From: Barry Song <21cnbao@gmail.com> To: Yu Zhao <yuzhao@google.com> Cc: Andrew Morton <akpm@linux-foundation.org>, Linus Torvalds <torvalds@linux-foundation.org>, Andi Kleen <ak@linux.intel.com>, Catalin Marinas <catalin.marinas@arm.com>, Dave Hansen <dave.hansen@linux.intel.com>, Hillf Danton <hdanton@sina.com>, Jens Axboe <axboe@kernel.dk>, Jesse Barnes <jsbarnes@google.com>, Johannes Weiner <hannes@cmpxchg.org>, Jonathan Corbet <corbet@lwn.net>, Matthew Wilcox <willy@infradead.org>, Mel Gorman <mgorman@suse.de>, Michael Larabel <Michael@michaellarabel.com>, Michal Hocko <mhocko@kernel.org>, Rik van Riel <riel@surriel.com>, Vlastimil Babka <vbabka@suse.cz>, Will Deacon <will@kernel.org>, Ying Huang <ying.huang@intel.com>, LAK <linux-arm-kernel@lists.infradead.org>, Linux Doc Mailing List <linux-doc@vger.kernel.org>, LKML <linux-kernel@vger.kernel.org>, Linux-MM <linux-mm@kvack.org>, page-reclaim@google.com, x86 <x86@kernel.org> Subject: Re: [PATCH v6 0/9] Multigenerational LRU Framework Date: Sun, 23 Jan 2022 18:43:06 +1300 [thread overview] Message-ID: <CAGsJ_4zULJ5vPwn73Z5Bap3eRkAX+Yv24c-n41+zC7fN8xG60g@mail.gmail.com> (raw) In-Reply-To: <20220104202227.2903605-1-yuzhao@google.com> On Wed, Jan 5, 2022 at 7:17 PM Yu Zhao <yuzhao@google.com> wrote: > > TLDR > ==== > The current page reclaim is too expensive in terms of CPU usage and it > often makes poor choices about what to evict. This patchset offers an > alternative solution that is performant, versatile and > straightforward. > > Design objectives > ================= > The design objectives are: > 1. Better representation of access recency > 2. Try to profit from spatial locality > 3. Clear fast path making obvious choices > 4. Simple self-correcting heuristics > > The representation of access recency is at the core of all LRU > approximations. The multigenerational LRU (MGLRU) divides pages into > multiple lists (generations), each having bounded access recency (a > time interval). Generations establish a common frame of reference and > help make better choices, e.g., between different memcgs on a computer > or different computers in a data center (for cluster job scheduling). > > Exploiting spatial locality improves the efficiency when gathering the > accessed bit. A rmap walk targets a single page and doesn't try to > profit from discovering an accessed PTE. A page table walk can sweep > all hotspots in an address space, but its search space can be too > large to make a profit. The key is to optimize both methods and use > them in combination. (PMU is another option for further exploration.) > > Fast path reduces code complexity and runtime overhead. Unmapped pages > don't require TLB flushes; clean pages don't require writeback. These > facts are only helpful when other conditions, e.g., access recency, > are similar. With generations as a common frame of reference, > additional factors stand out. But obvious choices might not be good > choices; thus self-correction is required (the next objective). > > The benefits of simple self-correcting heuristics are self-evident. > Again with generations as a common frame of reference, this becomes > attainable. Specifically, pages in the same generation are categorized > based on additional factors, and a closed-loop control statistically > compares the refault percentages across all categories and throttles > the eviction of those that have higher percentages. > > Patchset overview > ================= > 1. mm: x86, arm64: add arch_has_hw_pte_young() > 2. mm: x86: add CONFIG_ARCH_HAS_NONLEAF_PMD_YOUNG > Materializing hardware optimizations when trying to clear the accessed > bit in many PTEs. If hardware automatically sets the accessed bit in > PTEs, there is no need to worry about bursty page faults (emulating > the accessed bit). If it also sets the accessed bit in non-leaf PMD > entries, there is no need to search the PTE table pointed to by a PMD > entry that doesn't have the accessed bit set. > > 3. mm/vmscan.c: refactor shrink_node() > A minor refactor. > > 4. mm: multigenerational lru: groundwork > Adding the basic data structure and the functions to initialize it and > insert/remove pages. > > 5. mm: multigenerational lru: mm_struct list > An infra keeps track of mm_struct's for page table walkers and > provides them with optimizations, i.e., switch_mm() tracking and Bloom > filters. > > 6. mm: multigenerational lru: aging > 7. mm: multigenerational lru: eviction > "The page reclaim" is a producer/consumer model. "The aging" produces > cold pages, whereas "the eviction " consumes them. Cold pages flow > through generations. The aging uses the mm_struct list infra to sweep > dense hotspots in page tables. During a page table walk, the aging > clears the accessed bit and tags accessed pages with the youngest > generation number. The eviction sorts those pages when it encounters > them. For pages in the oldest generation, eviction walks the rmap to > check the accessed bit one more time before evicting them. During an > rmap walk, the eviction feeds dense hotspots back to the aging. Dense > hotspots flow through the Bloom filters. For pages not mapped in page > tables, the eviction uses the PID controller to statistically > determine whether they have higher refaults. If so, the eviction > throttles their eviction by moving them to the next generation (the > second oldest). > > 8. mm: multigenerational lru: user interface > The knobs to turn on/off MGLRU and provide the userspace with > thrashing prevention, working set estimation (the aging) and proactive > reclaim (the eviction). > > 9. mm: multigenerational lru: Kconfig > The Kconfig options. > > Benchmark results > ================= > Independent lab results > ----------------------- > Based on the popularity of searches [01] and the memory usage in > Google's public cloud, the most popular open-source memory-hungry > applications, in alphabetical order, are: > Apache Cassandra Memcached > Apache Hadoop MongoDB > Apache Spark PostgreSQL > MariaDB (MySQL) Redis > > An independent lab evaluated MGLRU with the most widely used benchmark > suites for the above applications. They posted 960 data points along > with kernel metrics and perf profiles collected over more than 500 > hours of total benchmark time. Their final reports show that, with 95% > confidence intervals (CIs), the above applications all performed > significantly better for at least part of their benchmark matrices. > > On 5.14: > 1. Apache Spark [02] took 95% CIs [9.28, 11.19]% and [12.20, 14.93]% > less wall time to sort three billion random integers, respectively, > under the medium- and the high-concurrency conditions, when > overcommitting memory. There were no statistically significant > changes in wall time for the rest of the benchmark matrix. > 2. MariaDB [03] achieved 95% CIs [5.24, 10.71]% and [20.22, 25.97]% > more transactions per minute (TPM), respectively, under the medium- > and the high-concurrency conditions, when overcommitting memory. > There were no statistically significant changes in TPM for the rest > of the benchmark matrix. > 3. Memcached [04] achieved 95% CIs [23.54, 32.25]%, [20.76, 41.61]% > and [21.59, 30.02]% more operations per second (OPS), respectively, > for sequential access, random access and Gaussian (distribution) > access, when THP=always; 95% CIs [13.85, 15.97]% and > [23.94, 29.92]% more OPS, respectively, for random access and > Gaussian access, when THP=never. There were no statistically > significant changes in OPS for the rest of the benchmark matrix. > 4. MongoDB [05] achieved 95% CIs [2.23, 3.44]%, [6.97, 9.73]% and > [2.16, 3.55]% more operations per second (OPS), respectively, for > exponential (distribution) access, random access and Zipfian > (distribution) access, when underutilizing memory; 95% CIs > [8.83, 10.03]%, [21.12, 23.14]% and [5.53, 6.46]% more OPS, > respectively, for exponential access, random access and Zipfian > access, when overcommitting memory. > > On 5.15: > 5. Apache Cassandra [06] achieved 95% CIs [1.06, 4.10]%, [1.94, 5.43]% > and [4.11, 7.50]% more operations per second (OPS), respectively, > for exponential (distribution) access, random access and Zipfian > (distribution) access, when swap was off; 95% CIs [0.50, 2.60]%, > [6.51, 8.77]% and [3.29, 6.75]% more OPS, respectively, for > exponential access, random access and Zipfian access, when swap was > on. > 6. Apache Hadoop [07] took 95% CIs [5.31, 9.69]% and [2.02, 7.86]% > less average wall time to finish twelve parallel TeraSort jobs, > respectively, under the medium- and the high-concurrency > conditions, when swap was on. There were no statistically > significant changes in average wall time for the rest of the > benchmark matrix. > 7. PostgreSQL [08] achieved 95% CI [1.75, 6.42]% more transactions per > minute (TPM) under the high-concurrency condition, when swap was > off; 95% CIs [12.82, 18.69]% and [22.70, 46.86]% more TPM, > respectively, under the medium- and the high-concurrency > conditions, when swap was on. There were no statistically > significant changes in TPM for the rest of the benchmark matrix. > 8. Redis [09] achieved 95% CIs [0.58, 5.94]%, [6.55, 14.58]% and > [11.47, 19.36]% more total operations per second (OPS), > respectively, for sequential access, random access and Gaussian > (distribution) access, when THP=always; 95% CIs [1.27, 3.54]%, > [10.11, 14.81]% and [8.75, 13.64]% more total OPS, respectively, > for sequential access, random access and Gaussian access, when > THP=never. > > Our lab results > --------------- > To supplement the above results, we ran the following benchmark suites > on 5.16-rc7 and found no regressions [10]. (These synthetic benchmarks > are popular among MM developers, but we prefer large-scale A/B > experiments to validate improvements.) > fs_fio_bench_hdd_mq pft > fs_lmbench pgsql-hammerdb > fs_parallelio redis > fs_postmark stream > hackbench sysbenchthread > kernbench tpcc_spark > memcached unixbench > multichase vm-scalability > mutilate will-it-scale > nginx > > [01] https://trends.google.com > [02] https://lore.kernel.org/linux-mm/20211102002002.92051-1-bot@edi.works/ > [03] https://lore.kernel.org/linux-mm/20211009054315.47073-1-bot@edi.works/ > [04] https://lore.kernel.org/linux-mm/20211021194103.65648-1-bot@edi.works/ > [05] https://lore.kernel.org/linux-mm/20211109021346.50266-1-bot@edi.works/ > [06] https://lore.kernel.org/linux-mm/20211202062806.80365-1-bot@edi.works/ > [07] https://lore.kernel.org/linux-mm/20211209072416.33606-1-bot@edi.works/ > [08] https://lore.kernel.org/linux-mm/20211218071041.24077-1-bot@edi.works/ > [09] https://lore.kernel.org/linux-mm/20211122053248.57311-1-bot@edi.works/ > [10] https://lore.kernel.org/linux-mm/20220104202247.2903702-1-yuzhao@google.com/ > > Read-world applications > ======================= > Third-party testimonials > ------------------------ > Konstantin wrote [11]: > I have Archlinux with 8G RAM + zswap + swap. While developing, I > have lots of apps opened such as multiple LSP-servers for different > langs, chats, two browsers, etc... Usually, my system gets quickly > to a point of SWAP-storms, where I have to kill LSP-servers, > restart browsers to free memory, etc, otherwise the system lags > heavily and is barely usable. > > 1.5 day ago I migrated from 5.11.15 kernel to 5.12 + the LRU > patchset, and I started up by opening lots of apps to create memory > pressure, and worked for a day like this. Till now I had *not a > single SWAP-storm*, and mind you I got 3.4G in SWAP. I was never > getting to the point of 3G in SWAP before without a single > SWAP-storm. > > The Arch Linux Zen kernel [12] has been using MGLRU since 5.12. Many > of its users reported their positive experiences to me, e.g., Shivodit > wrote: > I've tried the latest Zen kernel (5.14.13-zen1-1-zen in the > archlinux testing repos), everything's been smooth so far. I also > decided to copy a large volume of files to check performance under > I/O load, and everything went smoothly - no stuttering was present, > everything was responsive. > > Large-scale deployments > ----------------------- > We've rolled out MGLRU to tens of millions of Chrome OS users and > about a million Android users. Google's fleetwide profiling [13] shows > an overall 40% decrease in kswapd CPU usage, in addition to Hi Yu, Was the overall 40% decrease of kswap CPU usgae seen on x86 or arm64? And I am curious how much we are taking advantage of NONLEAF_PMD_YOUNG. Does it help a lot in decreasing the cpu usage? If so, this might be a good proof that arm64 also needs this hardware feature? In short, I am curious how much the improvement in this patchset depends on the hardware ability of NONLEAF_PMD_YOUNG. > improvements in other UX metrics, e.g., an 85% decrease in the number > of low-memory kills at the 75th percentile and an 18% decrease in > rendering latency at the 50th percentile. > > [11] https://lore.kernel.org/linux-mm/140226722f2032c86301fbd326d91baefe3d7d23.camel@yandex.ru/ > [12] https://github.com/zen-kernel/zen-kernel/ > [13] https://research.google/pubs/pub44271/ > Thanks Barry
WARNING: multiple messages have this Message-ID (diff)
From: Barry Song <21cnbao@gmail.com> To: Yu Zhao <yuzhao@google.com> Cc: Andrew Morton <akpm@linux-foundation.org>, Linus Torvalds <torvalds@linux-foundation.org>, Andi Kleen <ak@linux.intel.com>, Catalin Marinas <catalin.marinas@arm.com>, Dave Hansen <dave.hansen@linux.intel.com>, Hillf Danton <hdanton@sina.com>, Jens Axboe <axboe@kernel.dk>, Jesse Barnes <jsbarnes@google.com>, Johannes Weiner <hannes@cmpxchg.org>, Jonathan Corbet <corbet@lwn.net>, Matthew Wilcox <willy@infradead.org>, Mel Gorman <mgorman@suse.de>, Michael Larabel <Michael@michaellarabel.com>, Michal Hocko <mhocko@kernel.org>, Rik van Riel <riel@surriel.com>, Vlastimil Babka <vbabka@suse.cz>, Will Deacon <will@kernel.org>, Ying Huang <ying.huang@intel.com>, LAK <linux-arm-kernel@lists.infradead.org>, Linux Doc Mailing List <linux-doc@vger.kernel.org>, LKML <linux-kernel@vger.kernel.org>, Linux-MM <linux-mm@kvack.org>, page-reclaim@google.com, x86 <x86@kernel.org> Subject: Re: [PATCH v6 0/9] Multigenerational LRU Framework Date: Sun, 23 Jan 2022 18:43:06 +1300 [thread overview] Message-ID: <CAGsJ_4zULJ5vPwn73Z5Bap3eRkAX+Yv24c-n41+zC7fN8xG60g@mail.gmail.com> (raw) In-Reply-To: <20220104202227.2903605-1-yuzhao@google.com> On Wed, Jan 5, 2022 at 7:17 PM Yu Zhao <yuzhao@google.com> wrote: > > TLDR > ==== > The current page reclaim is too expensive in terms of CPU usage and it > often makes poor choices about what to evict. This patchset offers an > alternative solution that is performant, versatile and > straightforward. > > Design objectives > ================= > The design objectives are: > 1. Better representation of access recency > 2. Try to profit from spatial locality > 3. Clear fast path making obvious choices > 4. Simple self-correcting heuristics > > The representation of access recency is at the core of all LRU > approximations. The multigenerational LRU (MGLRU) divides pages into > multiple lists (generations), each having bounded access recency (a > time interval). Generations establish a common frame of reference and > help make better choices, e.g., between different memcgs on a computer > or different computers in a data center (for cluster job scheduling). > > Exploiting spatial locality improves the efficiency when gathering the > accessed bit. A rmap walk targets a single page and doesn't try to > profit from discovering an accessed PTE. A page table walk can sweep > all hotspots in an address space, but its search space can be too > large to make a profit. The key is to optimize both methods and use > them in combination. (PMU is another option for further exploration.) > > Fast path reduces code complexity and runtime overhead. Unmapped pages > don't require TLB flushes; clean pages don't require writeback. These > facts are only helpful when other conditions, e.g., access recency, > are similar. With generations as a common frame of reference, > additional factors stand out. But obvious choices might not be good > choices; thus self-correction is required (the next objective). > > The benefits of simple self-correcting heuristics are self-evident. > Again with generations as a common frame of reference, this becomes > attainable. Specifically, pages in the same generation are categorized > based on additional factors, and a closed-loop control statistically > compares the refault percentages across all categories and throttles > the eviction of those that have higher percentages. > > Patchset overview > ================= > 1. mm: x86, arm64: add arch_has_hw_pte_young() > 2. mm: x86: add CONFIG_ARCH_HAS_NONLEAF_PMD_YOUNG > Materializing hardware optimizations when trying to clear the accessed > bit in many PTEs. If hardware automatically sets the accessed bit in > PTEs, there is no need to worry about bursty page faults (emulating > the accessed bit). If it also sets the accessed bit in non-leaf PMD > entries, there is no need to search the PTE table pointed to by a PMD > entry that doesn't have the accessed bit set. > > 3. mm/vmscan.c: refactor shrink_node() > A minor refactor. > > 4. mm: multigenerational lru: groundwork > Adding the basic data structure and the functions to initialize it and > insert/remove pages. > > 5. mm: multigenerational lru: mm_struct list > An infra keeps track of mm_struct's for page