def test_invalid_unit_access_len(self):
''' Invalid unit_access len. '''
with self.assertRaisesRegexp(ValueError,
'NestedLoopDesc: .*unit_access.*'):
_ = NestedLoopDesc(loopcnt=(3, 8, 4),
usize_gbuf=(20, 30, 9),
usize_regf=(3, 3, 1),
unit_access=((19, 29, 9), (35, 45, 15), (1, 1,
2)),
data_loops=(DataDimLoops(le.IFM, le.OFM),
DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)),
unit_ops=7,
unit_time=7)
with self.assertRaisesRegexp(ValueError,
'NestedLoopDesc: .*unit_access.*'):
_ = NestedLoopDesc(loopcnt=(3, 8, 4),
usize_gbuf=(20, 30, 9),
usize_regf=(3, 3, 1),
unit_access=((19, 29, 9), (18, 28),
(35, 45, 15), (1, 1, 2)),
data_loops=(DataDimLoops(le.IFM, le.OFM),
DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)),
unit_ops=7,
unit_time=7)
def test_total_access_of_at(self):
''' Get total_access_of_at. '''
nld = NestedLoopDesc(loopcnt=(3, 8, 4),
usize_gbuf=(20, 30, 9),
usize_regf=(3, 3, 1),
unit_access=((19, 29, 9),
(18, 28, 8),
(35, 45, 15),
(1, 1, 2)),
data_loops=(DataDimLoops(le.IFM, le.OFM),
DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)),
unit_ops=7,
unit_time=7
)
self.assertEqual(nld.total_access_at_of(me.DRAM, de.FIL), 19 * 3 * 8)
self.assertEqual(nld.total_access_at_of(me.DRAM, de.IFM), 29 * 3 * 4)
self.assertEqual(nld.total_access_at_of(me.DRAM, de.OFM), 9 * 8 * 4)
self.assertEqual(nld.total_access_at_of(me.GBUF, de.FIL), 18 * 3 * 8)
self.assertEqual(nld.total_access_at_of(me.GBUF, de.IFM), 28 * 3 * 4)
self.assertEqual(nld.total_access_at_of(me.GBUF, de.OFM), 8 * 8 * 4)
self.assertEqual(nld.total_access_at_of(me.ITCN, de.FIL), 35 * 3 * 8)
self.assertEqual(nld.total_access_at_of(me.ITCN, de.IFM), 45 * 3 * 4)
self.assertEqual(nld.total_access_at_of(me.ITCN, de.OFM), 15 * 8 * 4)
self.assertEqual(nld.total_access_at_of(me.REGF, de.FIL), 1 * 3 * 8)
self.assertEqual(nld.total_access_at_of(me.REGF, de.IFM), 1 * 3 * 4)
self.assertEqual(nld.total_access_at_of(me.REGF, de.OFM), 2 * 8 * 4)
def test_invalid_data_loops_type(self):
''' Invalid data_loops type. '''
with self.assertRaisesRegex(TypeError,
'NestedLoopDesc: .*data_loops.*'):
_ = NestedLoopDesc(loopcnt=(3, 8, 4),
usize_gbuf=(20, 30, 9),
usize_regf=(3, 3, 1),
unit_access=((19, 29, 9),
(18, 28, 8),
(35, 45, 15),
(1, 1, 2)),
data_loops=[DataDimLoops(le.IFM, le.OFM),
DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)],
unit_ops=7,
unit_time=7
)
with self.assertRaisesRegex(TypeError,
'NestedLoopDesc: .*data_loops.*'):
_ = NestedLoopDesc(loopcnt=(3, 8, 4),
usize_gbuf=(20, 30, 9),
usize_regf=(3, 3, 1),
unit_access=((19, 29, 9),
(18, 28, 8),
(35, 45, 15),
(1, 1, 2)),
data_loops=((le.IFM, le.OFM),
(le.IFM, le.BAT),
(le.OFM, le.BAT)),
unit_ops=7,
unit_time=7
)
def test_total_ops(self):
''' Get total_ops. '''
nld = NestedLoopDesc(loopcnt=(3, 8, 4),
usize_gbuf=(20, 30, 9),
usize_regf=(3, 3, 1),
unit_access=((19, 29, 9), (18, 28, 8),
(35, 45, 15), (1, 1, 2)),
data_loops=(DataDimLoops(le.IFM, le.OFM),
DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)),
unit_ops=7,
unit_time=7)
self.assertEqual(nld.total_ops(), 7 * 3 * 8 * 4)
def test_usize_regf_of(self):
''' Accessor usize_regf. '''
nld = NestedLoopDesc(loopcnt=(3, 8, 4),
usize_gbuf=(20, 30, 9),
