IAA Compression Batching:
=========================
This patch-series introduces the use of the Intel Analytics Accelerator
(IAA) for parallel batch compression of pages in large folios to improve
zswap swapout latency, resulting in sys time reduction by 41% (usemem 30
processes) and by 24% (kernel compilation); as well as a 39% increase in
usemem30 throughput with IAA batching as compared to zstd.
The patch-series is organized as follows:
1) crypto acomp & iaa_crypto driver enablers for batching: Relevant
patches are tagged with "crypto:" in the subject:
Patch 1) Adds acomp_alg/crypto_acomp batch_compress() and
batch_decompress() interfaces, that swap modules can invoke
using the new batching API crypto_acomp_batch_compress() and
crypto_acomp_batch_decompress(). Additionally, crypto acomp
provides a new acomp_has_async_batching() interface to query
for these API before allocating batching resources for a given
compressor in zswap/zram.
Patch 2) New CRYPTO_ACOMP_REQ_POLL acomp_req flag to act as a gate for
async poll mode in iaa_crypto.
Patch 3) iaa-crypto driver implementations for async polling,
crypto_acomp_batch_compress() and crypto_acomp_batch_decompress().
The "iaa_acomp_fixed_deflate" algorithm registers these
implementations for its batch_compress and batch_decompress
interfaces respectively.
Patch 4) Modifies the default iaa_crypto driver mode to async.
Patch 5) Disables verify_compress by default, to facilitate users to
run IAA easily for comparison with software compressors.
Patch 6) Reorganizes the iaa_crypto driver code into logically related
sections and avoids forward declarations, in order to facilitate
Patch 7. This patch makes no functional changes.
Patch 7) Makes a major infrastructure change in the iaa_crypto driver,
to map IAA devices/work-queues to cores based on packages
instead of NUMA nodes. This doesn't impact performance on
the Sapphire Rapids system used for performance
testing. However, this change fixes problems found on Granite
Rapids in internal validation, where the number of NUMA nodes
is greater than the number of packages, which was resulting in
over-utilization of some IAA devices and non-usage of other
IAA devices as per the current NUMA based mapping
infrastructure. This patch also eliminates duplication of device
wqs in per-cpu wq_tables, thereby saving 140MiB on a 384 cores
Granite Rapids server with 8 IAAs. Submitting this change now
so that it can go through code reviews before it can be merged.
Patch 8) Builds upon the new infrastructure for mapping IAAs to cores
based on packages, and enables configuring a "global_wq" per
IAA, which can be used as a global resource for compress jobs
for the package. If the user configures 2WQs per IAA device,
the driver will distribute compress jobs from all cores on the
package to the "global_wqs" of all the IAA devices on that
package, in a round-robin manner. This can be used to improve
compression throughput for workloads that see a lot of swapout
activity.
2) zswap modifications to enable compress batching in zswap_batch_store()
of large folios (including pmd-mappable folios):
Patch 9) Changes the "struct crypto_acomp_ctx" to contain a configurable
number of acomp_reqs and buffers. Subsequently, the cpu
hotplug onlining code will query acomp_has_async_batching() to
allocate up to SWAP_CRYPTO_BATCH_SIZE (i.e. 8)
acomp_reqs/buffers if the acomp supports batching, and 1
acomp_req/buffer if not.
Patch 10) zswap_batch_store() IAA compress batching implementation
using the new crypto_acomp_batch_compress() iaa_crypto driver
API. swap_writepage() will call zswap_batch_store() for large
folios if zswap_can_batch().
With v4 of this patch series, the IAA compress batching feature will be
enabled seamlessly on Intel platforms that have IAA by selecting
'deflate-iaa' as the zswap compressor.
System setup for testing:
=========================
Testing of this patch-series was done with mm-unstable as of 11-18-2024,
commit 5a7056135bb6, without and with this patch-series.
Data was gathered on an Intel Sapphire Rapids server, dual-socket 56 cores
per socket, 4 IAA devices per socket, 503 GiB RAM and 525G SSD disk
partition swap. Core frequency was fixed at 2500MHz.
