From 991cbfcb1459f3adce13d70c383f9c4c6fd4ee96 Mon Sep 17 00:00:00 2001 From: =?utf8?q?Marek=20Ol=C5=A1=C3=A1k?= Date: Fri, 3 Jun 2016 16:20:17 +0200 Subject: [PATCH] radeonsi: add a performance tweak for 4 SE parts MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit Ported from Vulkan. Reviewed-by: Nicolai Hähnle --- src/gallium/drivers/radeonsi/si_state_draw.c | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/src/gallium/drivers/radeonsi/si_state_draw.c b/src/gallium/drivers/radeonsi/si_state_draw.c index aed509de9bb..0a85d5a0ca3 100644 --- a/src/gallium/drivers/radeonsi/si_state_draw.c +++ b/src/gallium/drivers/radeonsi/si_state_draw.c @@ -308,6 +308,17 @@ static unsigned si_get_ia_multi_vgt_param(struct si_context *sctx, (info->indirect || info->instance_count > 1)) wd_switch_on_eop = true; + /* Performance recommendation for 4 SE Gfx7-8 parts if + * instances are smaller than a primgroup. Ignore the fact + * primgroup_size is a primitive count, not vertex count. + * Don't do anything for indirect draws. + */ + if (sctx->b.chip_class <= VI && + sctx->b.screen->info.max_se >= 4 && + !info->indirect && + info->instance_count > 1 && info->count < primgroup_size) + wd_switch_on_eop = true; + /* Required on CIK and later. */ if (sctx->b.screen->info.max_se > 2 && !wd_switch_on_eop) ia_switch_on_eoi = true; -- 2.30.2