- optimized the core queue dispatching logic. Minor performance
increase.
- fixed a number of bugs in parray. Now parray<double, 0> will be purely
dynamic allocation with the number of elements passed in the
constructor.
- Added noexpr where appropriate.
- Simplified the queue
Checked with gcc-7 (ubuntu), gcc-9, clang-10, macosx clang 10, mingw
cross compile on linux.
- moved netlists out of driver code into audio/ or machine/ as
nl_xxx.cpp files.
- identified and documented extended validation
- updated arcade, mess and nl targets
- Fix SUBMODEL
- move to strongly typed matrix sort constant
- extend maximum matrix size to 512x512
- optionally do parallel processing based on total operations
- templatize GMRES solver loops
(nw) This has been a long time coming but it's here at last. It should
be easier now that logerror, popmessage and osd_printf_* behave like
string_format and stream_format. Remember the differences from printf:
* Any object with a stream out operator works with %s
* %d, %i, %o, %x, %X, etc. work out the size by magic
* No sign extending promotion to int for short/char
* No widening/narrowing conversions for characters/strings
* Same rules on all platforms, insulated from C runtime library
* No format warnings from compiler
* Assert in debug builds if number of arguments doesn't match format
(nw) Also removed a pile of redundant c_str and string_format, and some
workarounds for not being able to portably format 64-bit integers or
long long.
* Fixed a compiltion bug under clang-cl
Oddly, this problem does not seem to manifest under clang on
gcc.godbolt.org. I suspect that this might be related to the fact that
sizeof(std::size_t) != sizeof(long) on Windows.
* Couriersud feedback
Some unknown system library seems to force the use of the global locale
on OSX. This is not the case for other *nix or Windows builds. This
commit fixes this by forcing the C locale in pfmt.
* hp9825: fixed a bug in 9825t
* hp9845: TACO driver re-written from scratch, DC100 tape separated into
a new device, various adaptations
* hp9845: "new TACO" renamed to just "TACO"
This now works with both single density and double density floppy disks, and
dynamically identifies boot sector IDs required for 6800 booting, and supports
writing back to the 'dsk' image files.
The UniFLEX disk format is not compatible with the Flex format. Significantly it
does not use a mix of single density for booting on some double density disks
which makes it simpler - hardware required a new boot ROM to run UniFLEX.
Further, the UniFLEX sector size is 512 bytes versus 256 for Flex, and the
UniFLEX 'SIR' info sector record is completely different to the info on Flex
disk, and the file system format is also not at all compatible.
Thus the UniFlex format can rely largely on the WD17xx format, with an
overload to handle the sector numbering on the second side continuing from the
first side (one feature in common with the Flex format). This gives a quick
'save' capability and shares code.
Support for 8" disks is included as this was the initial distribution format
and the only one found so far.
The CPU tries to read data in a tight loop, so there must be
some way for it to be halted until data is ready. The current
solution works good enough to boot CP/M.
Also:
- Change FDC to KR1818VG93
- Change disk format from 3.5" to 5.25"
- Add single sided disk format
- Add softlist
Current TZX specification says: "Header blocks have 8063 and data blocks have 3223 pilot pulses"
It also says: "The pilot tone consists in 8063 pulses if the first data byte (flag byte) is < 128, 3223 otherwise"
(unlike tap format that if flag byte is 0 indicates header block and data block otherwise).
CMOS 40xx and 4316 power pins fixed.
Also fixed gcc-9 error. clang++ complains about unreachable code in
nl_base.cpp line 480 even if double parantheses are used. Assigning the
define to a local variable and testing this local variable works. Weird.
The truthtable implementation of 74107 (JK-Flipflop) is included
for educational purposes to demonstrate how to implement state
holding devices as truthtables. It will completely nuke
performance for pong and therefore is disabled.
- converted NL_MAX_LINK_RESOLVE_LOOPS into a netlist parameter.
- Reduced potential bit-rot.
- nltool -v --version now displays values of all compile time defines.
There are still far too many compile time defines. However, most of them
ensure and test future scalability.
* Start fixing OS9 disk handling
* Got it working, now need to refine and test
* Almost complete. Doing more testing...
* Tested aginst a varient of sotware. Solved bug. Cleaned up tabs.
* Turned on sector interleaving in OS9_DSK and retested.
funworld.cpp cleanup: (nw)
* remove revision history - we use version control for a reason
* don't #define things before inluding the PCH, and don't #define generic names before #including anything
Protected defines in nl_config.h with ifdefs. Added a define to disable
queue statistics during compile. This is only needed during development.
Documented performance improvement efforts so I don't try this again.
Runtime performance statistics can now be enabled with nltool
option "-s". To enable those with MAME you need to run
NL_STATS=1 ./mamenl64 -v -oslog game
This fixes an over simplification. Logic devices implicitly assumed that
GND/VDD actually is connected to GND(i.e. 0V). There is no immediate
benefit from this change. It is a preparation for the future
scalability. Now all power terminals (typically 7/14, 8/16) have to be
explicitly connected to the supply rails.
Also added a validation mode to the netlist core. This is not
intended for running, but solely to better indentify pins which
are not properly connected.
mame -validate now also checks all netlist devices. It does this
by constructing a temporary netlist.
