From: Joseph Greathouse <Joseph.Greathouse@amd.com>
To: amd-gfx@lists.freedesktop.org
Cc: Joseph Greathouse <Joseph.Greathouse@amd.com>
Subject: [PATCH] drm/amdkfd: Update hardware scheduling time quanta
Date: Mon, 29 Jun 2020 17:00:44 -0500 [thread overview]
Message-ID: <20200629220044.179230-1-Joseph.Greathouse@amd.com> (raw)
Update PROCESS_QUANTUM, the time the hardware scheduler allows
processes to run before switching to other processes when it becomes
over-subscribed. Increase this to 10ms, to allow processes to better
amortize their task switch times.
Update HQD Quantum, the amount of time that an active queue stays
attached to the CP before we forcibly switch it for another active
queue for fairness.
Setting these so that HQD < PROCESS makes it easier to ensure that
we get fairness when we have multiple active queues on the device.
Otherwise we may start process-swapping before we get to all the
queues in a CP.
Signed-off-by: Joseph Greathouse <Joseph.Greathouse@amd.com>
Change-Id: I5f46d268a82eb08e75bfaf0aed5333c3341b64bd
---
drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c | 2 +-
drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c | 2 +-
drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c | 2 +-
drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c | 2 +-
drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c | 2 +-
5 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c
index d1d68a51bfb8..18e08d82d978 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v10.c
@@ -113,7 +113,7 @@ static void init_mqd(struct mqd_manager *mm, void **mqd,
m->cp_hqd_quantum = 1 << CP_HQD_QUANTUM__QUANTUM_EN__SHIFT |
1 << CP_HQD_QUANTUM__QUANTUM_SCALE__SHIFT |
- 10 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
+ 1 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
if (q->format == KFD_QUEUE_FORMAT_AQL) {
m->cp_hqd_aql_control =
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
index 48cda3073b70..3b6f5963180d 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_v9.c
@@ -160,7 +160,7 @@ static void init_mqd(struct mqd_manager *mm, void **mqd,
m->cp_hqd_quantum = 1 << CP_HQD_QUANTUM__QUANTUM_EN__SHIFT |
1 << CP_HQD_QUANTUM__QUANTUM_SCALE__SHIFT |
- 10 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
+ 1 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
if (q->format == KFD_QUEUE_FORMAT_AQL) {
m->cp_hqd_aql_control =
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
index a5e8ff1e5945..31799e5f3b3c 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager_vi.c
@@ -117,7 +117,7 @@ static void init_mqd(struct mqd_manager *mm, void **mqd,
m->cp_hqd_quantum = 1 << CP_HQD_QUANTUM__QUANTUM_EN__SHIFT |
1 << CP_HQD_QUANTUM__QUANTUM_SCALE__SHIFT |
- 10 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
+ 1 << CP_HQD_QUANTUM__QUANTUM_DURATION__SHIFT;
set_priority(m, q);
m->cp_hqd_eop_rptr = 1 << CP_HQD_EOP_RPTR__INIT_FETCHER__SHIFT;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c
index bdca9dc5f118..dfaf771a42e6 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c
@@ -39,7 +39,7 @@ static int pm_map_process_v9(struct packet_manager *pm,
packet->header.u32All = pm_build_pm4_header(IT_MAP_PROCESS,
sizeof(struct pm4_mes_map_process));
packet->bitfields2.diq_enable = (qpd->is_debug) ? 1 : 0;
- packet->bitfields2.process_quantum = 1;
+ packet->bitfields2.process_quantum = 10;
packet->bitfields2.pasid = qpd->pqm->process->pasid;
packet->bitfields14.gds_size = qpd->gds_size & 0x3F;
packet->bitfields14.gds_size_hi = (qpd->gds_size >> 6) & 0xF;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c
index bed4d0ccb6b1..a852e0d7d804 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c
@@ -50,7 +50,7 @@ static int pm_map_process_vi(struct packet_manager *pm, uint32_t *buffer,
packet->header.u32All = pm_build_pm4_header(IT_MAP_PROCESS,
sizeof(struct pm4_mes_map_process));
packet->bitfields2.diq_enable = (qpd->is_debug) ? 1 : 0;
- packet->bitfields2.process_quantum = 1;
+ packet->bitfields2.process_quantum = 10;
packet->bitfields2.pasid = qpd->pqm->process->pasid;
packet->bitfields3.page_table_base = qpd->page_table_base;
packet->bitfields10.gds_size = qpd->gds_size;
--
2.20.1
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
next reply other threads:[~2020-06-29 22:01 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-29 22:00 Joseph Greathouse [this message]
2020-06-29 22:29 ` [PATCH] drm/amdkfd: Update hardware scheduling time quanta Felix Kuehling
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200629220044.179230-1-Joseph.Greathouse@amd.com \
--to=joseph.greathouse@amd.com \
--cc=amd-gfx@lists.freedesktop.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).