Check out the official site for more info.
Take home messages for the soon to be announced system
This system set new records for 10 of the 17 response-time benchmarks, including:
* 40% faster than the next best result in the high-bid search (STAC-M3.?1.1T.YRHIBID.LAT2)
* 30% faster than the next best result in the daily VWAB test (STAC-M3.?1.1T.VWAB-D.LAT2)
* 20% faster than the next best result in volume-curves test (STAC-M3.?1.1T.VOLCURV.LAT2)
This system set two new records in bytes read per second (STAC-M3.?1.1T.YRHIBID.BPS, STAC-M3.?1.1T.QTRHIBID.BPS)
Thus, the extent to which this SUT with 16 cores outperformed the next best SUT (KDB130603), which had 32 cores, declined with the number of requests. This SUT outperformed the higher core system by approximately 60% when limited to a single request, 50% with 10 requests, 20% with 50 requests, and at 100 requests was 10% slower than the system with double the cores (but only 10%!).
What is remarkable about this architecture is that despite these variances with previous configurations tested under STAC-M3, the Scalable Informatics Time Series Analytics appliance provided some of the fastest observed performance with the smallest consumption of key resources. Architecture and implementation matter in high performance use cases.
What is this magical beast you ask? What are its configuration limits? You’ll have to wait for the official unveiling.