This commit also fixes some memory leaks and a bad bug which
surfaced in validation.
LM3900 model 3 only has half the number of BJTs compared to model 1
but delivers comparable results for Money Money.
Model 3 follows the datasheet.
I left code for Model 0 and 2 in for educational reasons.
improvements.
- fix MB3614 parameter
- Added VARCLOCK which derives step size from function
- optimized function handling in CS and VS
- fixed a bug in ppreprocessor
- add trunc to pfunction
- added opamp_amplification_curve to derive characteristic
amplification curve
- Untie diode code
- Fix some typos
- add TYPE=2 opamp model which omits output voltage limitation
Useful for determining causes of non-convergence
- Fix MB3614 opamp parameters to match datasheet
modelling. [Couriersud]
Added global NETLIST.DEFAULT_MOS_CAPMODEL parameter. Setting this to
zero disables using capitance modelling in mos models.
On a per mos device basis this can be achieved by adding CAPMODEL=0 to
the model definition, e.g. MOSFET(X, "NMOS(CAPMODEL=0)")
Improve MOSFET convergence by using log-stepping.
This fixes a rounding issue in the ptime code. This bug surfaced when I
tested netlist with picosecond resolution in ptime.
This will have a small impact on every driver using netlist. For
breakout, it required to adjust the color overlay.
This is a significant improvement to the MOS transistor model. It adds
modelling of the Meyer capacitance model.
This is a somewhat academic addition since the effects occur on a
nanosecond time scale and have a huge impact on performance. I plan to
make the capacitance model selectable. Both on a model level as well as
by introducing a global solver parameter.
The model delivers comparable results to LTSpice.
This effectively reverts b380514764 and
c24473ddff, restoring the state at
598cd52272.
Before pushing, please check that what you're about to push is sane.
Check your local commit log and ensure there isn't anything out-of-place
before pushing to mainline. When things like this happen, it wastes
everyone's time. I really don't need this in a week when real work™ is
busting my balls and I'm behind where I want to be with preparing for
MAME release.
- added MOSFET model. Currently capacitances are not modelled.
This is a 3-pin model (Bulk connected to Source) with provisions to
extend it to 4-pin at a later stage.
- Add a capacitor generic model which is charge conserving.
Switch netlist to use this model instead of constant capacity model.
- Start putting constants into a central place.
Please expect minor timing differences due to a different numerical
path.
The cmos inverter example illustrates the analog implementation of a
cmos inverter gate. These were used a lot back in the 70s/80s to
generate sinus waves. The model should also be able to better emulate
4066 analog switches.
The addition of a relatively simple capacitor model is planned at a
later stage.
Expect everything from the MOSFET model at the current stage. Wrong
results as well as convergence issues and crashes.
MAME crashed if user was trying to load a dsk image with more heads or tracks than disk device supports.
Now the error 'Incompatible image format' is raised.
Despite the overhead - two devices more per transistor - this addition
significantly reduces computing time on switching conditions by reducing
the needed Newton-Raphson loops dramatically.
- 80 column card with SDX in CP/M mode.
- ROM/RAM banking fixed for CP/M, and MTX500 now correctly detected.
- Support for Type 03 and Type 07 .mfloppy images.
- Added alternate MTX2 romset (German).
- Keyboard ROM now selected in Configuration.
- Quickload .RUN files.
Change the sign of go (or in other terms a12 and a21 matrix stencil
elements). This should make further optimization of matrix population
easier.
In addition hopefully improve the readability of the code by sacrifying
overloads for more verbose member names.
Memory management in plib is now alignment-aware. All allocations
respect c++11 alignas. Selected classes like parray and aligned_vector
also provide hints (__builtin_assume_aligned) to g++ and clang.
The alignment optimizations have little impact on the current use cases.
They only become effective on bigger data processing.
What has a measurable impact is memory pooling. This speeds up netlist
games like breakout and pong by about 5%.
Tested with linux, macosx and windows cross builds. All features are
disabled since I can not rule out they may temporarily break more exotic
builds.
Set USE_MEMPOOL to 1 to try this (max 5% performance increase).
For mingw, there is no alignment support. This triggers -Wattribute
errors which due to -Werror crash the build.
At least on macosx memory used by an object seems to be invalidated
before the dtor is executed. This of course is deadly for child objects
with references to the parent-in-deletion which may call back into the
parent.
One of the worst issues I had to fix. Ever. Lesson learnt: No tricks in
dtors. Never.
startup strategies. This determines the order of device triggering.
0: Full - trigger all delegates. Next all devices not touched.
1: Backwards - trigger all devices backwards (only update delegate)
2: Forward - trigger all devices forward (only update delegate)
- convert macros to c++ code.
- order of device creation should not depend on std lib.
- some state saving cleanup.
- added support for clang-tidy to makefile.
- modifications triggered by clang-tidy-9.
Useful for debugging purposes in the end - but not performance.
/*! Store input values in logic_terminal_t.
*
* Set to 1 to store values in logic_terminal_t instead of
* accessing them indirectly by pointer from logic_net_t.
* This approach is stricter and should identify bugs in
* the netlist core faster.
* By default it is disabled since it is not as fast as
* the default approach.
