WH Issues

[kkwong10]

Single Op LLKs

• [x] #2258
• [x] #2259
• [x] #2283
• [x] #2292

Fused Op LLKs

• [x] #2260
• [x] #2261
• [x] #2557

Op Failure

• [x] rsqrt - #2315
• [x] power op #2314
• [ ] sign-bit #2317
• [x] exp2, expm1 #2319
• [x] log - #2397
• [x] leaky relu #2318 (fixed)
• [x] erf - erfc #2316 (fixed)

Unary + Binary Op Failure

• [ ] L1 memory config #2313

Test updates

• 5071

• 5072

Matrix multiply bugs

• python tests for large shapes hang/error for large shapes #2459
• tilize-untilize BMM mixed precision test hang/error for large shapes #2460

Reduce bugs

• reduce max w, sum w hang #2470

[davorchap]

fyi @ttmtrajkovic and @muthutt -- this a top-level WH LLK case linking to the individual cases.

[kkwong10]

@davorchap @ttmtrajkovic Creating artificial c++ tests to hit the error scenario may not prove fruitful since the C++ tests may not use the same mapping as the op-library python tests. I think the longer term solution would be to have more C++ based unit testing, but for now I am trying to enable versim on python tests and see what is needed to get waveforms from that. Once I have a process for that, we can pretty much have other people create waveforms and generate additional information if LLKs are not working.