diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 4d4f7e52891a0cb8477672fbaf14b29c067de0f9..58fb78251d04afb0e7f613aae237799237a010b9 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -117,6 +117,20 @@ Python 3.6 POCL Twice With Cache:
# except:
# - tags
+Python 3.6 POCL Examples:
+ script:
+ - export PY_EXE=python3.6
+ - export PYOPENCL_TEST=portable
+ - export EXTRA_INSTALL="numpy mako pyvisfile matplotlib jupyter nbconvert"
+ - ". ./build-py-project-and-run-examples.sh"
+ tags:
+ - python3.6
+ - pocl
+ - large-node
+ - ispc
+ except:
+ - tags
+
CentOS binary:
script:
- (cd build-helpers; ./make-linux-build-docker.sh --nodate)
diff --git a/build-py-project-and-run-examples.sh b/build-py-project-and-run-examples.sh
new file mode 100644
index 0000000000000000000000000000000000000000..e51a86d2085364ca142f5bfde3380a9fade0de01
--- /dev/null
+++ b/build-py-project-and-run-examples.sh
@@ -0,0 +1,37 @@
+#! /bin/bash
+
+set -e
+
+curl -L -O -k https://gitlab.tiker.net/inducer/ci-support/raw/master/build-py-project.sh
+source build-py-project.sh
+
+function run_examples()
+{
+ PATTERN=$1
+ CMDLINE=$2
+ for i in $(find examples -name "$PATTERN" -print ); do
+ echo "-----------------------------------------------------------------------"
+ echo "RUNNING $i"
+ echo "-----------------------------------------------------------------------"
+ dn=$(dirname "$i")
+ bn=$(basename "$i")
+ (cd $dn; echo $CMDLINE "$bn"; $CMDLINE "$bn")
+ done
+}
+
+function run_py_examples()
+{
+ run_examples "*.py" ${PY_EXE}
+}
+function run_ipynb_examples()
+{
+ run_examples "*.ipynb" "${PY_EXE} -m nbconvert --execute"
+}
+function run_floopy_examples()
+{
+ run_examples "*.floopy" "${PY_EXE} -m loopy"
+}
+
+run_py_examples
+run_ipynb_examples
+run_floopy_examples
diff --git a/examples/fortran/foo.floopy b/examples/fortran/foo.floopy
deleted file mode 100644
index 6b8741e1150a05226f3aa9de849db8abb1915002..0000000000000000000000000000000000000000
--- a/examples/fortran/foo.floopy
+++ /dev/null
@@ -1,30 +0,0 @@
-subroutine fill(out, a, n)
- implicit none
-
- real_type a, out(n)
- integer n, i
-
- do i = 1, n
- out(i) = a
- end do
- do i = 1, n
- out(i) = out(i) * factor
- end do
-end
-
-!$loopy begin
-!
-! SOURCE = lp.c_preprocess(SOURCE, [
-! "factor 4.0",
-! "real_type real*8",
-! ])
-! fill, = lp.parse_fortran(SOURCE, FILENAME)
-! fill = lp.split_iname(fill, "i", 128,
-! outer_tag="g.0", inner_tag="l.0")
-! fill = lp.split_iname(fill, "i_1", 128,
-! outer_tag="g.0", inner_tag="l.0")
-! RESULT = [fill]
-!