*
*/
#define USE_COPY_INSTEAD_OF_REFERENCE (0)
- Removed dead code.
- nltool now adds a define NLTOOL_VERSION. This can be tested in
netlists. It is used in kidniki to ensure I stop committing
debug parameters.
- Optimized the proposal for no-deactivate hints.
- Documented in breakout that hints were manually optimized.
- Minor optimizations in the order of 2% enhancement.
Still some work ahead to separate interface from execution. This is a
preparation to switch to another sparse matrix format easily which may
be better suited for parallel processing.
On the linear algebra side there are some nice additions:
- Two additional sort modes: One tries to obtain a upper left identity
matrix, the other prefers a diagonal band matrix structure. Both deliver
slightly better performance than just sorting.
- Parallel execution analysis for Gaussian elimination and LU solve.
This determines which operations may be done independently.
All of this is not really useful right now. The matrix sizes are below
100 nets. I estimate that we at least need four times more so that CPU
parallel processing overhead pays off. For GPU, add another order. But
it's nice to have code which may scale.
- src/lib/netlist/solver/mat_cr.h:143:32: error: call to 'abs' is ambiguous
- src/lib/netlist/solver/nld_ms_direct.h:62:19: error: non-type template argument evaluates to 18446744073709551488, which cannot be narrowed to type 'int' [-Wc++11-narrowing]
* Split up the different parts of ICO loading in the menus (locating files, scaling, drawing, etc.)
* Added icon support to software selection menu
* Added support for more ICO file variants, including PNG-in-ICO (new DIB parser is overkill for ICO but I can factor it out for BMP loading at some point)
* Added favourites filter for software menus - includes software that's favourited on any system, so GBC includes DMG favourties and vice versa
* Eliminated unnecessary member variables and O(n) walks in software selection menu
* Made the menus' cached texture structures a bit more efficient
This is an effort to separate netlist creation from netlist execution.
The primary target is to avoid that code which will only run during
execution is able to call setup code and thus create ugly hacks.
The schematic for the tp1983 contains an error. R19 is connected to GND.
This will never work since when Q goes low, the reset pulse will dead
lock Q to low. R19 needs to be connected to VCC (5V). This will generate
the proper, high-pass filtered pulse.
Further fixes for the 7497 as well.
The 7497 device should now work as described in the TI datasheet. This
datasheet contains an internal schematic with details on gates and D
flip flops used.
This change removes all string extensions like trim, rpad, left, right,
... from pstring and replaces them by function templates.
This aligns a lot better with the intentions of the standard library.
Someone needs to get MS QA to put some non-trivial modern C++
compliation tests in the acceptance tests for their C++ compiler. Maybe
MAME could even be a candidate. Well, that might be a plan if MS still
had any QA. At least this makes some lines shorter (at the cost of
needing more lines).
The existing disk image was replaced by a new one created from the
master disk.
dsk_dsk: Increase maximum cell count
This allows for slightly out of spec disk images to run, like Theatre
Europe on the Einstein.
[Lord Sméagol (Carl Lloyd-Parker)]
* Basically, initialisers go in the constructor arguments, and things for setting format go in set_format.
* Initialisation patterns can be specified with an enum discriminator or with a FUNC and optionally a tag.
* Formats can be specified with an enum discriminator or a size and function pointer.
* You must always supply the number of entries when setting the format.
* When initislising with a paletter initialisation member, you can specify the entries and indirecte entries together.
* The palette_device now has a standard constructor, so use .set_entries if you are specifying entry count with no format/initialisation.
* Also killed an overload on delegates that wasn't being useful.
* Add per-language compiler flag options to help with exotic setups
* Get rid of a potention buffer overrun in NuBus image card
* CHAR_WIDTH and LONG_WIDTH are preprocessor macros in limits.h with glibc if __GLIBC_USE (IEC_60559_BFP_EXT) is enabled - avoid using them as names
* Make formats/upd765_dsk.h slightly safer with defualt initialisers for key format members
* Don't rely on random BSS data being zero in imagedev/floppy.cpp
Not 100% confident about the variable type, and there may still be some debate over the function name (trying to avoid confusion with the length of the container, or with the number of empty slots), so appreciate review/comments.
This fixes all the non-const pointers with static lifetime I could find
with a cheap grep (in combination with the last commit). There are
likely more lurking that I didn't find, and things that aren't pointers
that should be made const.
There are still a few mutable static pointers that break the ability to
host multiple drivers but these require refactoring to fix:
src/devices/sound/sidvoice.cpp:static const uint8_t* waveform30;
src/devices/sound/sidvoice.cpp:static const uint8_t* waveform50;
src/devices/sound/sidvoice.cpp:static const uint8_t* waveform60;
src/devices/sound/sidvoice.cpp:static const uint8_t* waveform70;
src/mame/drivers/pockstat.cpp: static const char *gme_id = "123-456-STD";
src/mame/machine/namco51.cpp: static const game_driver *namcoio_51XX_driver = nullptr;
* Eliminates the need for the horizontal/vertical/LCD/SVG layout files
* Screens can now have orientation and physical aspect ratio specified
* RASTER/VECTOR defaults to 4:3, LCD/SVG defaults to square pixels at config time
* System orientation is applied on top of screen orientation
Automatically generated single-screen views and orientation flags in XML
output now work correctly for systems with multiple screens in different
geometries/orientations, e.g. housemnq, rocnms, stepstag, or netmerc.
