Add support for unsigned types with mlir

[pfultz2]

No description provided.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.18%. Comparing base (1cfd6c2) to head (75e4348).
Report is 7 commits behind head on develop.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #3582   +/-   ##
========================================
  Coverage    92.17%   92.18%           
========================================
  Files          513      513           
  Lines        21560    21576   +16     
========================================
+ Hits         19873    19889   +16     
  Misses        1687     1687           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[pfultz2]

This requires the PR to be merged in MLIR: ROCm/rocMLIR#1692

[TedThemistokleous]

Does this mean ops like the matmul and quant dot not need to have the shifts for uint8 anymore now?

[pfultz2]

Does this mean ops like the matmul and quant dot not need to have the shifts for uint8 anymore now?

I though the shifts was needed because one input was signed and the other unsigned.

[migraphx-bot]

Test	Batch	Rate new 75e434	Rate old b39a93	Diff	Compare
torchvision-resnet50	64	3,260.00	3,258.87	0.03%	✅
torchvision-resnet50_fp16	64	6,987.58	6,988.66	-0.02%	✅
torchvision-densenet121	32	2,434.76	2,432.78	0.08%	✅
torchvision-densenet121_fp16	32	4,098.57	4,101.38	-0.07%	✅
torchvision-inceptionv3	32	1,628.11	1,627.93	0.01%	✅
torchvision-inceptionv3_fp16	32	2,744.19	2,743.41	0.03%	✅
cadene-inceptionv4	16	765.37	764.71	0.09%	✅
cadene-resnext64x4	16	811.58	809.79	0.22%	✅
slim-mobilenet	64	7,468.25	7,462.30	0.08%	✅
slim-nasnetalarge	64	208.47	208.39	0.04%	✅
slim-resnet50v2	64	3,439.16	3,441.05	-0.05%	✅
bert-mrpc-onnx	8	1,146.84	1,151.55	-0.41%	✅
bert-mrpc-tf	1	468.29	469.89	-0.34%	✅
pytorch-examples-wlang-gru	1	427.17	438.68	-2.62%	✅
pytorch-examples-wlang-lstm	1	403.86	384.52	5.03%	🔆
torchvision-resnet50_1	1	808.88	773.59	4.56%	🔆
cadene-dpn92_1	1	399.61	402.80	-0.79%	✅
cadene-resnext101_1	1	383.28	383.13	0.04%	✅
onnx-taau-downsample	1	345.67	345.42	0.07%	✅
dlrm-criteoterabyte	1	33.32	33.30	0.06%	✅
dlrm-criteoterabyte_fp16	1	52.51	52.73	-0.42%	✅
agentmodel	1	8,289.40	8,187.33	1.25%	✅
unet_fp16	2	58.93	58.82	0.19%	✅
resnet50v1_fp16	1	949.30	933.18	1.73%	✅
resnet50v1_int8	1	997.79	1,017.62	-1.95%	✅
bert_base_cased_fp16	64	1,171.38	1,171.03	0.03%	✅
bert_large_uncased_fp16	32	363.46	363.30	0.04%	✅
bert_large_fp16	1	200.32	199.02	0.66%	✅
distilgpt2_fp16	16	2,204.62	2,201.62	0.14%	✅
yolov5s	1	528.44	534.34	-1.10%	✅
tinyllama	1	43.71	43.42	0.67%	✅
vicuna-fastchat	1	178.22	177.52	0.40%	✅
whisper-tiny-encoder	1	415.71	418.19	-0.59%	✅
whisper-tiny-decoder	1	426.35	427.01	-0.15%	✅

Check results before merge 🔆

[migraphx-bot]

     ✅ bert-mrpc-onnx: PASSED: MIGraphX meets tolerance

     ✅ bert-mrpc-tf: PASSED: MIGraphX meets tolerance

     ✅ pytorch-examples-wlang-gru: PASSED: MIGraphX meets tolerance

     ✅ pytorch-examples-wlang-lstm: PASSED: MIGraphX meets tolerance

     ✅ torchvision-resnet50_1: PASSED: MIGraphX meets tolerance

     ✅ cadene-dpn92_1: PASSED: MIGraphX meets tolerance

     ✅ cadene-resnext101_1: PASSED: MIGraphX meets tolerance

     ✅ dlrm-criteoterabyte: PASSED: MIGraphX meets tolerance

     ✅ agentmodel: PASSED: MIGraphX meets tolerance

     ✅ unet: PASSED: MIGraphX meets tolerance

     ✅ resnet50v1: PASSED: MIGraphX meets tolerance

     ✅ bert_base_cased_fp16: PASSED: MIGraphX meets tolerance

🔴bert_large_uncased_fp16: FAILED: MIGraphX is not within tolerance - check verbose output

     ✅ bert_large: PASSED: MIGraphX meets tolerance

     ✅ yolov5s: PASSED: MIGraphX meets tolerance

     ✅ tinyllama: PASSED: MIGraphX meets tolerance

     ✅ vicuna-fastchat: PASSED: MIGraphX meets tolerance

     ✅ whisper-tiny-encoder: PASSED: MIGraphX meets tolerance

     ✅ whisper-tiny-decoder: PASSED: MIGraphX meets tolerance

     ✅ distilgpt2_fp16: PASSED: MIGraphX meets tolerance