Memory reference reuse latency: Accelerated warmup for sampled microarchitecture simulation

Abstract

Abstract-This paper proposes to speedup sampled microprocessor simulations by reducing warmup times without sacrificing simulation accuracy. It exploiting the observation that of the memory references that precede a sample cluster, references that occur nearest to the cluster are more likely to be germane to the execution of the cluster itself. Hence, while modeling all cache and branch predictor interactions that precede a sample cluster would reliably establish their state, this is overkill and leads to longrunning simulations. Instead, accurately establishing simulated cache and branch predictor state can be accomplished quickly by only modeling a subset of the memory references and controlflow instructions immediately preceding a sample cluster.Our technique measures memory reference reuse latencies (MRRLs)-the number of completed instructions between consecutive references to each unique memory location-and uses these data to choose a point prior to each cluster to engage cache hierarchy and branch predictor modeling. By starting cache and branch predictor modeling late in the pre-cluster instruction stream, we were able to reduce overall simulation running times by an average of 90.62% of the maximum potential speedup (accomplished by performing no pre-cluster warmup at all), while generating an average error in IPC of less than 1%, both relative to the IPC generated by warming up all pre-cluster cache and branch predictor interactions.

Accelerating Multiprocessor Simulation with a Memory Timestamp Record

“…Memory Reference Reuse Latency (MRRL) can be used to bound the amount of "detailed warming" prior to a sample to achieve a desired accuracy [8]. In a uniprocessor model, MRRL can remove 90% of the warmup costs, but as structures are larger and duplicated in a multiprocessor, the remaining warmup time can still be significant.…”

Characterizing and predicting program behavior and its variability

“…Prior work can be categorized into methods that define and use metrics to dynamically identify phases for adaptive optimization [5,6,8,13,26], and into techniques that identify appropriate simulation points for the desired workloads [10,16,17,24,25].…”

Budgeted region sampling (BeeRS): do not separate sampling from warm-up, and then spend wisely your simulation budget

“…Several recent research works [3,4,8] propose various techniques for reducing the warm-up size before samples while maintaining a high accuracy, but they are all separate from the sampling techniques themselves. In sampling research works, warm-up has been treated in two different ways: either using functional [14,7] or checkpoint-based warm-up [13], or assuming warm-up is perfect based on the principle of separating warm-up and sampling issues [12,10,6].…”