Bf16 gpu support

[richagadgil]

No description provided.

[codecov]

Codecov Report

Attention: Patch coverage is 12.50000% with 7 lines in your changes missing coverage. Please review.

Project coverage is 92.14%. Comparing base (b39a938) to head (59eec66).
Report is 3 commits behind head on develop.

Files with missing lines	Patch %	Lines
src/quantization.cpp	0.00%	7 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #3630      +/-   ##
===========================================
- Coverage    92.18%   92.14%   -0.04%     
===========================================
  Files          513      513              
  Lines        21576    21584       +8     
===========================================
- Hits         19889    19888       -1     
- Misses        1687     1696       +9     

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

---

🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests

[pfultz2]

There is duplicate typedefs here and they are both wrong. It should be using bf16 = __hip_bfloat16.

[migraphx-bot]

Test	Batch	Rate new 59eec6	Rate old 0f36aa	Diff	Compare
torchvision-resnet50	64	3,247.51	3,261.99	-0.44%	✅
torchvision-resnet50_fp16	64	2,766.25	6,984.41	-60.39%	🔴
torchvision-densenet121	32	2,264.43	2,434.46	-6.98%	🔴
torchvision-densenet121_fp16	32	3,889.55	4,068.77	-4.40%	🔴
torchvision-inceptionv3	32	1,634.16	1,630.14	0.25%	✅
torchvision-inceptionv3_fp16	32	2,752.52	2,746.22	0.23%	✅
cadene-inceptionv4	16	761.60	765.59	-0.52%	✅
cadene-resnext64x4	16	806.14	809.78	-0.45%	✅
slim-mobilenet	64	7,356.64	7,474.57	-1.58%	✅
slim-nasnetalarge	64	208.59	208.58	0.00%	✅
slim-resnet50v2	64	3,436.22	3,441.49	-0.15%	✅
bert-mrpc-onnx	8	1,150.74	1,150.80	-0.01%	✅
bert-mrpc-tf	1	482.60	465.54	3.66%	🔆
pytorch-examples-wlang-gru	1	424.90	420.06	1.15%	✅
pytorch-examples-wlang-lstm	1	174.63	381.98	-54.28%	🔴
torchvision-resnet50_1	1	771.08	750.44	2.75%	✅
cadene-dpn92_1	1	405.44	398.35	1.78%	✅
cadene-resnext101_1	1	381.56	382.96	-0.37%	✅
onnx-taau-downsample	1	159.45	346.08	-53.93%	🔴
dlrm-criteoterabyte	1	33.34	33.35	-0.01%	✅
dlrm-criteoterabyte_fp16	1	52.74	52.68	0.10%	✅
agentmodel	1	8,250.55	8,091.53	1.97%	✅
unet_fp16	2	58.96	58.77	0.33%	✅
resnet50v1_fp16	1	1,004.59	943.16	6.51%	🔆
resnet50v1_int8	1	998.84	1,012.12	-1.31%	✅
bert_base_cased_fp16	64	1,170.38	1,169.97	0.04%	✅
bert_large_uncased_fp16	32	363.05	363.75	-0.19%	✅
bert_large_fp16	1	198.54	199.03	-0.25%	✅
distilgpt2_fp16	16	2,200.89	2,201.98	-0.05%	✅
yolov5s	1	531.48	539.79	-1.54%	✅
tinyllama	1	43.41	43.42	-0.01%	✅
vicuna-fastchat	1	168.79	175.75	-3.96%	🔴
whisper-tiny-encoder	1	417.55	418.02	-0.11%	✅
whisper-tiny-decoder	1	428.26	428.37	-0.03%	✅

This build is not recommended to 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