[v9,1/3] x86: Increase `non_temporal_threshold` to roughly `sizeof_L3 / 4`

Current `non_temporal_threshold` set to roughly '3/4 * sizeof_L3 /
ncores_per_socket'. This patch updates that value to roughly
'sizeof_L3 / 4`

The original value (specifically dividing the `ncores_per_socket`) was
done to limit the amount of other threads' data a `memcpy`/`memset`
could evict.

Dividing by 'ncores_per_socket', however leads to exceedingly low
non-temporal thresholds and leads to using non-temporal stores in
cases where REP MOVSB is multiple times faster.

Furthermore, non-temporal stores are written directly to main memory
so using it at a size much smaller than L3 can place soon to be
accessed data much further away than it otherwise could be. As well,
modern machines are able to detect streaming patterns (especially if
REP MOVSB is used) and provide LRU hints to the memory subsystem. This
in affect caps the total amount of eviction at 1/cache_associativity,
far below meaningfully thrashing the entire cache.

As best I can tell, the benchmarks that lead this small threshold
where done comparing non-temporal stores versus standard cacheable
stores. A better comparison (linked below) is to be REP MOVSB which,
on the measure systems, is nearly 2x faster than non-temporal stores
at the low-end of the previous threshold, and within 10% for over
100MB copies (well past even the current threshold). In cases with a
low number of threads competing for bandwidth, REP MOVSB is ~2x faster
up to `sizeof_L3`.

The divisor of `4` is a somewhat arbitrary value. From benchmarks it
seems Skylake and Icelake both prefer a divisor of `2`, but older CPUs
such as Broadwell prefer something closer to `8`. This patch is meant
to be followed up by another one to make the divisor cpu-specific, but
in the meantime (and for easier backporting), this patch settles on
`4` as a middle-ground.

Benchmarks comparing non-temporal stores, REP MOVSB, and cacheable
stores where done using:
https://github.com/goldsteinn/memcpy-nt-benchmarks

Sheets results (also available in pdf on the github):
https://docs.google.com/spreadsheets/d/e/2PACX-1vS183r0rW_jRX6tG_E90m9qVuFiMbRIJvi5VAE8yYOvEOIEEc3aSNuEsrFbuXw5c3nGboxMmrupZD7K/pubhtml
---
 sysdeps/x86/dl-cacheinfo.h | 70 +++++++++++++++++++++++---------------
 1 file changed, 43 insertions(+), 27 deletions(-)

