Details:
- NOTE: This is a merge commit of 'master' of git://github.com/amd/blis
into 'amd-master' of flame/blis.
- Fixed a bug in the downstream value of BLIS_NUM_ARCHS, which was
inadvertantly not incremented when the Zen2 subconfiguration was
added.
- In bli_gemm_front(), added a missing conditional constraint around the
call to bli_gemm_small() that ensures that the computation precision
of C matches the storage precision of C.
- In bli_syrk_front(), reorganized and relocated the notrans/trans logic
that existed around the call to bli_syrk_small() into bli_syrk_small()
to minimize the calling code footprint and also to bring that code
into stylistic harmony with similar code in bli_gemm_front() and
bli_trsm_front(). Also, replaced direct accessing of obj_t fields with
proper accessor static functions (e.g. 'a->dim[0]' becomes
'bli_obj_length( a )').
- Added #ifdef BLIS_ENABLE_SMALL_MATRIX guard around prototypes for
bli_gemm_small(), bli_syrk_small(), and bli_trsm_small(). This is
strictly speaking unnecessary, but it serves as a useful visual cue to
those who may be reading the files.
- Removed cpp macro-protected small matrix debugging code from
bli_trsm_front.c.
- Added a GCC_OT_9_1_0 variable to build/config.mk.in to facilitate gcc
version check for availability of -march=znver2, and added appropriate
support to configure script.
- Cleanups to compiler flags common to recent AMD microarchitectures in
config/zen/amd_config.mk, including: removal of -march=znver1 et al.
from CKVECFLAGS (since the -march flag is added within make_defs.mk);
setting CRVECFLAGS similarly to CKVECFLAGS.
- Cleanups to config/zen/bli_cntx_init_zen.c.
- Cleanups, added comments to config/zen/make_defs.mk.
- Cleanups to config/zen2/make_defs.mk, including making use of newly-
added GCC_OT_9_1_0 and existing GCC_OT_6_1_0 to choose the correct
set of compiler flags based on the version of gcc being used.
- Reverted downstream changes to test/test_gemm.c.
- Various whitespace/comment changes.
Details:
- Implemented a new sub-framework within BLIS to support the management
of code and kernels that specifically target matrix problems for which
at least one dimension is deemed to be small, which can result in long
and skinny matrix operands that are ill-suited for the conventional
level-3 implementations in BLIS. The new framework tackles the problem
in two ways. First the stripped-down algorithmic loops forgo the
packing that is famously performed in the classic code path. That is,
the computation is performed by a new family of kernels tailored
specifically for operating on the source matrices as-is (unpacked).
Second, these new kernels will typically (and in the case of haswell
and zen, do in fact) include separate assembly sub-kernels for
handling of edge cases, which helps smooth performance when performing
problems whose m and n dimension are not naturally multiples of the
register blocksizes. In a reference to the sub-framework's purpose of
supporting skinny/unpacked level-3 operations, the "sup" operation
suffix (e.g. gemmsup) is typically used to denote a separate namespace
for related code and kernels. NOTE: Since the sup framework does not
perform any packing, it targets row- and column-stored matrices A, B,
and C. For now, if any matrix has non-unit strides in both dimensions,
the problem is computed by the conventional implementation.
- Implemented the default sup handler as a front-end to two variants.
bli_gemmsup_ref_var2() provides a block-panel variant (in which the
2nd loop around the microkernel iterates over n and the 1st loop
iterates over m), while bli_gemmsup_ref_var1() provides a panel-block
variant (2nd loop over m and 1st loop over n). However, these variants
are not used by default and provided for reference only. Instead, the
default sup handler calls _var2m() and _var1n(), which are similar
to _var2() and _var1(), respectively, except that they defer to the
sup kernel itself to iterate over the m and n dimension, respectively.
In other words, these variants rely not on microkernels, but on
so-called "millikernels" that iterate along m and k, or n and k.
The benefit of using millikernels is a reduction of function call
and related (local integer typecast) overhead as well as the ability
for the kernel to know which micropanel (A or B) will change during
the next iteration of the 1st loop, which allows it to focus its
prefetching on that micropanel. (In _var2m()'s millikernel, the upanel
of A changes while the same upanel of B is reused. In _var1n()'s, the
upanel of B changes while the upanel of A is reused.)
- Added a new configure option, --[en|dis]able-sup-handling, which is
enabled by default. However, the default thresholds at which the
default sup handler is activated are set to zero for each of the m, n,
and k dimensions, which effectively disables the implementation. (The
default sup handler only accepts the problem if at least one dimension
is smaller than or equal to its corresponding threshold. If all
dimensions are larger than their thresholds, the problem is rejected
by the sup front-end and control is passed back to the conventional
implementation, which proceeds normally.)
- Added support to the cntx_t structure to track new fields related to
the sup framework, most notably:
- sup thresholds: the thresholds at which the sup handler is called.
- sup handlers: the address of the function to call to implement
the level-3 skinny/unpacked matrix implementation.
- sup blocksizes: the register and cache blocksizes used by the sup
implementation (which may be the same or different from those used
by the conventional packm-based approach).
- sup kernels: the kernels that the handler will use in implementing
the sup functionality.
- sup kernel prefs: the IO preference of the sup kernels, which may
differ from the preferences of the conventional gemm microkernels'
IO preferences.
- Added a bool_t to the rntm_t structure that indicates whether sup
handling should be enabled/disabled. This allows per-call control
of whether the sup implementation is used, which is useful for test
drivers that wish to switch between the conventional and sup codes
without having to link to different copies of BLIS. The corresponding
accessor functions for this new bool_t are defined in bli_rntm.h.
- Implemented several row-preferential gemmsup kernels in a new
directory, kernels/haswell/3/sup. These kernels include two general
implementation types--'rd' and 'rv'--for the 6x8 base shape, with
two specialized millikernels that embed the 1st loop within the kernel
itself.
