Alternatives
Products that do what I audited 500 K8s pods. Java wastes ~48% RAM, Go ~18% does
- 1SI
2021 · github.com
- 2HA
2017 · hasura.io
- 3KA
2017 · github.com
- 4MM
2015 · github.com
- 5AE
2018 · github.com
- 6IW
2023 · github.com
- 7KD
2017 · k6.io
- 8KA
2016 · github.com
- 9RG
Hi HN! We’re Yann, Edouard, and Bastien from Koyeb (https://www.koyeb.com/). We’re building a platform to let you deploy full-stack apps on high-performance hardware around the world, with zero configuration. We provide a “global serverless feeling”, without the hassle of re-writing all your apps or managing k8s complexity [1]. We built Scaleway, a cloud service provider where we designed ARM servers and provided them as cloud servers. During our time there, we saw customers struggle with the same issues while trying to deploy full-stack applications and APIs resiliently. As…
2023 · koyeb.com
- 10BI
2020 · blunders.io
- 11VD
2019 · valence.net
- 12KC
2021 · github.com
- 13MT
2019 · github.com
- 14SM
2020 · github.com
- 156K
2017 · github.com
- 16KB
2019 · kubestone.io
- 17AJ
2017 · github.com
- 18MW
2017 · github.com
- 19GL
2019 · github.com
- 20KF
2020 · github.com
- 21MW
2016 · github.com
- 22GG
2015 · gogitit.co
- 23AN
Kimi K3 has 2.78 trillion parameters and ships as 1.42 TB of weights. It clearly does not fit in the memory of a laptop. But K3 is a Mixture-of-Experts model. For each token, only a small fraction of its 896 experts per layer is activated. That changes the problem: the entire model does not need to be resident in RAM, as long as the weights required by each token can be reached quickly enough. We built WASTE — the Weight-Aware Streaming Tensor Engine — to explore that idea. WASTE keeps the dense, repeatedly used part of the model resident in memory, stores the routed experts in an…
Jul 2026
- 24TK
2021 · github.com
Ranked by how close each launch is in meaning, then by votes. Refine with a description →