-!$loopy end
-
-! vim:filetype=floopy
diff --git a/examples/fortran/ipython-integration-demo.ipynb b/examples/fortran/ipython-integration-demo.ipynb
index c2b34f1d13f7b4971c0c2e88a0ae5013ed12ee23..7a5c8257bf80fdfcc3d3b978a7dca2d401c48271 100644
--- a/examples/fortran/ipython-integration-demo.ipynb
+++ b/examples/fortran/ipython-integration-demo.ipynb
@@ -1,190 +1,142 @@
{
- "metadata": {
- "name": "",
- "signature": "sha256:c9f8334aa7aa4a5ad1437fa5871aafa52bbc9131271d9e90e7be47d22725cc94"
- },
- "nbformat": 3,
- "nbformat_minor": 0,
- "worksheets": [
+ "cells": [
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# Loopy IPython Integration Demo"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "%load_ext loopy.ipython_ext"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "## Without transform code"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "%%fortran_kernel\n",
+ "\n",
+ "subroutine fill(out, a, n)\n",
+ " implicit none\n",
+ "\n",
+ " real*8 a, out(n)\n",
+ " integer n, i\n",
+ "\n",
+ " do i = 1, n\n",
+ " out(i) = a\n",
+ " end do\n",
+ "end"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "print(fill)"
+ ]
+ },
{
- "cells": [
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "# Loopy IPython Integration Demo"
- ]
- },
- {
- "cell_type": "code",
- "collapsed": false,
- "input": [
- "%load_ext loopy.ipython_ext"
- ],
- "language": "python",
- "metadata": {},
- "outputs": [],
- "prompt_number": 1
- },
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "## Without transform code"
- ]
- },
- {
- "cell_type": "code",
- "collapsed": false,
- "input": [
- "%%fortran_kernel\n",
- "\n",
- "subroutine fill(out, a, n)\n",
- " implicit none\n",
- "\n",
- " real*8 a, out(n)\n",
- " integer n, i\n",
- "\n",
- " do i = 1, n\n",
- " out(i) = a\n",
- " end do\n",
- "end"
- ],
- "language": "python",
- "metadata": {},
- "outputs": [],
- "prompt_number": 2
- },
- {
- "cell_type": "code",
- "collapsed": false,
- "input": [
- "print(fill)"
- ],
- "language": "python",
- "metadata": {},
- "outputs": [
- {
- "output_type": "stream",
- "stream": "stdout",
- "text": [
- "---------------------------------------------------------------------------\n",
- "KERNEL: fill\n",
- "---------------------------------------------------------------------------\n",
- "ARGUMENTS:\n",
- "a: ValueArg, type: float64\n",
- "n: ValueArg, type: int32\n",
- "out: GlobalArg, type: float64, shape: (n), dim_tags: (N0:stride:1)\n",
- "---------------------------------------------------------------------------\n",
- "DOMAINS:\n",
- "[n] -> { [i] : i >= 0 and i <= -1 + n }\n",
- "---------------------------------------------------------------------------\n",
- "INAME IMPLEMENTATION TAGS:\n",
- "i: None\n",
- "---------------------------------------------------------------------------\n",
- "INSTRUCTIONS:\n",
- "[i] out[i] <- a # insn0\n",
- "---------------------------------------------------------------------------\n"
- ]
- }
- ],
- "prompt_number": 3
- },
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "## With transform code"
- ]
- },
- {
- "cell_type": "code",
- "collapsed": false,
- "input": [
- "split_amount = 128"
- ],
- "language": "python",
- "metadata": {},
- "outputs": [],
- "prompt_number": 4
- },
- {
- "cell_type": "code",
- "collapsed": false,
- "input": [
- "%%transformed_fortran_kernel\n",
- "\n",
- "subroutine tr_fill(out, a, n)\n",
- " implicit none\n",
- "\n",
- " real*8 a, out(n)\n",
- " integer n, i\n",
- "\n",
- " do i = 1, n\n",
- " out(i) = a\n",
- " end do\n",
- "end\n",
- "\n",
- "!$loopy begin\n",
- "!\n",
- "! tr_fill, = lp.parse_fortran(SOURCE)\n",
- "! tr_fill = lp.split_iname(tr_fill, \"i\", split_amount,\n",
- "! outer_tag=\"g.0\", inner_tag=\"l.0\")\n",
- "! RESULT = [tr_fill]\n",
- "!\n",
- "!$loopy end"
- ],
- "language": "python",
- "metadata": {},
- "outputs": [],
- "prompt_number": 5
- },
- {
- "cell_type": "code",
- "collapsed": false,
- "input": [
- "print(tr_fill)"
- ],
- "language": "python",
- "metadata": {},
- "outputs": [
- {
