# Does Executorch support int16x8 quantization?

**URL:** <https://discuss.pytorch.org/t/does-executorch-support-int16x8-quantization/211418>\
**Category:** ExecuTorch\
**Created:** [October 11, 2024, 9:09am UTC](https://discuss.pytorch.org/t/does-executorch-support-int16x8-quantization/211418 "2024-10-11T09:09:46Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![andresovela](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/andresovela/32/73277_2.png) [@andresovela](https://discuss.pytorch.org/u/andresovela)\
**Post date:** [October 11, 2024, 9:09am UTC](https://discuss.pytorch.org/t/does-executorch-support-int16x8-quantization/211418/1 "2024-10-11T09:09:46Z")

</div>

Hi,

I’m currently working on getting audio models to run on HiFi 4 DSPs, coming from TFLite Micro.

TFLM supports [int16x8 quantization](https://ai.google.dev/edge/litert/models/post_training_integer_quant_16x8) and I’m wondering if Executorch does too.

I just started looking into Executorch, so I’m pretty ignorant about PyTorch stuff at the moment. I read something about Mixed Precision support in PyTorch, but I couldn’t find anything about that in Executorch.

---

<div class="post-metadata">

**Author:** ![lucylq](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/lucylq/32/65270_2.png) [@lucylq](https://discuss.pytorch.org/u/lucylq)\
**Post date:** [October 16, 2024, 11:05pm UTC](https://discuss.pytorch.org/t/does-executorch-support-int16x8-quantization/211418/2 "2024-10-16T23:05:30Z")

</div>

Hey @andresovela, thanks for taking a look at ExecuTorch!

@jerryzh168, is there support for int16x8 quantization?

ExecuTorch supports various quantization schemes - see [Quantization Overview — ExecuTorch 0.3 documentation](https://pytorch.org/executorch/stable/quantization-overview.html#quantization-overview)

---

<div class="post-metadata">

**Author:** ![jerryzh168](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/jerryzh168/32/15217_2.png) [@jerryzh168](https://discuss.pytorch.org/u/jerryzh168)\
**Post date:** [October 16, 2024, 11:08pm UTC](https://discuss.pytorch.org/t/does-executorch-support-int16x8-quantization/211418/3 "2024-10-16T23:08:36Z")

</div>

it’s not right now probably, but it’s possible to express in quantizer API, and also will require xnnpack or some backend to implement an efficient op for it I think

---

<div class="post-metadata">

**Author:** ![andresovela](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/andresovela/32/73277_2.png) [@andresovela](https://discuss.pytorch.org/u/andresovela)\
**Post date:** [October 17, 2024, 1:27pm UTC](https://discuss.pytorch.org/t/does-executorch-support-int16x8-quantization/211418/4 "2024-10-17T13:27:00Z")

</div>

I see. Thanks for the response!