From: Lionel Landwerlin Date: Thu, 12 Dec 2019 10:14:09 +0000 (+0200) Subject: intel/perf: drop batchbuffer flushing at query begin X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=a575b3cd5c1e61a7e92fa2521ced95d24b64f392;p=mesa.git intel/perf: drop batchbuffer flushing at query begin This was initially intended to fix issues with the query timings going occassionally high. It turns out there was a bug in the attribution of OA reports to our context when parsing the OA data. This led to reports flagged with other context IDs to be included in our queries results. Signed-off-by: Lionel Landwerlin Reviewed-by: Kenneth Graunke --- diff --git a/src/intel/perf/gen_perf.c b/src/intel/perf/gen_perf.c index 4c47aa6dcf2..daa092c88c9 100644 --- a/src/intel/perf/gen_perf.c +++ b/src/intel/perf/gen_perf.c @@ -1851,14 +1851,6 @@ gen_perf_begin_query(struct gen_perf_context *perf_ctx, query->oa.begin_report_id = perf_ctx->next_query_start_report_id; perf_ctx->next_query_start_report_id += 2; - /* We flush the batchbuffer here to minimize the chances that MI_RPC - * delimiting commands end up in different batchbuffers. If that's the - * case, the measurement will include the time it takes for the kernel - * scheduler to load a new request into the hardware. This is manifested in - * tools like frameretrace by spikes in the "GPU Core Clocks" counter. - */ - perf_cfg->vtbl.batchbuffer_flush(perf_ctx->ctx, __FILE__, __LINE__); - /* Take a starting OA counter snapshot. */ perf_cfg->vtbl.emit_mi_report_perf_count(perf_ctx->ctx, query->oa.bo, 0, query->oa.begin_report_id);