The Computational Problem

There is a huge and growing mountain of seismic data - new and archived material requiring reprocessing - out there. To interpret it means that the demand for computational tools to support the 3D seismic interpretation process throughout the E&P (exploration and production) workflow has never been greater.

Even with Moore's Law, high performance clusters, current computational software, a data driven approach to 3D seismic interpretation can provide only provide step changes in productivity.  For geoscientists to fully exploit their data and deliver more accurate interpretations of the subsurface, improved understanding of uncertainty and better quantification of risk, processing power needs to scale at 10-25X..

The Solution

Enter RAI's SuperGPUcluster. The HR-100 introduces an innovative new way of obtaining extreme performance: hybrid-core computing. Hybrid-core computing is based on an asymmetric architecture that combines the economies and programmability of industry standard processors with the performance and efficiency of a hardware-based, application-specific design.

The RAI HR-100 server employs highly parallel coprocessors that augment the capabilities of commodity multi-core processors with processing elements optimized for performance-critical operations. Instructions executed by the coprocessor appear as extensions to the x86 instruction set; applications can contain both x86 and coprocessor instructions in a single instruction stream.

The coprocessors are based on standard Field Programmable Gate Arrays (FPGAs), Nvidia's Tesla and Fermi,  and is coupled with a standard multi-core Intel® Xeon® or AMD Istanbul processors. While each processor has its own physical memory, logically memory is globally addressable and cache-coherent, making software development as easy as that of a standard x86 environment.

RAI offers complete hybrid systems, consulting and tools to port your application and optimize it for the massively parallel environment using Nvidia's CUDA language.  For a limited time, RAI is offering a complete GPU starter Package to help you develop a hybrid prototype of your application, so you can  determine the ROI and level of effort entailed in porting all your critical applications.

See how others are using GPUs in the Oil and Gas business:

• Hess's use of GPU's in seismic processing -Video

• Hess GPU slide show - PDF

These videos and reference materials can help you understand more about what it takes to migrate your applications to take advantage of GPUs superior price/performance over conventional clusters:

• Developers share their initial experiences using CUDA
• Introduction to Nvidia's Next Gen GPGPU - FERMI
• Nvidia's Fermi Architecture White Paper
• Nvidia's Fermi Slide Show

 

GPGPU Hardware Subsystems

GPU-Integrated Servers (1U)

		6016GT-TF-TM2 • Dual Quad/Dual-Core Intel® Xeon® processors 5500 series • Intel® 5520 chipset with QPI up to 6.4GT/s • Up to 96GB of Reg. ECC DDR3 DIMM 1333/1066/800 MHz SDRAM • 2x NVIDIA Tesla M1060 GPU Cards    PCI-E 2.0 x4 (in x16 slot - Low-Profile) • 3x hot-swap 3.5" drive bays • Matrox G200eW graphics controller • 8x counter-rotating fans with optimal fan speed control • IPMI 2.0 with virtual media over LAN and KVM-ove-LAN • Dual LAN with Intel 82576 Gigabit Ethernet controller • 1400W high-efficiency Gold Level (93%) Power Supply with PMbus
		6016GT-TF-TC2 • Dual Quad/Dual-Core Intel® Xeon® processors 5500 series • Intel® 5520 chipset with QPI up to 6.4GT/s • Up to 96GB of Reg. ECC DDR3 DIMM 1333/1066/800 MHz SDRAM • 2x NVIDIA Tesla C1060 GPU Cards    PCI-E 2.0 x4 (in x16 slot - Low-Profile) • 3x hot-swap 3.5" drive bays • Matrox G200eW graphics controller • 8x counter-rotating fans with optimal fan speed control • IPMI 2.0 with virtual media over LAN and KVM-ove-LAN • Dual LAN with Intel 82576 Gigabit Ethernet controller • 1400W high-efficiency Gold Level (93%) Power Supply with PMbus

    GPU-Ready Servers (1U)

		6016GT-TF • Dual Quad/Dual-Core Intel® Xeon® processors 5500 series • Intel® 5520 chipset with QPI up to 6.4GT/s • Up to 96GB of Reg. ECC DDR3 DIMM 1333/1066/800 MHz SDRAM • Supports up to two of any PCI-E 2.0 x16 compatible GPUs    2x PCI-E 2.0 x16 (Full-height Full-length)    PCI-E 2.0 x4 (in x16 slot - Low-Profile) • 3x hot-swap 3.5" drive bays • Matrox G200eW graphics controller • 8x counter-rotating fans with optimal fan speed control • IPMI 2.0 with virtual media over LAN and KVM-ove-LAN • Dual LAN with Intel 82576 Gigabit Ethernet controller • 1400W high-efficiency Gold Level (93%) Power Supply with PMbus Note: Graphics card is not included
		6016T-GIBQF / 6016T-GIBXF / 6016T-GTF • Dual Quad/Dual-Core Intel® Xeon® processors 5500 series • Intel® 5520 chipset with QPI up to 6.4GT/s • Up to 48GB of Reg. ECC DDR3 DIMM 1333/1066/800 MHz SDRAM • Supports one PCI-E 2.0 x16 compatible GPU    6016T-GIBQF: ConnectX QDR InfiniBand with QSFP connector    6016T-GIBXF: ConnectX DDR InfiniBand with QSFP connector • 3x hot-swap 3.5" drive bays • Matrox G200eW graphics controller • 4x counter-rotating fans with optimal fan speed control • IPMI 2.0 with virtual media over LAN and KVM-ove-LAN • Dual LAN with Intel 82576 Gigabit Ethernet controller • 1400W high-efficiency Gold Level (93%) Power Supply with PMbus Note: Graphics card is not included

    Support for 5 Add-on Cards (1U)

6016XT-TF • Dual Quad/Dual-Core Intel® Xeon® processors 5500 series • Intel® 5520 chipset with QPI up to 6.4GT/s • Up to 96GB of Reg. ECC DDR3 DIMM 1333/1066/800 MHz SDRAM • Supports 5x PCI-E 2.0 add-on cards in 1U    4x PCI-E 2.0 x8 (in x16 slots - Full-height Full-length)    PCI-E 2.0 x4 (in x16 slot - Low-Profile) • 3x hot-swap 3.5" drive bays • Matrox G200eW graphics controller • 8x counter-rotating fans with optimal fan speed control • IPMI 2.0 with virtual media over LAN and KVM-ove-LAN • Dual LAN with Intel 82576 Gigabit Ethernet controller • 1400W high-efficiency Gold Level (93%) Power Supply with PMbus Note: Graphics card is not included     All of the above Server SKUs use Chassis 818G-1400B