usize_regf=(3, 3, 1),
unit_access=((19, 29, 9), (18, 28, 8),
(35, 45, 15), (1, 1, 2)),
data_loops=(DataDimLoops(le.IFM, le.OFM),
DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)),
unit_ops=7,
unit_time=7)
self.assertEqual(nld.usize_regf_of(de.FIL), 3, 'usize_regf: FIL')
self.assertEqual(nld.usize_regf_of(de.IFM), 3, 'usize_regf: IFM')
self.assertEqual(nld.usize_regf_of(de.OFM), 1, 'usize_regf: OFM')
def test_valid_args(self):
''' Valid arguments. '''
nld = NestedLoopDesc(loopcnt=(3, 8, 4),
usize_gbuf=(20, 30, 9),
usize_regf=(3, 3, 1),
unit_access=((19, 29, 9),
(18, 28, 8),
(35, 45, 15),
(1, 1, 2)),
data_loops=(DataDimLoops(le.IFM, le.OFM),
DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)),
unit_ops=7,
unit_time=7
)
self.assertEqual(nld.loopcnt, (3, 8, 4), 'loopcnt')
self.assertEqual(nld.usize_gbuf, (20, 30, 9), 'usize_gbuf')
self.assertEqual(nld.usize_regf, (3, 3, 1), 'usize_regf')
self.assertEqual(nld.unit_access, ((19, 29, 9),
(18, 28, 8),
(35, 45, 15),
(1, 1, 2)), 'unit_access')
self.assertEqual(nld.data_loops[de.FIL], DataDimLoops(le.IFM, le.OFM),
'data_loops: FIL')
self.assertEqual(nld.data_loops[de.IFM], DataDimLoops(le.IFM, le.BAT),
'data_loops: IFM')
self.assertEqual(nld.data_loops[de.OFM], DataDimLoops(le.OFM, le.BAT),
'data_loops: OFM')
self.assertEqual(nld.unit_ops, 7, 'unit_ops')
self.assertEqual(nld.unit_time, 7, 'unit_time')
def test_unit_access_at_of(self):
''' Accessor unit_access. '''
nld = NestedLoopDesc(loopcnt=(3, 8, 4),
usize_gbuf=(20, 30, 9),
usize_regf=(3, 3, 1),
unit_access=((19, 29, 9), (18, 28, 8),
(35, 45, 15), (1, 1, 2)),
data_loops=(DataDimLoops(le.IFM, le.OFM),
DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)),
unit_ops=7,
unit_time=7)
self.assertEqual(nld.unit_access_at_of(me.DRAM), 19 + 29 + 9,
'unit_access: DRAM')
self.assertEqual(nld.unit_access_at_of(me.ITCN), 35 + 45 + 15,
'unit_access: ITCN')
self.assertEqual(nld.unit_access_at_of(me.GBUF, de.OFM), 8,
'unit_access: GBUF, OFM')
self.assertEqual(nld.unit_access_at_of(me.REGF, de.FIL), 1,
'unit_access: REGF, FIL')
def setUp(self):
# Workload.
self.layer = {}
self.layer['BASE'] = ConvLayer(12, 10, 28, 3)
self.layer['LGFIL'] = ConvLayer(2, 4, 28, 20)
self.layer['POOL'] = PoolingLayer(32, 28, 2)
self.layer['PAR'] = ConvLayer(24, 36, 56, 3)
self.batch_size = 4
# Resource.
self.resource = {}
dim_array = PhyDim2(16, 16)
proc_region = NodeRegion(origin=PhyDim2(0, 0),
dim=PhyDim2(1, 1),
type=NodeRegion.PROC)
data_region = NodeRegion(origin=PhyDim2(0, 0),
dim=PhyDim2(1, 1),
type=NodeRegion.DRAM)
# Typical resource.
self.resource['BASE'] = Resource(proc_region=proc_region,
dram_region=data_region,
src_data_region=data_region,
dst_data_region=data_region,
dim_array=dim_array,
size_gbuf=65536,
size_regf=64,
array_bus_width=float('inf'),
dram_bandwidth=float('inf'),
no_time_mux=False)
# Larger resource with sufficient capacity, to make all schemes valid.
self.resource['LG'] = Resource(proc_region=proc_region,
dram_region=data_region,
src_data_region=data_region,
dst_data_region=data_region,
dim_array=dim_array,
size_gbuf=1024**3,
size_regf=1024**3,
array_bus_width=float('inf'),
dram_bandwidth=float('inf'),
no_time_mux=False)
