TDA2EXEVM: TIDL limitation

Part Number: TDA2EXEVM

I have a prototxt like below，there are some limitation:

1. tidl do NOT support transpose layer, so I use dma to implement the layer, BUT I do NOT know how to integrate it to TIDL.

2. the tidl user guide says the channel is only up to 1024, BUT the input shape of "conv_2" is (4094, 1, 64). how to support it ? use dense conv ? use group=4 ?

3. is there some other limitation about this prototxt ?

name: "model_A"
input: "data"
input_shape {
dim: 1
dim: 3
dim: 32
dim: 64
}
layer {
name: "conv_1"
type: "Convolution"
bottom: "data"
top: "conv_1"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 128
bias_term: true
pad: 1
kernel_size: 3
stride: 1
weight_filler {
type: "msra"
}
bias_filler {
type: "constant"
value: 0
}
}
}
layer {
name: "transpose_1"
type: "Permute"
bottom: "conv_1"
top: "transpose_1"
permute_param {
order: 0
order: 2
order: 1
order: 3
}
}
layer {
name: "reshape_1"
type: "Reshape"
bottom: "transpose_1"
top: "reshape_1"
reshape_param {
shape {
dim: 1
dim: 4096
dim: 1
dim: 64
}
}
}
layer {
name: "conv_2"
type: "Convolution"
bottom: "reshape_1"
top: "conv_2"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 128
bias_term: true
pad: 0
kernel_size: 1
stride: 1
weight_filler {
type: "msra"
}
bias_filler {
type: "constant"
value: 0
}
}
}