Loading...
Searching...
No Matches
Collaboration diagram for QgemmKernels:
Classes | |
| class | QgemmScalar< TT, TTPARAM, M, K, N > |
| Scalar implementation for MK*KN, stores weights and biases, QgemmScalar<a,a,1,80,32> takes 17340 cycles. More... | |
| class | Qgemm< TT, TTPARAM, M, K, N > |
| Vector implementation for MK*KN, stores weights and biases, requires N%16=0 Qgemm<a,a,1,80,32> takes 227 cycles. More... | |
| class | QgemmStream< TT, TTPARAM, M, K, N > |
| Vector implementation for MK*KN, stores weights and biases, requires N%16=0 Qgemm<a,a,1,80,32> takes cycles. More... | |
Detailed Description
- y = saturate ((x / y_scale) + y_zero_point)
- Bias must be quantized using scale = x_scale * w_scale and zero_point = 0
Computation
- x = (qx - qx_zero) * qx_scale
- bias = qbias * x_scale * w_scale
- y = x*w + bias
(qy-qy_zero)*qy_scale
- = (qx-qx_zero)*qx_scale * (qw-qw_zero)*qw_scale + qbias*qx_scale*qw_scale
- = [(qx-qx_zero) * (qw-qw_zero) + qbias] * qx_scale*qw_scale