# Small resource.
self.resource['SM'] = Resource(proc_region=proc_region,
dram_region=data_region,
src_data_region=data_region,
dst_data_region=data_region,
dim_array=dim_array,
size_gbuf=4096,
size_regf=16,
array_bus_width=float('inf'),
dram_bandwidth=float('inf'),
no_time_mux=False)
# Multi-node parallel resource.
self.resource['PAR'] = Resource(proc_region=NodeRegion(
origin=PhyDim2(0, 0), dim=PhyDim2(4, 2), type=NodeRegion.PROC),
dram_region=data_region,
src_data_region=data_region,
dst_data_region=data_region,
dim_array=dim_array,
size_gbuf=25000,
size_regf=64,
array_bus_width=float('inf'),
dram_bandwidth=float('inf'),
no_time_mux=False)
# Resource with no data regions.
proc_data_region = NodeRegion(origin=PhyDim2(1, 1),
dim=PhyDim2(1, 1),
type=NodeRegion.PROC)
self.resource['SRCNOTDATA'] = Resource(
proc_region=proc_region,
dram_region=data_region,
src_data_region=proc_data_region,
dst_data_region=data_region,
dim_array=dim_array,
size_gbuf=1024**3,
size_regf=1024**3,
array_bus_width=float('inf'),
dram_bandwidth=float('inf'),
no_time_mux=False)
self.resource['DSTNOTDATA'] = Resource(
proc_region=proc_region,
dram_region=data_region,
src_data_region=data_region,
dst_data_region=proc_data_region,
dim_array=dim_array,
size_gbuf=1024**3,
size_regf=1024**3,
array_bus_width=float('inf'),
dram_bandwidth=float('inf'),
no_time_mux=False)
self.resource['DATALOCAL'] = Resource(proc_region=proc_region,
dram_region=data_region,
src_data_region=proc_region,
dst_data_region=proc_region,
dim_array=dim_array,
size_gbuf=1024**3,
size_regf=1024**3,
array_bus_width=float('inf'),
dram_bandwidth=float('inf'),
no_time_mux=False)
# Filter pinning.
self.resource['FILPIN'] = Resource(proc_region=proc_region,
dram_region=data_region,
src_data_region=data_region,
dst_data_region=data_region,
dim_array=dim_array,
size_gbuf=1024**3,
size_regf=1024**3,
array_bus_width=float('inf'),
dram_bandwidth=float('inf'),
no_time_mux=True)
# Nested loop description after mapping.
self.nld = {}
self.nld['BASE'] = next(
MapStrategyEyeriss(self.layer['BASE'], self.batch_size, 1,
dim_array).gen_nested_loop_desc())
self.nld['LGFIL'] = next(
MapStrategyEyeriss(self.layer['LGFIL'], self.batch_size, 1,
dim_array).gen_nested_loop_desc())
self.nld['POOL'] = next(
MapStrategyEyeriss(self.layer['POOL'], self.batch_size, 1,
dim_array).gen_nested_loop_desc())
# Fake nested loop, with zero filter size.
self.nld['ZERO_FIL'] = NestedLoopDesc(
loopcnt=(12, 10, 4),
usize_gbuf=(0, 1000, 800),
usize_regf=(0, 3, 1),
unit_access=((0, 1000, 800), (0, 1000, 800), (3, 9, 7), (1, 1, 1)),
data_loops=(DataDimLoops(le.IFM,
le.OFM), DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)),
unit_ops=1,
unit_time=1)
# Fake nested loop, with zero ifmap size.
self.nld['ZERO_IFM'] = NestedLoopDesc(
loopcnt=(12, 10, 4),
usize_gbuf=(9, 0, 800),
usize_regf=(3, 0, 1),
unit_access=((9, 0, 800), (9, 0, 800), (3, 9, 7), (1, 1, 1)),
data_loops=(DataDimLoops(le.IFM,
le.OFM), DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)),
unit_ops=1,
unit_time=1)
# Fake partition scheme.
self.part = PartitionScheme(range(pe.NUM), ((1, 1), ) * pe.NUM)
# Fake buffer sharing scheme.
self.bufshr = BufShrScheme(proc_region, self.part)
# Options.
self.options = {}
# Basic.
self.options['BASE'] = Option(ntops=2**30)
# Multiprocessing.
self.options['MP'] = Option(ntops=2**30, nprocesses=8)
# Limited top schemes.
self.options['NTOPS'] = Option(ntops=10)
# Bypass.
self.options['BYP'] = Option(sw_gbuf_bypass=(True, ) * 3, ntops=2**30)