- "output_type": "stream",
- "stream": "stdout",
- "text": [
- "---------------------------------------------------------------------------\n",
- "KERNEL: tr_fill\n",
- "---------------------------------------------------------------------------\n",
- "ARGUMENTS:\n",
- "a: ValueArg, type: float64\n",
- "n: ValueArg, type: int32\n",
- "out: GlobalArg, type: float64, shape: (n), dim_tags: (N0:stride:1)\n",
- "---------------------------------------------------------------------------\n",
- "DOMAINS:\n",
- "[n] -> { [i_outer, i_inner] : i_inner >= -128i_outer and i_inner <= -1 + n - 128i_outer and i_inner >= 0 and i_inner <= 127 }\n",
- "---------------------------------------------------------------------------\n",
- "INAME IMPLEMENTATION TAGS:\n",
- "i_inner: l.0\n",
- "i_outer: g.0\n",
- "---------------------------------------------------------------------------\n",
- "INSTRUCTIONS:\n",
- "[i_inner,i_outer] out[i_inner + i_outer*128] <- a # insn0\n",
- "---------------------------------------------------------------------------\n"
- ]
- }
- ],
- "prompt_number": 6
- },
- {
- "cell_type": "code",
- "collapsed": false,
- "input": [],
- "language": "python",
- "metadata": {},
- "outputs": []
- }
- ],
- "metadata": {}
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "## With transform code"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {
+ "collapsed": true
+ },
+ "outputs": [],
+ "source": [
+ "split_amount = 128"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "%%transformed_fortran_kernel\n",
+ "\n",
+ "subroutine tr_fill(out, a, n)\n",
+ " implicit none\n",
+ "\n",
+ " real*8 a, out(n)\n",
+ " integer n, i\n",
+ "\n",
+ " do i = 1, n\n",
+ " out(i) = a\n",
+ " end do\n",
+ "end\n",
+ "\n",
+ "!$loopy begin\n",
+ "!\n",
+ "! tr_fill, = lp.parse_fortran(SOURCE)\n",
+ "! tr_fill = lp.split_iname(tr_fill, \"i\", split_amount,\n",
+ "! outer_tag=\"g.0\", inner_tag=\"l.0\")\n",
+ "! RESULT = [tr_fill]\n",
+ "!\n",
+ "!$loopy end"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "print(tr_fill)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {
+ "collapsed": true
+ },
+ "outputs": [],
+ "source": []
}
- ]
-}
\ No newline at end of file
+ ],
+ "metadata": {
+ "kernelspec": {
+ "display_name": "Python 3",
+ "language": "python",
+ "name": "python3"
+ },
+ "language_info": {
+ "codemirror_mode": {
+ "name": "ipython",
+ "version": 3
+ },
+ "file_extension": ".py",
+ "mimetype": "text/x-python",
+ "name": "python",
+ "nbconvert_exporter": "python",
+ "pygments_lexer": "ipython3",
+ "version": "3.6.4"
+ }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 1
+}
diff --git a/examples/fortran/matmul.floopy b/examples/fortran/matmul.floopy
index 3352449d7bfcc406e5349f36c5cb7ebb3e3613b6..8ee05643b926f722a3419378c52d254e43617708 100644
--- a/examples/fortran/matmul.floopy
+++ b/examples/fortran/matmul.floopy
@@ -1,6 +1,6 @@
subroutine dgemm(m,n,l,alpha,a,b,c)
implicit none
- real*8 temp, a(m,l),b(l,n),c(m,n), alpha
+ real*8 a(m,l),b(l,n),c(m,n), alpha
integer m,n,k,i,j,l
do j = 1,n
diff --git a/examples/fortran/tagging.floopy b/examples/fortran/tagging.floopy
index 40b487528cdd788f26fb4b600869e4bf81f090a8..87aacba68ae2fc6f3b7052325fcd2378e9880e47 100644
--- a/examples/fortran/tagging.floopy
+++ b/examples/fortran/tagging.floopy
@@ -1,7 +1,7 @@
subroutine fill(out, a, n)
implicit none
- real*8 a, out(n)
+ real_type a, out(n)
integer n, i
!$loopy begin tagged: init
@@ -9,17 +9,28 @@ subroutine fill(out, a, n)
out(i) = a
end do
!$loopy end tagged: init
+
+!$loopy begin tagged: mult
do i = 1, n
- out(i) = out(i) * 2
+ out(i) = out(i) * factor
end do
+!$loopy end tagged: mult
end
!$loopy begin
+!
+! SOURCE = lp.c_preprocess(SOURCE, [
+! "factor 4.0",
+! "real_type real*8",
+! ])
! fill, = lp.parse_fortran(SOURCE, FILENAME)
+! fill = lp.add_barrier(fill, "tag:init", "tag:mult", "gb1")
! fill = lp.split_iname(fill, "i", 128,
! outer_tag="g.0", inner_tag="l.0")
! fill = lp.split_iname(fill, "i_1", 128,
! outer_tag="g.0", inner_tag="l.0")
! RESULT = [fill]
+!
