3 ms·
Yes, it's currently used by two of our customers for grid computing applications. Development started in late 2009 but there was a couple of "let's redo this p
by shin_lao 16y ago
Yes, it's currently used by two of our customers for grid computing applications.
Development started in late 2009 but there was a couple of "let's redo this part" and "ho, wait, it's never going to work!".
- eengstrom 16y agoHave you had a chance to measure operational throughput vs hardware utilization yet? I'd be very interested in loose metrics and stats instead of a general "faster than" statement.
- shin_lao 16y agoWe manage to saturate the memory bandwidth. We can do 3,000,000 get / sec on a Core i7 machine with entries of roughly 1 kib in size. Having benchmarks is our #1 priority
- eengstrom 16y agoSorry for my slow response; how does the process handle scaling above 1kb? Traditionally there's a predictable curve of expected performance as the scale of request size increases. Happy to take this conversation private; spent the last 10 years in performance and scalability as a indie consultant for Fortune companies. I know a lot of people looking for replacement mechanisms for high volume, variable sized transaction enterprise buses. Also interested to understand the data access methods you've tested and what works best with your code. eengstrom >at< gmail.