Removing conditional check for -arch=native flag.

[Treece-Burgess]

Pull Request Description

This PR removes the conditional block which determined if the -arch=native flag was used. There is no apparent difference to setting this flag or not. See below for test output after removal of the conditional block.

Test running simpleMultiGPU in Cuda Component

Command line:

./simpleMultiGPU cuda:::dram__bytes.avg

Output:

Starting simpleMultiGPU
PAPI version: 7.1.0
CUDA-capable device count: 8
CUDA Device 0: Tesla V100-SXM2-32GB : computeCapability 7.0 runtimeVersion 12.1 driverVersion 12.5
CUDA Device 1: Tesla V100-SXM2-32GB : computeCapability 7.0 runtimeVersion 12.1 driverVersion 12.5
CUDA Device 2: Tesla V100-SXM2-32GB : computeCapability 7.0 runtimeVersion 12.1 driverVersion 12.5
CUDA Device 3: Tesla V100-SXM2-32GB : computeCapability 7.0 runtimeVersion 12.1 driverVersion 12.5
CUDA Device 4: Tesla V100-SXM2-32GB : computeCapability 7.0 runtimeVersion 12.1 driverVersion 12.5
CUDA Device 5: Tesla V100-SXM2-32GB : computeCapability 7.0 runtimeVersion 12.1 driverVersion 12.5
CUDA Device 6: Tesla V100-SXM2-32GB : computeCapability 7.0 runtimeVersion 12.1 driverVersion 12.5
CUDA Device 7: Tesla V100-SXM2-32GB : computeCapability 7.0 runtimeVersion 12.1 driverVersion 12.5
Generating input data...
Setup PAPI counters internally (PAPI)
Found CUDA Component at id 2
Add event success: 'cuda:::dram__bytes.avg:device=0' GPU 0
Add event success: 'cuda:::dram__bytes.avg:device=1' GPU 1
Add event success: 'cuda:::dram__bytes.avg:device=2' GPU 2
Add event success: 'cuda:::dram__bytes.avg:device=3' GPU 3
Add event success: 'cuda:::dram__bytes.avg:device=4' GPU 4
Add event success: 'cuda:::dram__bytes.avg:device=5' GPU 5
Add event success: 'cuda:::dram__bytes.avg:device=6' GPU 6
Add event success: 'cuda:::dram__bytes.avg:device=7' GPU 7
Computing with 8 GPUs...
Process GPU results on 8 GPUs...
PAPI counterValue        26491 		 --> cuda:::dram__bytes.avg:device=0 
PAPI counterValue        24344 		 --> cuda:::dram__bytes.avg:device=1 
PAPI counterValue        24144 		 --> cuda:::dram__bytes.avg:device=2 
PAPI counterValue        23583 		 --> cuda:::dram__bytes.avg:device=3 
PAPI counterValue        22175 		 --> cuda:::dram__bytes.avg:device=4 
PAPI counterValue        22783 		 --> cuda:::dram__bytes.avg:device=5 
PAPI counterValue        21681 		 --> cuda:::dram__bytes.avg:device=6 
PAPI counterValue        21450 		 --> cuda:::dram__bytes.avg:device=7 
  GPU Processing time: 3.043000 (ms)
Computing the same result with Host CPU...
  CPU Processing time: 4.527000 (ms) (speedup 1.49X)
Comparing GPU and Host CPU results...
  GPU sum: 777239.750000
  CPU sum: 777239.896331
  Relative difference: 1.882697E-07 
PASSED

Author Checklist

• Description
  Why this PR exists. Reference all relevant information, including background, issues, test failures, etc
• Commits
  Commits are self contained and only do one thing
  Commits have a header of the form: module: short description
  Commits have a body (whenever relevant) containing a detailed description of the addressed problem and its solution
• Tests
  The PR needs to pass all the tests

[jagode]

Now that you removed the initial setting of NVCFLAGS, you should also remove the + from line:
NVCFLAGS += -g -ccbin='$(CC)' $(PAPI_FLAG) so that it looks like:
NVCFLAGS = -g -ccbin='$(CC)' $(PAPI_FLAG)