!$loopy end
+
! vim:filetype=floopy
diff --git a/examples/fortran/volumeKernelSimple.floopy b/examples/fortran/volumeKernelSimple.floopy
deleted file mode 100644
index afc3321b8fa18ff532b3c2898dbaea8f99e3700c..0000000000000000000000000000000000000000
--- a/examples/fortran/volumeKernelSimple.floopy
+++ /dev/null
@@ -1,38 +0,0 @@
-subroutine volumeKernel(elements, Nfields, Ngeo, Ndim, Dop, geo, Q, rhsQ )
-
- implicit none
-
- integer elements, Nfields, Ngeo, Ndim
-
- real*4 Dop(Nq,Nq)
- real*4 Q(Nq,Nq,Nq,Nfields,elements)
- real*4 geo(Nq,Nq,Nq,Ngeo,elements)
- real*4 rhsQ(Nq,Nq,Nq,Nfields,elements)
-
- integer e,i,j,k,d,n,cnt
-
- real*4 u,v,w,p, dFdr, dFds, dFdt, divF
- real*4 F(Nq,Ndim)
-
-
- do e=1,elements
- do i=1,Nq
-
- F(i,1) = 5
- F(i,2) = 7
-
- end do
-
- end do
-
-end subroutine volumeKernel
-
-!$loopy begin
-!
-! volumeKernel, = lp.parse_fortran(SOURCE, FILENAME)
-! volumeKernel = lp.fix_parameters(volumeKernel,
-! Nq=5, Ndim=3)
-! volumeKernel = lp.tag_inames(volumeKernel, dict(i="l.0"))
-! RESULT = [volumeKernel]
-!
-!$loopy end
diff --git a/examples/python/find-centers.py b/examples/python/find-centers.py
index c5e5e916156fd44b5a37cdb3cd41718916461a06..1822d3b7990d27f14d04e3b86a1d3c2d0b5c26ee 100644
--- a/examples/python/find-centers.py
+++ b/examples/python/find-centers.py
@@ -26,7 +26,7 @@ knl = lp.make_kernel(
<> post_dist_sq = if(matches, dist_sq, HUGE)
end
- <> min_dist_sq, <> min_ictr = argmin(ictr, post_dist_sq)
+ <> min_dist_sq, <> min_ictr = argmin(ictr, ictr, post_dist_sq)
tgt_to_qbx_center[itgt] = if(min_dist_sq < HUGE, min_ictr, -1)
end
diff --git a/examples/python/global_barrier_removal.py b/examples/python/global_barrier_removal.py
index b71c561195647b7e35bc5538cfd1d601e270f3c2..7ab049cd1906f703b0efc39808ff68a63b91ff37 100644
--- a/examples/python/global_barrier_removal.py
+++ b/examples/python/global_barrier_removal.py
@@ -6,14 +6,17 @@ import pyopencl.array
knl = lp.make_kernel(
"{ [i,k]: 0<=i<n and 0<=k<3 }",
"""
- c[k,i] = a[k, i + 1]
- out[k,i] = c[k,i]
- """)
+ for i, k
+ ... gbarrier
+ c[k,i] = a[k, i + 1]
+ ... gbarrier
+ out[k,i] = c[k,i]
+ end
+ """, seq_dependencies=True)
# transform
knl = lp.split_iname(knl, "i", 128, outer_tag="g.0", inner_tag="l.0")
-from loopy.kernel.tools import add_dtypes
-knl = add_dtypes(knl,
+knl = lp.add_and_infer_dtypes(knl,
{"a": np.float32, "c": np.float32, "out": np.float32, "n": np.int32})
# schedule
diff --git a/examples/python/hello-loopy-lp.py b/examples/python/hello-loopy.loopy
similarity index 100%
rename from examples/python/hello-loopy-lp.py
rename to examples/python/hello-loopy.loopy
diff --git a/examples/python/rank-one.py b/examples/python/rank-one.py
index 726c44dea927c7dd9f7305fe59b12a27bc820269..b22edfcfa6da39ae9de9b2bb9bc496d2c56c36b5 100644
--- a/examples/python/rank-one.py
+++ b/examples/python/rank-one.py
