-
Hi,
On and off I've been plagued with the following problem, but never knew why. Well, Cary gave me a way to track it down when it happened again:
>> BTW, I've been getting error messages pointing to lines
>> that are 10-20 lines away from the actual error. I
>> doubt if I can make a test case, because when you simplify
>> it, it usually goes away. I have lots of
>> `ifdefs. I...
2009-07-18 01:29:22 UTC in Icarus Verilog
-
IVL evaluates negedge cnt[3] whenever cnt[0] changes, rather than at negedge cnt[3]. See included test case.
.
task automatic auto_task;
begin
@ (negedge cnt[3])
//@ (negedge clk)
$display("auto_task: Time is %5t", $time);
end
endtask
task normal_task;
begin
@ (negedge cnt[3])
//@ (negedge clk)
$display("normal_task: Time is %5t", $time);
end...
2009-07-15 03:01:50 UTC in Icarus Verilog
-
Hi,
In version:
vvp.tgt: Icarus Verilog VVP Code Generator 0.10.0 (devel) (s20090225-101-g291dc52)
IVL incorrectly calculates the real result of a negative integer divided by a real. This used to work relatively recently. I've attached a test case.
[rmm@endo orion]$ iverilog -Wall real_calc.v
[rmm@endo orion]$ ./a.out
FAILED: -1/0.000001 gave 7000000.000000, not expected...
2009-07-09 02:59:36 UTC in Icarus Verilog
-
Yes, I agree. I sent a note to Tony regarding the possibility of adding a flush capability to the LXT API, and mentioned the performance difference between generating/displaying LXT and LXT2. I just converted an LXT2 file to VCD, then VZT:
LXT: 85.0MB
LXT2: 93.6MB
VCD: 793MB
VZT: 244MB
The VZT file took 1m35s to load into GTKWave, whereas the LXT file took 6s.
2009-03-28 00:13:52 UTC in Icarus Verilog
-
Thank you for clarifying that distinction. It seems like adding the capability to LXT2 or VCD would not be that hard. Great!
Off-topic ...
Despite lacking a flush capability, LXT1 is still very useful for longer simulations because it's faster to generate, makes a smaller waveform file, and loads into GTKWave 3.2.0 much faster than LXT2. I just ran some quick benchmarks with iverilog 0.9.1...
2009-03-27 21:23:45 UTC in Icarus Verilog
-
It would be nice to be able to make PDF printout directly. Currently, I output a .ps file, then use ps2pdf14 to make the conversion to PDF. If ps2pdf or gs is in the path, perhaps GTKWave provide a PDF option and do the conversion internally.
2009-03-27 20:31:07 UTC in gtkwave
-
Hi,
It would be tremendously handy to be able to break into a long simulation and take a look at the progress of a simulation that is outputting LXT-format waveforms. Currently, GTKWave fails to open an LXT file that is not properly $finished (or 'finish', in interactive mode). Ideally, when the user hits Ctrl-C (or $stop), the LXT file would be in a state such that it can be viewed with...
2009-03-27 00:15:00 UTC in Icarus Verilog
-
iverilog 0.9.1 compiler crashes when $displaying a real or realtime array element.
From modified real_array.v (attached):
...
if (passed) $display("PASSED, got real array[0] = %g, real array[1] = %g", rvar[0], rvar[1]);
...
gives:
vvp: array.cc:822: vvp_vector4_t array_get_word(__vpiArray*, unsigned int): Assertion `arr->nets != 0' failed.
Aborted
Thanks,
Dick.
2009-03-26 23:52:00 UTC in Icarus Verilog
-
Hi,
I'm confused about PCMCIA IDE cards. Do they integrate IDE I/F chips capable of supporting two hard drives (or other devices) independent of any motherboard IDE interface? Or are they (as an Apricorn tech told me) merely an interface to an already existing IDE channel on the motherboard? In the later case, you'd probably be limited to one external drive (if that) assuming the internal...
2001-01-25 22:07:32 UTC in Linux PCMCIA Card Services
