<FromGitter>
<proyb6> But there is some errors request which seem to me, is it better to replace with the newer TFB status as some devs would assume that errors is a flaw or problem for which they aren't aware of the database connection was only rectified recently
<FromGitter>
<proyb6> Not a good way to market Amber when the results wasn't su
<FromGitter>
<proyb6> ...exactly what seem to be
snsei has quit [Ping timeout: 250 seconds]
<FromGitter>
<proyb6> My past experience with Vapor in Swift have request errors and I dropped that web framework for that reason
<FromGitter>
<vlazar> These are Round 17 results and there is no Round 18 yet. The results between rounds are intermediate. There might be errors in reports like banner on top show about one specific run here https://tfb-status.techempower.com
<FromGitter>
<vlazar> What's included in Round N is more trustworthy. It's reviewed by TechEmpower's guys.
<FromGitter>
<vlazar> Sure, the current results look much better :) https://www.techempower.com/benchmarks/#section=test&runid=2711d6a5-fbc0-4989-8606-8c532da9b088&hw=ph&test=fortune&b=4&d=9&o=e&f=zfyy1j-ziiiv3-suqynz-yelncf-v241rz-y9bz6n-h6zji7-ziimf1-18l4v
<FromGitter>
<vlazar> But I believe all Crystal implementations need some more DB pool tweaking and optimizations. The throughput looks good, and that's what people see first, but I think latency is probably too high. I think most people would probably care more about good latency, not highest possible throughput unless you are optimizing for throughput of course.
<FromGitter>
<proyb6> Yeah, we need to get pipeling for Crystal-pg to get more throughput but I still the author might have issue on how to get the pipeling going with the implementation. Who knows could contribute is welcome
<FromGitter>
<vlazar> I don't, but I'll keep tweaking TFB tests for more useful data.