The "core rotation options" only interact with system orientation.
Allowing multi-screen systems to work well with one monitor per emulated
screen is a complex topic. System orientation also affects the GFX
viewer while screen orientation doesn't. The orientation displayed in
the system selection menu is from the system orientation.
Let me know if I've broken any systems or use cases.
Also, add save state support for std::array/C array nested to any depth.
* Bug fix for multi-channel reads in cassette_get_samples().
Use the unused "sample_bytes" arg the same way as cassette_put_samples()
Can be used with multi-channel files now!
* Rename sample_bytes to more descriptive sample_spacing
Add some comments about sample functions and sample_spacing variable
* Keeping the 35 tracks drive as an option
* Reducing the crashes a bit more.
This helps matching the form_factor and the parameters found in the TeleDisk images, thus exiting before a serious crash.
It is NOT enough to improve the 8" disk image handling, nor it solves all the stability problems, but it is a tiny step forwards.
* Fixed building using system utf8proc
* Fixed building using system portaudio
* Allow using system-wide asio headers (1.11.0 or higher required).
* Allow using system-wide glm headers
* Allow using system-wide rapidjson headers
Disassemblers are now independant classes. Not only the code is
cleaner, but unidasm has access to all the cpu cores again. The
interface to the disassembly method has changed from byte buffers to
objects that give a result to read methods. This also adds support
for lfsr and/or paged PCs.
* Created a more flexible imgtool::datetime structure for use within Imgtool
This is intended to replace most usage of time_t
* Changing the granularity of imgtool_clock from 1ms to 100ns, as per Vas' suggestion
* Created arbitrary_datetime in timeconv.h to facilitate interpretation of datetime info
I concluded that invoking std::mktime on manually assembled std::tm is bad, because it is indeterminate how the std::tm members may be "dominant". This required that I go further in imgtool, and update a number of drivers and eliminate the parameter of imgtool::datetime that takes std::tm.
- Removed ACORN_CPM_FORMAT, same geometry as ACORN_SSD_FORMAT/ACORN_DSD_FORMAT
- Improved find_size/identify for SSD/DSD formats
- ACORN_DOS_FORMAT 640K now handled as ACORN_ADFS_OLD_FORMAT
- Fixed geometry for ACORN_DOS_FORMAT 800K
bbc_acorn8271: Added Amcom DFS A7874
bbc_opus8272: New Opus 8272 FDC device with DDOS 3.00, not yet working.
wd177x_dsk: Apply sector_count and gaps to track description.
* Unify code for copying PNG data into bitmap for MAME and pngcmp
* Fix upsampling of monochrome PNGs (need to splat across byte)
* Add support for greyscale+alpha
* Detect more unsupported conditions rather than just behaving badly
(nw) This is the path of least resistance, and I plan to fix it up
later, I just wanted to get it to actually work first. Decompression
and unfiltering is fully supported, at least for all the pixel formats
that previously worked. Expanding 1/2/4bpp to 8bpp should work
properly, too. Bitmap mapping for Adam7 is only implemented in
rendutil.cpp which is whate everything in MAME uses. The function in
png.cpp (used by pngcmp) has not been updated. At some point I'll unify
at least one of the functions in rendutil.cpp with the one in png.cpp
and we can go from three functions that need to do the mapping down to
two at the most.
Refactor server_{ws,http}.hpp into separate interface and implementation headers.
When shutting down the HTTP server, also explicitly stop the asio::io_context.
sound board, completing the tromba circuit
(nw) I'm not sure whether the model works properly or not, but in the
circuit where it's used, I don't think it can work properly with the
current TTL output model. A capacitor is charged by the Q output of a
74LS74 flipflop (U3A) until the voltage passes the Schmitt trigger's
threshold, causing it to reset the flipflop. However, the positive
trigger voltage of the Schmitt trigger is 1.6V, but our TTL output model
has a high output voltage of 1.0V (see nl_base.cpp:89). I realise the
simplified model of TTL logic with high impedance inputs and outputs
behaving as though thery're loaded is convenient and fast to simulate,
but it's not detailed enough for applications like this where
7400-series chips are used in analog circuitry. This is what held me up
last time I tried adding a netlist for this sound board.
- OPENMP refactored. All OPENMP operations are now templatized in pomp.h
- We don't need thread-safe priority queue. Event code updating analog
outputs now runs outside the parallel code.
(nw)
0: Parallel processing of solvers disabled
1: One processor parallel processing. Can be used to measure OPENMP
overhead
>1: Solve n analog subnets in parallel.
Previously, all available processors were used which caused performance
to degrade on hyperthreading.
[Couriersud]
* Tromba (trumpet) sound is not working - requires Schmitt trigger device
* Connecting cassa (bass drum) swamps other instruments so it's disconnected for now
* Mixing melody sound with speech/SFX is not done in netlist (should be)
* Relative levels of melody/speech/SFX are probably still wrong
(nw) A good test case for this is the Money Money driver (monymony).
