Our research group operates a multi-tiered high-performance computing infrastructure composed of dedicated in-house compute nodes, a GPU workstation, and national supercomputing resources.
Our Group Compute Cluster
We maintain a dedicated in-house compute cluster established during the TUBITAK Projects 117M430 and 120M671, with compute nodes connected via 40 Gbps Infiniband for high-speed parallel data transfer. The architecture includes a login node for administration and two compute nodes with RAID configurations optimized for storage and I/O performance.
| Node |
CPU |
Memory |
Storage |
| Compute Node 1 |
2 × Intel Xeon E5-2660 v4 @ 2.00GHz |
256 GB RDIMM |
1.6 TB SSD (RAID0) + 4 × 4TB HDD (RAID5) |
| Compute Node 2 |
2 × Intel Xeon Silver 4214R @ 2.40GHz |
192 GB RDIMM |
2 × 480GB SSD (RAID1) + 4 × 960GB SSD (RAID5) + 2 × 16TB HDD |
GPU Workstation
A new GPU-enabled workstation was established under TUBITAK Project 124M416 to support simulation-visualization tasks and real-time rendering with virtual reality devices.
| Component |
Specification |
| CPU |
Intel i9-14900K |
| Memory |
96 GB DDR5 |
| GPU |
NVIDIA GeForce RTX 4070 |
| Storage |
2 × 4TB SSD (Run Space) + 2 × 10TB HDD |
| VR Devices |
2 × Meta Quest |
National HPC Resources (UHeM)
In addition to our group cluster, we actively utilize national supercomputing resources provided by
UHeM at ITU. We access two major clusters:
- ALTAY: 1.5 PFLOPS CPU cluster based on AMD EPYC processors
- SARIYER: GPU cluster with NVIDIA A100 and V100 accelerators
Our projects have been allocated thousands of CPU-hours over the years.
| Project Title |
Duration |
CPU Hours Used |
| DVT Modeling |
2018–2022 |
368,495 |
| Development of Quantum Machine Learning (QML) Algorithms for the solution of Partial Differential Equations |
2021–2025 |
128,816 |
| Computational Hemodynamic Modelling of Prosthesis Heart and Venous Valves |
2021–2025 |
1,180,641 |
| Hemodynamic Simulation of Supra Left Ventricular Heart Blood Flow Coupled with Coronary Arteries |
2025–2028 |
185,000 |