According to the tests conducted by engineers at Sandia National Laboratories, New Mexico, who have simulated future high-performance computers containing the 8-core, 16‑core, and 32-core microprocessors, it appears that after 8 cores put to process the more cores added the worse applications perform especially bad for data-intensive apps.

As the chart below shows, while keep adding more cores to the tests the memory bandwidth remains flat, and that is the cause of the trouble.

More cores per chip will slow some programs (red) unless there’s a big boost in memory bandwidth (yellow). Credit: Sandia

The problem, called “Memory Wall”, is not new referring to the growing gap between how fast a CPU can operate on data and how fast it can get the data it needs.

Recently, finding no other way to further improve the performance of processors, chip makers manage to put more and more processor cores on the same chip. However, while the number of cores per processor is increasing, the number of connections from the chip to the rest of the computer is not.

“The key to solving this bottleneck is tighter, and maybe smarter, integration of memory and processors,” says James Peery, director of computation and mathematics at Sandia.

Sandia itself is exploring a solution for the problem by stacking memory chips atop processors to improve memory bandwidth, according to an IEEE Spectrum article.