Remove zero point parameter for dequantizelinear when its zero

[pfultz2]

This removes the zero point from dequantizelinear when its zero. We dont do this for quantizelinear since its necessary for deducing the output type.

[lakhinderwalia]

This is no longer a variable called zero. It is initialized to a 1. Maybe it should be called zp? Also, are you changing its value so it doesn't get eliminated?

[lakhinderwalia]

For case of quantizelinear, could this not replace a bunch of 0s with a scalar broadcast..? Thanks.

[TedThemistokleous]

Isn't zero point based on the input type here for dequantize? If so why are we using float_equal then? Is this more to cover the case of say fp8 as well as int8, int4, etc?

[pfultz2]

The visitor will visit all data types, so we need the float_equal if the zero point is a floating point(which is the case for fp8).

[TedThemistokleous]

One question, otherwise I get what you're doing here

[TedThemistokleous]

Fix CI but I think this is fine with what you're doing here

[migraphx-bot]

Test	Batch	Rate new 427fee	Rate old b73def	Diff	Compare
torchvision-resnet50	64	3,260.08	3,257.93	0.07%	✅
torchvision-resnet50_fp16	64	6,991.66	6,992.99	-0.02%	✅
torchvision-densenet121	32	2,438.54	2,432.26	0.26%	✅
torchvision-densenet121_fp16	32	4,058.84	4,038.39	0.51%	✅
torchvision-inceptionv3	32	1,637.48	1,638.89	-0.09%	✅
torchvision-inceptionv3_fp16	32	2,763.91	2,761.69	0.08%	✅
cadene-inceptionv4	16	776.01	776.39	-0.05%	✅
cadene-resnext64x4	16	810.74	811.37	-0.08%	✅
slim-mobilenet	64	7,538.67	7,532.73	0.08%	✅
slim-nasnetalarge	64	211.54	211.42	0.06%	✅
slim-resnet50v2	64	3,504.60	3,507.25	-0.08%	✅
bert-mrpc-onnx	8	1,148.10	1,147.76	0.03%	✅
bert-mrpc-tf	1	462.55	469.91	-1.57%	✅
pytorch-examples-wlang-gru	1	421.08	514.96	-18.23%	🔴
pytorch-examples-wlang-lstm	1	385.54	386.61	-0.28%	✅
torchvision-resnet50_1	1	805.56	772.05	4.34%	🔆
cadene-dpn92_1	1	396.22	398.73	-0.63%	✅
cadene-resnext101_1	1	384.50	383.67	0.22%	✅
onnx-taau-downsample	1	342.75	342.33	0.12%	✅
dlrm-criteoterabyte	1	33.45	33.33	0.36%	✅
dlrm-criteoterabyte_fp16	1	52.74	52.70	0.06%	✅
agentmodel	1	8,499.64	8,056.20	5.50%	🔆
unet_fp16	2	58.86	58.92	-0.10%	✅
resnet50v1_fp16	1	944.31	950.32	-0.63%	✅
resnet50v1_int8	1	1,003.77	1,000.02	0.37%	✅
bert_base_cased_fp16	64	1,170.09	1,169.24	0.07%	✅
bert_large_uncased_fp16	32	363.92	363.69	0.06%	✅
bert_large_fp16	1	198.80	198.89	-0.04%	✅
distilgpt2_fp16	16	2,206.09	2,203.09	0.14%	✅
yolov5s	1	532.52	540.85	-1.54%	✅
tinyllama	1	43.72	43.43	0.65%	✅
vicuna-fastchat	1	181.32	170.64	6.26%	🔆
whisper-tiny-encoder	1	419.46	418.21	0.30%	✅
whisper-tiny-decoder	1	427.68	426.10	0.37%	✅

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