- Added ref_kernels/3/bli_gemmsup_ref.c, which provides reference
gemmsup microkernels. NOTE: These microkernels, unlike the current
crop of conventional (pack-based) microkernels, do not use constant
loop bounds. Additionally, their inner loop iterates over the k
dimension.
- Defined new typedef enums:
- stor3_t: captures the effective storage combination of the level-3
problem. Valid values are BLIS_RRR, BLIS_RRC, BLIS_RCR, etc. A
special value of BLIS_XXX is used to denote an arbitrary combination
which, in practice, means that at least one of the operands is
stored according to general stride.
- threshid_t: captures each of the three dimension thresholds.
- Changed bli_adjust_strides() in bli_obj.c so that bli_obj_create()
can be passed "-1, -1" as a lazy request for row storage. (Note that
"0, 0" is still accepted as a lazy request for column storage.)
- Added support for various instructions to bli_x86_asm_macros.h,
including imul, vhaddps/pd, and other instructions related to integer
vectors.
- Disabled the older small matrix handling code inserted by AMD in
bli_gemm_front.c, since the sup framework introduced in this commit
is intended to provide a more generalized solution.
- Added test/sup directory, which contains standalone performance test
drivers, a Makefile, a runme.sh script, and an 'octave' directory
containing scripts compatible with GNU Octave. (They also may work
with matlab, but if not, they are probably close to working.)
- Reinterpret the storage combination string (sc_str) in the various
level-3 testsuite modules (e.g. src/test_gemm.c) so that the order
of each matrix storage char is "cab" rather than "abc".
- Comment updates in level-3 BLAS API wrappers in frame/compat.
Details:
- Implemented a sophisticated data structure and set of APIs that track
the small blocks of memory (around 80-100 bytes each) used when
creating nodes for control and thread trees (cntl_t and thrinfo_t) as
well as thread communicators (thrcomm_t). The purpose of the small
block allocator, or sba, is to allow the library to transition into a
runtime state in which it does not perform any calls to malloc() or
free() during normal execution of level-3 operations, regardless of
the threading environment (potentially multiple application threads
as well as multiple BLIS threads). The functionality relies on a new
data structure, apool_t, which is (roughly speaking) a pool of
arrays, where each array element is a pool of small blocks. The outer
pool, which is protected by a mutex, provides separate arrays for each
application thread while the arrays each handle multiple BLIS threads
for any given application thread. The design minimizes the potential
for lock contention, as only concurrent application threads would
need to fight for the apool_t lock, and only if they happen to begin
their level-3 operations at precisely the same time. Thanks to Kiran
Varaganti and AMD for requesting this feature.
- Added a configure option to disable the sba pools, which are enabled
by default; renamed the --[dis|en]able-packbuf-pools option to
--[dis|en]able-pba-pools; and rewrote the --help text associated with
this new option and consolidated it with the --help text for the
option associated with the sba (--[dis|en]able-sba-pools).
- Moved the membrk field from the cntx_t to the rntm_t. We now pass in
a rntm_t* to the bli_membrk_acquire() and _release() APIs, just as we
do for bli_sba_acquire() and _release().
- Replaced all calls to bli_malloc_intl() and bli_free_intl() that are
used for small blocks with calls to bli_sba_acquire(), which takes a
rntm (in addition to the bytes requested), and bli_sba_release().
These latter two functions reduce to the former two when the sba pools
are disabled at configure-time.
- Added rntm_t* arguments to various cntl_t and thrinfo_t functions, as
required by the new usage of bli_sba_acquire() and _release().
- Moved the freeing of "old" blocks (those allocated prior to a change
in the block_size) from bli_membrk_acquire_m() to the implementation
of the pool_t checkout function.
- Miscellaneous improvements to the pool_t API.
- Added a block_size field to the pblk_t.
- Harmonized the way that the trsm_ukr testsuite module performs packing
relative to that of gemmtrsm_ukr, in part to avoid the need to create
a packm control tree node, which now requires a rntm_t that has been
initialized with an sba and membrk.
- Re-enable explicit call bli_finalize() in testsuite so that users who
run the testsuite with memory tracing enabled can check for memory
leaks.
- Manually imported the compact/minor changes from 61441b24 that cause
the rntm to be copied locally when it is passed in via one of the
expert APIs.
- Reordered parameters to various bli_thrcomm_*() functions so that the
thrcomm_t* to the comm being modified is last, not first.
- Added more descriptive tracing for allocating/freeing small blocks and
formalized via a new configure option: --[dis|en]able-mem-tracing.
- Moved some unused scalm code and headers into frame/1m/other.
- Whitespace changes to bli_pthread.c.
- Regenerated build/libblis-symbols.def.
Details:
- Removed explicit reference to The University of Texas at Austin in the
third clause of the license comment blocks of all relevant files and
replaced it with a more all-encompassing "copyright holder(s)".
- Removed duplicate words ("derived") from a few kernels' license
comment blocks.
- Homogenized license comment block in kernels/zen/3/bli_gemm_small.c
with format of all other comment blocks.
Details:
- Added a num_t datatype bitfield to the obj_t in the form of a new
info2 field in the obj_t. This change was made primarily so that in
the case of mixed-datatype gemm, the alpha scalar would not need to
be cast to the storage datatype of B (or A) before then being cast to
the computation datatype just before the macrokernel is called. This
double-casting regime could result in loss of precision if the storage
datatype of B (or A) is less than the computation precision. In
practice, it was likely not going to be a big deal since most usage of
alpha is for -1.0, 0.0, and 1.0 (or integer multiples thereof), which
can all be represented exactly in single or double precision.
- The type of objbits_t was changed to uint32_t, so the new format
potentially takes up the same space as the previous obj_t definition,
assuming no padding inserted by the compiler. Shrinking info to 32
bits and spilling over into a second field was chosen over using the
high 32 bits of a single 64-bit objbits_t info field because many of
the bitwise operations are performed with enums such as num_t, dom_t,
and prec_t, which may take on the type of 32-bit ints. It's easier to
just keep all of those bitwise operations in 32 bits than perform a
million typecasts throughout bli_type_defs.h and bli_obj_macro_defs.h
to ensure that the integers are treated as 64-bit for the purposes of
the ANDs, ORs, and bitshifts.
