Commit Graph

1684 Commits

Author SHA1 Message Date
Stephen Williams e932c7a637 Support get of vpiIntVal from vec4 stack. 2014-03-04 19:02:21 -08:00
Stephen Williams 4068570f43 Fix %pow and %pow/s to properly account for 'x operands. 2014-03-01 16:11:47 -08:00
Stephen Williams 7ceb18fb37 Merge branch 'master' into vec4-stack
Conflicts:
	netmisc.cc
	tgt-vvp/eval_expr.c
	vvp/vthread.cc
2014-03-01 09:02:14 -08:00
Cary R 13c1925045 Fix space issue 2014-02-28 17:19:35 -08:00
Martin Whitaker 1f81d4c081 Fix for br947.
When performing the initial assignment for a procedural continuous
assignment, any previous continuous assignment to the destination
signal must be unlinked first, otherwise the initial value for the
assignment will propagate to any other nets that are driven by the
original source signal.
2014-02-26 08:39:33 +00:00
Stephen Williams 91b7c6ab55 Merge branch 'master' into vec4-stack
Conflicts:
	vvp/vthread.cc
2014-02-21 18:04:16 -08:00
Stephen Williams ebed793dec Merge branch 'master' into vec4-stack
Conflicts:
	vvp/vthread.cc

