st.codegen.backend.cuda.BackendCUDA Class Reference

Inheritance diagram for st.codegen.backend.cuda.BackendCUDA:

Collaboration diagram for st.codegen.backend.cuda.BackendCUDA:

Public Member Functions
def	__init__ (self, VECLEN=32, LID=None, ocl=False)
def	checkConfig (self)
def	setCodeGen (self, codegen)
def	genVectorLoop (self, CodeBlock group)
def	gen_lhs (self, Buffer buf, List[int] offset, rel=None, dimrels=None)
def	gen_rhs (self, Expr comp, List[int] shift, List[int] offset, rel=None, dimrels=None)
def	declare_reg (self, name, CodeBlock block)
def	declare_vec (self, name, CodeBlock block)
def	declare_gridref (self, Grid grid, CodeBlock block)
def	store_vecbuf (self, vecbuf_name, reg_name, CodeBlock block)
def	declare_buf (self, Buffer buf, CodeBlock block)
def	genStoreLoop (self, CodeBlock group)
def	storeTile (self, Buffer buf, CodeBlock group)
def	genStoreTileLoop (self, CodeBlock group, dims)
def	store (self, Buffer buf, CodeBlock group)
def	read_aligned (self, Grid grid, offset, str name, CodeBlock block, rel=None)
def	merge (self, rego, regl, regr, dim, shift, CodeBlock block)
Public Member Functions inherited from st.codegen.backend.base.Backend
def	__init__ (self)
def	setCodeGen (self, codegen)
def	setLayout (self, layout)
def	prequel (self, toplevel)
def	checkConfig (self)
def	declare_buf (self, Buffer buf, CodeBlock block)
def	declare_gridref (self, Grid grid, CodeBlock block)
def	stride (self, dim)
def	declare_vecbuf (self, Grid grid, vec_shift, CodeBlock block)
def	genVectorLoop (self, CodeBlock group)
def	genStoreLoop (self, CodeBlock group)
def	store (self, Buffer buf, CodeBlock group)
def	storeTile (self, Buffer buf, CodeBlock group)
def	genStoreLoc (self, Grid grid, shift, offset, rel, dimrels)
def	genStoreTileLoop (self, CodeBlock group, dims)
def	gen_lhs (self, Buffer buf, List[int] offset, rel=None, dimrels=None)
def	gen_rhs (self, Expr comp, List[int] shift, List[int] offset, rel=None, dimrels=None)
def	declare_reg (self, name, CodeBlock block)
def	declare_vec (self, name, CodeBlock block)
def	store_vecbuf (self, vecbuf_name, reg_name, CodeBlock block)
def	merge (self, rego, regl, regr, dim, shift, CodeBlock block)
def	read_aligned (self, Grid grid, offset, str name, CodeBlock block, rel=None)

Public Attributes
	VECLEN
	ocl
	LID
	STRIDE
	codegen
	printer
Public Attributes inherited from st.codegen.backend.base.Backend
	printer
	codegen
	prec
	layout
	STRIDE
	VECLEN
	ALIGNED

Static Public Attributes
str	LID = "threadIdx.x"
Static Public Attributes inherited from st.codegen.backend.base.Backend
str	preffix = "_cg"

Additional Inherited Members
Static Public Member Functions inherited from st.codegen.backend.base.Backend
def	gridref_name (grid)
def	vectmp_name (idx)
def	vecbuf_name (grid, vec_shift)
def	vecreg_name (grid, vec_shift)
def	index_name (idx)
def	rel_name (idx=None)

Constructor & Destructor Documentation

__init__()

def st.codegen.backend.cuda.BackendCUDA.__init__	(		self,
			VECLEN = 32,
			LID = None,
			ocl = False
	)

Reimplemented from st.codegen.backend.base.Backend.

Reimplemented in st.codegen.backend.cuda.BackendCuFlex.

Here is the call graph for this function:

Member Function Documentation

checkConfig()

def st.codegen.backend.cuda.BackendCUDA.checkConfig

(

self

)

Reimplemented from st.codegen.backend.base.Backend.

Reimplemented in st.codegen.backend.cuda.BackendCuFlex.

declare_buf()

def st.codegen.backend.cuda.BackendCUDA.declare_buf	(		self,
		Buffer	buf,
		CodeBlock	block
	)

Reimplemented from st.codegen.backend.base.Backend.

declare_gridref()

def st.codegen.backend.cuda.BackendCUDA.declare_gridref	(		self,
		Grid	grid,
		CodeBlock	block
	)

Reimplemented from st.codegen.backend.base.Backend.

