Even though this failed with non regression bounds on fishtest, merging
it as is to improve the current situation.
Failed Non Regression:
https://tests.stockfishchess.org/tests/view/6988814db0f3ca5200aafb50
LLR: -2.94 (-2.94,2.94) <-1.75,0.25>
Total: 185792 W: 48109 L: 48565 D: 89118
Ptnml(0-2): 553, 21228, 49811, 20730, 574
Previously we had code similar to this
```
alignas(64) std::uint8_t foo[1024];
const auto* int_ptr = reinterpret_cast<const std::int32_t*>(foo);
```
Which is a strict aliasing violation once we dereference int_ptr, which
invokes UB and causes newer gcc version to emit incorrect code.
fixes https://github.com/official-stockfish/Stockfish/issues/6598
fixes https://github.com/official-stockfish/Stockfish/issues/4932
This violation was introduced in PR #3287 and merged in commit 23c385e.
Due to aggressive compiler optimizations, users may need to include
CXXFLAGS=-fno-strict-aliasing as a build argument to ensure a correctly
functioning binary.
Also thanks to @anematode for writing parts of this and realizing this
strict aliasing violation.
closes https://github.com/official-stockfish/Stockfish/pull/6599
Bench: 2668754
Co-Authored-By: Timothy Herchen <timothy.herchen@gmail.com>
This enables different Stockfish processes that use the same weights to use the
same memory. The approach establishes equivalence by memory content, and is
compatible with NUMA replication. The benefit of sharing is reduced memory usage
and a speedup thanks to improved (inter-process) caching of the network in the
CPUs cache, and thus reduced bandwidth usage to main memory. Even though this
change doesn't benefit a user running a single process, this helps on fishtest
or e.g. for Lichess, when multiple games run concurrently, or multiple
positions are analyzed in parallel.
This concept was probably first introduced in the Monty engine
(https://github.com/official-monty/Monty/pull/62), after a discussion in
https://github.com/official-stockfish/fishtest/issues/2077 on the issue of
memory pressure. Measurements based on Torch
(https://github.com/user-attachments/files/21386224/verbatim.pdf) further
suggested that large gains were possible. Multiple other engines have
adopted this 'verbatim' format as well.
The implementation here adds the flexibility needed for SF, for example, retains
the ability to bundle compressed networks with the binary, to load nets by uci
option, and to distribute the shared nets to the proper NUMA region. This
flexibility comes with a fair amount of complexity in the implementation, such
as OS specific code, and fallback code.
For most users this should be transparent. However, for example, those running
docker containers should ensure the `--ipc` flag is set correctly, and
`--shm-size` is sufficiently large.
The benefits of this patch significantly depend on hardware, with systems with
many cores and a large (O(150MB), the net size) L3 cache benefitting typically
most. On such systems SF speedups (as measured via nps playing games with
large concurrency but just 1 thread) can be 38%, which results in master vs.
patch Elo which gains about 25 Elo.
```
# PLAYER : RATING ERROR POINTS PLAYED (%)
1 shared_memoryPR : 24.8 1.9 39432.0 73728 53
2 master : 0.0 ---- 34296.0 73728 47
```
In a multithreaded setup, where weights are already shared, that benefit is smaller,
for example on the same HW as above, but with 8t for each side.
```
# PLAYER : RATING ERROR POINTS PLAYED (%)
1 shared_memoryPR : 5.2 3.5 9351.0 18432 51
2 master : 0.0 ---- 9081.0 18432 49
```
On fishtest with a typical hardware mix of our contributors, the following was measured:
STC, 60k games
https://tests.stockfishchess.org/tests/view/69074a49ea4b268f1fac236c
Elo: 4.69 ± 1.4 (95%) LOS: 100.0%
Total: 60000 W: 16085 L: 15275 D: 28640
Ptnml(0-2): 154, 6440, 16053, 7148, 205
nElo: 9.38 ± 2.8 (95%) PairsRatio: 1.12
To verify correctness with a single process on a NUMA architecture,
speedtest was used, confirming near equivalence:
```
master: Average (over 10): 296236186
shared_memory: Average (over 10): 295769332
```
Currently, using large pages for the shared network weights is not always possible,
which can lead to a small slowdown (1-2%), in case a single process is run.
