MLECO-1901: Document MAC config mismatch error
Change-Id: I9b6b63724d136f36f57a9402059327f8fab27bf1
Signed-off-by: George Gekov <george.gekov@arm.com>
diff --git a/docs/documentation.md b/docs/documentation.md
index b7b6b30..3ccdcbd 100644
--- a/docs/documentation.md
+++ b/docs/documentation.md
@@ -308,6 +308,7 @@
- [Troubleshooting](./sections/troubleshooting.md)
- [Inference results are incorrect for my custom files](./sections/troubleshooting.md#inference-results-are-incorrect-for-my-custom-files)
- [The application does not work with my custom model](./sections/troubleshooting.md#the-application-does-not-work-with-my-custom-model)
+ - [NPU configuration mismatch error when running inference](./sections/troubleshooting.md#npu-configuration-mismatch-error-when-running-inference)
## Appendix
diff --git a/docs/sections/troubleshooting.md b/docs/sections/troubleshooting.md
index 8ab81eb..0ba6ced 100644
--- a/docs/sections/troubleshooting.md
+++ b/docs/sections/troubleshooting.md
@@ -3,6 +3,7 @@
- [Troubleshooting](#troubleshooting)
- [Inference results are incorrect for my custom files](#inference-results-are-incorrect-for-my-custom-files)
- [The application does not work with my custom model](#the-application-does-not-work-with-my-custom-model)
+ - [NPU configuration mismatch error when running inference](#npu-configuration-mismatch-error-when-running-inference)
## Inference results are incorrect for my custom files
@@ -23,4 +24,46 @@
> from: <https://pypi.org/project/ethos-u-vela/>. The source code is hosted on
> <https://git.mlplatform.org/ml/ethos-u/ethos-u-vela.git/>.
-The next section of the documentation is: [Appendix](appendix.md).
+## NPU configuration mismatch error when running inference
+
+If you see the following error message when an inference is executed:
+
+```commandline
+NPU config mismatch: npu.macs_per_cc=8 optimizer.macs_per_cc=7
+NPU config mismatch: npu.shram_size=48 optimizer.shram_size=24
+ERROR - Invoke failed.
+ERROR - Inference failed.
+```
+
+It shows that the configuration of the Vela compiled `.tflite` file doesn't match the number of MAC units on the FVP.
+
+The Vela configuration parameter `accelerator-config` used for producing the .`tflite` file that is used
+while building the application should match the MAC configuration that the FVP is emulating.
+For example, if the `accelerator-config` from the Vela command was `ethos-u55-128`, the FVP should be emulating the
+128 MAC configuration of the Ethos-U55 block(default FVP configuration). If the `accelerator-config` used was
+`ethos-u55-256`, the FVP must be executed with additional command line parameter to instruct it to emulate the
+256 MAC configuration instead.
+
+The [deploying on an FVP emulating MPS3](./deployment.md#deploying-on-an-fvp-emulating-mps3) page provides guidance
+on how to instruct the FVP to change the number of MAC units.
+
+Note that when the FVP is launched and the application starts executing, various parameters about the system are
+logged over UART. These include the MAC/cc configuration of the FVP.
+
+```log
+INFO - MPS3 core clock has been set to: 32000000Hz
+INFO - CPU ID: 0x410fd220
+INFO - CPU: Cortex-M55 r0p0
+INFO - Ethos-U55 device initialised
+INFO - Ethos-U55 version info:
+INFO - Arch: v1.0.6
+INFO - Driver: v0.16.0
+INFO - MACs/cc: 128
+INFO - Cmd stream: v0
+INFO - SHRAM size: 24
+INFO - Arm Corstone-300 (SSE-300) platform initialised
+INFO - ARM ML Embedded Evaluation Kit for MPS3 FPGA and FastModel
+INFO - Target system design: Arm Corstone-300 (SSE-300)
+```
+
+Next section of the documentation: [Appendix](appendix.md).