We outline an approach for predicting the performance of High-Performance Computing (HPC) workloads based on fine-grained architectural metrics. The methodology and workflow presented consists of gathering data from runtime hardware counters across a range of HPC applications and benchmarks and developing an artificial intelligence model based on ensemble tree algorithms. This model is capable of forecasting the performance of other unseen HPC applications. The workflow presented is based on automatic instrumentation without manual code changes. It uses the traces used for performance analysis to train the model, is fast to run and integrates the prediction results with the performance tools, providing an easy way to iterate and fine-tune the model. Through this approach, we prove that a prediction of the instructions per cycle (IPC) metric of unseen applications is possible based on hardware counters that can be obtained with standard performance tools.