There's a bit of buzzing on this one as well. The problem with the
cassa could be caused by running into non-ideal characteristics of opams
again (the LM3900 seems to ignore the V+ value supplied to it). When
the netlist library gets Schmitt trigger support, the tromba can be
completed. Unfortunately, the tromba is a key part of the
characteristic sound of these boards, so you really notice when it's
lacking.
(nw) It doesn't work quite right yet. The "Hammer" and "Pest" sounds
are generated by free-running 555/556 timers and gated with LM324
applifiers. For whatever reason, the netlist system produces a kind of
buzzing from the "Hammer" circuit when it's supposed to be suppressed,
and it doesn't think the pest sound should be suppressed completely so
you can always hear it at a low level in the background. The "Cheese"
circuit is a bit weird - either they're using the base-emitter junction
of a 2SC945 as a signal diode, or there's an error in the schematic
(collector is shown unconnected). Connecting this part of the circuit
causes the netlist system to hang, so R2/R3/C8/Q2 are not connected for
now.
The core changes are:
* Short name, full name and source file are no longer members of device_t, they are part of the device type
* MACHINE_COFIG_START no longer needs a driver class
* MACHINE_CONFIG_DERIVED_CLASS is no longer necessary
* Specify the state class you want in the GAME/COMP/CONS line
* The compiler will work out the base class where the driver init member is declared
* There is one static device type object per driver rather than one per machine configuration
Use DECLARE_DEVICE_TYPE or DECLARE_DEVICE_TYPE_NS to declare device type.
* DECLARE_DEVICE_TYPE forward-declares teh device type and class, and declares extern object finders.
* DECLARE_DEVICE_TYPE_NS is for devices classes in namespaces - it doesn't forward-declare the device type.
Use DEFINE_DEVICE_TYPE or DEFINE_DEVICE_TYPE_NS to define device types.
* These macros declare storage for the static data, and instantiate the device type and device finder templates.
The rest of the changes are mostly just moving stuff out of headers that shouldn't be there, renaming stuff for consistency, and scoping stuff down where appropriate.
Things I've actually messed with substantially:
* More descriptive names for a lot of devices
* Untangled the fantasy sound from the driver state, which necessitates breaking up sound/flip writes
* Changed DECO BSMT2000 ready callback into a device delegate
* Untangled Microprose 3D noise from driver state
* Used object finders for CoCo multipak, KC85 D002, and Irem sound subdevices
* Started to get TI-99 stuff out of the TI-990 directory and arrange bus devices properly
* Started to break out common parts of Samsung ARM SoC devices
* Turned some of FM, SID, SCSP DSP, EPIC12 and Voodoo cores into something resmbling C++
* Tried to make Z180 table allocation/setup a bit safer
* Converted generic keyboard/terminal to not use WRITE8 - space/offset aren't relevant
* Dynamically allocate generic terminal buffer so derived devices (e.g. teleprinter) can specify size
* Imporved encapsulation of Z80DART channels
* Refactored the SPC7110 bit table generator loop to make it more readable
* Added wrappers for SNES PPU operations so members can be made protected
* Factored out some boilerplate for YM chips with PSG
* toaplan2 gfx
* stic/intv resolution
* Video System video
* Out Run/Y-board sprite alignment
* GIC video hookup
* Amstrad CPC ROM box members
* IQ151 ROM cart region
* MSX cart IRQ callback resolution time
* SMS passthrough control devices starting subslots
I've smoke-tested several drivers, but I've probably missed something. Things I've missed will likely blow up spectacularly with failure to bind errors and the like. Let me know if there's more subtle breakage (could have happened in FM or Voodoo).
And can everyone please, please try to keep stuff clean. In particular, please stop polluting the global namespace. Keep things out of headers that don't need to be there, and use things that can be scoped down rather than macros.
It feels like an uphill battle trying to get this stuff under control while more of it's added.
* Resurrected auxverb_cleanup_and_romident_bugfix
* Changed usage for -romident and minor cleanups
* Supporting auxverbs in any order
The previous patch was supporting 'mame64 -listsource pacman' but not 'mame64 pacman -listsource'
* Bug fix to -romident and aux verb cleanup
Made the following changes:
1. Fixed a bug where resolved slot/image options would choke -romident (reproducible in MAME 0.185 with 'mame64 -romident coco.zip')
2. 'mame64 -romident' no longer crashes (though it doesn't do anything useful)
3. Changed the aux verb functions to take 'const std::string &'
* Further cleanups to auxillary verb code, as per Vas
Specifically:
1. The commands themselves now take 'const std::vector<std::string> &' for their argument lists
2. util::core_options now collects command arguments into a separate vector rather than treating them as unadorned arguments
* Vas Crabb feedback
* Now only using trim_spaces_and_quotes() when parsing INIs
Vas pointed out that it is inappropriate to trim spaces and quotes when parsing command line options
This reverts commit 536990e77b.
Conflicts:
src/frontend/mame/mame.cpp
Sorry, but this change was half-baked. It breaks a lot of existing
functionality and clearly hasn't been tested in more than a tiny subset
of use cases. Please play this work back onto your own branch, and test
it before submitting another PR.