# Bypass solver.
self.options['BYPSOL'] = Option(sw_gbuf_bypass=(True, ) * 3,
sw_solve_loopblocking=True,
ntops=2**30)
# Access forwarding.
self.options['ACCFWD'] = Option(hw_access_forwarding=True, ntops=2**30)
# Buffer sharing.
self.options['BUFSHR'] = Option(hw_gbuf_sharing=True, ntops=2**30)
# Buffer sharing with bypassing.
self.options['BUFSHR-BYP'] = Option(sw_gbuf_bypass=(True, ) * 3,
hw_gbuf_sharing=True,
ntops=2**30)
# Constraint.
self.none_cstr = SchedulingConstraint()
self.cstr = SchedulingConstraint(topifm=1, topbat=1)
# Cost.
self.cost = Cost(mac_op=1,
mem_hier=(200, 6, 2, 1),
noc_hop=50,
idl_unit=50)
def setUp(self):
# Workload.
self.layer = {}
self.layer['BASE'] = ConvLayer(12, 10, 28, 3)
self.layer['LGFIL'] = ConvLayer(2, 4, 28, 20)
self.layer['POOL'] = PoolingLayer(32, 28, 2)
self.batch_size = 4
# Resource.
self.resource = {}
dim_array = PhyDim2(16, 16)
proc_region = NodeRegion(origin=PhyDim2(0, 0),
dim=PhyDim2(1, 1),
type=NodeRegion.PROC)
data_regions = (NodeRegion(origin=PhyDim2(0, 0),
dim=PhyDim2(1, 1),
type=NodeRegion.DATA), )
# Typical resource.
self.resource['BASE'] = Resource(proc_region=proc_region,
data_regions=data_regions,
dim_array=dim_array,
size_gbuf=65536,
size_regf=64)
# Larger resource with sufficient capacity, to make all schemes valid.
self.resource['LG'] = Resource(proc_region=proc_region,
data_regions=data_regions,
dim_array=dim_array,
size_gbuf=1024**3,
size_regf=1024**3)
# Small resource.
self.resource['SM'] = Resource(proc_region=proc_region,
data_regions=data_regions,
dim_array=dim_array,
size_gbuf=4096,
size_regf=16)
# Nested loop description after mapping.
self.nld = {}
self.nld['BASE'] = next(
MapStrategyEyeriss(self.layer['BASE'], self.batch_size,
dim_array).gen_nested_loop_desc())
self.nld['LGFIL'] = next(
MapStrategyEyeriss(self.layer['LGFIL'], self.batch_size,
dim_array).gen_nested_loop_desc())
self.nld['POOL'] = next(
MapStrategyEyeriss(self.layer['POOL'], self.batch_size,
dim_array).gen_nested_loop_desc())
# Fake nested loop, with zero filter size.
self.nld['ZERO_FIL'] = NestedLoopDesc(
loopcnt=(12, 10, 4),
usize_gbuf=(0, 1000, 800),
usize_regf=(0, 3, 1),
unit_access=((0, 1000, 800), (0, 1000, 800), (3, 9, 7), (1, 1, 1)),
data_loops=(DataDimLoops(le.IFM,
le.OFM), DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)),
unit_ops=1,
unit_time=1)
# Fake nested loop, with zero ifmap size.
self.nld['ZERO_IFM'] = NestedLoopDesc(
loopcnt=(12, 10, 4),
usize_gbuf=(9, 0, 800),
usize_regf=(3, 0, 1),
unit_access=((9, 0, 800), (9, 0, 800), (3, 9, 7), (1, 1, 1)),
data_loops=(DataDimLoops(le.IFM,
le.OFM), DataDimLoops(le.IFM, le.BAT),
DataDimLoops(le.OFM, le.BAT)),
unit_ops=1,
unit_time=1)
# Options.
self.options = {}
# Basic.
self.options['BASE'] = Option(ntops=2**30)
# Multiprocessing.
self.options['MP'] = Option(ntops=2**30, nprocesses=8)
# Limited top schemes.
self.options['NTOPS'] = Option(ntops=10)
# Bypass.
self.options['BYP'] = Option(sw_gbuf_bypass=(True, ) * 3, ntops=2**30)
# Bypass solver.
self.options['BYPSOL'] = Option(sw_gbuf_bypass=(True, ) * 3,
sw_solve_loopblocking=True,
ntops=2**30)
# Cost.
self.cost = Cost(mac_op=1,
mem_hier=(200, 6, 2, 1),
noc_hop=50,
unit_static=50)
# Partition occupation.
self.part_occ = 0.91