table walkers and > provides them with optimizations, i.e., switch_mm() tracking and Bloom > filters. > > 6. mm: multigenerational lru: aging > 7. mm: multigenerational lru: eviction > "The page reclaim" is a producer/consumer model. "The aging" produces > cold pages, whereas "the eviction " consumes them. Cold pages flow > through generations. The aging uses the mm_struct list infra to sweep > dense hotspots in page tables. During a page table walk, the aging > clears the accessed bit and tags accessed pages with the youngest > generation number. The eviction sorts those pages when it encounters > them. For pages in the oldest generation, eviction walks the rmap to > check the accessed bit one more time before evicting them. During an > rmap walk, the eviction feeds dense hotspots back to the aging. Dense > hotspots flow through the Bloom filters. For pages not mapped in page > tables, the eviction uses the PID controller to statistically > determine whether they have higher refaults. If so, the eviction > throttles their eviction by moving them to the next generation (the > second oldest). > > 8. mm: multigenerational lru: user interface > The knobs to turn on/off MGLRU and provide the userspace with > thrashing prevention, working set estimation (the aging) and proactive > reclaim (the eviction). > > 9. mm: multigenerational lru: Kconfig > The Kconfig options. > > Benchmark results > ================= > Independent lab results > ----------------------- > Based on the popularity of searches [01] and the memory usage in > Google's public cloud, the most popular open-source memory-hungry > applications, in alphabetical order, are: > Apache Cassandra Memcached > Apache Hadoop MongoDB > Apache Spark PostgreSQL > MariaDB (MySQL) Redis > > An independent lab evaluated MGLRU with the most widely used benchmark > suites for the above applications. They posted 960 data points along > with kernel metrics and perf profiles collected over more than 500 > hours of total benchmark time. Their final reports show that, with 95% > confidence intervals (CIs), the above applications all performed > significantly better for at least part of their benchmark matrices. > > On 5.14: > 1. Apache Spark [02] took 95% CIs [9.28, 11.19]% and [12.20, 14.93]% > less wall time to sort three billion random integers, respectively, > under the medium- and the high-concurrency conditions, when > overcommitting memory. There were no statistically significant > changes in wall time for the rest of the benchmark matrix. > 2. MariaDB [03] achieved 95% CIs [5.24, 10.71]% and [20.22, 25.97]% > more transactions per minute (TPM), respectively, under the medium- > and the high-concurrency conditions, when overcommitting memory. > There were no statistically significant changes in TPM for the rest > of the benchmark matrix. > 3. Memcached [04] achieved 95% CIs [23.54, 32.25]%, [20.76, 41.61]% > and [21.59, 30.02]% more operations per second (OPS), respectively, > for sequential access, random access and Gaussian (distribution) > access, when THP=always; 95% CIs [13.85, 15.97]% and > [23.94, 29.92]% more OPS, respectively, for random access and > Gaussian access, when THP=never. There were no statistically > significant changes in OPS for the rest of the benchmark matrix. > 4. MongoDB [05] achieved 95% CIs [2.23, 3.44]%, [6.97, 9.73]% and > [2.16, 3.55]% more operations per second (OPS), respectively, for > exponential (distribution) access, random access and Zipfian > (distribution) access, when underutilizing memory; 95% CIs > [8.83, 10.03]%, [21.12, 23.14]% and [5.53, 6.46]% more OPS, > respectively, for exponential access, random access and Zipfian > access, when overcommitting memory. > > On 5.15: > 5. Apache Cassandra [06] achieved 95% CIs [1.06, 4.10]%, [1.94, 5.43]% > and [4.11, 7.50]% more operations per second (OPS), respectively, > for exponential (distribution) access, random access and Zipfian > (distribution) access, when swap was off; 95% CIs [0.50, 2.60]%, > [6.51, 8.77]% and [3.29, 6.75]% more OPS, respectively, for > exponential access, random access and Zipfian access, when swap was > on. > 6. Apache Hadoop [07] took 95% CIs [5.31, 9.69]% and [2.02, 7.86]% > less average wall time to finish twelve parallel TeraSort jobs, > respectively, under the medium- and the high-concurrency > conditions, when swap was on. There were no statistically > significant changes in average wall time for the rest of the > benchmark matrix. > 7. PostgreSQL [08] achieved 95% CI [1.75, 6.42]% more transactions per > minute (TPM) under the high-concurrency condition, when swap was > off; 95% CIs [12.82, 18.69]% and [22.70, 46.86]% more TPM, > respectively, under the medium- and the high-concurrency > conditions, when swap was on. There were no statistically > significant changes in TPM for the rest of the benchmark matrix. > 8. Redis [09] achieved 95% CIs [0.58, 5.94]%, [6.55, 14.58]% and > [11.47, 19.36]% more total operations per second (OPS), > respectively, for sequential access, random access and Gaussian > (distribution) access, when THP=always; 95% CIs [1.27, 3.54]%, > [10.11, 14.81]% and [8.75, 13.64]% more total OPS, respectively, > for sequential access, random access and Gaussian access, when > THP=never. > > Our lab results > --------------- > To supplement the above results, we ran the following benchmark suites > on 5.16-rc7 and found no regressions [10]. (These synthetic benchmarks > are popular among MM developers, but we prefer large-scale A/B > experiments to validate improvements.) > fs_fio_bench_hdd_mq pft > fs_lmbench pgsql-hammerdb > fs_parallelio redis > fs_postmark stream > hackbench sysbenchthread > kernbench tpcc_spark > memcached unixbench > multichase vm-scalability > mutilate will-it-scale > nginx > > [01] https://trends.google.com > [02] https://lore.kernel.org/linux-mm/20211102002002.92051-1-bot@edi.works/ > [03] https://lore.kernel.org/linux-mm/20211009054315.47073-1-bot@edi.works/ > [04] https://lore.kernel.org/linux-mm/20211021194103.65648-1-bot@edi.works/ > [05] https://lore.kernel.org/linux-mm/20211109021346.50266-1-bot@edi.works/ > [06] https://lore.kernel.org/linux-mm/20211202062806.80365-1-bot@edi.works/ > [07] https://lore.kernel.org/linux-mm/20211209072416.33606-1-bot@edi.works/ > [08] https://lore.kernel.org/linux-mm/20211218071041.24077-1-bot@edi.works/ > [09] https://lore.kernel.org/linux-mm/20211122053248.57311-1-bot@edi.works/ > [10] https://lore.kernel.org/linux-mm/20220104202247.2903702-1-yuzhao@google.com/ > > Read-world applications > ======================= > Third-party testimonials > ------------------------ > Konstantin wrote [11]: > I have Archlinux with 8G RAM + zswap + swap. While developing, I > have lots of apps opened such as multiple LSP-servers for different > langs, chats, two browsers, etc... Usually, my system gets quickly > to a point of SWAP-storms, where I have to kill LSP-servers, > restart browsers to free memory, etc, otherwise the system lags > heavily and is barely usable. > > 1.5 day ago I migrated from 5.11.15 kernel to 5.12 + the LRU > patchset, and I started up by opening lots of apps to create memory > pressure, and worked for a day like this. Till now I had *not a > single SWAP-storm*, and mind you I got 3.4G in SWAP. I was never > getting to the point of 3G in SWAP before without a single > SWAP-storm. > > The Arch Linux Zen kernel [12] has been using MGLRU since 5.12. Many > of its users reported their positive experiences to me, e.g., Shivodit > wrote: > I've tried the latest Zen kernel (5.14.13-zen1-1-zen in the > archlinux testing repos), everything's been smooth so far. I also > decided to copy a large volume of files to check performance under > I/O load, and everything went smoothly - no stuttering was present, > everything was responsive. > > Large-scale deployments > ----------------------- > We've rolled out MGLRU to tens of millions of Chrome OS users and > about a million Android users. Google's fleetwide profiling [13] shows > an overall 40% decrease in kswapd CPU usage, in addition to Hi Yu, Was the overall 40% decrease of kswap CPU usgae seen on x86 or arm64? And I am curious how much we are taking advantage of NONLEAF_PMD_YOUNG. Does it help a lot in decreasing the cpu usage? If so, this might be a good proof that arm64 also needs this hardware feature? In short, I am curious how much the improvement in this patchset depends on the hardware ability of NONLEAF_PMD_YOUNG. > improvements in other UX metrics, e.g., an 85% decrease in the number > of low-memory kills at the 75th percentile and an 18% decrease in > rendering latency at the 50th percentile. > > [11] https://lore.kernel.org/linux-mm/140226722f2032c86301fbd326d91baefe3d7d23.camel@yandex.ru/ > [12] https://github.com/zen-kernel/zen-kernel/ > [13] https://research.google/pubs/pub44271/ > Thanks Barry _______________________________________________ linux-arm-kernel mailing list linux-arm-kernel@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
