AWS c3.8xlarge - CPU utilization is very less

We are using AWS c3.8xlarge instance[32 core processor with 60 GB RAM] for one of the batch job execution.

We are running the same job[2 instances] to find out CPU utilization. The input file volume is huge (150 GB * 2 = 300 GB = 2 Billion total records)

When we run the job as single instance its completed in 2 hours.

But when we try to run the job parallel it's taking around 6 hours[The main reason is it's not utilizing the CPU at some point, it's using less than 10% at some point].

Is anyone else faced the kind of issues?

Any help on this greatly appreciated.

Regards, Shankar