From 2455c2bb0e35bb476ec073174b42f8c3adddc183 Mon Sep 17 00:00:00 2001
From: Sioni Summers <14807534+thesps@users.noreply.github.com>
Date: Thu, 24 Nov 2022 16:37:07 +0100
Subject: [PATCH] Move VHDL backend simulators to separate module. Use C++
 fixed point adapter in VHDL backend decision_function input/output handling

---
 conifer/backends/vhdl/__init__.py         |   5 +-
 conifer/backends/vhdl/simulators.py       |  99 ++++++++++++++++
 conifer/backends/vhdl/writer.py           | 131 +++-------------------
 conifer/utils/fixed_point.py              |  11 +-
 conifer/utils/fixed_point_conversions.cpp |   7 ++
 5 files changed, 133 insertions(+), 120 deletions(-)
 create mode 100644 conifer/backends/vhdl/simulators.py

diff --git a/conifer/backends/vhdl/__init__.py b/conifer/backends/vhdl/__init__.py
index 4b198818..8307a926 100644
--- a/conifer/backends/vhdl/__init__.py
+++ b/conifer/backends/vhdl/__init__.py
@@ -1,2 +1,3 @@
-from .writer import write, auto_config, sim_compile, decision_function, build, Simulators
-simulator = Simulators.xsim
+from conifer.backends.vhdl.writer import write, auto_config, sim_compile, decision_function, build
+from conifer.backends.vhdl.simulators import Modelsim, GHDL, Xsim
+simulator = Xsim
diff --git a/conifer/backends/vhdl/simulators.py b/conifer/backends/vhdl/simulators.py
new file mode 100644
index 00000000..3a222dbc
--- /dev/null
+++ b/conifer/backends/vhdl/simulators.py
@@ -0,0 +1,99 @@
+import os
+import logging
+logger = logging.getLogger(__name__)
+
+
+def _compile_sim(simulator, odir):
+  logger.info(f'Compiling simulation for {simulator.__name__.lower()} simulator')
+  logger.debug(f'Compiling simulation with command "{simulator._compile_cmd}"')
+  cwd = os.getcwd()
+  os.chdir(odir)
+  success = os.system(simulator._compile_cmd)
+  os.chdir(cwd)
+  if(success > 0):
+      logger.error(f"'sim_compile' failed, check {simulator.__name__.lower()}_compile.log")
+  return success == 0
+
+def _run_sim(simulator, odir):
+    logger.info(f'Running simulation for {simulator.__name__.lower()} simulator')
+    logger.debug(f'Running simulation with command "{simulator._run_cmd}"')
+    cwd = os.getcwd()
+    os.chdir(odir)
+    success = os.system(simulator._run_cmd)
+    os.chdir(cwd)
+    if(success > 0):
+      logger.error(f"'sim_compile' failed, check {simulator.__name__.lower()}.log")
+    return success == 0
+
+class Modelsim:
+  _compile_cmd = 'sh modelsim_compile.sh > modelsim_compile.log'
+  _run_cmd = 'vsim -c -do "vsim -L BDT -L xil_defaultlib xil_defaultlib.testbench; run -all; quit -f" > vsim.log'
+  
+  def write_scripts(outputdir, filedir, n_classes):
+    f = open(os.path.join(filedir,'./scripts/modelsim_compile.sh'),'r')
+    fout = open(f'{outputdir}/modelsim_compile.sh','w')
+    for line in f.readlines():
+      if 'insert arrays' in line:
+        for i in range(n_classes):
+          newline = f'vcom -2008 -work BDT ./firmware/Arrays{i}.vhd\n'
+          fout.write(newline)
+      else:
+        fout.write(line)
+    f.close()
+    fout.close()
+
+    f = open(f'{outputdir}/test.tcl', 'w')
+    f.write('vsim -L BDT -L xil_defaultlib xil_defaultlib.testbench\n')
+    f.write('run 100 ns\n')
+    f.write('quit -f\n')
+    f.close()
+
+  def compile(odir):
+    return _compile_sim(Modelsim, odir)
+
+  def run_sim(odir):
+    return _run_sim(Modelsim, odir)
+
+class GHDL:
+  _compile_cmd = 'sh ghdl_compile.sh > ghdl_compile.log'
+  _run_cmd = 'ghdl -r --std=08 --work=xil_defaultlib testbench > ghdl.log'
+  def write_scripts(outputdir, filedir, n_classes):
+    f = open(os.path.join(filedir, './scripts/ghdl_compile.sh'), 'r')
+    fout = open(f'{outputdir}/ghdl_compile.sh', 'w')
+    for line in f.readlines():
+      if 'insert arrays' in line:
+        for i in range(n_classes):
+          newline = f'ghdl -a --std=08 --work=BDT ./firmware/Arrays{i}.vhd\n'
+          fout.write(newline)
+      else:
+        fout.write(line)
+    f.close()
+    fout.close()
+
+  def compile(odir):
+    return _compile_sim(GHDL, odir)
+
+  def run_sim(odir):
+    return _run_sim(GHDL, odir)
+
+class Xsim:
+  _compile_cmd = 'sh xsim_compile.sh > xsim_compile.log'
+  _run_cmd = 'xsim -R bdt_tb > xsim.log'
+  def write_scripts(outputdir, filedir, n_classes):
+    f = open(os.path.join(filedir, './scripts/xsim_compile.sh'), 'r')
+    fout = open(f'{outputdir}/xsim_compile.sh', 'w')
+    for line in f.readlines():
+      if 'insert arrays' in line:
+        for i in range(n_classes):
+          newline = f'xvhdl -2008 -work BDT ./firmware/Arrays{i}.vhd\n'
+          fout.write(newline)
+      else:
+        fout.write(line)
+    f.close()
+    fout.close()
+
+  def compile(odir):
+    return _compile_sim(Xsim, odir)
+
+  def run_sim(odir):
+    return _run_sim(Xsim, odir)
diff --git a/conifer/backends/vhdl/writer.py b/conifer/backends/vhdl/writer.py
index 126fa161..b537c91f 100644
--- a/conifer/backends/vhdl/writer.py
+++ b/conifer/backends/vhdl/writer.py
@@ -8,11 +8,6 @@
 import logging
 logger = logging.getLogger(__name__)
 
