- From: Ilya Grigorik <igrigorik@google.com>
- Date: Thu, 25 Jun 2015 12:14:43 +0100
- To: Eli Perelman <eperelman@mozilla.com>
- Cc: public-web-perf <public-web-perf@w3.org>
- Message-ID: <CADXXVKonczA+UKX7aZ0HfjYOO0M6GELnCxrF9OPfiq5Y-qSBHQ@mail.gmail.com>
Hi Eli. Thanks for kicking this off, this is topic we've been exploring on Chrome side as well. FWIW, I've been collecting use cases and notes from various memory-related conversations in this doc: https://docs.google.com/document/d/1tFCEOMOUg4zmqeHNg1Xo11Xpdm7Bmxl5y98_ESLCLgM/edit Would be curious to hear your thoughts on above. ig On Wed, Jun 24, 2015 at 7:32 PM, Eli Perelman <eperelman@mozilla.com> wrote: > In working to find holes in our performance testing in Firefox OS, I > currently see a lack of functionality around the memory space. I'd like to > bring up a couple use cases and see where things go from there. > > 1. Trigger moment to capture memory information. > > Firefox OS uses a custom command-line tool on-device to capture memory > information about a particular web application. We trigger this tool upon > receiving a certain User Timing marker. It would be great if there was a > specific API for capturing the memory at a specific moment in time. > > performance.memory.mark()? > > 2. Fetch peak memory usage > > Another useful API for us would be to know metrics about what the peak > memory usage was for a particular web application up until the method call. > > performance.memory.peak()? > > 3. Receive better memory information > > In our command-line tool we capture USS, PSS, and RSS of a web > application. I see that Chrome has a proprietary performance.memory API > that captures JS-specific memory information, but expanding this to capture > other memory details would be useful if it is supported on the platform. > > Thoughts? > > Thanks! > > Eli Perelman > Mozilla >
Received on Thursday, 25 June 2015 11:15:50 UTC