Other kernel configuration parameters:
zswap compressor : zstd, deflate-iaa
zswap allocator : zsmalloc
vm.page-cluster : 0, 2
IAA "compression verification" is disabled and IAA is run in the async
poll mode (the defaults with this series). 2WQs are configured per IAA
device. Compress jobs from all cores on a socket are distributed among all
4 IAA devices on the same socket.
I ran experiments with these workloads:
1) usemem 30 processes with these large folios enabled to "always":
- 16k/32k/64k
- 2048k
2) Kernel compilation allmodconfig with 2G max memory, 32 threads, run in
tmpfs with these large folios enabled to "always":
- 16k/32k/64k
Performance testing (usemem30):
===============================
The vm-scalability "usemem" test was run in a cgroup whose memory.high
was fixed at 150G. The is no swap limit set for the cgroup. 30 usemem
processes were run, each allocating and writing 10G of memory, and sleeping
for 10 sec before exiting:
usemem --init-time -w -O -s 10 -n 30 10g
16k/32/64k folios: usemem30:
============================
-------------------------------------------------------------------------------
mm-unstable-11-18-2024 v4 of this patch-series
-------------------------------------------------------------------------------
zswap compressor zstd deflate-iaa deflate-iaa IAA Batching
vm.page-cluster 2 2 2 vs. vs.
Seq zstd
-------------------------------------------------------------------------------
Total throughput (KB/s) 6,284,634 7,149,906 8,392,830 17% 39%
Avg throughput (KB/s) 209,487 238,330 279,761 17% 39%
elapsed time (sec) 107.64 84.38 79.88 -5% -29%
sys time (sec) 2,566.69 1,844.32 1,592.02 -14% -41%
-------------------------------------------------------------------------------
memcg_high 477,219 616,897 683,170
memcg_swap_fail 1,040 2,734 2,330
zswpout 48,931,670 55,520,017 57,467,487
zswpin 384 491 415
pswpout 0 0 0
pswpin 0 0 0
thp_swpout 0 0 0
thp_swpout_fallback 0 0 0
16kB-swpout_fallback 0 0 0
32kB_swpout_fallback 0 0 0
64kB_swpout_fallback 1,040 2,734 2,330
pgmajfault 3,258 3,314 3,251
swap_ra 95 128 112
swap_ra_hit 46 49 61
ZSWPOUT-16kB 2 4 3
ZSWPOUT-32kB 0 2 0
ZSWPOUT-64kB 3,057,203 3,467,400 3,589,487
SWPOUT-16kB 0 0 0
SWPOUT-32kB 0 0 0
SWPOUT-64kB 0 0 0
-------------------------------------------------------------------------------
2M folios: usemem30:
====================
-------------------------------------------------------------------------------
mm-unstable-11-18-2024 v4 of this patch-series
-------------------------------------------------------------------------------
zswap compressor zstd deflate-iaa deflate-iaa IAA Batching
vm.page-cluster 2 2 2 vs. vs.
Seq zstd
-------------------------------------------------------------------------------
Total throughput (KB/s) 6,466,700 7,245,936 9,107,731 26% 39%
Avg throughput (KB/s) 215,556 241,531 303,591 26% 39%
elapsed time (sec) 106.80 84.44 74.37 -12% -30%
sys time (sec) 2,420.88 1,753.41 1,450.21 -17% -41%
-------------------------------------------------------------------------------
memcg_high 60,926 79,259 90,314
memcg_swap_fail 44 139 182
zswpout 48,892,828 57,701,156 59,051,023
zswpin 391 419 411
pswpout 0 0 0
pswpin 0 0 0
thp_swpout 0 0 0
thp_swpout_fallback 44 139 182
pgmajfault 4,907 11,542 30,492
swap_ra 5,070 24,613 80,933
swap_ra_hit 5,024 24,555 80,856
ZSWPOUT-2048kB 95,442 112,515 114,996
SWPOUT-2048kB 0 0 0
-------------------------------------------------------------------------------
Performance testing (Kernel compilation, allmodconfig):
=======================================================
The experiments with kernel compilation test, 32 threads, in tmpfs use the
"allmodconfig" that takes ~12 minutes, and has considerable swapout
activity. The cgroup's memory.max is set to 2G.
16k/32k/64k folios: Kernel compilation/allmodconfig:
====================================================
-------------------------------------------------------------------------------
mm-unstable-11-18-2024 v4 of this patch-series
-------------------------------------------------------------------------------
zswap compressor zstd deflate-iaa deflate-iaa IAA Batching
vm.page-cluster 0 0 0 vs. vs.
Seq zstd
-------------------------------------------------------------------------------
real_sec 783.15 792.78 789.65
user_sec 15,763.86 15,779.60 15,775.48
sys_sec 5,198.29 4,215.74 3,930.92 -7% -24%
-------------------------------------------------------------------------------
Max_Res_Set_Size_KB 1,872,932 1,873,444 1,872,896
-------------------------------------------------------------------------------
memcg_high 0 0 0
memcg_swap_fail 0 0 0
zswpout 88,824,270 109,828,718 109,402,157
zswpin 25,371,781 32,647,096 32,174,520
pswpout 121 360 297
pswpin 122 337 288
thp_swpout 0 0 0
thp_swpout_fallback 0 0 0
16kB_swpout_fallback 0 0 0
32kB_swpout_fallback 0 0 0
64kB_swpout_fallback 924 19,203 5,206
pgmajfault 27,124,258 35,120,147 34,545,319
swap_ra 0 0 0
swap_ra_hit 2,561 3,131 2,380
ZSWPOUT-16kB 1,246,641 1,499,293 1,469,160
ZSWPOUT-32kB 675,242 865,310 827,968
ZSWPOUT-64kB 2,886,860 3,596,899 3,638,188
SWPOUT-16kB 0 0 0
SWPOUT-32kB 1 0 0
SWPOUT-64kB 7 19 18
-------------------------------------------------------------------------------
Summary:
========
The performance testing data with usemem 30 processes and kernel
compilation test show 39% throughput gains and 41% sys time reduction
(usemem30) and 24% sys time reduction (kernel compilation) with
zswap_batch_store() large folios using IAA compress batching as compared to
zstd.
The iaa_crypto wq stats will show almost the same number of compress calls
for wq.1 of all IAA devices. wq.0 will handle decompress calls exclusively.
We see a latency reduction of 2.5% by distributing compress jobs among all
IAA devices on the socket (based on v1 data).
We can expect to see even more significant performance and throughput
improvements if we use the parallelism offered by IAA to do reclaim
batching of 4K/large folios (really any-order folios), and using the
zswap_batch_store() high throughput compression to batch-compress pages
comprising these folios, not just batching within large folios. This is the
reclaim batching patch 13 in v1, which will be submitted in a separate
patch-series.
Our internal validation of IAA compress/decompress batching in highly
contended Sapphire Rapids server setups with workloads running on 72 cores
for ~25 minutes under stringent memory limit constraints have shown up to
50% reduction in sys time and 3.5% reduction in workload run time as
compared to software compressors.
Changes since v3:
=================
1) Rebased to mm-unstable as of 11-18-2024, commit 5a7056135bb6.
2) Major re-write of iaa_crypto driver's mapping of IAA devices to cores,
based on packages instead of NUMA nodes.
3) Added acomp_has_async_batching() API to crypto acomp, that allows
zswap/zram to query if a crypto_acomp has registered batch_compress and
batch_decompress interfaces.
4) Clear the poll bits on the acomp_reqs passed to
iaa_comp_a[de]compress_batch() so that a module like zswap can be
confident about the acomp_reqs[0] not having the poll bit set before
calling the fully synchronous API crypto_acomp_[de]compress().
Herbert, I would appreciate it if you can review changes 2-4; in patches
1-8 in v4. I did not want to introduce too many iaa_crypto changes in
v4, given that patch 7 is already making a major change. I plan to work
on incorporating the request chaining using the ahash interface in v5
(I need to understand the basic crypto ahash better). Thanks Herbert!
5) Incorporated Johannes' suggestion to not have a sysctl to enable
compress batching.
6) Incorporated Yosry's suggestion to allocate batching resources in the
cpu hotplug onlining code, since there is no longer a sysctl to control
batching. Thanks Yosry!
7) Incorporated Johannes' suggestions related to making the overall
sequence of events between zswap_store() and zswap_batch_store() similar
as much as possible for readability and control flow, better naming of
procedures, avoiding forward declarations, not inlining error path
procedures, deleting zswap internal details from zswap.h, etc. Thanks
Johannes, really appreciate the direction!
I have tried to explain the minimal future-proofing in terms of the
zswap_batch_store() signature and the definition of "struct
zswap_batch_store_sub_batch" in the comments for this struct. I hope the
new code explains the control flow a bit better.
Changes since v2:
=================
1) Rebased to mm-unstable as of 11-5-2024, commit 7994b7ea6ac8.
2) Fixed an issue in zswap_create_acomp_ctx() with checking for NULL
returned by kmalloc_node() for acomp_ctx->buffers and for
acomp_ctx->reqs.
3) Fixed a bug in zswap_pool_can_batch() for returning true if
pool->can_batch_comp is found to be equal to BATCH_COMP_ENABLED, and if
the per-cpu acomp_batch_ctx tests true for batching resources having
been allocated on this cpu. Also, changed from per_cpu_ptr() to
raw_cpu_ptr().
4) Incorporated the zswap_store_propagate_errors() compilation warning fix
suggested by Dan Carpenter. Thanks Dan!
5) Replaced the references to SWAP_CRYPTO_SUB_BATCH_SIZE in comments in
zswap.h, with SWAP_CRYPTO_BATCH_SIZE.
Changes since v1:
=================
1) Rebased to mm-unstable as of 11-1-2024, commit 5c4cf96cd702.
2) Incorporated Herbert's suggestions to use an acomp_req flag to indicate
async/poll mode, and to encapsulate the polling functionality in the
iaa_crypto driver. Thanks Herbert!
3) Incorporated Herbert's and Yosry's suggestions to implement the batching
API in iaa_crypto and to make its use seamless from zswap's
perspective. Thanks Herbert and Yosry!
4) Incorporated Yosry's suggestion to make it more convenient for the user
to enable compress batching, while minimizing the memory footprint
cost. Thanks Yosry!
5) Incorporated Yosry's suggestion to de-couple the shrink_folio_list()
reclaim batching patch from this series, since it requires a broader
discussion.
I would greatly appreciate code review comments for the iaa_crypto driver
and mm patches included in this series!
Thanks,
Kanchana
Kanchana P Sridhar (10):
crypto: acomp - Define two new interfaces for compress/decompress
batching.
crypto: iaa - Add an acomp_req flag CRYPTO_ACOMP_REQ_POLL to enable
async mode.
crypto: iaa - Implement batch_compress(), batch_decompress() API in
iaa_crypto.
crypto: iaa - Make async mode the default.
crypto: iaa - Disable iaa_verify_compress by default.
crypto: iaa - Re-organize the iaa_crypto driver code.
crypto: iaa - Map IAA devices/wqs to cores based on packages instead
of NUMA.
crypto: iaa - Distribute compress jobs from all cores to all IAAs on a
package.
mm: zswap: Allocate pool batching resources if the crypto_alg supports
batching.
mm: zswap: Compress batching with Intel IAA in zswap_batch_store() of
large folios.
crypto/acompress.c | 2 +
drivers/crypto/intel/iaa/iaa_crypto.h | 18 +-
drivers/crypto/intel/iaa/iaa_crypto_main.c | 1637 ++++++++++++++------
include/crypto/acompress.h | 96 ++
include/crypto/internal/acompress.h | 16 +
include/linux/zswap.h | 19 +
mm/page_io.c | 16 +-
mm/zswap.c | 759 ++++++++-
8 files changed, 2090 insertions(+), 473 deletions(-)
base-commit: 5a7056135bb69da2ce0a42eb8c07968c1331777b
--
2.27.0