-class Simulators(Enum):
-   modelsim = 0
-   xsim = 1
-   ghdl = 2
-
 def write(model):
 
   model.save()
@@ -72,6 +67,8 @@ def write(model):
   mult = 2**dtype_frac
 
   fp = FixedPointConverter(cfg['Precision'])
+  # TODO this should be attached to the model differently
+  model._fp_converter = fp
 
   # binary classification only uses one set of trees
   n_classes = 1 if ensembleDict['n_classes'] == 2 else ensembleDict['n_classes']
@@ -82,16 +79,12 @@ def write(model):
   for i in range(n_classes):
     fout[i].write(array_header_text)
     fout[i].write('package Arrays{} is\n\n'.format(i))
-    #fout[i].write('    constant initPredict : ty := to_ty({});\n'.format(int(np.round(ensembleDict['init_predict'][i] * mult))))
     fout[i].write('    constant initPredict : ty := to_ty({});\n'.format(fp.to_int(np.float64(ensembleDict['init_predict'][i]))))
-    #fout[i].write('    constant initPredict : ty := to_ty({});\n'.format(int(np.round(ensembleDict['init_predict'][i] * mult))))
-
 
   # Loop over fields (childrenLeft, childrenRight, threshold...)
   for field in ensembleDict['trees'][0][0].keys():
     # Write the type for this field to each classes' file
     for iclass in range(n_classes):
-      #dtype = 'txArray2D' if field == 'threshold' else 'tyArray2D' if field == 'value' else 'intArray2D'
       fieldName = field
       # The threshold and value arrays are declared as integers, then cast
       # So need a separate constant
@@ -99,7 +92,6 @@ def write(model):
         fieldName += '_int'
         # Convert the floating point values to integers
         for ii, trees in enumerate(ensembleDict['trees']):
-          #ensembleDict['trees'][ii][iclass][field] = np.round(np.array(ensembleDict['trees'][ii][iclass][field]) * mult).astype('int')
           ensembleDict['trees'][ii][iclass][field] = np.array([fp.to_int(x) for x in ensembleDict['trees'][ii][iclass][field]])
       nElem = 'nLeaves' if field == 'iLeaf' else 'nNodes'
       fout[iclass].write('    constant {} : intArray2D{}(0 to nTrees - 1) := ('.format(fieldName, nElem))
@@ -117,7 +109,8 @@ def write(model):
     fout[i].write('end Arrays{};'.format(i))
     fout[i].close()
 