On Sat, May 13, 2023 at 12:19 AM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>
> Current `non_temporal_threshold` set to roughly '3/4 * sizeof_L3 /
> ncores_per_socket'. This patch updates that value to roughly
> 'sizeof_L3 / 4`
>
> The original value (specifically dividing the `ncores_per_socket`) was
> done to limit the amount of other threads' data a `memcpy`/`memset`
> could evict.
>
> Dividing by 'ncores_per_socket', however leads to exceedingly low
> non-temporal thresholds and leads to using non-temporal stores in
> cases where REP MOVSB is multiple times faster.
>
> Furthermore, non-temporal stores are written directly to main memory
> so using it at a size much smaller than L3 can place soon to be
> accessed data much further away than it otherwise could be. As well,
> modern machines are able to detect streaming patterns (especially if
> REP MOVSB is used) and provide LRU hints to the memory subsystem. This
> in affect caps the total amount of eviction at 1/cache_associativity,
> far below meaningfully thrashing the entire cache.
>
> As best I can tell, the benchmarks that lead this small threshold
> where done comparing non-temporal stores versus standard cacheable
> stores. A better comparison (linked below) is to be REP MOVSB which,
> on the measure systems, is nearly 2x faster than non-temporal stores
> at the low-end of the previous threshold, and within 10% for over
> 100MB copies (well past even the current threshold). In cases with a
> low number of threads competing for bandwidth, REP MOVSB is ~2x faster
> up to `sizeof_L3`.
>
> The divisor of `4` is a somewhat arbitrary value. From benchmarks it
> seems Skylake and Icelake both prefer a divisor of `2`, but older CPUs
> such as Broadwell prefer something closer to `8`. This patch is meant
> to be followed up by another one to make the divisor cpu-specific, but
> in the meantime (and for easier backporting), this patch settles on
> `4` as a middle-ground.
>
> Benchmarks comparing non-temporal stores, REP MOVSB, and cacheable
> stores where done using:
> https://github.com/goldsteinn/memcpy-nt-benchmarks
>
> Sheets results (also available in pdf on the github):
> https://docs.google.com/spreadsheets/d/e/2PACX-1vS183r0rW_jRX6tG_E90m9qVuFiMbRIJvi5VAE8yYOvEOIEEc3aSNuEsrFbuXw5c3nGboxMmrupZD7K/pubhtml
> ---
>  sysdeps/x86/dl-cacheinfo.h | 70 +++++++++++++++++++++++---------------
>  1 file changed, 43 insertions(+), 27 deletions(-)
>
> diff --git a/sysdeps/x86/dl-cacheinfo.h b/sysdeps/x86/dl-cacheinfo.h
> index ec88945b39..4a1a5423ff 100644
> --- a/sysdeps/x86/dl-cacheinfo.h
> +++ b/sysdeps/x86/dl-cacheinfo.h
> @@ -407,7 +407,7 @@ handle_zhaoxin (int name)
>  }
>
>  static void
> -get_common_cache_info (long int *shared_ptr, unsigned int *threads_ptr,
> +get_common_cache_info (long int *shared_ptr, long int * shared_per_thread_ptr, unsigned int *threads_ptr,
>                  long int core)
>  {
>    unsigned int eax;
> @@ -426,6 +426,7 @@ get_common_cache_info (long int *shared_ptr, unsigned int *threads_ptr,
>    unsigned int family = cpu_features->basic.family;
>    unsigned int model = cpu_features->basic.model;
>    long int shared = *shared_ptr;
> +  long int shared_per_thread = *shared_per_thread_ptr;
>    unsigned int threads = *threads_ptr;
>    bool inclusive_cache = true;
>    bool support_count_mask = true;
> @@ -441,6 +442,7 @@ get_common_cache_info (long int *shared_ptr, unsigned int *threads_ptr,
>        /* Try L2 otherwise.  */
>        level  = 2;
>        shared = core;
> +      shared_per_thread = core;
>        threads_l2 = 0;
>        threads_l3 = -1;
>      }
> @@ -597,29 +599,28 @@ get_common_cache_info (long int *shared_ptr, unsigned int *threads_ptr,
>          }
>        else
>          {
> -intel_bug_no_cache_info:
> -          /* Assume that all logical threads share the highest cache
> -             level.  */
> -          threads
> -            = ((cpu_features->features[CPUID_INDEX_1].cpuid.ebx >> 16)
> -              & 0xff);
> -        }
> -
> -        /* Cap usage of highest cache level to the number of supported
> -           threads.  */
> -        if (shared > 0 && threads > 0)
> -          shared /= threads;
> +       intel_bug_no_cache_info:
> +         /* Assume that all logical threads share the highest cache
> +            level.  */
> +         threads = ((cpu_features->features[CPUID_INDEX_1].cpuid.ebx >> 16)
> +                    & 0xff);
> +
> +         /* Get per-thread size of highest level cache.  */
> +         if (shared_per_thread > 0 && threads > 0)
> +           shared_per_thread /= threads;
> +       }
>      }
>
>    /* Account for non-inclusive L2 and L3 caches.  */
>    if (!inclusive_cache)
>      {
>        if (threads_l2 > 0)
> -        core /= threads_l2;
> +       shared_per_thread += core / threads_l2;
>        shared += core;
>      }
>
>    *shared_ptr = shared;
> +  *shared_per_thread_ptr = shared_per_thread;
>    *threads_ptr = threads;
>  }
>
> @@ -629,6 +630,7 @@ dl_init_cacheinfo (struct cpu_features *cpu_features)