closes https://github.com/official-stockfish/Stockfish/pull/6173
No functional change
Co-authored-by: disservin <disservin.social@gmail.com>
Co-authored-by: Joost VandeVondele <Joost.VandeVondele@gmail.com>
Removes some max calls
Some speedup stats, courtesy of @AndyGrant (albeit measured in an alternate implementation)
Dev 749240 nps
Base 748495 nps
Gain 0.100%
289936 games
STC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 203040 W: 52213 L: 52179 D: 98648
Ptnml(0-2): 480, 20722, 59139, 20642, 537
https://tests.stockfishchess.org/tests/view/664805fe6dcff0d1d6b05f2ccloses#5261
No functional change
This introduces clang-format to enforce a consistent code style for Stockfish.
Having a documented and consistent style across the code will make contributing easier
for new developers, and will make larger changes to the codebase easier to make.
To facilitate formatting, this PR includes a Makefile target (`make format`) to format the code,
this requires clang-format (version 17 currently) to be installed locally.
Installing clang-format is straightforward on most OS and distros
(e.g. with https://apt.llvm.org/, brew install clang-format, etc), as this is part of quite commonly
used suite of tools and compilers (llvm / clang).
Additionally, a CI action is present that will verify if the code requires formatting,
and comment on the PR as needed. Initially, correct formatting is not required, it will be
done by maintainers as part of the merge or in later commits, but obviously this is encouraged.
fixes https://github.com/official-stockfish/Stockfish/issues/3608
closes https://github.com/official-stockfish/Stockfish/pull/4790
Co-Authored-By: Joost VandeVondele <Joost.VandeVondele@gmail.com>
deal with the general case
About a 8.6% speedup (for general arch)
Results for 200 tests for each version:
Base Test Diff
Mean 141741 153998 -12257
StDev 2990 3042 3742
p-value: 0.999
speedup: 0.086
closes https://github.com/official-stockfish/Stockfish/pull/4786
No functional change
Also make two get_weight_index() static methods constexpr, for
consistency with the other static get_hash_value() method right above.
Tested for speed by user Torom (thanks).
closes https://github.com/official-stockfish/Stockfish/pull/4708
No functional change
Architecture:
The diagram of the "SFNNv4" architecture:
https://user-images.githubusercontent.com/8037982/153455685-cbe3a038-e158-4481-844d-9d5fccf5c33a.png
The most important architectural changes are the following:
* 1024x2 [activated] neurons are pairwise, elementwise multiplied (not quite pairwise due to implementation details, see diagram), which introduces a non-linearity that exhibits similar benefits to previously tested sigmoid activation (quantmoid4), while being slightly faster.
* The following layer has therefore 2x less inputs, which we compensate by having 2 more outputs. It is possible that reducing the number of outputs might be beneficial (as we had it as low as 8 before). The layer is now 1024->16.
* The 16 outputs are split into 15 and 1. The 1-wide output is added to the network output (after some necessary scaling due to quantization differences). The 15-wide is activated and follows the usual path through a set of linear layers. The additional 1-wide output is at least neutral, but has shown a slightly positive trend in training compared to networks without it (all 16 outputs through the usual path), and allows possibly an additional stage of lazy evaluation to be introduced in the future.
Additionally, the inference code was rewritten and no longer uses a recursive implementation. This was necessitated by the splitting of the 16-wide intermediate result into two, which was impossible to do with the old implementation with ugly hacks. This is hopefully overall for the better.