next prev parent reply other threads:[~2022-01-23 5:43 UTC|newest] Thread overview: 223+ messages / expand[flat|nested] mbox.gz Atom feed top 2022-01-04 20:22 [PATCH v6 0/9] Multigenerational LRU Framework Yu Zhao 2022-01-04 20:22 ` Yu Zhao 2022-01-04 20:22 ` [PATCH v6 1/9] mm: x86, arm64: add arch_has_hw_pte_young() Yu Zhao 2022-01-04 20:22 ` Yu Zhao 2022-01-05 10:45 ` Will Deacon 2022-01-05 10:45 ` Will Deacon 2022-01-05 20:47 ` Yu Zhao 2022-01-05 20:47 ` Yu Zhao 2022-01-06 10:30 ` Will Deacon 2022-01-06 10:30 ` Will Deacon 2022-01-07 7:25 ` Yu Zhao 2022-01-07 7:25 ` Yu Zhao 2022-01-11 14:19 ` Will Deacon 2022-01-11 14:19 ` Will Deacon 2022-01-11 22:27 ` Yu Zhao 2022-01-11 22:27 ` Yu Zhao 2022-01-04 20:22 ` [PATCH v6 2/9] mm: x86: add CONFIG_ARCH_HAS_NONLEAF_PMD_YOUNG Yu Zhao 2022-01-04 20:22 ` Yu Zhao 2022-01-04 21:24 ` Linus Torvalds 2022-01-04 21:24 ` Linus Torvalds 2022-01-04 20:22 ` [PATCH v6 3/9] mm/vmscan.c: refactor shrink_node() Yu Zhao 2022-01-04 20:22 ` Yu Zhao 2022-01-04 20:22 ` [PATCH v6 4/9] mm: multigenerational lru: groundwork Yu Zhao 2022-01-04 20:22 ` Yu Zhao 2022-01-04 21:34 ` Linus Torvalds 2022-01-04 21:34 ` Linus Torvalds 2022-01-11 8:16 ` Aneesh Kumar K.V 2022-01-11 8:16 ` Aneesh Kumar K.V 2022-01-12 2:16 ` Yu Zhao 2022-01-12 2:16 ` Yu Zhao 2022-01-04 20:22 ` [PATCH v6 5/9] mm: multigenerational lru: mm_struct list Yu Zhao 2022-01-04 20:22 ` Yu Zhao 2022-01-07 9:06 ` Michal Hocko 2022-01-07 9:06 ` Michal Hocko 2022-01-08 0:19 ` Yu Zhao 2022-01-08 0:19 ` Yu Zhao 2022-01-10 15:21 ` Michal Hocko 2022-01-10 15:21 ` Michal Hocko 2022-01-12 8:08 ` Yu Zhao 2022-01-12 8:08 ` Yu Zhao 2022-01-04 20:22 ` [PATCH v6 6/9] mm: multigenerational lru: aging Yu Zhao 2022-01-04 20:22 ` Yu Zhao 2022-01-06 16:06 ` Michal Hocko 2022-01-06 16:06 ` Michal Hocko 2022-01-06 21:27 ` Yu Zhao 2022-01-06 21:27 ` Yu Zhao 2022-01-07 8:43 ` Michal Hocko 2022-01-07 8:43 ` Michal Hocko 2022-01-07 21:12 ` Yu Zhao 2022-01-07 21:12 ` Yu Zhao 2022-01-06 16:12 ` Michal Hocko 2022-01-06 16:12 ` Michal Hocko 2022-01-06 21:41 ` Yu Zhao 2022-01-06 21:41 ` Yu Zhao 2022-01-07 8:55 ` Michal Hocko 2022-01-07 8:55 ` Michal Hocko 2022-01-07 9:00 ` Michal Hocko 2022-01-07 9:00 ` Michal Hocko 2022-01-10 3:58 ` Yu Zhao 2022-01-10 3:58 ` Yu Zhao 2022-01-10 14:37 ` Michal Hocko 2022-01-10 14:37 ` Michal Hocko 2022-01-13 9:43 ` Yu Zhao 2022-01-13 9:43 ` Yu Zhao 2022-01-13 12:02 ` Michal Hocko 2022-01-13 12:02 ` Michal Hocko 2022-01-19 6:31 ` Yu Zhao 2022-01-19 6:31 ` Yu Zhao 2022-01-19 9:44 ` Michal Hocko 2022-01-19 9:44 ` Michal Hocko 2022-01-10 15:01 ` Michal Hocko 2022-01-10 15:01 ` Michal Hocko 2022-01-10 16:01 ` Vlastimil Babka 2022-01-10 16:01 ` Vlastimil Babka 2022-01-10 16:25 ` Michal Hocko 2022-01-10 16:25 ` Michal Hocko 2022-01-11 23:16 ` Yu Zhao 2022-01-11 23:16 ` Yu Zhao 2022-01-12 10:28 ` Michal Hocko 2022-01-12 10:28 ` Michal Hocko 2022-01-13 9:25 ` Yu Zhao 2022-01-13 9:25 ` Yu Zhao 2022-01-07 13:11 ` Michal Hocko 2022-01-07 13:11 ` Michal Hocko 2022-01-07 23:36 ` Yu Zhao 2022-01-07 23:36 ` Yu Zhao 2022-01-10 15:35 ` Michal Hocko 2022-01-10 15:35 ` Michal Hocko 2022-01-11 1:18 ` Yu Zhao 2022-01-11 1:18 ` Yu Zhao 2022-01-11 9:00 ` Michal Hocko 2022-01-11 9:00 ` Michal Hocko [not found] ` <1641900108.61dd684cb0e59@mail.inbox.lv> 2022-01-11 12:15 ` Michal Hocko 2022-01-11 12:15 ` Michal Hocko 2022-01-13 17:00 ` Alexey Avramov 2022-01-13 17:00 ` Alexey Avramov 2022-01-11 14:22 ` Alexey Avramov 2022-01-11 14:22 ` Alexey Avramov 2022-01-07 14:44 ` Michal Hocko 2022-01-07 14:44 ` Michal Hocko 2022-01-10 4:47 ` Yu Zhao 2022-01-10 4:47 ` Yu Zhao 2022-01-10 10:54 ` Michal Hocko 2022-01-10 10:54 ` Michal Hocko 2022-01-19 7:04 ` Yu Zhao 2022-01-19 7:04 ` Yu Zhao 2022-01-19 9:42 ` Michal Hocko 2022-01-19 9:42 ` Michal Hocko 2022-01-23 21:28 ` Yu Zhao 2022-01-23 21:28 ` Yu Zhao 2022-01-24 14:01 ` Michal Hocko 2022-01-24 14:01 ` Michal Hocko 2022-01-10 16:57 ` Michal Hocko 2022-01-10 16:57 ` Michal Hocko 2022-01-12 1:01 ` Yu Zhao 2022-01-12 1:01 ` Yu Zhao 2022-01-12 10:17 ` Michal Hocko 2022-01-12 10:17 ` Michal Hocko 2022-01-12 23:43 ` Yu Zhao 2022-01-12 23:43 ` Yu Zhao 2022-01-13 11:57 ` Michal Hocko 2022-01-13 11:57 ` Michal Hocko 2022-01-23 21:40 ` Yu Zhao 2022-01-23 21:40 ` Yu Zhao 2022-01-04 20:22 ` [PATCH v6 7/9] mm: multigenerational lru: eviction Yu Zhao 2022-01-04 20:22 ` Yu Zhao 2022-01-11 10:37 ` Aneesh Kumar K.V 2022-01-11 10:37 ` Aneesh Kumar K.V 2022-01-12 8:05 ` Yu Zhao 2022-01-12 8:05 ` Yu Zhao 2022-01-04 20:22 ` [PATCH v6 8/9] mm: multigenerational lru: user interface Yu Zhao 2022-01-04 20:22 ` Yu Zhao 2022-01-10 10:27 ` Mike Rapoport 2022-01-10 10:27 ` Mike Rapoport 2022-01-12 8:35 ` Yu Zhao 2022-01-12 8:35 ` Yu Zhao 2022-01-12 10:31 ` Michal Hocko 2022-01-12 10:31 ` Michal Hocko 2022-01-12 15:45 ` Mike Rapoport 2022-01-12 15:45 ` Mike Rapoport 2022-01-13 9:47 ` Yu Zhao 2022-01-13 9:47 ` Yu Zhao 2022-01-13 10:31 ` Aneesh Kumar K.V 2022-01-13 10:31 ` Aneesh Kumar K.V 2022-01-13 23:02 ` Yu Zhao 2022-01-13 23:02 ` Yu Zhao 2022-01-14 5:20 ` Aneesh Kumar K.V 2022-01-14 5:20 ` Aneesh Kumar K.V 2022-01-14 6:50 ` Yu Zhao 2022-01-14 6:50 ` Yu Zhao 2022-01-04 20:22 ` [PATCH v6 9/9] mm: multigenerational lru: Kconfig Yu Zhao 2022-01-04 20:22 ` Yu Zhao 2022-01-04 21:39 ` Linus Torvalds 2022-01-04 21:39 ` Linus Torvalds 2022-01-04 20:22 ` [PATCH v6 0/9] Multigenerational LRU Framework Yu Zhao 2022-01-04 20:30 ` Yu Zhao 2022-01-04 20:30 ` Yu Zhao 2022-01-04 21:43 ` Linus Torvalds 2022-01-04 21:43 ` Linus Torvalds 2022-01-05 21:12 ` Yu Zhao 2022-01-05 21:12 ` Yu Zhao 2022-01-07 9:38 ` Michal Hocko 2022-01-07 9:38 ` Michal Hocko 2022-01-07 18:45 ` Yu Zhao 2022-01-07 18:45 ` Yu Zhao 2022-01-10 15:39 ` Michal Hocko 2022-01-10 15:39 ` Michal Hocko 2022-01-10 22:04 ` Yu Zhao 2022-01-10 22:04 ` Yu Zhao 2022-01-10 22:46 ` Jesse Barnes 2022-01-10 22:46 ` Jesse Barnes 2022-01-11 1:41 ` Linus Torvalds 2022-01-11 1:41 ` Linus Torvalds 2022-01-11 10:40 ` Michal Hocko 2022-01-11 10:40 ` Michal Hocko 2022-01-11 8:41 ` Yu Zhao 2022-01-11 8:41 ` Yu Zhao 2022-01-11 8:53 ` Holger Hoffstätte 2022-01-11 8:53 ` Holger Hoffstätte 2022-01-11 9:26 ` Jan Alexander Steffens (heftig) 2022-01-11 16:04 ` Shuang Zhai 2022-01-11 16:04 ` Shuang Zhai 2022-01-12 1:46 ` Suleiman Souhlal 2022-01-12 1:46 ` Suleiman Souhlal 2022-01-12 6:07 ` Sofia Trinh 2022-01-12 6:07 ` Sofia Trinh 2022-01-12 16:17 ` Daniel Byrne 2022-01-18 9:21 ` Yu Zhao 2022-01-18 9:21 ` Yu Zhao 2022-01-18 9:36 ` Donald Carr 2022-01-18 9:36 ` Donald Carr 2022-01-19 20:19 ` Steven Barrett 2022-01-19 20:19 ` Steven Barrett 2022-01-19 22:25 ` Brian Geffon 2022-01-19 22:25 ` Brian Geffon 2022-01-05 2:44 ` Shuang Zhai 2022-01-05 2:44 ` Shuang Zhai 2022-01-05 8:55 ` SeongJae Park 2022-01-05 8:55 ` SeongJae Park 2022-01-05 10:53 ` Yu Zhao 2022-01-05 10:53 ` Yu Zhao 2022-01-05 11:12 ` Borislav Petkov 2022-01-05 11:12 ` Borislav Petkov 2022-01-05 11:25 ` SeongJae Park 2022-01-05 11:25 ` SeongJae Park 2022-01-05 21:06 ` Yu Zhao 2022-01-05 21:06 ` Yu Zhao 2022-01-10 14:49 ` Alexey Avramov 2022-01-10 14:49 ` Alexey Avramov 2022-01-11 10:24 ` Alexey Avramov 2022-01-11 10:24 ` Alexey Avramov 2022-01-12 20:56 ` Oleksandr Natalenko 2022-01-12 20:56 ` Oleksandr Natalenko 2022-01-13 8:59 ` Yu Zhao 2022-01-13 8:59 ` Yu Zhao 2022-01-23 5:43 ` Barry Song [this message] 2022-01-23 5:43 ` Barry Song 2022-01-25 6:48 ` Yu Zhao 2022-01-25 6:48 ` Yu Zhao 2022-01-28 8:54 ` Barry Song 2022-01-28 8:54 ` Barry Song 2022-02-08 9:16 ` Yu Zhao 2022-02-08 9:16 ` Yu Zhao
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=CAGsJ_4zULJ5vPwn73Z5Bap3eRkAX+Yv24c-n41+zC7fN8xG60g@mail.gmail.com \ --to=21cnbao@gmail.com \ --cc=Michael@michaellarabel.com \ --cc=ak@linux.intel.com \ --cc=akpm@linux-foundation.org \ --cc=axboe@kernel.dk \ --cc=catalin.marinas@arm.com \ --cc=corbet@lwn.net \ --cc=dave.hansen@linux.intel.com \ --cc=hannes@cmpxchg.org \ --cc=hdanton@sina.com \ --cc=jsbarnes@google.com \ --cc=linux-arm-kernel@lists.infradead.org \ --cc=linux-doc@vger.kernel.org \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-mm@kvack.org \ --cc=mgorman@suse.de \ --cc=mhocko@kernel.org \ --cc=page-reclaim@google.com \ --cc=riel@surriel.com \ --cc=torvalds@linux-foundation.org \ --cc=vbabka@suse.cz \ --cc=will@kernel.org \ --cc=willy@infradead.org \ --cc=x86@kernel.org \ --cc=ying.huang@intel.com \ --cc=yuzhao@google.com \ /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: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.