Here is the call graph for this function:

declare_reg()

def st.codegen.backend.cuda.BackendCUDA.declare_reg	(		self,
			name,
		CodeBlock	block
	)

Reimplemented from st.codegen.backend.base.Backend.

declare_vec()

def st.codegen.backend.cuda.BackendCUDA.declare_vec	(		self,
			name,
		CodeBlock	block
	)

Reimplemented from st.codegen.backend.base.Backend.

Reimplemented in st.codegen.backend.cuda.BackendCuFlex.

gen_lhs()

def st.codegen.backend.cuda.BackendCUDA.gen_lhs	(		self,
		Buffer	buf,
		List[int]	offset,
			rel = None,
			dimrels = None
	)

Reimplemented from st.codegen.backend.base.Backend.

gen_rhs()

def st.codegen.backend.cuda.BackendCUDA.gen_rhs	(		self,
		Expr	comp,
		List[int]	shift,
		List[int]	offset,
			rel = None,
			dimrels = None
	)

:param comp: The expression to print
:param shift: The shift of scatter
:param offset: Scattered from
:param rel: Added offset when using loops
:return:

Reimplemented from st.codegen.backend.base.Backend.

genStoreLoop()

def st.codegen.backend.cuda.BackendCUDA.genStoreLoop	(		self,
		CodeBlock	group
	)

Reimplemented from st.codegen.backend.base.Backend.

genStoreTileLoop()

def st.codegen.backend.cuda.BackendCUDA.genStoreTileLoop	(		self,
		CodeBlock	group,
			dims
	)

Reimplemented from st.codegen.backend.base.Backend.

Here is the call graph for this function:

genVectorLoop()

def st.codegen.backend.cuda.BackendCUDA.genVectorLoop	(		self,
		CodeBlock	group
	)

Reimplemented from st.codegen.backend.base.Backend.

merge()

def st.codegen.backend.cuda.BackendCUDA.merge	(		self,
			rego,
			regl,
			regr,
			dim,
			shift,
		CodeBlock	block
	)

Reimplemented from st.codegen.backend.base.Backend.

Reimplemented in st.codegen.backend.cuda.BackendCuFlex.

read_aligned()

def st.codegen.backend.cuda.BackendCUDA.read_aligned	(		self,
		Grid	grid,
			offset,
		str	name,
		CodeBlock	block,
			rel = None
	)

Reimplemented from st.codegen.backend.base.Backend.

Reimplemented in st.codegen.backend.cuda.BackendCuFlex.

setCodeGen()

def st.codegen.backend.cuda.BackendCUDA.setCodeGen	(		self,
			codegen
	)

Reimplemented from st.codegen.backend.base.Backend.

store()

def st.codegen.backend.cuda.BackendCUDA.store	(		self,
		Buffer	buf,
		CodeBlock	group
	)

Reimplemented from st.codegen.backend.base.Backend.

Here is the call graph for this function:

store_vecbuf()

def st.codegen.backend.cuda.BackendCUDA.store_vecbuf	(		self,
			vecbuf_name,
			reg_name,
		CodeBlock	block
	)

Reimplemented from st.codegen.backend.base.Backend.

Reimplemented in st.codegen.backend.cuda.BackendCuFlex.

storeTile()

def st.codegen.backend.cuda.BackendCUDA.storeTile	(		self,
		Buffer	buf,
		CodeBlock	group
	)

Reimplemented from st.codegen.backend.base.Backend.

Here is the call graph for this function:

Member Data Documentation

codegen

st.codegen.backend.cuda.BackendCUDA.codegen

LID [1/2]

str st.codegen.backend.cuda.BackendCUDA.LID = "threadIdx.x"

static

LID [2/2]

st.codegen.backend.cuda.BackendCUDA.LID

ocl

st.codegen.backend.cuda.BackendCUDA.ocl

printer

st.codegen.backend.cuda.BackendCUDA.printer

STRIDE

st.codegen.backend.cuda.BackendCUDA.STRIDE

VECLEN

st.codegen.backend.cuda.BackendCUDA.VECLEN

---

The documentation for this class was generated from the following file:

• codegen/st/codegen/backend/cuda.py