-  write_sim_scripts(cfg, filedir, n_classes)
+  from conifer.backends.vhdl import simulator
+  simulator.write_scripts(cfg['OutputDir'], filedir, n_classes)
 
   f = open('{}/SimulationInput.txt'.format(cfg['OutputDir']), 'w')
   f.write(' '.join(map(str, [0] * ensembleDict['n_features'])))
@@ -193,67 +186,24 @@ def auto_config():
 
 def sim_compile(model):
   from conifer.backends.vhdl import simulator
-  config = copy.deepcopy(model.config)
-  xsim_cmd = 'sh xsim_compile.sh > xsim_compile.log'
-  msim_cmd = 'sh modelsim_compile.sh > modelsim_compile.log'
-  ghdl_cmd = 'sh ghdl_compile.sh > ghdl_compile.log'
-  cmdmap = {Simulators.modelsim : msim_cmd,
-            Simulators.xsim : xsim_cmd,
-            Simulators.ghdl : ghdl_cmd}
-  cmd = cmdmap[simulator]
-  logger.info(f'Compiling simulation for {simulator} simulator')
-  logger.debug(f'Compiling simulation with command "{cmd}"')
-  cwd = os.getcwd()
-  os.chdir(config['OutputDir'])
-  success = os.system(cmd)
-  os.chdir(cwd)
-  if(success > 0):
-      logger.error("'sim_compile' failed, check {}_compile.log".format(simulator.name))
-  return
+  return simulator.compile(model.config['OutputDir'])
 
 def decision_function(X, model, trees=False):
     from conifer.backends.vhdl import simulator
 
     config = copy.deepcopy(model.config)
-    msim_cmd = 'vsim -c -do "vsim -L BDT -L xil_defaultlib xil_defaultlib.testbench; run -all; quit -f" > vsim.log'
-    xsim_cmd = 'xsim -R bdt_tb > xsim.log'
-    ghdl_cmd = 'ghdl -r --std=08 --work=xil_defaultlib testbench > ghdl.log'
-    cmdmap = {Simulators.modelsim : msim_cmd,
-              Simulators.xsim : xsim_cmd,
-              Simulators.ghdl : ghdl_cmd}
-    cmd = cmdmap[simulator]
-    msim_log = 'vsim.log'
-    xsim_log = 'xsim.log'
-    ghdl_log = 'ghdl.log'
-    logmap = {Simulators.modelsim : msim_log,
-              Simulators.xsim : xsim_log,
-              Simulators.ghdl : ghdl_log}
-    logfile = logmap[simulator]
 
-    logger.info(f'Running simulation for {simulator} simulator')
-
-    dtype = config['Precision']
-    if not 'ap_fixed' in dtype:
-        logger.error("Only ap_fixed is currently supported, exiting")
-        return
-    dtype = dtype.replace('ap_fixed<', '').replace('>', '')
-    dtype_n = int(dtype.split(',')[0].strip()) # total number of bits
-    dtype_int = int(dtype.split(',')[1].strip()) # number of integer bits
-    dtype_frac = dtype_n - dtype_int # number of fractional bits
-    mult = 2**dtype_frac
-    Xint = (X *  mult).astype('int')
-    logger.debug(f'Converting X ({X.dtype}), to integers with scale factor {mult} from {config["Precision"]}')
+    Xint = np.array([model._fp_converter.to_int(x) for x in X.ravel()]).reshape(X.shape)
     np.savetxt('{}/SimulationInput.txt'.format(config['OutputDir']),
                Xint, delimiter=' ', fmt='%d')
-    cwd = os.getcwd()
-    os.chdir(config['OutputDir'])
-    logger.debug(f'Running simulation with command "{cmd}"')
-    success = os.system(cmd)
-    os.chdir(cwd)
-    if(success > 0):
-        logger.error("'decision_function' failed, see {}.log".format(logfile))
-        return
-    y = np.loadtxt('{}/SimulationOutput.txt'.format(config['OutputDir'])) * 1. / mult
+    success = simulator.run_sim(config['OutputDir'])
+    if not success:
+      return 
+    y = np.loadtxt('{}/SimulationOutput.txt'.format(config['OutputDir'])).astype(np.int32)
+    y = np.array([model._fp_converter.from_int(yi) for yi in y.ravel()]).reshape(y.shape)
+    if np.ndim(y) == 1:
+      y = np.expand_dims(y, 1)
+
     if trees:
         logger.warn("Individual tree output (trees=True) not yet implemented for this backend")
     return y
@@ -273,55 +223,4 @@ def build(config, **kwargs):
         logger.error("build failed, check build.log")
         return False
     return True
-            
-def write_sim_scripts(cfg, filedir, n_classes):
-  from conifer.backends.vhdl import simulator
-  fmap = {Simulators.modelsim : write_modelsim_scripts,
-          Simulators.xsim : write_xsim_scripts,
-          Simulators.ghdl : write_ghdl_scripts,}
-  fmap[simulator](cfg, filedir, n_classes)
-
-def write_modelsim_scripts(cfg, filedir, n_classes):
-  f = open(os.path.join(filedir,'./scripts/modelsim_compile.sh'),'r')
-  fout = open('{}/modelsim_compile.sh'.format(cfg['OutputDir']),'w')
-  for line in f.readlines():
-    if 'insert arrays' in line:
-      for i in range(n_classes):
-        newline = 'vcom -2008 -work BDT ./firmware/Arrays{}.vhd\n'.format(i)
-        fout.write(newline)
-    else:
-      fout.write(line)
-  f.close()
-  fout.close()
-
-  f = open('{}/test.tcl'.format(cfg['OutputDir']), 'w')
-  f.write('vsim -L BDT -L xil_defaultlib xil_defaultlib.testbench\n')
-  f.write('run 100 ns\n')
-  f.write('quit -f\n')
-  f.close()
-
-def write_xsim_scripts(cfg, filedir, n_classes):
-  f = open(os.path.join(filedir, './scripts/xsim_compile.sh'), 'r')
-  fout = open('{}/xsim_compile.sh'.format(cfg['OutputDir']), 'w')
-  for line in f.readlines():
-    if 'insert arrays' in line:
-      for i in range(n_classes):
-        newline = 'xvhdl -2008 -work BDT ./firmware/Arrays{}.vhd\n'.format(i)
-        fout.write(newline)
-    else:
-      fout.write(line)
-  f.close()
-  fout.close()
-
-def write_ghdl_scripts(cfg, filedir, n_classes):
-  f = open(os.path.join(filedir, './scripts/ghdl_compile.sh'), 'r')
-  fout = open('{}/ghdl_compile.sh'.format(cfg['OutputDir']), 'w')
-  for line in f.readlines():
-    if 'insert arrays' in line:
-      for i in range(n_classes):
-        newline = 'ghdl -a --std=08 --work=BDT ./firmware/Arrays{}.vhd\n'.format(i)
-        fout.write(newline)
-    else:
-      fout.write(line)
-  f.close()
-  fout.close()
\ No newline at end of file
+            
\ No newline at end of file
diff --git a/conifer/utils/fixed_point.py b/conifer/utils/fixed_point.py
index 769e063f..1df67445 100644
--- a/conifer/utils/fixed_point.py
+++ b/conifer/utils/fixed_point.py
@@ -1,5 +1,6 @@
 import os
-import shutil
+import logging
+logger = logging.getLogger(__name__)
 