@@ -6,15 +6,16 @@ import loopy as lp
ctx = cl.create_some_context()
queue = cl.CommandQueue(ctx)
-knl = lp.make_kernel(queue.device,
- "{[i,j]: 0<=i,j<n}",
+knl = lp.make_kernel(
+ "{[i, j]: 0<=i<n and 0<=j<n}",
"c[i, j] = a[i]*b[j]",
assumptions="n >= 16")
a = np.arange(200, dtype=np.float32)
b = np.arange(200, dtype=np.float32)
-evt, (c,) = knl(queue, a=a, b=b, options="write_cl")
+knl = lp.set_options(knl, write_code=True)
+evt, (c,) = knl(queue, a=a, b=b)
# SETUPEND
orig_knl = knl
@@ -26,7 +27,8 @@ knl = lp.split_iname(knl, "j", 16,
outer_tag="g.1", inner_tag="l.1")
# SPLITEND
-evt, (c,) = knl(queue, a=a, b=b, options="write_cl")
+knl = lp.set_options(knl, write_code=True)
+evt, (c,) = knl(queue, a=a, b=b)
split_knl = knl
@@ -35,7 +37,8 @@ knl = lp.add_prefetch(knl, "a")
knl = lp.add_prefetch(knl, "b")
# PREFETCH1END
-evt, (c,) = knl(queue, a=a, b=b, options="write_cl")
+knl = lp.set_options(knl, write_code=True)
+evt, (c,) = knl(queue, a=a, b=b)
knl = split_knl
@@ -44,7 +47,8 @@ knl = lp.add_prefetch(knl, "a", ["i_inner"])
knl = lp.add_prefetch(knl, "b", ["j_inner"])
# PREFETCH2END
-evt, (c,) = knl(queue, a=a, b=b, options="write_cl")
+knl = lp.set_options(knl, write_code=True)
+evt, (c,) = knl(queue, a=a, b=b)
knl = orig_knl
@@ -68,4 +72,5 @@ knl = lp.split_iname(knl, "a_dim_0", 16,
outer_tag="l.1", inner_tag="l.0")
# PREFETCH3END
-evt, (c,) = knl(queue, a=a, b=b, options="write_cl")
+knl = lp.set_options(knl, write_code=True)
+evt, (c,) = knl(queue, a=a, b=b)
diff --git a/examples/python/sparse.py b/examples/python/sparse.py
index bb143d62bfa39c7ba4b569dd5c85782050e3d6ee..a13ba34ae7329ef6a7a72a968a5783f9496fa05e 100644
--- a/examples/python/sparse.py
+++ b/examples/python/sparse.py
@@ -5,10 +5,12 @@ k = lp.make_kernel([
"{ [i] : 0 <= i < m }",
"{ [j] : 0 <= j < length }"],
"""
- <> rowstart = rowstarts[i]
- <> rowend = rowstarts[i]
- <> length = rowend - rowstart
- y[i] = sum(j, values[rowstart+j] * x[colindices[rowstart + j]])
+ for i
+ <> rowstart = rowstarts[i]
+ <> rowend = rowstarts[i]
+ <> length = rowend - rowstart
+ y[i] = sum(j, values[rowstart+j] * x[colindices[rowstart + j]])
+ end
""")
k = lp.add_and_infer_dtypes(k, {
diff --git a/loopy/ipython_ext.py b/loopy/ipython_ext.py
index 12fd0354267e86707cc91de0027319e3584a08f0..ec1b10f1f512e18079f44b94b298e876776cae35 100644
--- a/loopy/ipython_ext.py
+++ b/loopy/ipython_ext.py
@@ -9,7 +9,7 @@ import loopy as lp
class LoopyMagics(Magics):
@cell_magic
def fortran_kernel(self, line, cell):
- result = lp.parse_fortran(cell.encode())
+ result = lp.parse_fortran(cell)
for knl in result:
self.shell.user_ns[knl.name] = knl
@@ -17,7 +17,7 @@ class LoopyMagics(Magics):
@cell_magic
def transformed_fortran_kernel(self, line, cell):
result = lp.parse_transformed_fortran(
- cell.encode(),
+ cell,
transform_code_context=self.shell.user_ns)
for knl in result: