| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
|
|
|
| |
copyright owners. Much thanks for that Couriersud, the main creator
and contributor, but also to Jonathan Gevaryahu and Sergey Svishchev.
There are small remnants in machine/NL_*, specifially breakout, pong
doubles and rebound that are also copyrighted by the DICE team,
whoever that means. They're not critical since they only concern
these drivers and not an important core subsystem.
|
| | |
|
| | |
|
| |
|
|
|
| |
* These are used during creation and reset only. No need to
maintain them on net_t level.
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
* Under cirtum circumstances the splitter would create "ghost" solvers
consisting of terminals already used in another and complete
solver. This may impact all netlist which use opamps and thus
is committed early in the cycle.
* This commit adds functionality to instruct the splitter code to
include terminals which will not create matrix elements into
the parsing of net groups for solvers.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
* Devices like BJTs and FETs which are defined but not used will now
cause an error.
* An unused device will create an additional solver with a singular
matrix.
- This is adding unnecessary performance overhead.
- It complicates debugging because the unused device will cause an
arithmetic signal if used with --fperr (nltool).
* Fixed all validation errors.
|
| | |
|
| |
|
|
| |
* Consequently prefer to save on compile time and size.
|
| |
|
| |
* also fix an issue with netlist vs build
|
| |
|
|
|
|
|
|
|
|
|
| |
* rename some misleading type names
* remove callback_t and replace by better scalable approach
* hide implementations details
* move sources classes from putil.h to psources.h
* reduce code complexity
* improve parsing performance, parsing netlists now is twice as fast.
* fix issues around multi-byte string support
* moved psplit into pstrutil.h
|
| |
|
|
|
|
|
|
|
| |
* removed include directory src/lib/netlist from various genie files to
avoid potential issues.
* Code using netlist should use #include "netlist/*".
* Updated includes.
* Fixed standalone makefile depend target to properly deal with relative
paths.
|
| |
|
|
|
| |
* This is an infrastructure change to enable better error reporting
including file/source and line numbers in the future
|
| |
|
|
|
|
|
|
|
|
| |
* Example: PARAM(Solver.Solver_0.METHOD, "GMRES")
* Will use the GMRES solver instead of the default MAT_CR solver.
* Same applies to all Solver parameters.
* Please use with care. If you change your netlist (e.g. using
frontiers) the allocation of nets to solvers and the number of solvers
may change. Thus this type of tweaking should only be used after the
netlist completely works.
|
| | |
|
| |
|
|
|
|
|
| |
* The GMRES solver for larger matrices (>>100) can outperform Gaussian
elimination. Including it so it gets wider attention.
* As far as I know netlist is the only SPICE-like circuit simulation
providing a GMRES solver.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
* All solver scheduling is now handled by nld_solver.
* Previously, for dynamic timestepping the sub solvers would be
responsible for their scheduling themselves.
- This prevented any attemps to use parallel execution of solvers.
* Now the route is free towards experiments to use parallel execution of
solvers.
* Uses ptimed_queue_linear in solver scheduling
* Improved netlist queue implementation (template now)
* Added const delegates.
* Added subsolver stats
* Removed dead code.
|
| | |
|
| |
|
| |
* all inputs now explicitly specify a signal handler.
|
| |
|
|
|
|
|
|
|
|
|
| |
* palloc.h/pmatrix2d.h: Fix static_assert warnings at the origin.
* Rework hints to broaden their use and fix NC hint.
* 74377: use NC hint
* plists.h: Fix debugging in MSVC
* Include cleanup: Move everything not needed by netlists from
nl_setup.h into core/setup.h
* Fix some clang tidy warnings
* srcclean
|
| |
|
|
|
|
|
|
| |
The purpose of this ongoing exercise is to remove unnecessary
dependencies in header files. netlist implementations should only have
access to what they need. The same applies to device implementations.
Core stuff will be moved to the core subdirectory going forward.
|
| |
|
|
|
|
| |
* decrease use of reinterpret_cast.
* change some defaults for better ttl game optimization.
* various code cleanup.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Kidniki now achieves up to 910% when run with static solvers and with
nltool. That is significant better than the 860% we have seen
previously.
This increase is driven by using a global memory pool in the solver
code.
In addition the following refactoring and code maintenance work is
included. Please excuse the large commit, some of this took interfered
with other work and the detail development steps were ugly.
- gsl support: This commit adds pgsl.h which implements a very limited
number of the functionality of the gsl header described in the c++ core
guidelines.
- clang-tidy fixes
- A significant refactoring of palloc.h. Aligned hints were removed,
they added complexity without a significant performance gain. Vector
operations should better be done on special spans/views.
The code has been tested on linux with g++-7, g++-9, clang-11.
On Windows mingw-10 and VS2019, OSX clang-11.
|
| | |
|
| |
|
|
|
|
| |
- more c++14, use enable_if_t instead of enable_if
- cleaned up the use of memory allocation arenas
- reduce MACRO usage, use std::conditional where possible
|
| |
|
| |
moving over to clang-tidy-11 and enabling more warnings.
|
| |
|
|
| |
Identified a number of code locations which could accept const
arguments.
|
| |
|
|
|
|
| |
Calls to the solver used different paths. These changes unify these
calls and provide a common call for discontinuous state changes on
analog components (change_state).
|
| |
|
|
|
|
|
|
|
|
|
|
| |
"Edge cases" are compiles were the floting point type used for the core
may be different from the one used by solvers, e.g. double/float.
This does not happen in MAME, here currently this is always
double/double. Since floats are better suited for GPUs whose double
performance is limited it was time to review the code.
This commit fixes warnings about type conversions and precision loss.
In addition there is better support for INT128 and FLOAT128.
The MAT solver has seen some minor performance increases.
|
| | |
|
| |
|
|
|
| |
Added a define NL_USE_ACADEMIC_SOLVERS and disabled it in the build.
This will not compile in solvers which are either illustrative or only
perform for large sparse matrices like GMRES.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit simplifies the creation of static solvers. For this to
happen the following changes were made:
- nltool does no longer support the "-f netlist_file.cpp" option.
All parts of netlist which used the option now expect the file
to be specified after all other options.
Before: nltool -t 10 -f somenetlist.cpp
Now : nltool -t 10 somenetlist.cpp
- The static compile command now supports an "--output" option
to specify the file to be created and accepts multiple input
files. To be create the static solver file for MAME use the
script provided or:
./nltool --cmd static
--output=src/lib/netlist/generated/static_solvers.cpp
src/mame/audio/nl_*.cpp src/mame/machine/nl_*.cpp
- Updated documentation and examples provided by nltool --help
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit introduces precompiled static solver code. Due to
additional optimizations the compiler can use because the detail
calculation steps for the solution are known e.g. the kidniki netlist
sees a 100% speed increase.
In all environments (windows/*nix/osx) the source for the static
solver code can be created using
bash src/lib/netlist/nl_create_mame_solvers.sh
This will create src/lib/netlist/generated/static_solvers.cpp which is
compiled into the mame binary.
The script is just a temporary workaround. The intention is that nltool
whill be able to create this file with one call.
There are other improvements in this commit speeding up the processing
of timestep and dynamic calculations.
|
| | |
|
| |
|
|
|
| |
- Fixes performance regression and net_splitter struct.
- Fixes nltool time measurements
- pstream simplification
|
| |
|
|
|
|
|
|
|
| |
- Fixed some clang lint warnings
- Removed dead code
- Experimental parser code to allow calculations in parameter value.
This already works for compiled netlists. These changes are
currently disabled. Updated pong netlist (and CRC/SHA) to work
with this new code.
|
| |
|
|
|
|
|
|
|
|
|
| |
This commit fixes an issue with parsing netlists. To cut a long story
short: The separation of nets into independant groups of nets failed
under some circumstances for four terminal devices (controlled
voltage/current devices). Everything worked as long as an external
feedback loop existed. Without external feedback loop, the separation
would either fail or create "false" separated nets.
This fix also highlighted an issue with cmos switches (4066/4016).
There is a slight chance that other bugs may surface due to this change.
|
| |
|
|
|
|
| |
- remove pthrow trampline as proposed by Vas.
- identify throwing code by adding noexcept(false)
- move "connected term" information to setup code.
- srcclean
|
| |
|
|
|
|
|
|
|
| |
Introduce an additional absolute time type netlist_time_ext to identify
whether absolute or relative time is used in the netlist code.
Extend ptime code to allow operations between ptime derived types with
different internal types.
In addition rewrote main queue serve loops. Adds a very small
performance increase.
|
| |
|
|
|
|
|
| |
- Fix automatic header generation
- clang lint fixes.
- srcclean
- remove dead code
|
| |
|
|
|
|
| |
- more doxygen \file annotations
- moved MAINCLOCK back to nl_base.h
- remove some const from simple function parameters
|
| |
|
|
|
|
|
| |
- comment style migration continues.
- Fixed a two bugs in the truthtable ignore inputs code
- refactored the truthtable code a bit for better readability.
- updated netlist specific gitignore.
|
| |
|
|
|
|
| |
- Separate code out into pmath.h and pstonum.h.
- Fix VC build error
- optimize pfmtlog.h a bit
|
| |
|
|
|
| |
__float128 is a gnu extension delivering true 128 bit floating point
support. Currently not supported by clang. In addition, the quadmath
library needs to be linked. For the time being therefore disabled.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
- Added new solver parameter FPTYPE. This determines in which floating
point domain the linear system is solved. May be one of "FLOAT",
"DOUBLE" or "LONGDOUBLE"
- Added option "--fperr" to nltool. This enables floating point
exceptions. This helps debugging the code under gdb.
The purpose of this going forward is to have more choice in
optimization. Non-dynamic systems should be just fine in the float
domain. Dynamic systems (i.e. diodes, bjts, mosfets) should in general
work with double. Certain edge cases may require long double resolution.
|
| |
|
|
|
|
|
|
|
|
|
| |
Added ability to compile using float instead of double. Specifically the
the solver as well as the infrastructure now can have their own floating
point type. Currently this is only an academic exercise since
numerically demanding circuits like kidniki only work with double/double
support. Using float here is pushing numerical stability over the
limits.
The long term design goal is too have the matrix type (double/float)
being a parameter.
|
| |
|
|
|
|
|
| |
- solver: align matrix population along the various solvers
- solver: delete dead code
- renamed nl_double to nl_fptype and use nl_fptype where previously
double has been used.
- renamed param_double_t to param_fp_t
|
| |
|
|
|
| |
- new namespace "solver"
- minor ptime modifications
|