Cadence HiFi4 NN Library v2.2.0 update #40943

pnikam-cad · 2020-06-30T12:10:35Z

The following changes are done to the HiFi4 implementation of TF Lite Micro

Update the xtensa_hifi kernel files and the make setup to use HiFi4 NN Library v2.2.0 .
Update the kernel files as per the latest reference implementation.
Add xtenas_hifi kernel implementations for the add and mul operators.

googlebot · 2020-06-30T12:10:50Z

Thanks for your pull request. It looks like this may be your first contribution to a Google open source project (if not, look below for help). Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

📝 Please visit https://cla.developers.google.com/ to sign.

Once you've signed (or fixed any issues), please reply here with @googlebot I signed it! and we'll verify it.

What to do if you already signed the CLA

Individual signers

It's possible we don't have your GitHub username or you're using a different email address on your commit. Check your existing CLA data and verify that your email is set on your git commits.

Corporate signers

Your company has a Point of Contact who decides which employees are authorized to participate. Ask your POC to be added to the group of authorized contributors. If you don't know who your Point of Contact is, direct the Google project maintainer to go/cla#troubleshoot (Public version).
The email used to register you as an authorized contributor must be the email used for the Git commit. Check your existing CLA data and verify that your email is set on your git commits.
The email used to register you as an authorized contributor must also be attached to your GitHub account.

ℹ️ Googlers: Go here for more info.

googlebot · 2020-06-30T13:16:43Z

CLAs look good, thanks!

ℹ️ Googlers: Go here for more info.

njeffrie

I have a few style nits. Other than that, looks good.

njeffrie · 2020-07-14T19:16:09Z

tensorflow/lite/micro/kernels/xtensa_hifi/activations.cc

-      err = xa_nn_vec_activation_min_max_asym8_asym8(
-          out_data_ptr, inp_data_ptr, 0, 255, flat_size);  // Is 255 right?
+      err = xa_nn_vec_activation_min_max_asym8_asym8(out_data_ptr, inp_data_ptr,
+                                                     zero, 255, flat_size);


nit: can we use std::numeric_limits<uint8_t>::min() and std::numeric_limits<uint8_t>::max() here?

Updated as per suggestion.

njeffrie · 2020-07-14T19:17:37Z

tensorflow/lite/micro/kernels/xtensa_hifi/add.cc

+                           &output_activation_max);
+  tflite::ArithmeticParams op_params;
+  SetActivationParams(output_activation_min, output_activation_max, &op_params);
+#define TF_LITE_ADD(opname)                                                   \


Prefer to avoid defining this and only using once. Please inline it.

Here we need the macro at couple of places, so keeping it. When the vector floating point unit (VFPU) hardware is not available, we need to fall back to the reference implementation.

Looks like I missed some of the uses below. Looks good.

njeffrie · 2020-07-14T19:21:36Z

tensorflow/lite/micro/kernels/xtensa_hifi/conv.cc

      return kTfLiteError;
    }

    ALLOCATE_XTENSA_NNLIB_SCRATCH_MEM;
-    p_scratch = xtensa_nnlib_scratch_buf;
+    p_scratch = (char*)xtensa_nnlib_scratch_buf;


please use c++ style static_cast if possible or reinterpret_cast here and elsewhere in this file / change.

Updated the code in conv.cc and depthwise_conv.cc to clean up type casts.

The following changes are done to the HiFi4 implementation of TFLM 1. Update the kernel files and the make setup to use HiFi4 NN Library v2.2.0. 2. Update the kernel files as per the latest reference implementation. 3. Add xtenas_hifi kernel implementations for the add and mul operators.

pnikam-cad · 2020-07-15T12:00:00Z

I have updated the PR to address the above suggestions, please review.

njeffrie · 2020-07-21T16:43:44Z

tensorflow/lite/micro/kernels/xtensa_hifi/add.cc

+                           &output_activation_max);
+  tflite::ArithmeticParams op_params;
+  SetActivationParams(output_activation_min, output_activation_max, &op_params);
+#define TF_LITE_ADD(opname)                                                   \