This is an overhaul to how MAME handles options to provide a better foundation for what MAME is already doing in practice. Previously, core_options was designed to provide an input/output facility for reading options from the command line and INI files. However, the current needs (image/slot/get_default_card_software calculus and MewUI) go way beyond that.
Broadly, this PR makes the following changes:
* core_options now has an extensibility mechanism, so one can register options that behave dramatically differently
* With that foundation, emu_options now encapsulates all of the funky image/slot/get_default_card_software calculus that were previously handled by static methods in mameopts.cpp. Changes to emu_options should not automatically cascade in such a way so that it stays in a consistent state
* emu_options no longer provides direct access to the slot_options/image_options maps; there are simpler API functions that control these capabilities
* Many core_options functions that expose internal data structures (e.g. - priority) that were only really needed because of previous (now obsolete) techniques have been removed.
* core_options is now exception based (rather than dumping text to an std::string). The burden is on the caller to catch these, and discern between warnings and errors as needed.
Obviously this is a risky change; that's why this is being submitted at the start of the dev cycle.
The +1 was previously needed becasue std::string::assign(char *) expects
the string to be NUL-terminated. The final NUL is not part of the
result. It's not needed when adjusting the length of the string
directly. Can people please be careful when refactoring, and alo when
reviewing pull requests? This stood out immediately.
This should address outstanding concerns with PR#2231. I'm trying to turn emu_options into a self contained structure that encapsulates behaviors related to options, including the gymnastics pertaining to image/slot loading and interactions with get_default_card_software() and "just works".
When the MAME 0.186 development cycle starts up, I hope to take this further. I want to make core_options::entry an abstract base class so that the entries associated with image options and slot options can derive from it. This will eliminate the current need for emu_options to directly expose maps for image and slot options.
For now, I'm in stabilization mode, and I hope to get things working for a stable 0.185 release.
This fix really doesn't go far enough. I added hooks so that options specified at the command line can also be responded to when parsed from INI files, but in the long run much of the logic that is currently in mame_options should go into emu_options so that when an option is specified, all of the wacko logic around slot/image specification "just works" because it is encapsulated within emu_options.
We have a release 11 days away; I want to be in stabilization mode.
- more use of c++ features
- some CRTP in pfmtlog
- demangled code for truthtables
- use more constexpr
- rewrite main loop
- use default constructors and assignment operators were applicable.
- optimized 7448 and 9316
All of this has decreased startup time by approx. 25% to 30%. Complex
netlists like pong or kidniki are parsed, analyzed and constructed in
around 15 ms. Run performance has increased by about 5%.
All in all not to bad. A game like pong uses a clock of 7 MHz (after
division by 2). Thats 14 MHz clock invocations. Running at over 200%, 28
MHz. On a 3.9 GHz Machine about 140 cycles/clock change.
[Couriersud]
* Introduced an 'util::arbitrary_clock' template class, to represent a clock that "knows" when the epoch starts
Also:
- Converted the NTFS filetime code to use util::arbitrary_clock
- Converted the Mac datetime code to use util::atribrary_clock
This is in preparation for a bigger change to Imgtool where I eliminate usage of time_t
pstring:
- added support for UTF16LE to pstring.
- renamed size() to mem_t_size()
- renmaed len() to length()
- added size() == length()
- added empty()
- added simple compare()
pfmtlog:
- Simplified pfmtlog, added more c++
pdynlib:
- add a dynproc type to dynlib to wrap dynamic library calls.
various:
- fix two coverty scan issue.
- various clang warnings fixed.
(nw)
This removes all allocation code from pstring. const_iterator is
consequently now based on pstring::const_iterator.
Removed pstring_buffer. This was class wasn't a good idea.
Vas was right: This change did not impact runtime performance. Startup
performance (string intensive) increased. (nw)
There are a few changes to achieve this:
* Rename to const_iterator since it's immutable
* Typedef iterator to const_iterator for now as there's no mutable iterator
* Add default constrcutor and operator-> required by concept, const-qualify operators
* Remove operator+ and operator+= since it's not a random-access iterator (use std::next and std::advance instead)
* Return reference/pointer to a proxy rather than a code_t value from opertator*/operator->
The final change is required to meet the requirement that operator* for
two equivalent forward iterators return an equivalent reference. The
pstring doesn't actually contain a sequence of code_t, so there's no way
to return a reference to code_t directly. Instead, a reference to a
proxy object aliased on the string storage is returned. The proxy is
implicitly convertible to code_t. The most noticeable side effect is
that auto c = *s.begin() or for (auto c : s) won't work. You need to do
for (auto &c : s) or for (code_t c : s) instead.
Fixed -Winconsistent-missing-destructor-override warnings. Made
some constructors of template classes and classes with virtual .. = 0
methods protected. Fixed src/lib/netlist/build/makefile (nw)
Small improvement for 7493.
Convert 9316 from subdevice to delegate.
Convert 74107 from subdevice style to delegate.
