https://exafunction.com/blog Efficient deep learning at scale en-us hello@exafunction.com (Exafunction Team) hello@exafunction.com (Exafunction Team) Mon, 10 Oct 2022 07:00:00 GMT https://exafunction.com/blog/motive-case-study https://exafunction.com/blog/motive-case-study How ExaDeploy has helped Motive to save money and accelerate their development. Mon, 10 Oct 2022 07:00:00 GMT hello@exafunction.com (Exafunction Team) product https://exafunction.com/blog/60-berts-1-gpu https://exafunction.com/blog/60-berts-1-gpu Using partial runners to load 60 BERTs onto a T4 with no prespecified colocation. Mon, 26 Sep 2022 07:00:00 GMT hello@exafunction.com (Exafunction Team) technical https://exafunction.com/blog/tensorflow-gpu-inputs-outputs https://exafunction.com/blog/tensorflow-gpu-inputs-outputs Passing inputs and outputs directly through GPU memory in TensorFlow. Fri, 16 Sep 2022 07:00:00 GMT hello@exafunction.com (Exafunction Team) technical https://exafunction.com/blog/are-gpus-worth-it https://exafunction.com/blog/are-gpus-worth-it How to get the best cost and latency on ML workloads from a mix of GPUs and CPUs. Fri, 26 Aug 2022 07:00:00 GMT hello@exafunction.com (Exafunction Team) product https://exafunction.com/blog/ml-deployment-problems https://exafunction.com/blog/ml-deployment-problems Dissecting the issues that pop up when trying to deploy an ML workload. Wed, 17 Aug 2022 07:00:00 GMT hello@exafunction.com (Exafunction Team) product https://exafunction.com/blog/ecc-gpu https://exafunction.com/blog/ecc-gpu Many cloud GPUs are configured by default in a way that reduces the total amount of GPU memory available. Thu, 28 Jul 2022 07:00:00 GMT hello@exafunction.com (Exafunction Team) technical https://exafunction.com/blog/series-a-announcement https://exafunction.com/blog/series-a-announcement We are excited to announce our $25 million Series A financing, led by Greenoaks with participation from Founders Fund. Thu, 28 Apr 2022 07:00:00 GMT hello@exafunction.com (Exafunction Team) funding