- Many comment updates.
- Thanks to Devin Matthews and Devangi Parikh for their feedback and
involvement during this commit cycle.
Details:
- Implemented support for gemm where A, B, and C may have different
storage datatypes, as well as a computational precision (and implied
computation domain) that may be different from the storage precision
of either A or B. This results in 128 different combinations, all
which are implemented within this commit. (For now, the mixed-datatype
functionality is only supported via the object API.) If desired, the
mixed-datatype support may be disabled at configure-time.
- Added a memory-intensive optimization to certain mixed-datatype cases
that requires a single m-by-n matrix be allocated (temporarily) per
call to gemm. This optimization aims to avoid the overhead involved in
repeatedly updating C with general stride, or updating C after a
typecast from the computation precision. This memory optimization may
be disabled at configure-time (provided that the mixed-datatype
support is enabled in the first place).
- Added support for testing mixed-datatype combinations to testsuite.
The user may test gemm with mixed domains, precisions, both, or
neither.
- Added a standalone test driver directory for building and running
mixed-datatype performance experiments.
- Defined a new variation of castm, castnzm, which operates like castm
except that imaginary values are not touched when casting a real
operand to a complex operand. (By contrast, in these situations castm
sets the imaginary components of the destination matrix to zero.)
- Defined bli_obj_imag_is_zero() and substituted calls in lieu of all
usages of bli_obj_imag_equals() that tested against BLIS_ZERO, and
also simplified the implementation of bli_obj_imag_equals().
- Fixed bad behavior from bli_obj_is_real() and bli_obj_is_complex()
when given BLIS_CONSTANT objects.
- Disabled dt_on_output field in auxinfo_t structure as well as all
accessor functions. Also commented out all usage of accessor
functions within macrokernels. (Typecasting in the microkernel is
still feasible, though probably unrealistic for now given the
additional complexity required.)
- Use void function pointer type (instead of void*) for storing function
pointers in bli_l0_fpa.c.
- Added documentation for using gemm with mixed datatypes in
docs/MixedDatatypes.md and example code in examples/oapi/11gemm_md.c.
- Defined level-1d operation xpbyd and level-1m operation xpbym.
- Added xpbym test module to testsuite.
- Updated frame/include/bli_x86_asm_macros.h with additional macros
(courtsey of Devin Matthews).
Details:
- Removed four trailing spaces after "BLIS" that occurs in most files'
commented-out license headers.
- Added UT copyright lines to some files. (These files previously had
only AMD copyright lines but were contributed to by both UT and AMD.)
- In some files' copyright lines, expanded 'The University of Texas' to
'The University of Texas at Austin'.
- Fixed various typos/misspellings in some license headers.
Details:
- Defined a new struct datatype, rntm_t (runtime), to house the thrloop
field of the cntx_t (context). The thrloop array holds the number of
ways of parallelism (thread "splits") to extract per level-3
algorithmic loop until those values can be used to create a
corresponding node in the thread control tree (thrinfo_t structure),
which (for any given level-3 invocation) usually happens by the time
the macrokernel is called for the first time.
- Relocating the thrloop from the cntx_t remedies a thread-safety issue
when invoking level-3 operations from two or more application threads.
The race condition existed because the cntx_t, a pointer to which is
usually queried from the global kernel structure (gks), is supposed to
be a read-only. However, the previous code would write to the cntx_t's
thrloop field *after* it had been queried, thus violating its read-only
status. In practice, this would not cause a problem when a sequential
application made a multithreaded call to BLIS, nor when two or more
application threads used the same parallelization scheme when calling
BLIS, because in either case all application theads would be using
the same ways of parallelism for each loop. The true effects of the
race condition were limited to situations where two or more application
theads used *different* parallelization schemes for any given level-3
call.
- In remedying the above race condition, the application or calling
library can now specify the parallelization scheme on a per-call basis.
All that is required is that the thread encode its request for
parallelism into the rntm_t struct prior to passing the address of the
rntm_t to one of the expert interfaces of either the typed or object
APIs. This allows, for example, one application thread to extract 4-way
parallelism from a call to gemm while another application thread
requests 2-way parallelism. Or, two threads could each request 4-way
parallelism, but from different loops.
- A rntm_t* parameter has been added to the function signatures of most
of the level-3 implementation stack (with the most notable exception
being packm) as well as all level-1v, -1d, -1f, -1m, and -2 expert
APIs. (A few internal functions gained the rntm_t* parameter even
though they currently have no use for it, such as bli_l3_packm().)
This required some internal calls to some of those functions to
be updated since BLIS was already using those operations internally
via the expert interfaces. For situations where a rntm_t object is
not available, such as within packm/unpackm implementations, NULL is
passed in to the relevant expert interfaces. This is acceptable for
now since parallelism is not obtained for non-level-3 operations.
- Revamped how global parallelism is encoded. First, the conventional
environment variables such as BLIS_NUM_THREADS and BLIS_*_NT are only
read once, at library initialization. (Thanks to Nathaniel Smith for
suggesting this to avoid repeated calls getenv(), which can be slow.)
Those values are recorded to a global rntm_t object. Public APIs, in
bli_thread.c, are still available to get/set these values from the
global rntm_t, though now the "set" functions have additional logic
to ensure that the values are set in a synchronous manner via a mutex.
If/when NULL is passed into an expert API (meaning the user opted to
not provide a custom rntm_t), the values from the global rntm_t are
copied to a local rntm_t, which is then passed down the function stack.
Calling a basic API is equivalent to calling the expert APIs with NULL
for the cntx and rntm parameters, which means the semantic behavior of
these basic APIs (vis-a-vis multithreading) is unchanged from before.
- Renamed bli_cntx_set_thrloop_from_env() to bli_rntm_set_ways_for_op()
and reimplemented, with the function now being able to treat the
incoming rntm_t in a manner agnostic to its origin--whether it came
from the application or is an internal copy of the global rntm_t.
- Removed various global runtime APIs for setting the number of ways of
parallelism for individual loops (e.g. bli_thread_set_*_nt()) as well
as the corresponding "get" functions. The new model simplifies these
interfaces so that one must either set the total number of threads, OR
set all of the ways of parallelism for each loop simultaneously (in a
single function call).
- Updated sandbox/ref99 according to above changes.
- Rewrote/augmented docs/Multithreading.md to document the three methods
(and two specific ways within each method) of requesting parallelism
in BLIS.
- Removed old, disabled code from bli_l3_thrinfo.c.
- Whitespace changes to code (e.g. bli_obj.c) and docs/BuildSystem.md.
Details:
- Converted most C preprocessor macros in bli_param_macro_defs.h and
bli_obj_macro_defs.h to static functions.
- Reshuffled some functions/macros to bli_misc_macro_defs.h and also
between bli_param_macro_defs.h and bli_obj_macro_defs.h.
- Changed obj_t-initializing macros in bli_type_defs.h to static
functions.
- Removed some old references to BLIS_TWO and BLIS_MINUS_TWO from
bli_constants.h.
- Whitespace changes in select files (four spaces to single tab).
Details:
- Added bli_setgetijm.c, which defines bli_setijm(), bli_getijm(), and
related functions that can be used to read and write individual
elements of an obj_t.
- Defined a new function, bli_obj_create_conf_to(), in bli_obj.c that will
create a new object with dimensions conformal to an existing object.
Transposition and conjugation states on the existing object are ignored,
as are structure and uplo fields.
- Defined a new function, bli_datatype_string(), in bli_obj.c that returns
a char* to a string representation of the name of each num_t datatype.
For example, BLIS_DOUBLE is "double" and BLIS_DCOMPLEX is "dcomplex".
BLIS_INT is included (as "int"), but BLIS_CONSTANT is not, and thus is
not a valid input argument to bli_datatype_string().
- Added calls to bli_init_once() to various functions in bli_obj.c, the
most important of which was bli_obj_create_without_buffer().
- Removed unintended/extra newline from the end of printv output.
- Whitespace changes to
- frame/base/bli_machval.c
- frame/base/bli_machval.h
- frame/0/copysc/bli_copysc.c
- Trivial changes to README.md and common.mk.
Details:
- Fixed a race condition in self-initialization whereby the bli_is_init
static variable could be erroneously read as TRUE by thread 1 while
thread 0 is still executing bli_init_apis(), thus allowing thread 1 to
use the library before it is actually ready. Thanks to to Minh Quan Ho
and Devin Matthews for pointing out this issue.
- Part of the solution to the aforementioned race condition was involved
replacing the runtime initialization of the global scalar constants
(e.g., BLIS_ONE, BLIS_ZERO, etc.) in bli_const.c with a static
initialization of those same constants. This eliminates the need for
bli_const_init() altogether. (The static initialization is made concise
via preprocess macros.)
- Defined bli_gks_query_cntx_noinit(), which behaves just like
bli_gks_query_cntx(), except that it does not call bli_init_once(). This
function is called in lieu of bli_gks_query_cntx() in bli_ind_init() and
bli_memsys_init() so as to not result in any recursion into
bli_init_once().
- Removed BLIS_ONE_HALF, BLIS_MINUS_ONE_HALF global scalar constants.
They have no use in BLIS or its test products, and we have little reason
to believe they are used by others.
- Removed testsuite/out file, which was accidentally committed as part
of 70640a3.
Details:
- Defined two new functions in bli_init.c: bli_init_once() and
bli_finalize_once(). Each is implemented with pthread_once(), which
guarantees that, among the threads that pass in the same pthread_once_t
data structure, exactly one thread will execute a user-defined function.
(Thus, there is now a runtime dependency against libpthread even when
multithreading is not enabled at configure-time.)
- Added calls to bli_init_once() to top-level user APIs for all
computational operations as well as many other functions in BLIS to
all but guarantee that BLIS will self-initialize through the normal
use of its functions.
- Rewrote and simplified bli_init() and bli_finalize() and related
functions.
- Added -lpthread to LDFLAGS in common.mk.
- Modified the bli_init_auto()/_finalize_auto() functions used by the
BLAS compatibility layer to take and return no arguments. (The
previous API that tracked whether BLIS was initialized, and then
only finalized if it was initialized in the same function, was too
cute by half and borderline useless because by default BLIS stays
initialized when auto-initialized via the compatibility layer.)
- Removed static variables that track initialization of the sub-APIs in
bli_const.c, bli_error.c, bli_init.c, bli_memsys.c, bli_thread, and
bli_ind.c. We don't need to track initialization at the sub-API level,
especially now that BLIS can self-initialize.
- Added a critical section around the changing of the error checking
level in bli_error.c.
- Deprecated bli_ind_oper_has_avail() as well as all functions
bli_<opname>_ind_get_avail(), where <opname> is a level-3 operation
name. These functions had no use cases within BLIS and likely none
outside of BLIS.
- Commented out calls to bli_init() and bli_finalize() in testsuite's
main() function, and likewise for standalone test drivers in 'test'
directory, so that self-initialization is exercised by default.
Details:
- Altered control tree node struct definitions so that all nodes have the
same struct definition, whose primary fields consist of a blocksize id,
a variant function pointer, a pointer to an optional parameter struct,
and a pointer to a (single) sub-node. This unified control tree type is
now named cntl_t.
- Changed the way control tree nodes are connected, and what computation
they represent, such that, for example, packing operations are now
associated with nodes that are "inline" in the tree, rather than off-
shoot braches. The original tree for the classic Goto gemm algorithm was
expressed (roughly) as:
blk_var2 -> blk_var3 -> blk_var1 -> ker_var2
| |
-> packb -> packa
and now, the same tree would look like:
blk_var2 -> blk_var3 -> packb -> blk_var1 -> packa -> ker_var2
Specifically, the packb and packa nodes perform their respective packing
operations and then recurse (without any loop) to a subproblem. This means
there are now two kinds of level-3 control tree nodes: partitioning and
non-partitioning. The blocked variants are members of the former, because
they iteratively partition off submatrices and perform suboperations on
those partitions, while the packing variants belong to the latter group.
(This change has the effect of allowing greatly simplified initialization
of the nodes, which previously involved setting many unused node fields to
NULL.)
- Changed the way thrinfo_t tree nodes are arranged to mirror the new
connective structure of control trees. That is, packm nodes are no longer
off-shoot branches of the main algorithmic nodes, but rather connected
"inline".
- Simplified control tree creation functions. Partitioning nodes are created
concisely with just a few fields needing initialization. By contrast, the
packing nodes require additional parameters, which are stored in a
packm-specific struct that is tracked via the optional parameters pointer
within the control tree struct. (This parameter struct must always begin
with a uint64_t that contains the byte size of the struct. This allows
us to use a generic function to recursively copy control trees.) gemm,
herk, and trmm control tree creation continues to be consolidated into
a single function, with the operation family being used to select
among the parameter-agnostic macro-kernel wrappers. A single routine,
bli_cntl_free(), is provided to free control trees recursively, whereby
the chief thread within a groups release the blocks associated with
mem_t entries back to the memory broker from which they were acquired.
- Updated internal back-ends, e.g. bli_gemm_int(), to query and call the
function pointer stored in the current control tree node (rather than
index into a local function pointer array). Before being invoked, these
function pointers are first cast to a gemm_voft (for gemm, herk, or trmm
families) or trsm_voft (for trsm family) type, which is defined in
frame/3/bli_l3_var_oft.h.
- Retired herk and trmm internal back-ends, since all execution now flows
through gemm or trsm blocked variants.
- Merged forwards- and backwards-moving variants by querying the direction
from routines as a function of the variant's matrix operands. gemm and
herk always move forward, while trmm and trsm move in a direction that
is dependent on which operand (a or b) is triangular.
- Added functions bli_thread_get_range_mdim(), bli_thread_get_range_ndim(),
each of which takes additional arguments and hides complexity in managing
the difference between the way ranges are computed for the four families
of operations.
- Simplified level-3 blocked variants according to the above changes, so that
the only steps taken are:
1. Query partitioning direction (forwards or backwards).
2. Prune unreferenced regions, if they exist.
3. Determine the thread partitioning sub-ranges.
<begin loop>
4. Determine the partitioning blocksize (passing in the partitioning
direction)
5. Acquire the curren iteration's partitions for the matrices affected
by the current variants's partitioning dimension (m, k, n).
6. Call the subproblem.
<end loop>
- Instantiate control trees once per thread, per operation invocation.
(This is a change from the previous regime in which control trees were
treated as stateless objects, initialized with the library, and shared
as read-only objects between threads.) This once-per-thread allocation
is done primarily to allow threads to use the control tree as as place
to cache certain data for use in subsequent loop iterations. Presently,
the only application of this caching is a mem_t entry for the packing
blocks checked out from the memory broker (allocator). If a non-NULL
control tree is passed in by the (expert) user, then the tree is copied
by each thread. This is done in bli_l3_thread_decorator(), in
bli_thrcomm_*.c.
- Added a new field to the context, and opid_t which tracks the "family"
of the operation being executed. For example, gemm, hemm, and symm are
all part of the gemm family, while herk, syrk, her2k, and syr2k are
all part of the herk family. Knowing the operation's family is necessary
when conditionally executing the internal (beta) scalar reset on on
C in blocked variant 3, which is needed for gemm and herk families,
but must not be performed for the trmm family (because beta has only
been applied to the current row-panel of C after the first rank-kc
iteration).
- Reexpressed 3m3 induced method blocked variant in frame/3/gemm/ind
to comform with the new control tree design, and renamed the macro-
kernel codes corresponding to 3m2 and 4m1b.
- Renamed bli_mem.c (and its APIs) to bli_memsys.c, and renamed/relocated
bli_mem_macro_defs.h from frame/include to frame/base/bli_mem.h.
- Renamed/relocated bli_auxinfo_macro_defs.h from frame/include to
frame/base/bli_auxinfo.h.
- Fixed a minor bug whereby the storage-to-ukr-preference matching
optimization in the various level-3 front-ends was not being applied
properly when the context indicated that execution would be via an
induced method. (Before, we always checked the native micro-kernel
corresponding to the datatype being executed, whereas now we check
the native micro-kernel corresponding to the datatype's real projection,
since that is the micro-kernel that is actually used by induced methods.
- Added an option to the testsuite to skip the testing of native level-3
complex implementations. Previously, it was always tested, provided that
the c/z datatypes were enabled. However, some configurations use
reference micro-kernels for complex datatypes, and testing these
implementations can slow down the testsuite considerably.
Details:
- Replaced all instances of bli_malloc() and bli_free() with one of:
- bli_malloc_pool()/bli_free_pool()
- bli_malloc_user()/bli_free_user()
- bli_malloc_intl()/bli_free_intl()
each of which can be configured to call malloc()/free() substitutes,
so long as the substitute functions have the same function type
signatures as malloc() and free() defined by C's stdlib.h. The _pool()
function is called when allocating blocks for the memory pools (used
for packing buffers, primarily), the _user() function is called when
obj_t's are created (via bli_obj_create() and friends), and the _intl()
function is called for internal use by BLIS, such as when creating
control tree nodes or temporary buffers for manipulating internal data
structures. Substitutes for any of the three types of bli_malloc() may
be specified by #defining the following pairs of cpp macros in
bli_kernel.h:
- BLIS_MALLOC_POOL/BLIS_FREE_POOL
- BLIS_MALLOC_USER/BLIS_FREE_USER
- BLIS_MALLOC_INTL/BLIS_FREE_INTL
to be the name of the substitute functions. (Obviously, the object
code that contains these functions must be provided at link-time.)
These macros default to malloc() and free(). Subsitute functions are
also automatically prototyped by BLIS (in bli_malloc_prototypes.h).
- Removed definitions for bli_malloc() and bli_free().
- Note that bli_malloc_pool() and bli_malloc_user() are now defined in
terms of a new function, bli_malloc_align(), which aligns memory to an
arbitrary (power of two) alignment boundary, but does so manually,
whereas before alignment was performed behind the scenes by
posix_memalign(). Currently, bli_malloc_intl() is defined in terms
of bli_malloc_noalign(), which serves as a simple wrapper to the
designated function that is passed in (e.g. BLIS_MALLOC_INTL).
Similarly, there are bli_free_align() and bli_free_noalign(), which
are used in concert with their bli_malloc_*() counterparts.
Details:
- Applied a patch submitted by Devin Matthews that:
- implements subtle changes to handling of somewhat unusual cases of
row and column strides to accommodate certail tensor cases, which
includes adding dimension parameters to _is_col_tilted() and
_is_row_tilted() macros,
- simplifies how buffers are sized when requested BLIS-allocated
objects,
- re-consolidates bli_adjust_strides_*() into one function, and
- defines 'restrict' keyword as a "nothing" macro for C++ and pre-C99
environments.
Details:
- Separated bli_adjust_strides() into _alloc() and _attach() flavors so
that the latter can avoid a test performed by the former, in which the
rs and cs are overridden and set to zero if either matrix dimension is
zero. Actually, we also disable this overridding behavior, even for the
_alloc() case, since keeping the original strides (probably) does not
hurt anything. The original code has been kept commented-out, though,
in case an unintended consequence is later discovered.
- Fixed a typo in an error check for general stride cases where rs == cs.
Details:
- Expanded/updated interface for bli_get_range_weighted() and
bli_get_range() so that the direction of movement is specified in the
function name (e.g. bli_get_range_l2r(), bli_get_range_weighted_t2b())
and also so that the object being partitioned is passed instead of an
uplo parameter. Updated invocations in level-3 blocked variants, as
appropriate.
- (Re)implemented bli_get_range_*() and bli_get_range_weighted_*() to
carefully take into account the location of the diagonal when computing
ranges so that the area of each subpartition (which, in all present
level-3 operations, is proportional to the amount of computation
engendered) is as equal as possible.
- Added calls to a new class of routines to all non-gemm level-3 blocked
variants:
bli_<oper>_prune_unref_mparts_[mnk]()
where <oper> is herk, trmm, or trsm and [mnk] is chosen based on which
dimension is being partitioned. These routines call a more basic
routine, bli_prune_unref_mparts(), to prune unreferenced/unstored
regions from matrices and simultaneously adjust other matrices which
share the same dimension accordingly.
- Simplified herk_blk_var2f, trmm_blk_var1f/b as a result of more the
new pruning routines.
- Fixed incorrect blocking factors passed into bli_get_range_*() in
bli_trsm_blk_var[12][fb].c
- Added a new test driver in test/thread_ranges that can exercise the new
bli_get_range_*() and bli_get_range_weighted_*() under a range of
conditions.
- Reimplemented m and n fields of obj_t as elements in a "dim"
array field so that dimensions could be queried via index constant
(e.g. BLIS_M, BLIS_N). Adjusted/added query and modification
macros accordingly.
- Defined mdim_t type to enumerate BLIS_M and BLIS_N indexing values.
- Added bli_round() macro, which calls C math library function round(),
and bli_round_to_mult(), which rounds a value to the nearest multiple
of some other value.
- Added miscellaneous pruning- and mdim_t-related macros.
- Renamed bli_obj_row_offset(), bli_obj_col_offset() macros to
bli_obj_row_off(), bli_obj_col_off().
Details:
- Added an imaginary stride field ("is") to obj_t.
- Renamed bli_obj_set_incs() macro to bli_obj_set_strides().
- Defined bli_obj_imag_stride() and bli_obj_set_imag_stride() and
added invocations in key locations.
- Added some basic error-checking related to imaginary stride.
- For now, imaginary stride will not be exposed into the most-used
BLIS APIs such as bli_obj_create(), and certainly not the
computational APIs such as bli_dgemm().
Details:
- Relaxed a long-held requirement in register blocksizes that required
the kernel programmer to choose a KC that was divisible by both MR
and NR. This was very constraining on some architectures that did not
use register blocksizes that were powers of two. The constraint is
now enforced only for trmm and trsm, where it is needed, and it is
now handled by "nudging" kc upward at runtime, if necessary, to be a
multiple of MR or NR, as needed.
- Defined bli_trmm_determine_kc_[fb]() and bli_trsm_determine_kc_[fb](),
which determine blocksizes for trmm and trsm, taking special care to
"nudge" the kc dimension up to a multiple of MR or NR, as needed.
- Changed bli_trmm_blk_var3[fb].c to call bli_trmm_determine_kc_[fb]()
instead of bli_determine_blocksize_[fb]().
- Added safeguard to bli_align_dim_to_mult() that returns the dimension
unmodified if the dimension multiple is zero (to avoid division by
zero).
- Removed cpp guard/check for KC % MR == 0 and KC % NR == 0 from
bli_kernel_macro_defs.h.
- Whitespace, variable name changes to bli_blocksize.c.
- Removed old commented code from bli_gemm_cntl.c.
Details:
- Fixed a bug in the test suite for the trsm_ukr and gemmtrsm_ukr test
modules whereby the uplo bits of some packed matrix objects were not
being set properly, resulting in false FAILURE results for those
tests. Thanks to Tyler Smith for bringing this issue to my attention.
