Hi John, tested on real PVC and the estimated min size needed was ~115K. Even without modelling, we can safely say that an imaginary device with a tile that is 4x bigger would still be half than 1 MB. That said I shall proceed with a re-rev that shall include dropping the size of guc-log-error-capture-region down to 1MB in addition to sticking with the "drm_warn(...mayb too small)" and ofc rebasing to latest drm-tip. Shall rerun some tests just in case - so will get it out later this week. ...alan On Mon, 2022-10-03 at 18:51 -0700, Harrison, John C wrote: > On 10/3/2022 17:46, Teres Alexis, Alan Previn wrote: > > So as per the last response and the offline conversation we had we agreed that: > > > > 1. we shall stick with drm_warn( ... maybe too small...) if the allocation didn't meet min_size. > > 2. I'll model for PVC (since its better to look at the spec as opposed to trying to hunt for a free machine with the > > most engines and DSS (for those steering registers that are counted multiple times). > > 3. If #2 yields us with substantial headroom (i.e. a model's PVC would be less than 700K min_size), then lets drop to > > 1MB allocation. > > > > ...alan > Sounds good to me. > > John. > > > > > > > On Mon, 2022-10-03 at 16:51 -0700, Harrison, John C wrote: > > > On 10/3/2022 14:10, Teres Alexis, Alan Previn wrote: > > > > On Mon, 2022-10-03 at 12:47 -0700, Harrison, John C wrote: > > > > > On 10/3/2022 11:28, Teres Alexis, Alan Previn wrote: > > > > > > On Fri, 2022-09-30 at 15:35 -0700, Harrison, John C wrote: > > > > > > > On 9/30/2022 14:08, Teres Alexis, Alan Previn wrote: > > > > > > > >
