Output is the sum (the subtraction) of the input buffers.
vector<float,8> fpadd (vector<float,8> acc, vector<float,32> xbuf, int xstart, unsigned int xoffs)
Parameter |
Comment |
|---|---|
acc |
First addition input buffer. It has the same type as the output. |
xbuf |
Second addition input buffer. |
xstart |
Starting offset for all lanes of X. |
xoffs |
4 bits per lane: Additional lane-dependent offset for X. |
The executed operation is:
for (i = 0 ; i < 8 ; i++)
ret[i] = acc[i] + xbuf[xstart + xoffs[i]]
The following datatypes are allowed:
acc:
vector<float,8>, vector<cfloat,4>xbuf:
vector<float,32>, vector<float,16>, vector<cfloat,16>, vector<cfloat,8>