Also refactored inconsistencies in other parts of the code. (nw)
This is currently not used since performance drops by about 40%. The
typical use case would be circuits a lot more complex than those we
currently emulate where the 2*log(n) advantage really applies. (nw)
This change also changes around how command line arguments are passed around; specifically I changed argc/argv to be std::vector<std::string>
Note this is not passed around 'const', the reason being that the command line processing will now "eat" the vector
* Eliminates the need for emu_options::update_cached_options() by providing a hook for when option values change
* This is a preliminary fix to the issue identified in PR#2065
* More softlist related refactoring:
- We now only parse the command line (with core_options::parse_command_line()) once
- Options that are set up during slot and image setup go through a 'value_specifier' function
- Eliminated the command line postprocessing
- Moved 9312 and 74279 to ttl macro library.
- Renamed TTL_9312_* to DM9312. This is more appropriate.
- Fixed a number of warnings from latest ubuntu clang-5.0.
* Use size_t for sizes and <algorithm> for algorithms
* Fix up some files that were getting linked into multiple libs
* Add missing virtual method to sh2 peripheral class
* Put shortname in driver struct for locality
* Use shared pointers in config LRU cache for safety
Per Vas' request. If the compile fails for you (i'm thinking osx and
windows native debuggers here in particular), add '#include "emu.h"'
as first include of the cpp files that fail.
Due to our use of precompilation and forced inclusion, emu.h must be
included as the very first non-comment thing we do if we want to be
sure msvc compiles are identical to gcc/clang ones. Doing it directly
instead of through an include increases the correctness probability by
a magnitude.
... for updates. This will make device implementation more flexible and
faster. A nice side-effect is that there was some minor (<5%)
performance increase already. Each input is now assigned a notification
handler. Currently this is update, but going forward this may be a
custom handler. In addition
- fixed MEMPOOL on OSX
- removed dead code
- avoid bit-rot
- added delegate support for emscripten and arm processors
- added delegate support for VS 2015 x64
[Couriersud]
- Fixed crashes on terminals without nets (i.e. connected to a rail)
- Reviewed "FIXMEs" and corrected some minor ones.
- Made m_cur_analog protected.
- Fixed pmf delegates to work with msvc.
- More optimizations to the solver code.
- Started work on a better signal pipeline in nlwav
- Only generate documentation for entities which are documented.
[Couriersud]
Device implementations (all cpp files in netlist/devices) now should
only include nl_base.h.
Netlist implementation sources should only include "net_lib.h".
Refactored netlist.h and netlist.cpp to avoid namespace congestion in
netlist.h.
Fixed VC2015 build. (nw)
For netlist device debugging one can now use
#define LOG(...) log().info(__VA_ARGS__)
to use debugging and the known
#define LOG(...) do {} while (0)
do disable debugging on device level.
To avoid bitrot one could as well use
#define LOG(...) log().info.log<true>(__VA_ARGS__)
and
#define LOG(...) log().info.log<false>(__VA_ARGS__)
The later disables debugging. If the compiler can assume that there are
no side effects from e.g. using foo(a/b), 'LOG("abc {1:04x}",
foo(a/b));' should be completely optimized away.
Log channels available are info, verbose, warning, error and fatal.
Don't use debug, it is enabled only on specific debug builds.
Use would be e.g.
LOG("abc {1:04x}", 2);
The format specifier in the string are enclosed in "{}". "{2}" is the
second parameter after the format string. Types are determined
automatically. "{3:04x}" would format a number as a hexadecimal with 4
leading zeros.
[Couriersud]
- Remove virtual from some destructors and make them protected.
- Various cleanups.
- Small performance improvement.
- Fixed some inconsistencies.
- More c++ refactoring. (nw)
Make streams provide binary access only. Use putf8_reader and
putf8_writer to actually access streams. Replace some char * parameters
with pstring where appropriate. Minor code refactoring and move
functionality were it belongs. (nw)
At the same time, any char pointer has to be explicitly converted to
pstring by specifying an encoding. Not yet optimal, but certainly better
than what was there before.
Removed unneeded methods from pstring. (nw)
- refactored reverse polish notation evaluator into own source files.
- added function parameter to current and voltage sources VS and CS.
You can now use those to e.g. produce a sine wave.
- Changed code to allow devices to optionally be treated as dynamic or
timestepping devices.
[Couriersud]
nltool now is able to create all defines from the factory definitions.
This will reduce the number of places needed to touch when adding
devices and always ensure that the parser and statically compiled
netlist code use the same syntax. This will enable us to delete most
device include files, e.g. nld_74107.h.
Netlist usage to create this header file:
./nltool -c header > src/lib/netlist/devices/nld_devinc.h
This is not yet used in production. It will be enabled after additional
tests.
[Couriersud]
- Fixed OPENMP compile
For congo bongo, using Solver.PARALLEL=1 significantly increases
performance from 270% to 380%. However, this has to be taken
with a grain of salt. Enabling this on predominantly logic netlists
can severly kill performance.
- Increased readability of timed queue code.
into these files as well. The code uses <cfenv> which is part of c++11
standard. Non-standard glib extensions are currently only used on linux
and (i386 or x86_64). (nw)
1. Changed the Win32 code to use _O_U8TEXT()
2. Reordered system #includes
3. Changed a number of 'std::wcout << ...' to use util::stream_format()
I'm getting massive code compilation issues, and it isn't clear to me what the problem is. I'm committing my WIP right now, with the intention of reviewing the specifics soon.