Looks like I missed some of the uses below. Looks good.

google-ml-butler bot added the size:XL CL Change Size:Extra Large label Jun 30, 2020

googlebot added the cla: no label Jun 30, 2020

pnikam-cad force-pushed the hifi4_nnlib_v2_2_0_update branch from 865cf87 to 2e274bb Compare June 30, 2020 13:16

googlebot added cla: yes and removed cla: no labels Jun 30, 2020

gbaned self-assigned this Jun 30, 2020

gbaned added this to Assigned Reviewer in PR Queue via automation Jun 30, 2020

gbaned requested review from srjoglekar246 and njeffrie June 30, 2020 14:04

gbaned added comp:lite TF Lite related issues comp:micro Related to TensorFlow Lite Microcontrollers labels Jun 30, 2020

gbaned removed the request for review from srjoglekar246 June 30, 2020 14:06

gbaned added the awaiting review Pull request awaiting review label Jul 6, 2020

pnikam-cad force-pushed the hifi4_nnlib_v2_2_0_update branch from 7ac6995 to b06b94c Compare July 13, 2020 17:42

njeffrie previously approved these changes Jul 14, 2020

View reviewed changes

PR Queue automation moved this from Assigned Reviewer to Approved by Reviewer Jul 14, 2020

google-ml-butler bot added kokoro:force-run Tests on submitted change ready to pull PR ready for merge process labels Jul 14, 2020

kokoro-team removed kokoro:force-run Tests on submitted change labels Jul 14, 2020

pnikam-cad dismissed njeffrie’s stale review via 3b0642f July 15, 2020 11:52

pnikam-cad force-pushed the hifi4_nnlib_v2_2_0_update branch from b06b94c to 3b0642f Compare July 15, 2020 11:52

PR Queue automation moved this from Approved by Reviewer to Reviewer Requested Changes Jul 15, 2020

google-ml-butler bot removed the ready to pull PR ready for merge process label Jul 15, 2020

gbaned requested a review from njeffrie July 15, 2020 15:26

tensorflowbutler removed the awaiting review Pull request awaiting review label Jul 17, 2020

gbaned added the awaiting review Pull request awaiting review label Jul 17, 2020

njeffrie approved these changes Jul 21, 2020

View reviewed changes

PR Queue automation moved this from Reviewer Requested Changes to Approved by Reviewer Jul 21, 2020

google-ml-butler bot added kokoro:force-run Tests on submitted change ready to pull PR ready for merge process labels Jul 21, 2020

kokoro-team removed the kokoro:force-run Tests on submitted change label Jul 21, 2020

gbaned removed the awaiting review Pull request awaiting review label Jul 22, 2020

tensorflow-copybara merged commit bf3b14f into tensorflow:master Jul 22, 2020

PR Queue automation moved this from Approved by Reviewer to Merged Jul 22, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Cadence HiFi4 NN Library v2.2.0 update #40943

Cadence HiFi4 NN Library v2.2.0 update #40943

pnikam-cad commented Jun 30, 2020

googlebot commented Jun 30, 2020

googlebot commented Jun 30, 2020

njeffrie left a comment

njeffrie Jul 14, 2020

pnikam-cad Jul 15, 2020

njeffrie Jul 14, 2020

pnikam-cad Jul 15, 2020

njeffrie Jul 21, 2020

njeffrie Jul 14, 2020

pnikam-cad Jul 15, 2020

pnikam-cad commented Jul 15, 2020

njeffrie Jul 21, 2020

Cadence HiFi4 NN Library v2.2.0 update #40943

Cadence HiFi4 NN Library v2.2.0 update #40943

Conversation

pnikam-cad commented Jun 30, 2020

googlebot commented Jun 30, 2020

What to do if you already signed the CLA

Individual signers

Corporate signers

googlebot commented Jun 30, 2020

njeffrie left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

pnikam-cad commented Jul 15, 2020

Choose a reason for hiding this comment