- Fixed a bug in bli_obj_alloc_buffer() that caused an unnecessary
"not yet implemented" abort() when creating a 1x1 object with non-unit
strides.
Details:
- Updated copyright headers to include "at Austin" in the name of the
University of Texas.
- Updated the copyright years of a few headers to 2014 (from 2011 and
2012).
Details:
- Fixed various bugs in packm_*_cxk(), the 4m/3m micro-kernels, and
elsewhere in the framework that were not yet set up to work properly
when BLIS_ENABLE_C99_COMPLEX is defined in bli_config.h
- Extensive changes to f2c-derived files in frame/compat/f2c to allow
C99 complex storage. Most of these changes center around accessing
real and imaginary components via bli_?real()/bli_?imag() accessor
macros, and setting of values via bli_?sets() assignment macros.
(Thanks to Vladimir Sukarev for pointing out that _ENABLE_C99_COMPLEX
was broken.)
Details:
- Added set of basic scalar macros that take arguments' real and
imaginary components separately, named like the previous set except
with the "ris" (instead of "s") suffix.
- Redefined the previous set of scalar macros (those that take arguments
"whole") in terms of the new "ri" set.
- Renamed setris and getris macros to sets and gets.
- Renamed setimag0 macros to seti0s.
- Use bli_?1 macro instead of a local constant in bla_trmv.c, bla_trsv.c.
Details:
- Added infrastructure to support a new scalar representation, whereby
every object contains an internal scalar that defaults to 1.0. This
facilitates passing scalars around without having to house them in
separate objects. These "attached" scalars are stored in the internal
atom_t field of the obj_t struct, and are always stored to be the same
datatype as the object to which they are attached. Level-3 variants no
longer take scalar arguments, however, level-3 internal back-ends stll
do; this is so that the calling function can perform subproblems such
as C := C - alpha * A * B on-the-fly without needing to change either
of the scalars attached to A or B.
- Removed scalar argument from packm_int().
- Observe and apply attached scalars in scalm_int(), and removed scalar
from interface of scalm_unb_var1().
- Renamed the following functions (and corresponding invocations):
bli_obj_init_scalar_copy_of()
-> bli_obj_scalar_init_detached_copy_of()
bli_obj_init_scalar() -> bli_obj_scalar_init_detached()
bli_obj_create_scalar_with_attached_buffer()
-> bli_obj_create_1x1_with_attached_buffer()
bli_obj_scalar_equals() -> bli_obj_equals()
- Defined new functions:
bli_obj_scalar_detach()
bli_obj_scalar_attach()
bli_obj_scalar_apply_scalar()
bli_obj_scalar_reset()
bli_obj_scalar_has_nonzero_imag()
bli_obj_scalar_equals()
- Placed all bli_obj_scalar_* functions in a new file, bli_obj_scalar.c.
- Renamed the following macros:
bli_obj_scalar_buffer() -> bli_obj_buffer_for_1x1()
bli_obj_is_scalar() -> bli_obj_is_1x1()
- Defined new macros to set and copy internal scalars between objects:
bli_obj_set_internal_scalar()
bli_obj_copy_internal_scalar()
- In level-3 internal back-ends, added conditional blocks where alpha and
beta are checked for non-unit-ness. Those values for alpha and beta are
applied to the scalars attached to aliases of A/B/C, as appropriate,
before being passed into the variant specified by the control tree.
- In level-3 blocked variants, pass BLIS_ONE into subproblems instead of
alpha and/or beta.
- In level-3 macro-kernels, changed how scalars are obtained. Now, scalars
attached to A and B are multiplied together to obtain alpha, while beta
is obtained directly from C.
- In level-3 front-ends, removed old function calls meant to provide
future support for mixed domain/precision. These can be added back later
once that functionality is given proper treatment. Also, removed the
creating of copy-casts of alpha and beta since typecasting of scalars
is now implicitly handled in the internal back-ends when alpha and
beta are applied to the attached scalars.
Details:
- Redefined dim_t and inc_t in terms of gint_t (instead of guint_t).
This will facilitate interoperability with Fortran in the future.
(Fortran does not support unsigned integers.)
- Redefined many instances of stride-related macros so that they return
or use the absolute value of the strides, rather than the raw strides
which may now be signed. Added new macros bli_is_row_stored_f() and
bli_is_col_stored_f(), which assume positive (forward-oriented) strides,
and changed the packm_blk_var[23] variants to use these macros instead
of the existing bli_is_row_stored(), bli_is_col_stored().
- Added/adjusted typecasting to to various functions/macros, including
bli_obj_alloc_buffer(), bli_obj_buffer_at_off(), and various pointer-
related macros in bli_param_macro_defs.h.
- Redefined bli_convert_blas_incv() macro so that the BLAS compatibility
layer properly handles situations where vector increments are negative.
Thanks to Vladimir Sukharev for pointing out this issue.
- Changed type of increment parameters in bli_adjust_strides() from dim_t
to inc_t. Likewise in bli_check_matrix_strides().
- Defined bli_check_matrix_object(), which checks for negative strides.
- Redefined bli_check_scalar_object() and bli_check_vector_object() so
that they also check for negative stride.
- Added instances of bli_check_matrix_object() to various operations'
_check routines.
Details:
- Fixed bugs in scalm and setm that resulted in segmentation faults when
beta is not the same type as the matrix operand. Thanks to Vladimir
Sukharev for reporting this bug.
- Changed axpym and scal2m front-ends in fashion similar to that of scalm
and setm; namely, the alpha scalar is copy-cast the type of the first
matrix operand.
- Changed the template and reference configurations' bli_config.h files
so that the number of memory allocator blocks of A and B are set based
on BLIS_MAX_NUM_THREADS.
- Comment updates to bli_obj.c and variable rename in bla_nrm2.c.
