Reevaluating Amdahl's law

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Optimal Distributed Online Prediction Using Mini-Batches (2010, methodology): “Therefore, we obtain an asymptotically linear speed-up, which is the ideal result that one would hope for in parallelizing the optimization process (see Gustafson, 1988).”
The workload on parallel supercomputers: modeling the characteristics of rigid jobs (2003, background): “Alternative scalability models assert that the added parallelism is not used to solve the same problem faster, but rather to solve larger problems [9,22].”
Sourcebook of parallel computing (2003, background): “This leads us to consider a second definition of scalability, called scaled speedup— an application will be said to be scalable if, when the number of processors and the problem size are increased by a factor of n, the running time remains the same [418].”
Fast ℓ1-SPIRiT Compressed Sensing Parallel Imaging MRI: Scalable Parallel Implementation and Clinically Feasible Runtime (2012, background): “Per Gustafson’s law [44], efficient utilization of future architectures will require increased problem size.”
Amdahl's Law in the Multicore Era (2008, background): “On the other hand, more cores may advantageously enable greater parallelism from larger problem sizes, as envisioned by Gustafson [4].”