this returns the m_matchcount; can be used for auto-sizing the bind array in match. it needs to be at least as large as the get_mathcount of all the M

return the priority for thi pass

return the priority for thi pass

* @brief given a scale in float and a recip shift amount * return a quantized scale multiplier and change recip shamt inplace * */ Now with correc

* @brief given a scale in float and a recip shift amount * return a quantized scale multiplier and change recip shamt inplace * */ Now with correc

the following 2 methods only work if nsplit < SPLITPER they are only intended for results of AUTOTHREAD

the following 2 methods only work if nsplit < SPLITPER they are only intended for results of AUTOTHREAD

this is so we can organize rules based on the root opname.

this is so we can organize rules based on the root opname.

hide the slower implementations in the base class...

hide the slower implementations in the base class...

hide the slower implementations of these in the base class...

hide the slower implementations in the base class...

hide the slower implementations in the base class...

hide the slower implementations of these in the base class...

* @brief returns the parameters for scaling. * bit 31-24: left shift amount * bit 23-16: right shift amout * bit 15- 0: scale factor * * Input mu

* @brief returns the parameters for scaling. * bit 31-24: left shift amount * bit 23-16: right shift amout * bit 15- 0: scale factor * * Input mu

This used to inform serialize_header() (and similar operations) what part of a multi-section file they are generating. 0 - no multi-section operation

This used to inform serialize_header() (and similar operations) what part of a multi-section file they are generating. 0 - no multi-section operation

{return false} in op.cc

{return false} in op.cc

By convention, op names that start with '#' are constant regardless of input This is useful (for example) to get quantization parameters out of output

By convention, op names that start with '#' are constant regardless of input This is useful (for example) to get quantization parameters out of output