Details:
- Fixed various warnings output by gcc 4.6.3-1, including removing some
set-but-not-used variables and addressing some instances of typecasting
of pointer types to integer types of different sizes.
Details:
- Changed the way bli_type_defs.h defines integer types so that dim_t,
inc_t, doff_t, etc. are all defined in terms of gint_t (general signed
integer) or guint_t (general unsigned integer).
- Renamed Fortran types fchar and fint to f77_char and f77_int.
- Define f77_int as int64_t if a new configuration variable,
BLIS_ENABLE_BLIS2BLAS_INT64, is defined, and int32_t otherwise.
These types are defined in stdint.h, which is now included in blis.h.
- Renamed "complex" type in f2c files to "singlecomplex" and typedef'ed
in terms of scomplex.
- Renamed "char" type in f2c files to "character" and typedef'ed in terms
of char.
- Updated bla_amax() wrappers so that the return type is defined directly
as f77_int, rather than letting the prototype-generating macro decide
the type. This was the only use of GENTFUNC2I/GENTPROT2I-related macros,
so I removed them. Also, changed the body of the wrapper so that a
gint_t is passed into abmaxv, which is THEN typecast to an f77_int
before returning the value.
- Updated f2c code that accessed .r and .i fields of complex and
doublecomplex types so that they use .real and .imag instead (now that
we are using scomplex and dcomplex).
Details:
- Made substantial changes throughout the framework to decouple the leading
dimension (row or column stride) used within each packed micro-panel from
the corresponding register blocksize. It appears advantageous on some
systems to use, for example, packed micro-panels of A where the column
stride is greater than MR (whereas previously it was always equal to MR).
- Changes include:
- Added BLIS_EXTEND_[MNK]R_? macros, which specify how much extra padding
to use when packing micro-panels of A and B.
- Adjusted all packing routines and macro-kernels to use PACKMR and PACKNR
where appropriate, instead of MR and NR.
- Added pd field (panel dimension) to obj_t.
- New interface to bli_packm_cntl_obj_create().
- Renamed bli_obj_packed_length()/_width() macros to
bli_obj_padded_length()/_width().
- Removed local #defines for cache/register blocksizes in level-3 *_cntl.c.
- Print out new cache and register blocksize extensions in test suite.
- Also added new BLIS_EXTEND_[MNK]C_? macros for future use in using a larger
blocksize for edge cases, which can improve performance at the margins.
Details:
- Modified bli_obj_alloc_buffer() to allow allocating an empty buffer, and
modified bli_adjust_strides() to explicitly handle m = n = 0.
- Updated bli_check_matrix_strides() to allow cases where m = n = 0.
Details:
- Relaxed type checking in getsc so that the input object could be a constant
and not just a proper floating-point type. (If it is a constant, default to
extracting the dcomplex values.) Thanks to Bryan Marker for reporting this
bug.
- Added definition for bli_is_constant() in bli_param_macro_defs.h
- Comment updates to various level-0 scalar routines.
Details:
- Added new memory alignment constants:
BLIS_HEAP_STRIDE_ALIGN_SIZE (previously assumed to be same as SYSTEM_MEM)
BLIS_CONTIG_ADDR_ALIGN_SIZE (previously assumed to be same as PAGE_SIZE)
BLIS_STACK_BUF_ALIGN_SIZE (previously not enforced)
and renamed existing ones
BLIS_SYSTEM_MEM_ALIGN_SIZE -> BLIS_HEAP_ADDR_ALIGN_SIZE
BLIS_CONTIG_MEM_ALIGN_SIZE -> BLIS_CONTIG_STRIDE_ALIGN_SIZE
to better convey what the alignment factor is used for (and what it is
not used for).
- Removed BLIS_ENABLE_SYSTEM_MEM_ALIGN. Dynamic memory alignment is now
disabled by setting BLIS_HEAP_STRIDE_ALIGN_SIZE to 1.
- Inserted instances of __attribute__((aligned(BLIS_STACK_BUF_ALIGN_SIZE)))
into macro-kernels to specify stack alignment of temporary buffers.
- Modified test suite driver to output new constants.
- Removed bli_align_dim_to_sys() and bli_align_dim_to_cmem(). Instead, we now
use bli_align_dim_to_size(), which takes a third argument (the desired
alignment).
Details:
- Replaced scalar constant macro definitions in bli_const_defs.h with a single,
simplier macro in bli_obj_macro_defs.h.
- Updated invocations of old macros accordingly.
- Removed bli_const_defs.h.
Details:
- Renamed BLIS_MEMORY_ALIGNMENT_SIZE to BLIS_CONTIG_MEM_ALIGN_SIZE.
- Renamed BLIS_ENABLE_MEMORY_ALIGNMENT to BLIS_ENABLE_SYSTEM_MEM_ALIGN.
- Added BLIS_SYSTEM_MEM_ALIGN_SIZE, which controls only the alignment
passed into posix_memalign() or equivalent.
- Defined new function, bli_align_dim_to_cmem(), which applies the
contiguous memory alignment (rather than the system/malloc alignment).
Details:
- Pass panel strides through bli_align_dim_to_sys() to ensure that each
subsequent packed panel of A and B begins at an aligned address. (The
first panel is presumably aligned to system alignment because it is
aligned to a page boundary, which is typically much larger.)
- Rearranged code in packm_init_pack() to prevent additional conditional
blocks as a result of the aforementioned change.
- Adjusted contiguous memory allocator so that the system memory alignment
is used to allocate enough space for each block no matter what kind of
register blocking is used (even if register blocksize is unit and every
row/column needs maximal padding).
- Adjusted default blocksizes in reference configuration so that MC*KC
and KC*NC result in identical footprints for all datatypes.
Details:
- Changed all filename and function prefixes from 'bl2' to 'bli'.
- Changed the "blis2.h" header filename to "blis.h" and changed all
corresponding #include statements accordingly.
- Fixed incorrect association for Fran in CREDITS file.
