Remove rocmlir unsupported reduce types

[dhernandez0]

We are going to enable f16 for reduce in this rocmlir PR: ROCm/rocMLIR#1722

However, the code here assumes bf16 and fp8 are supported and that's not the case.

[umangyadav]

These would require tests in test/verify/ section

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.29%. Comparing base (720bc53) to head (a540de5).
Report is 18 commits behind head on develop.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #3766      +/-   ##
===========================================
+ Coverage    92.24%   92.29%   +0.05%     
===========================================
  Files          519      519              
  Lines        22170    22233      +63     
===========================================
+ Hits         20450    20520      +70     
+ Misses        1720     1713       -7     

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

[dhernandez0]

By the way, reduce f16 should not be enabled for navi. As it does not support atomic_add with packed_f16, so it would be slow. Leaving this comment here for whoever is assigned the PR. @causten

[CharlieL7]

What's the status on this PR? Is there additional work needed or is the removal of the allowed types the only thing that needs review?

[dhernandez0]

What's the status on this PR? Is there additional work needed or is the removal of the allowed types the only thing that needs review?

I thought this was all that needed to be done, so I created the PR. However, we need some tests and make sure reduce f16 is not enabled for Navi (except for gfx12). I think @causten will assign somebody from migraphx team to finish this.

[migraphx-bot]

Test	Batch	Rate new ec1df6	Rate old 879f30	Diff	Compare
torchvision-resnet50	64	3,253.19	3,235.22	0.56%	✅
torchvision-resnet50_fp16	64	6,917.99	6,875.82	0.61%	✅
torchvision-densenet121	32	2,452.71	2,435.92	0.69%	✅
torchvision-densenet121_fp16	32	4,172.34	4,191.85	-0.47%	✅
torchvision-inceptionv3	32	1,629.82	1,614.66	0.94%	✅
torchvision-inceptionv3_fp16	32	2,714.65	2,690.95	0.88%	✅
cadene-inceptionv4	16	763.10	749.91	1.76%	✅
cadene-resnext64x4	16	812.73	809.52	0.40%	✅
slim-mobilenet	64	7,461.17	6,664.85	11.95%	🔆
slim-nasnetalarge	64	208.63	199.04	4.82%	🔆
slim-resnet50v2	64	3,443.42	3,427.74	0.46%	✅
bert-mrpc-onnx	8	1,145.68	1,145.10	0.05%	✅
bert-mrpc-tf	1	484.41	477.68	1.41%	✅
pytorch-examples-wlang-gru	1	487.10	473.56	2.86%	✅
pytorch-examples-wlang-lstm	1	446.35	443.33	0.68%	✅
torchvision-resnet50_1	1	812.59	799.32	1.66%	✅
cadene-dpn92_1	1	427.87	432.43	-1.05%	✅
cadene-resnext101_1	1	385.37	390.06	-1.20%	✅
onnx-taau-downsample	1	374.60	373.13	0.39%	✅
dlrm-criteoterabyte	1	33.34	31.81	4.81%	🔆
dlrm-criteoterabyte_fp16	1	52.35	51.06	2.53%	✅
agentmodel	1	8,682.59	8,763.63	-0.92%	✅
unet_fp16	2	58.55	57.90	1.12%	✅
resnet50v1_fp16	1	1,027.37	1,046.48	-1.83%	✅
resnet50v1_int8	1	1,021.70	799.12	27.85%	🔆
bert_base_cased_fp16	64	1,180.44	1,173.21	0.62%	✅
bert_large_uncased_fp16	32	365.20	362.51	0.74%	✅
bert_large_fp16	1	199.45	200.49	-0.52%	✅
distilgpt2_fp16	16	2,227.37	2,218.26	0.41%	✅
yolov5s	1	529.84	530.33	-0.09%	✅
tinyllama	1	43.61	43.59	0.04%	✅
vicuna-fastchat	1	173.56	174.80	-0.71%	✅
whisper-tiny-encoder	1	416.58	411.98	1.12%	✅
whisper-tiny-decoder	1	432.27	411.87	4.95%	🔆

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