Serban Georgescu

Molecular cloning of human homolog of yeast GAA1 which is required for attachment of glycosylphosphatidylinositols to proteins

Anchoring proteins to cell surface membranes by glycosylphosphatidylinositols (GPIs) is important. We have isolated a component of the putative transamidase machinery, hGaa1p (human GPI anchor attachment protein). hGAA1 cDNA is approximately 2 kb in length and codes 621 amino acids. The amino acid sequence of hGaa1p is 25% identical and 57% homologous to that of yeast Gaa1p. Moreover, Kite‐Dolittle hydrophobicity plots of both proteins show marked similarity. hGAA1 gene is expressed ubiquitously and mRNA levels are higher in the undifferentiated state. Overexpression of antisense hGAA1 in human K562 cells significantly reduced the production of a reporter GPI‐anchored protein.

GPU accelerated CAE using open solvers and the cloud

After more than five years since GPUs were first used as accelerators for general scientific computations, the field of General Purpose GPU computing or GPGPU has finally reached mainstream. Developers have now access to a mature hardware and software ecosystem. On the software side, several major open-source packages now support GPU acceleration while on the hardware side cloud-based solutions provide a simple way to access powerful machines with the latest GPUs at low cost. In this context, we look at the GPU acceleration of CAE, with a focus on the matrix solvers. We compare the performance that can be achieved using the open-source solver package PETSc ran on GPU-enabled Amazon EC2 hardware with that of an optimized legacy FEM code ran on a last generation 12-core blade server. Our results show that, although good performance can be achieved, some development is still needed to achieve peak performance.

Software design for decoupled parallel meshing of CAD models

The creation of Finite Element (FE) meshes is one of the most time-consuming steps in FE analysis. While the exponential increase in computational power, following Moores law, has gradually reduced the time spent in the FE solver, this has not generally been the case for FE mesh creation software. There are two main reason why this has been the case: most FE meshers are still serial and human intervention is generally required. In this paper we present the design of a system that tackles both these issues. More specifically, this paper proposes a system that, in combination with an unmodified off-the-shelf serial meshing program and an off-the-shelf CAD kernel, results in a fast and scalable tool capable of meshing complex CAD models, such as the ones used in industry, with reduced user intervention. To achieve scalability, our system uses two levels of parallelism: assembly level parallelism - across the multiple parts found in an assembly-type CAD model, and part level parallelism - obtained by partitioning individual CAD solids in multiple sections at the CAD level. We show preliminary results for the parallel meshing of a complex laptop model via which we highlight both some of the achieved benefits and the main challenges that need to be addressed in order to obtain good scalability.

Automatically Tuned Mixed-Precision Conjugate Gradient Solver

Linear algebra solvers such as the conjugate gradients method are at the base of numerous scientific and industrial applications. Among these, Krylov type iterative solvers are highly memory bounded, meaning that their bottleneck lies in the transfer of data from memory. The tremendous computational power current multicore processors and hardware accelerators, such as the graphic processing unit (GPU), are capable of, puts a great strain on interconnects, both external (e.g., networks) and internal (e.g., memory and PCIe buses). By using a lower precision for most of the computations, the technique of iterative refinement enables one to reduce the amount of memory being transferred at all levels, thus increasing computational performance. Moreover, it makes possible the use of more cost-effective accelerators, such as cheap GPUs, that lack support for native double precision, for tasks requiring double precision accuracy. In this chapter, we propose two heuristics for automatically setting the two parameters needed for using iterative refinement: the inner residual reduction target and the stopping criteria. Although the heuristics prove effective for most matrices from our test collection, we find many cases in which the increase in iterations due to the restarting of the solver leads to an actual decrease in performance.