>    /* Find out what brand of processor.  */
>    long int data = -1;
>    long int shared = -1;
> +  long int shared_per_thread = -1;
>    long int core = -1;
>    unsigned int threads = 0;
>    unsigned long int level1_icache_size = -1;
> @@ -649,6 +651,7 @@ dl_init_cacheinfo (struct cpu_features *cpu_features)
>        data = handle_intel (_SC_LEVEL1_DCACHE_SIZE, cpu_features);
>        core = handle_intel (_SC_LEVEL2_CACHE_SIZE, cpu_features);
>        shared = handle_intel (_SC_LEVEL3_CACHE_SIZE, cpu_features);
> +      shared_per_thread = shared;
>
>        level1_icache_size
>         = handle_intel (_SC_LEVEL1_ICACHE_SIZE, cpu_features);
> @@ -672,13 +675,14 @@ dl_init_cacheinfo (struct cpu_features *cpu_features)
>        level4_cache_size
>         = handle_intel (_SC_LEVEL4_CACHE_SIZE, cpu_features);
>
> -      get_common_cache_info (&shared, &threads, core);
> +      get_common_cache_info (&shared, &shared_per_thread, &threads, core);
>      }
>    else if (cpu_features->basic.kind == arch_kind_zhaoxin)
>      {
>        data = handle_zhaoxin (_SC_LEVEL1_DCACHE_SIZE);
>        core = handle_zhaoxin (_SC_LEVEL2_CACHE_SIZE);
>        shared = handle_zhaoxin (_SC_LEVEL3_CACHE_SIZE);
> +      shared_per_thread = shared;
>
>        level1_icache_size = handle_zhaoxin (_SC_LEVEL1_ICACHE_SIZE);
>        level1_icache_linesize = handle_zhaoxin (_SC_LEVEL1_ICACHE_LINESIZE);
> @@ -692,13 +696,14 @@ dl_init_cacheinfo (struct cpu_features *cpu_features)
>        level3_cache_assoc = handle_zhaoxin (_SC_LEVEL3_CACHE_ASSOC);
>        level3_cache_linesize = handle_zhaoxin (_SC_LEVEL3_CACHE_LINESIZE);
>
> -      get_common_cache_info (&shared, &threads, core);
> +      get_common_cache_info (&shared, &shared_per_thread, &threads, core);
>      }
>    else if (cpu_features->basic.kind == arch_kind_amd)
>      {
>        data = handle_amd (_SC_LEVEL1_DCACHE_SIZE);
>        core = handle_amd (_SC_LEVEL2_CACHE_SIZE);
>        shared = handle_amd (_SC_LEVEL3_CACHE_SIZE);
> +      shared_per_thread = shared;
>
>        level1_icache_size = handle_amd (_SC_LEVEL1_ICACHE_SIZE);
>        level1_icache_linesize = handle_amd (_SC_LEVEL1_ICACHE_LINESIZE);
> @@ -715,6 +720,9 @@ dl_init_cacheinfo (struct cpu_features *cpu_features)
>        if (shared <= 0)
>          /* No shared L3 cache.  All we have is the L2 cache.  */
>         shared = core;
> +
> +      if (shared_per_thread <= 0)
> +       shared_per_thread = shared;
>      }
>
>    cpu_features->level1_icache_size = level1_icache_size;
> @@ -730,17 +738,25 @@ dl_init_cacheinfo (struct cpu_features *cpu_features)
>    cpu_features->level3_cache_linesize = level3_cache_linesize;
>    cpu_features->level4_cache_size = level4_cache_size;
>
> -  /* The default setting for the non_temporal threshold is 3/4 of one
> -     thread's share of the chip's cache. For most Intel and AMD processors
> -     with an initial release date between 2017 and 2020, a thread's typical
> -     share of the cache is from 500 KBytes to 2 MBytes. Using the 3/4
> -     threshold leaves 125 KBytes to 500 KBytes of the thread's data
> -     in cache after a maximum temporal copy, which will maintain
> -     in cache a reasonable portion of the thread's stack and other
> -     active data. If the threshold is set higher than one thread's
> -     share of the cache, it has a substantial risk of negatively
> -     impacting the performance of other threads running on the chip. */
> -  unsigned long int non_temporal_threshold = shared * 3 / 4;
> +  /* The default setting for the non_temporal threshold is 1/4 of size
> +     of the chip's cache. For most Intel and AMD processors with an
> +     initial release date between 2017 and 2023, a thread's typical
> +     share of the cache is from 18-64MB. Using the 1/4 L3 is meant to
> +     estimate the point where non-temporal stores begin outcompeting
> +     REP MOVSB. As well the point where the fact that non-temporal
> +     stores are forced back to main memory would already occurred to the
> +     majority of the lines in the copy. Note, concerns about the
> +     entire L3 cache being evicted by the copy are mostly alleviated
> +     by the fact that modern HW detects streaming patterns and
> +     provides proper LRU hints so that the maximum thrashing
> +     capped at 1/associativity. */
> +  unsigned long int non_temporal_threshold = shared / 4;
> +  /* If no ERMS, we use the per-thread L3 chunking. Normal cacheable stores run
> +     a higher risk of actually thrashing the cache as they don't have a HW LRU
> +     hint. As well, there performance in highly parallel situations is
> +     noticeably worse.  */
> +  if (!CPU_FEATURE_USABLE_P (cpu_features, ERMS))
> +    non_temporal_threshold = shared_per_thread * 3 / 4;
>    /* SIZE_MAX >> 4 because memmove-vec-unaligned-erms right-shifts the value of
>       'x86_non_temporal_threshold' by `LOG_4X_MEMCPY_THRESH` (4) and it is best
>       if that operation cannot overflow. Minimum of 0x4040 (16448) because the
> --
> 2.34.1
>

