andrey/k210-sdk-stuff
1
0
Fork 0
mirror of https://github.com/laanwj/k210-sdk-stuff.git synced 2024-11-22 01:16:20 +04:00
Code Issues Projects Releases Wiki Activity

doc: Typo fixes in kpu.md

Browse Source
This commit is contained in:
Wladimir J. van der Laan 2020-01-21 18:36:35 +01:00
parent bd40892845
commit dff0540fc6
1 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
doc/kpu.md
View File

@ -42,7 +42,7 @@ segments (`kpu_activate_table_t`).
> (flash size - software size)
The flash size specs are somewhat of a red herring as they relate to software
instead of hardare: the KPU does not have logic for loading parameters from
instead of hardware: the KPU does not have logic for loading parameters from
flash.
Some other source mentions:
@ -76,10 +76,10 @@ Looking at `lib/drivers/kpu.c` in the SDK, function `ai_step`, many types of CNN
implemented in software instead of executed by the KPU. I suppose they accelerated the
most common multiplication-intensive layers in hardware, which is `KL_K210_CONV`.
Peripehral layout
Peripheral layout
=================
The register layout of the peripheral is as folllows. Source: `lib/drivers/include/kpu.h`.
The register layout of the peripheral is as follows. Source: `lib/drivers/include/kpu.h`.
All registers are 64-bit.
| Ofs | Name | Description |
@ -114,13 +114,13 @@ src/common/include/kernels/k210/k210_kernels.h (emulation)
bit name
------ ----------------------
0 `int_en` Generate interuupt after layer computation finished
0 `int_en` Generate interrupt after layer computation finished
1 `ram_flag` ?
2 `full_add` Set in `kpu_conv2d_output_full_add`
3 `depth_wise_layer` Is a "depth-wise" layer (1 if enabled)
4..63 reserved
"depth-wise" affects meny of the computations: it likely means that the layer
"depth-wise" affects many of the computations: it likely means that the layer
computation mixes multiple channels so that they cannot be processed one by
one.