@@ -21,7 +21,7 @@ Core frequency capping is implemented for the Karolina supercomputer:
...
@@ -21,7 +21,7 @@ Core frequency capping is implemented for the Karolina supercomputer:
The performance impact depends on the [arithmetic intensity][1] of the job.
The performance impact depends on the [arithmetic intensity][1] of the job.
The [arithmetic intensity][2] is a measure of floating-point operations (FLOPs) performed by a given code (or code section) relative to the amount of memory accesses (Bytes) that are required to support those operations. It is defined as a FLOP per Byte ratio (F/B). It is a characteristics of the particular computational algorithm.
The [arithmetic intensity][2] is a measure of floating-point operations (FLOPs) performed by a given code (or code section) relative to the amount of memory accesses (Bytes) that are required to support those operations. It is defined as a FLOP per Byte ratio (F/B). It is a characteristics of the particular computational algorithm.
In general, the processor frequency capping has low performance impact for memory bound computations with intensity below the [ridge point][2]. For CPU bound computations with intensity above the [ridge point][2], the impact is proportional to the frequency reduction.
In general, the processor frequency capping has low performance impact for memory bound computations with intensity below the [ridge point][2]. For processor bound computations with intensity above the [ridge point][2], the impact is proportional to the frequency reduction.
On Karolina, relative time increase factor **up to 1.16** is observed for arithmeticaly intensive workloads on CPU and **up to 1.1** on GPU. **No slowdown** is observed for memory bound workloads.
On Karolina, relative time increase factor **up to 1.16** is observed for arithmeticaly intensive workloads on CPU and **up to 1.1** on GPU. **No slowdown** is observed for memory bound workloads.