Carlos, any update on reproducing this on ICX?

On 5/15/23 14:29, Noah Goldstein via Libc-alpha wrote:
> Carlos, any update on reproducing this on ICX?
 
Not yet. DJ is helping me test this on a variety of pieces of hardware.

We're also looking at STREAMS benchmark results to make sure we get no
worse on this since it models certain workloads well.

I think we should have more data by the end of the week.

Ignoring the benchmarks for now, technical review...

LGTM.
Reviewed-by: DJ Delorie <dj@redhat.com>


>  static void
> -get_common_cache_info (long int *shared_ptr, unsigned int *threads_ptr,
> +get_common_cache_info (long int *shared_ptr, long int * shared_per_thread_ptr, unsigned int *threads_ptr,
>                  long int core)

Adding a new parameter, ok.

> +  long int shared_per_thread = *shared_per_thread_ptr;

No NULL check, but all callers pass address-of-variable, so OK.

> +      shared_per_thread = core;

Ok.

> @@ -597,29 +599,28 @@ get_common_cache_info (long int *shared_ptr, unsigned int *threads_ptr,
>          }
>        else
>          {
> -intel_bug_no_cache_info:
> -          /* Assume that all logical threads share the highest cache
> -             level.  */
> -          threads
> -            = ((cpu_features->features[CPUID_INDEX_1].cpuid.ebx >> 16)
> -	       & 0xff);
> -        }
> -
> -        /* Cap usage of highest cache level to the number of supported
> -           threads.  */
> -        if (shared > 0 && threads > 0)
> -          shared /= threads;
> +	intel_bug_no_cache_info:
> +	  /* Assume that all logical threads share the highest cache
> +	     level.  */
> +	  threads = ((cpu_features->features[CPUID_INDEX_1].cpuid.ebx >> 16)
> +		     & 0xff);
> +
> +	  /* Get per-thread size of highest level cache.  */
> +	  if (shared_per_thread > 0 && threads > 0)
> +	    shared_per_thread /= threads;
> +	}
>      }

Mostly whitespace, but real change is to modify shared_per_thread
instead of shared.  Ok.

>        if (threads_l2 > 0)
> -        core /= threads_l2;
> +	shared_per_thread += core / threads_l2;
>        shared += core;

shared is still unscaled, shared_per_thread has the scaled version.  Ok.