Signed-off-by: Stephen Williams <steve@icarus.com>
2014-02-21 17:34:59 -08:00
Martin Whitaker a76be3571d Fix for br946 - procedural continuous assignment causes vvp internal error.
In the case that the RHS of a procedural continuous assignment is a simple
vector that is wider than the LHS, changes to the RHS vector cause the
entire vector to be sent to port 1 of the LHS vvp_fun_signal object. This
vector needs to be coerced to the size of the LHS. Note that this is a
stopgap fix until vvp handles arbitrary expressions on the RHS of a
procedural continuous assignment.
2014-02-21 22:48:50 +00:00
Martin Whitaker 345c9cf21c Implement correct behaviour for signed vector power operations in vvp.
Signed vector power operations were being implemented using the double
pow() function. This gave inaccurate results when the operands or
result were not exactly representable by a 64-bit floating point number.
2014-02-19 20:11:57 +00:00
Martin Whitaker 5853f7d867 Fix for GitHub issue 9 part 1 : Efficiency of vvp_vector2_t::pow() function.
The vvp_vector2_t::pow() function is recursive, and performs a multiplication
operation on each step. The multiplication operator was expanding the result
vector to accomodate the maximum possible result value for the given operand
vectors, thus causing the execution time of the power operation to be
exponentially proportional to the exponent value. Both in this case and
in general, it is unnecessary for the multiplication result vector to be
expanded, as the compiler has already determined the required vector width
during elaboration, and sizes the operand vectors to match.
2014-02-15 21:40:55 +00:00
Martin Whitaker 68f8de28af Fix for GitHub issue 7 : Undef propagation in power operator.
The vvp_vector2_t constructor that takes a vvp_vector4_t value was
documented as creating a NaN value if the supplied vector contained
any X or Z bits, but instead used the standard Verilog 4-state to
2-state conversion semantics (X or Z translate to 0). I've added an
optional second parameter to the constructor to allow the user to
choose which semantics they want, as both are needed.
2014-02-15 13:14:45 +00:00
Stephen Williams 801e795112 Better job of matching adder operand sizes for vec4 code generator. 2014-02-10 18:06:56 -08:00
Stephen Williams 1d63875e5d Merge branch 'master' into vec4-stack
Conflicts:
	elab_lval.cc
2014-02-10 16:26:22 -08:00
Stephen Williams 554fb7ebdd Various internal vec4 size mismatches fixed. 2014-02-07 17:50:13 -08:00
Stephen Williams c9e8392dc7 Fix dangling vec4 stack when force assignment is suppressed.
Also improve the %debug/thr instruction.
2014-02-07 16:10:28 -08:00
Stephen Williams 60d37e1f53 More vec4 support for various things. 2014-02-07 11:24:41 -08:00
Stephen Williams 401fccdf6e vec4 handling of DARRAY of vec4 vectors. 2014-02-06 15:05:26 -08:00
Stephen Williams f5564a195f vec4 versions of compressed assignment statements. 2014-02-06 09:55:25 -08:00
Stephen Williams 7f59c51ca2 Handle proceedural writes to UWIRE objects.
The elaborator allows this only if it determines that the bits
that are procedural written are not also continuously written.
2014-02-02 16:43:48 -08:00
Stephen Williams 1805598eae Fix some vec4 handling of putc, len, and other string methods. 2014-01-31 19:01:28 -08:00
Stephen Williams 0eca210722 Merge branch 'master' into vec4-stack 2014-01-30 17:07:54 -08:00
Larry Doolittle 3e95966d70 More spelling fixes
Mostly comments
Changs "initilzers" in a string visible with debug_elaborate.
Includes a few British->American changes in the root directory only.
2014-01-30 16:43:17 -08:00
Stephen Williams fdc9e8590d Merge branch 'master' into vec4-stack 2014-01-30 15:59:51 -08:00
Larry Doolittle be17bfc0e9 Spelling fixes
Mostly comments.
One user-visible string ("Evalutated to ") changed in the debug_eval_tree case.
2014-01-30 15:34:20 -08:00
Stephen Williams 23ba0bc019 various vec4 fixes. 2014-01-25 19:25:21 -08:00
Stephen Williams 9cfb15a302 Implement assign/vec4/off/e 2014-01-23 18:39:29 -08:00
Stephen Williams 0e13726e83 fix broken call-out of real-valued system functions. 2014-01-21 13:31:49 -08:00
Stephen Williams 54346d7095 vec4 implementations of real ternary and vec4 to real casts. 2014-01-21 12:02:59 -08:00
Stephen Williams f1855eed0e Fix system function call in net types to get proper vector width.
This means using some of the new vec4 infrastructure to get at the
data, instead of using the old thread bit pointers. In the process,
remove the vbit and vwid members that pointed to thread bits. Those
bits no longer exist.
2014-01-20 19:07:11 -08:00
Stephen Williams 0f9828f179 Fix vec4 overflow of times when they are >32bits. 2014-01-19 14:28:39 -08:00
Stephen Williams d2c2d6d8dc vec4 versions of %cassign/vec4/off and %force/vec4/off send flags[4]. 2014-01-18 17:08:36 -08:00
Stephen Williams dfaed128c0 Fix %store/vec4 overlap of vectors written with extreme indices. 2014-01-15 16:30:34 -08:00
Stephen Williams ae1f8be277 Wrap up vpi access to vec4 stack items. 2014-01-14 17:10:03 -08:00
Stephen Williams d8533104b0 Cleanup vec4 handling of suppressed store/assign due to bad offset/index math. 2014-01-14 12:00:39 -08:00
Stephen Williams c6c697bdb7 WRap up vec4 support for system function calls. 2014-01-14 09:22:23 -08:00
Stephen Williams 660d59a7a9 vec4 versions of the %force instructions. 2014-01-13 18:57:03 -08:00
Stephen Williams aaa054a932 Fix botched implementations of vec4 based shift. 2014-01-13 17:46:14 -08:00
Stephen Williams b0a9430e98 Wrap up vecc4 support for left/right shift expressions. 2014-01-13 16:11:47 -08:00
Stephen Williams ec6009dd8f Compile out the T<...> format in vvp.
This ifdef's the code out. Leave it in place for a little longer
for reference.
2014-01-06 20:36:26 -08:00
Stephen Williams 4820d46353 Implement some vpi_get_value formats for vec4 stack values. 2014-01-06 20:27:14 -08:00
Stephen Williams a2845cee70 Update %pow instructions to use vec4 stack. 2014-01-06 10:34:07 -08:00
Stephen Williams c897ca017c Update %div instruction to vec4 version. 2014-01-06 10:14:49 -08:00
Stephen Williams aee540d8bb Fix sort of %event command in vvp compile list.
The sort is important, the names must be sorted because the table is
searched using binary search.
2014-01-06 09:12:42 -08:00
Stephen Williams 19402aeef2 get vpiStringVal from vec4 stack positions. 2014-01-06 08:35:49 -08:00
Stephen Williams e5eb754150 vec4 versions of a bunch of unary operators. 2014-01-05 14:12:27 -08:00
Stephen Williams 063c6d6065 Add the %event instruction, remove %ix/get and %ix/get/s. 2014-01-05 12:39:52 -08:00
Stephen Williams 2fc8ce8a16 Implement vec4 basec %assign delay and event / vpiTimeVal for functions. 2014-01-05 12:04:16 -08:00
Stephen Williams 1a3adbe9cd Vec4 store to memories.
This also reworks the working of the %store/vec4 instruction to
take a part offset, and eliminate the %store/vec4/off instruction.
2014-01-05 10:30:59 -08:00
Stephen Williams f89dbd48c8 Add vec4 support for bitwise or/nor/nand/xor/xnor. 2014-01-04 23:48:16 +00:00
Stephen Williams 9fc3e84e93 vec4 cassign to part selects. 2014-01-04 22:58:58 +00:00