isclose:", np.isclose(np.sum(input_data), jlsum(input_data)))
+    np_timer = timeit.Timer(lambda: np.sum(input_data))
+    jl_timer = timeit.Timer(lambda: jlsum(input_data))
+    print(f"Numpy    time (ms): {np.min(np_timer.repeat(number=5, repeat=500)) / 5 * 1000}")
+    print(f"FHist.jl time (ms): {np.min(jl_timer.repeat(number=5, repeat=500)) / 5 * 1000}")

C_FHist/pixi.toml

@@ -0,0 +1,13 @@
+[project]
+authors = ["Jerry Ling <[email protected]>"]
+channels = ["conda-forge"]
+description = "Add a short description here"
+name = "C_FHist"
+platforms = ["linux-64", "osx-arm64"]
+version = "0.1.0"
+
+[tasks]
+
+[dependencies]
+numpy = ">=2.3.1,<3"
+ipython = ">=9.4.0,<10"

C_FHist/test.py

@@ -0,0 +1,41 @@
+import ctypes
+import numpy as np
+import timeit
+import time
+
+lib = ctypes.CDLL('./libfhistjl.so')
+lib.hist1d.argtypes = [
+        ctypes.POINTER(ctypes.c_double), ctypes.c_long, # input and length
+        ctypes.POINTER(ctypes.c_double), ctypes.c_long, # bincoutns and length
+        ctypes.c_double, ctypes.c_double, ctypes.c_double # binedges
+        ]
+lib.hist1d.restype = ctypes.c_voidp
+
+def jlhist(a, bins, range):
+    bincounts = np.zeros(bins)
+    step = (range[1] - range[0]) / bins
+
+    lib.hist1d(
+            a.ctypes.data_as(ctypes.POINTER(ctypes.c_double)), 
+            ctypes.c_long(len(a)), 
+            bincounts.ctypes.data_as(ctypes.POINTER(ctypes.c_double)), 
+            ctypes.c_long(len(bincounts)), 
+            ctypes.c_double(range[0]), 
+            ctypes.c_double(step),
+            ctypes.c_double(range[1])
+            )
+    return bincounts
+
+
+for N_input in [10**3, 10**4, 10**5, 10**6]:
+    input_data = np.random.rand(N_input)
+    bins = 20
+    print("=====================================")
+    print(f"Input size: {N_input}")
+    print("All close:", np.allclose(np.histogram(input_data, bins=bins, range=(0.0, 1.0))[0], jlhist(input_data,
+                                                                                                     bins=bins, range=(0.0,
+                                                                                                                       1.0))))
+    np_timer = timeit.Timer(lambda: np.histogram(input_data, bins=bins, range=(0.0, 1.0)))
+    jl_timer = timeit.Timer(lambda: jlhist(input_data, bins=bins, range=(0.0, 1.0)))
+    print(f"Numpy    time (ms): {np.min(np_timer.repeat(number=5, repeat=200)) / 5 * 1000}")
+    print(f"FHist.jl time (ms): {np.min(jl_timer.repeat(number=5, repeat=200)) / 5 * 1000}")

src/hist1d.jl

@@ -1,8 +1,8 @@
-function _fast_bincounts!(h::Hist1D, TA::Tuple{T}, r::AbstractRange, weights::Nothing) where T
-    A = TA[1]
+_fast_bincounts!(h::Hist1D, TA::Tuple{T}, r::AbstractRange, ::Nothing) where T = _fast_bincounts_1d!(h, TA[1], r)
+
+function _fast_bincounts_1d!(h::Hist1D, A, r::AbstractRange)
     overflow = h.overflow
     counts = bincounts(h)
-    s2 = sumw2(h)
     firstr = first(r)
     invstep = inv(step(r))
     L = nbins(h)