>      }
>  
>    *shared_ptr = shared;
> +  *shared_per_thread_ptr = shared_per_thread;

Ok.

> @@ -629,6 +630,7 @@ dl_init_cacheinfo (struct cpu_features *cpu_features)
>    /* Find out what brand of processor.  */
>    long int data = -1;
>    long int shared = -1;
> +  long int shared_per_thread = -1;

Ok.

> @@ -649,6 +651,7 @@ dl_init_cacheinfo (struct cpu_features *cpu_features)
>        data = handle_intel (_SC_LEVEL1_DCACHE_SIZE, cpu_features);
>        core = handle_intel (_SC_LEVEL2_CACHE_SIZE, cpu_features);
>        shared = handle_intel (_SC_LEVEL3_CACHE_SIZE, cpu_features);
> +      shared_per_thread = shared;

Ok.

> -      get_common_cache_info (&shared, &threads, core);
> +      get_common_cache_info (&shared, &shared_per_thread, &threads, core);

Ok.

>        shared = handle_zhaoxin (_SC_LEVEL3_CACHE_SIZE);
> +      shared_per_thread = shared;

Ok.

> -      get_common_cache_info (&shared, &threads, core);
> +      get_common_cache_info (&shared, &shared_per_thread, &threads, core);

Ok.

>        shared = handle_amd (_SC_LEVEL3_CACHE_SIZE);
> +      shared_per_thread = shared;

Ok.

> @@ -715,6 +720,9 @@ dl_init_cacheinfo (struct cpu_features *cpu_features)
>        if (shared <= 0)
>          /* No shared L3 cache.  All we have is the L2 cache.  */
>  	shared = core;
> +
> +      if (shared_per_thread <= 0)
> +	shared_per_thread = shared;

Reasonable default, ok.

> @@ -730,17 +738,25 @@ dl_init_cacheinfo (struct cpu_features *cpu_features)
>    cpu_features->level3_cache_linesize = level3_cache_linesize;
>    cpu_features->level4_cache_size = level4_cache_size;
>  
> -  /* The default setting for the non_temporal threshold is 3/4 of one
> -     thread's share of the chip's cache. For most Intel and AMD processors
> -     with an initial release date between 2017 and 2020, a thread's typical
> -     share of the cache is from 500 KBytes to 2 MBytes. Using the 3/4
> -     threshold leaves 125 KBytes to 500 KBytes of the thread's data
> -     in cache after a maximum temporal copy, which will maintain
> -     in cache a reasonable portion of the thread's stack and other
> -     active data. If the threshold is set higher than one thread's
> -     share of the cache, it has a substantial risk of negatively
> -     impacting the performance of other threads running on the chip. */
> -  unsigned long int non_temporal_threshold = shared * 3 / 4;

> +  /* The default setting for the non_temporal threshold is 1/4 of size
> +     of the chip's cache. For most Intel and AMD processors with an
> +     initial release date between 2017 and 2023, a thread's typical
> +     share of the cache is from 18-64MB. Using the 1/4 L3 is meant to
> +     estimate the point where non-temporal stores begin outcompeting
> +     REP MOVSB. As well the point where the fact that non-temporal
> +     stores are forced back to main memory would already occurred to the
> +     majority of the lines in the copy. Note, concerns about the
> +     entire L3 cache being evicted by the copy are mostly alleviated
> +     by the fact that modern HW detects streaming patterns and
> +     provides proper LRU hints so that the maximum thrashing
> +     capped at 1/associativity. */
> +  unsigned long int non_temporal_threshold = shared / 4;

Changing from 3/4 of a scaled amount, to 1/4 of the total amount.  Ok.

> +  /* If no ERMS, we use the per-thread L3 chunking. Normal cacheable stores run
> +     a higher risk of actually thrashing the cache as they don't have a HW LRU
> +     hint. As well, there performance in highly parallel situations is
> +     noticeably worse.  */
> +  if (!CPU_FEATURE_USABLE_P (cpu_features, ERMS))
> +    non_temporal_threshold = shared_per_thread * 3 / 4;

This is the same as the old default, so OK.