OpenCores

Floating Point Adder and Multiplier

Issue List
fpadd_pipeline #1
Open fpgamaster opened this issue almost 14 years ago
fpgamaster commented almost 14 years ago

I utilize your IP fpadd_pipeline with Quartus2- Altera. But it doesn't wotk well. I think fpadd_normalize_struct .vhd has a bug at the line 94 "IF (signed(count) > signed(EXP_in)) THEN". I think "singed" before "(count)" and "(Exp_in)" is not necessary. How do you think?

fpgamaster commented almost 14 years ago

Type your text here

fbeldianu commented over 12 years ago

Same issue encountered: Add these two numbers with the last revision of fpadd_normalize_struct.vhd:

  a <= 32'hc5b3db3a;
  b <= 32'h451f0dd4;

I switched to previous one.

fbeldianu commented over 12 years ago

The hex operands are: c5b3db3a; 451f0dd4;

fbeldianu commented over 12 years ago

Lane IF (signed(count) > signed(EXP_in)) THEN

should be replaced with: IF ((count) > (EXP_in)) THEN

gmarcus commented over 12 years ago

there is currently a bug related to the normalized numbers. It does work better with the previous version, I hope to have time in the next two weeks to finally close that issue (has been hanging for some time now...).

Granted, signed does not seems to be needed at this point. It will certainly assume the wrong sign from the count for some values, and when sign extending will return the wrong value. Will check it further. Thanks!

ashwini commented about 6 years ago

Can i get a Verilog code for master an slave of i2c in Verilog

gmarcus commented about 6 years ago

Ashwini, I think you posted in the wrong place.

nitinndg commented about 4 years ago

is this issue solved. if so i need it urgently


Assignee
No one
Labels
Bug