 class FixedPointConverter:
   '''
@@ -12,6 +13,7 @@ def __init__(self, type_string):
     args:
       type_string : string for the ap_ type, e.g. ap_fixed<16,6,AP_RND,AP_SAT>
     '''
+    logger.info(f'Constructing converter for {type_string}')
     self.type_string = type_string
     self.sani_type = type_string.replace('<','_').replace('>','').replace(',','_')
     filedir = os.path.dirname(os.path.abspath(__file__))
@@ -32,6 +34,7 @@ def __init__(self, type_string):
     curr_dir = os.getcwd()
     os.chdir(cpp_filedir)
     cmd = f"g++ -O3 -shared -std=c++11 -fPIC $(python3 -m pybind11 --includes) -I/cvmfs/cms.cern.ch/slc7_amd64_gcc900/external/hls/2019.08/include/ {self.sani_type}.cpp -o {self.sani_type}.so"
+    logger.debug(f'Compiling with command {cmd}')
     try:
       ret_val = os.system(cmd)
       if ret_val != 0:
@@ -40,6 +43,7 @@ def __init__(self, type_string):
       os.chdir(curr_dir)
     
     os.chdir(cpp_filedir)
+    logger.debug(f'Importing compiled module {self.sani_type}.so')
     try:
       import importlib.util
       spec = importlib.util.spec_from_file_location('fixed_point', f'{self.sani_type}.so')
@@ -54,4 +58,7 @@ def to_int(self, x):
     return self.lib.to_int(x)
 
   def to_double(self, x):
-    return self.lib.to_double(x)
\ No newline at end of file
+    return self.lib.to_double(x)
+
+  def from_int(self, x):
+    return self.lib.from_int(x)
\ No newline at end of file
diff --git a/conifer/utils/fixed_point_conversions.cpp b/conifer/utils/fixed_point_conversions.cpp
index c3a258bf..f9e7419b 100644
--- a/conifer/utils/fixed_point_conversions.cpp
+++ b/conifer/utils/fixed_point_conversions.cpp
@@ -13,9 +13,16 @@ double to_double(double x){
   return y.to_double();
 }
 
+double from_int(int x){
+  T y;
+  y.V = x;
+  return y.to_double();
+}
+
 namespace py = pybind11;
 PYBIND11_MODULE(fixed_point, m) {
     m.doc() = "fixed point conversion";
     m.def("to_int", &to_int, "Get the integer representation of the ap_fixed");
     m.def("to_double", &to_double, "Get the double representation of the ap_fixed");
+    m.def("from_int", &from_int, "Set the underlying bits of the ap_fixed from int, return the double");
 }
\ No newline at end of file