I really don't like the prevalence of '#ifdef WIN32' in this change, both the _setmode() and bypassing codecvt. I strongly suspect that the latter is the consequence of some mistake that in practice doesn't cause problems in MSVC. I welcome all eyes.
Mario driver now uses netlist audio implementation instead of discrete
implementation. The previous discrete sound emulation has not been
removed yet because it still contains a lot of documentation.
[Couriersud]
The identifier is used to load data from a source_t implementation. This
allos a consistent approach in netlists independent of netlist
implementation. Both sources code and parameter code needed quite some
rewrite to support this. [Couriersud]
Renamed DM9334 to 9334 to align with naming convention used for other
93XX chips.
Added proper connection information for a number of chips to
net_lib.cpp. This allows to parse and run nl_hazelvid.cpp using nltool:
nltool -t 1 -f nl_hazelvid.cpp
* 7473: Made device only transition on a falling clock.
* 74161: Inverted Clear and Clock inputs to match datasheet.
* 74260: Fixed number of inputs.
* Am2847: Fixed shift register size (was 160 bits, should have been 80 bits)
* DM9334: Inverted C and E inputs to match datasheet.
-netlist: Added new devices: [Ryan Holtz]
* Intel 2102A 1Kbit (1024 x 1) Static RAM
* 74365 Hex Bus Driver with 3-State Outputs
* Generic 2- and 3-terminal Tristate device
* Note: Tristate device and 74365 do not actually tristate, they are simply a way of combining multiple outputs + chip enables.
selmenu: use LRU cache so icons don't all need to be reloaded on scroll
uismall.bdf: set default character for missing glyphs
rendfont.cpp:
* encapsulate many BDF and BDC handling details
* make file I/O 64-bit clean, check for allocation errors
* more solid BDF parser with error messages and trace logging
* fix heap smash when building bitmaps for BDF fonts
* extend BDC format to support high planes and default character
* render default character if glyph not found for BDF/BDC
* move stuff to namespace util::xml
* scope down some enums
* split config load/save delegate types
* make config load take const so it can't mangle data
* New abbreviated types are in osd and util namespaces, and also in global namespace for things that #include "emu.h"
* Get rid of import of cstdint types to global namespace (C99 does this anyway)
* Remove the cstdint types from everything in emu
* Get rid of U64/S64 macros
* Fix a bug in dps16 caused by incorrect use of macro
* Fix debugcon not checking for "do " prefix case-insensitively
* Fix a lot of messed up tabulation
* More constexpr
* Fix up many __names
* an empty tag and a self-closing tag are semantically equivalent, so std::string can be used saving some manual allocation
* a valid tag can't have an empty name, so that can be std::string too
* still expose empty strings as nullptr to API users
* scope down the int_format enum
* # is not £ ffs
* instantiate static constant members for ODR context
* declare static members of template classes properly
* declare external template instantiations
* fix blind use of autocomplete ("treats" really?)
* fix ptype_traits on platforms where char is unsigned
* signed char is never an alias for char even if char is signed
* even if eclipse doesn't like using type, surely it supports typedef
- Make most class methods constexpr
- Make color constants (white, black, etc.) into constexpr factory methods, in order to fix a static initialization problem discussed on the MAMEWorld forums.
(Note that while C++14 allows constexpr member variables to be initialized outside classes, current compilers' support for C++14 constexpr rules has proven to be lamentably deficient.)
- Create bitmap_rgb32::erase to simplify syntax in update handlers
Use standard uint64_t, uint32_t, uint16_t or uint8_t instead of UINT64, UINT32, UINT16 or UINT8
also use standard int64_t, int32_t, int16_t or int8_t instead of INT64, INT32, INT16 or INT8
Caveats:
1. Because of how this trick works, it is no longer possible to declare an option guide as static, so I had to make a bunch of changes
2. I'm going to want the hardcore C++ guys (i.e. - Vas) to review this with a fine toothed comb
* Make iterators actually meet requirements of ForwardIterator (and by consequence, ForwardIterator, Iterator and EqualityComparable)
* Don't use function statics if they can be avoided - it isn't thread-safe
* Remove leftover crud from when dynamic_buffer and friends were templates in lib/util
It's still dangerous that the const behaviour of iterators doesn't match STL.
Also, simple_list members with similar functionality to STL container members should be renamed.
The main point of this change is to C++-ify option_guide. It was changed from a struct array to a class, namespaced etc, with the ultimate hope of incorporating an in-emulation image creation UI.
Imgtool got hit with a number of changes; I'll probably have to bring that off of the backburner and touch that up too
This is just better error checking. You can see this if you create a garbage file named 'foo.csw' and invoke the following command:
mame bbcb -cass1 foo.csw
With this change you get an invalid image error. Without it, you get this:
Caught unhandled St12length_error exception: vector::_M_default_append
Use macro NL_NOEXCEPT if debug builds would throw an exception (e.g. by
using nl_assert). This is not entirely safe. Going forward, nl_assert
should call log first and than throw. (nw)
In the interests of removing burdens from client code, I also changed the following:
- filetype() will always return the file extension in lower case
- device_image_interface::filetype() will return the correct extension for files loaded as a part of softlists
- The code for extracting a file extension is now encapsulated in core_filename_extract_extension()
Client code has been updated
