In my schedule there are two ops. One is to calculate the result using gemm and
the other is to reshape it . The function is like this:
```
for (i.outer.outer, 0, 98) {
for (j.outer.outer, 0, 16) {
for (ii, 0, 8) {
for (jj, 0, 8) {
gemm_C[i.outer.outer*1024) +
[quote="alopez_13, post:7, topic:6578"]
This is part of the Relay code:
```
%0 = layout_transform(%input, src_layout="NHWC", dst_layout="NCHW");
%1 = layout_transform(%v_param_1, src_layout="HWIO", dst_layout="OIHW");
%2 = qnn.conv2d(%0, %1, 128, 122, 0.0078125f, 0.0339689f, strides=[2, 2]
Just to confirm, can you please double check your script?
We specify input shape and dtype for the model while parsing (`from_tflite`).
So, even though most of the AutoTVM script can be same, there needs to be a
small change while passing on the input shape and dtype for FP32 and quantized
mo
IIUC, simple compilation (no auto-tuning) of both FP32 and quantized models
work.
But, the auto-tuning + compilation fails for quantized model (while the same
script works for FP32), right?
---
[Visit
Topic](https://discuss.tvm.ai/t/autotvm-task-extract-from-program-in-tflite/6578/11)
t
@anijain2305 Thanks for the prompt reply. Yes I am setting `dtype_input =
"uint8"` Also I just verified that optimization of a non-quantized TFlite model
does work. In summary, the same optimization script will work for an FP32
version but not for a quantized version. Both models come from
h
Are you giving the right input dtypes to the model. Tflite quantized models
need `uint8` dtype.
---
[Visit
Topic](https://discuss.tvm.ai/t/autotvm-task-extract-from-program-in-tflite/6578/9)
to respond.
You are receiving this because you enabled mailing list mode.
To unsubscribe from th
I'm not familiar with the QNN module so I'm calling @anijain2305 for help.
I would suggest opening another topic with a proper title for a new problem
next time; otherwise it's easy to be ignored.
---
[Visit
Topic](https://discuss.tvm.ai/t/autotvm-task-extract-from-program-in-tflite/6578/8
In the topi, we could get the target information during schedule using
`tvm.target.Target.current()`. But we don't have `target_host` information as
far as I know.
But seems that you could do in `def _build_for_device` (we could add pass
inside it like other passes). However, you should doub
Hi @FrozenGene,
In the TOPI library the target and target_host information is not available and
hence the bind pipeline pass cannot be implemented.
Do you think a relay pass should be implemented to take care of it ? If yes,
could you please share an example codebase within tvm that i can r
After trying multiple quantized models the schedule is finally produced. For
testing purposes I am using the quantized models for MobileNetV2 from
https://www.tensorflow.org/lite/guide/hosted_models However, now I get at
least two kinds of errors when generating the binary:
```
an internal in
It's simple: We support only intra-operator parallelism, not inter-operator
parallelism. We use threads for paralizing the outer most loop of convolution,
for example.
---
[Visit
Topic](https://discuss.tvm.ai/t/execution-order-of-operators-at-runtime-in-tvm/6572/10)
to respond.
You are
I guess the module.so can be seen as a hash map with func name as key and
function implement as value.
The following code can be found in graph_runtime_codegen.cc.
```
LoweredOutput Codegen(relay::Function func) {
auto pf = GetPackedFunc("relay.backend.GraphPlanMemory");
storage_device
@hht - thank you again. Now it makes a kind of sense. Could you please clarify
what do you mean by "Parallelism only exists in the module."? My understanding
is that there is only one Module, and module contains multiple graph nodes that
can run in parallel.
[quote="hht, post:5, topic:657
@hht -- this is definitely interesting. In my given example, add1 and add2 are
Op types, and thus, I'd expect them to be run in parallel in a HW that is
capable of running two adders ("+") in parallel.
[quote="hht, post:5, topic:6572"]
There is no strategy to enforce parallelism to the op_e
14 matches
Mail list logo
