- Compiler TLS is now used only used when the compiler supports it
- If compiler TLS is unsupported, we use platform-specific TLS
- Only one variable (an index) is now in TLS
- We only access TLS once per alloc, and never when freeing
- Allocation / release info is now stored within the allocation itself, by
over-allocating; this saves having external structures do the bookkeeping, and
reduces some of the redundant data that was being stored (such as addresses)
- We never hit the alloc lock when not using SMP or when using OpenMP (that was
my fault)
- Now that there are fewer tracking structures I think this is a bit easier to
read than before
a few places in the gemm scheduler code were missing barriers;
the code likely worked OK due to heavy use of volatile / _Atomic
but there's no reason to get this incorrect
The use of _Atomic leads to really bad code generation in the compiler
(on x86, you get 2 "mfence" memory barriers around each access with gcc8, despite
x86 being ordered and cache coherent). But there's a fallback in the code that
just uses volatile which is more than plenty in practice.
If we're nervous about cross thread synchronization for these variables, we should
make the YIELD function be a compiler/memory barrier instead.
performance before (after last commit)
Matrix SGEMM cycles MPC DGEMM cycles MPC
48 x 48 10630.0 10.6 0.7% 18112.8 6.2 -0.7%
64 x 64 20374.8 13.0 1.9% 40487.0 6.5 0.4%
65 x 65 141955.2 1.9 -428.3% 146708.8 1.9 -179.2%
80 x 80 178921.1 2.9 -369.6% 186032.7 2.8 -156.6%
96 x 96 205436.2 4.3 -233.4% 224513.1 3.9 -97.0%
112 x 112 244408.2 5.8 -162.7% 262158.7 5.4 -47.1%
128 x 128 321334.5 6.5 -141.3% 333829.0 6.3 -29.2%
Performance with this patch (roughly a 2x improvement):
Matrix SGEMM cycles MPC DGEMM cycles MPC
48 x 48 10756.0 10.5 -0.5% 18296.7 6.1 -1.7%
64 x 64 20490.0 12.9 1.4% 40615.0 6.5 0.0%
65 x 65 83528.3 3.3 -210.9% 96319.0 2.9 -83.3%
80 x 80 101453.5 5.1 -166.3% 128021.7 4.0 -76.6%
96 x 96 149795.1 5.9 -143.1% 168059.4 5.3 -47.4%
112 x 112 191481.2 7.3 -105.8% 204165.0 6.9 -14.6%
128 x 128 265019.2 7.9 -99.0% 272006.4 7.7 -5.3%
The jobs array is getting initialized in O(compiled cpus^2) complexity.
Distros and people with bigger systems will use pretty high values
(128 or 256 or more) for this value, leading to interesting bubbles
in performance.
Baseline (single threaded performance) gets roughly 13 - 15 multiplications per cycle
in the interesting range (threading kicks in at 65x65 mult by 65x65).
The hardware is capable of 32 multiplications per cycle theoretically.
Matrix SGEMM cycles MPC DGEMM cycles MPC
48 x 48 10703.9 10.6 0.0% 17990.6 6.3 0.0%
64 x 64 20778.4 12.8 0.0% 40629.2 6.5 0.0%
65 x 65 26869.9 10.3 0.0% 52545.7 5.3 0.0%
80 x 80 38104.5 13.5 0.0% 72492.7 7.1 0.0%
96 x 96 61626.4 14.4 0.0% 113983.8 7.8 0.0%
112 x 112 91803.8 15.3 0.0% 180987.3 7.8 0.0%
128 x 128 133161.4 15.8 0.0% 258374.3 8.1 0.0%
When threading is turned on
TARGET=SKYLAKEX F_COMPILER=GFORTRAN SHARED=1 DYNAMIC_THREADS=1 USE_OPENMP=0 NUM_THREADS=128
Matrix SGEMM cycles MPC DGEMM cycles MPC
48 x 48 10725.9 10.5 -0.2% 18134.9 6.2 -0.8%
64 x 64 20500.6 12.9 1.3% 40929.1 6.5 -0.7%
65 x 65 2040832.1 0.1 -7495.2% 2097633.6 0.1 -3892.0%
80 x 80 2063129.1 0.2 -5314.4% 2119925.2 0.2 -2824.3%
96 x 96 2070374.5 0.4 -3259.6% 2173604.4 0.4 -1806.9%
112 x 112 2111721.5 0.7 -2169.6% 2263330.8 0.6 -1170.0%
128 x 128 2276181.5 0.9 -1609.3% 2377228.9 0.9 -820.1%
There is a deep deep cliff once you hit 65x65
With this patch
Matrix SGEMM cycles MPC DGEMM cycles MPC
48 x 48 10630.0 10.6 0.7% 18112.8 6.2 -0.7%
64 x 64 20374.8 13.0 1.9% 40487.0 6.5 0.4%
65 x 65 141955.2 1.9 -428.3% 146708.8 1.9 -179.2%
80 x 80 178921.1 2.9 -369.6% 186032.7 2.8 -156.6%
96 x 96 205436.2 4.3 -233.4% 224513.1 3.9 -97.0%
112 x 112 244408.2 5.8 -162.7% 262158.7 5.4 -47.1%
128 x 128 321334.5 6.5 -141.3% 333829.0 6.3 -29.2%
The cliff is very significantly reduced.
(more to follow)
* Allow choosing the OpenMP scheduler and add range hint for GEMM_MULTITHREAD_THRESHOLD
* Amended description of GEMM_MULTITHREAD_THRESHOLD
to reflect #742 making it track floating point operations rather than matrix size
This patch adds the basic infrastructure for adding the SkylakeX (Intel Skylake server)
target. The SkylakeX target will use the AVX512 (AVX512VL level) instruction set,
which brings 2 basic things:
1) 512 bit wide SIMD (2x width of AVX2)
2) 32 SIMD registers (2x the number on AVX2)
This initial patch only contains a trivial transofrmation of the Haswell SGEMM kernel
to AVX512VL; more will follow later but this patch aims to get the infrastructure
in place for this "later".
Full performance tuning has not been done yet; with more registers and wider SIMD
it's in theory possible to retune the kernels but even without that there's an
interesting enough performance increase (30-40% range) with just this change.
For #1416 and #1529, more information about the options OpenBLAS was built with is needed. Additionally we may want to add this data to the openblas.pc file (but not all projects use pkgconfig, and as far as I am aware the cmake module for accessing it does not make such "private" declarations available)
Even if we're directly using the win32 threading driver and not pthreads,
pthread_atfork still works fine to register a pre-fork handler, and is
necessary to restore the threading server to a pre-initialized state.
Martin Kroeker
Craig Donner
oon3m0oo
Arjan van de Ven
Matthew Brett
zhiyong.dang
Zhang Xianyi
Alex Arslan
Erik M. Bray
Andrew