However, it is commonly neglected to assess the additional integration costs for the entire system. The common part of all these works is the focus on improving compression performance, albeit for various reasons. Most published works about compression utilizing accelerators are focusing on the compression performance, including work on GPU , addressing SIMD in general , and FPGAs . Of particular interest in this context is the recent introduction of dedicated accelerators for compression tasks, including accelerators provided by AHA , Intel QuickAssist or Microsoft's Project Corsica . Overall run time benefits should be possible, even though this might substantially increase the amount of computations. As a consequence, especially for big data applications, it becomes more and more promising to trade computations for communication in order to diminish the implications of data movements. Furthermore, the amount of data generated and processed, preferably as fast and interactive as possible, is growing dramatically. The current computational landscape is dominated by increasingly costly data movements, both in terms of energy and time, while computations continuously decrease in these costs . Considering that compression is only a single task of a larger data processing pipeline, this overhead cannot be neglected. These results suggest that, given the right orchestration of compression and data movement tasks, the overhead of offloading compression is limited but present. The results imply that on average the zlib implementation on the accelerator achieves a comparable compression ratio to zlib level 2 on a CPU, while having up to 17 times the throughput and utilizing over 80 % less CPU resources. Among others, High Energy Physics data are used as a prime example of big data sources. This work evaluates the integration costs compared to a solely software-based solution considering multiple compression algorithms. In particular, such offloading is most beneficial for overlap with other tasks, if the associated overhead on the main processor is negligible. Yet, one lacks the understanding of the overhead of compression when offloading tasks. Most recently, a couple of accelerators have been introduced to offload compression tasks from the main processor, for instance by AHA, Intel and Microsoft.
![compressor 4 setup for a 3.4 ghz intel core i7 compressor 4 setup for a 3.4 ghz intel core i7](https://overclock3d.net/gfx/articles/2018/11/29085624972l.jpg)
For these reasons there is already a plethora of related works on compression from various domains. Second, performance is determined by energy efficiency, and the overall power consumption is dominated by the consumption of data movements.
![compressor 4 setup for a 3.4 ghz intel core i7 compressor 4 setup for a 3.4 ghz intel core i7](https://i5.walmartimages.com/asr/0356ee2a-c4a2-4cf4-8429-bc903158cf41_1.ebd9c373b4a9ac4b029d7dd8bdecffe2.jpeg)
![compressor 4 setup for a 3.4 ghz intel core i7 compressor 4 setup for a 3.4 ghz intel core i7](https://i.ebayimg.com/images/g/~2UAAOSwkfRd~s6K/s-l960.jpg)
There are two main reasons: First, the gap between processing speed and I/O continues to grow, and technology trends indicate a continuation of this. Exploring compression is increasingly promising as trade-off between computations and data movement.