First session:
The first session was training a network from scratch (random initialization). The exact trainer used was slightly different (older) from the one used in the second session, but it should not have a measurable effect. The purpose of this session is to establish a strong network base for the second session. Small deviations in strength do not harm the learnability in the second session.
The training was done using the following command:
python3 train.py \
/home/sopel/nnue/nnue-pytorch-training/data/nodes5000pv2_UHO.binpack \
/home/sopel/nnue/nnue-pytorch-training/data/nodes5000pv2_UHO.binpack \
--gpus "$3," \
--threads 4 \
--num-workers 4 \
--batch-size 16384 \
--progress_bar_refresh_rate 20 \
--random-fen-skipping 3 \
--features=HalfKAv2_hm^ \
--lambda=1.0 \
--gamma=0.992 \
--lr=8.75e-4 \
--max_epochs=400 \
--default_root_dir ../nnue-pytorch-training/experiment_$1/run_$2
Every 20th net was saved and its playing strength measured against some baseline at 25k nodes per move with pure NNUE evaluation (modified binary). The exact setup is not important as long as it's consistent. The purpose is to sift good candidates from bad ones.
The dataset can be found https://drive.google.com/file/d/1UQdZN_LWQ265spwTBwDKo0t1WjSJKvWY/view
Second session:
The second training session was done starting from the best network (as determined by strength testing) from the first session. It is important that it's resumed from a .pt model and NOT a .ckpt model. The conversion can be performed directly using serialize.py
The LR schedule was modified to use gamma=0.995 instead of gamma=0.992 and LR=4.375e-4 instead of LR=8.75e-4 to flatten the LR curve and allow for longer training. The training was then running for 800 epochs instead of 400 (though it's possibly mostly noise after around epoch 600).
The training was done using the following command:
The training was done using the following command:
python3 train.py \
/data/sopel/nnue/nnue-pytorch-training/data/T60T70wIsRightFarseerT60T74T75T76.binpack \
/data/sopel/nnue/nnue-pytorch-training/data/T60T70wIsRightFarseerT60T74T75T76.binpack \
--gpus "$3," \
--threads 4 \
--num-workers 4 \
--batch-size 16384 \
--progress_bar_refresh_rate 20 \
--random-fen-skipping 3 \
--features=HalfKAv2_hm^ \
--lambda=1.0 \
--gamma=0.995 \
--lr=4.375e-4 \
--max_epochs=800 \
--resume-from-model /data/sopel/nnue/nnue-pytorch-training/data/exp295/nn-epoch399.pt \
--default_root_dir ../nnue-pytorch-training/experiment_$1/run_$run_id
In particular note that we now use lambda=1.0 instead of lambda=0.8 (previous nets), because tests show that WDL-skipping introduced by vondele performs better with lambda=1.0. Nets were being saved every 20th epoch. In total 16 runs were made with these settings and the best nets chosen according to playing strength at 25k nodes per move with pure NNUE evaluation - these are the 4 nets that have been put on fishtest.
The dataset can be found either at ftp://ftp.chessdb.cn/pub/sopel/data_sf/T60T70wIsRightFarseerT60T74T75T76.binpack in its entirety (download might be painfully slow because hosted in China) or can be assembled in the following way:
Get the https://github.com/official-stockfish/Stockfish/blob/5640ad48ae5881223b868362c1cbeb042947f7b4/script/interleave_binpacks.py script.
Download T60T70wIsRightFarseer.binpack https://drive.google.com/file/d/1_sQoWBl31WAxNXma2v45004CIVltytP8/view
Download farseerT74.binpack http://trainingdata.farseer.org/T74-May13-End.7z
Download farseerT75.binpack http://trainingdata.farseer.org/T75-June3rd-End.7z
Download farseerT76.binpack http://trainingdata.farseer.org/T76-Nov10th-End.7z
Run python3 interleave_binpacks.py T60T70wIsRightFarseer.binpack farseerT74.binpack farseerT75.binpack farseerT76.binpack T60T70wIsRightFarseerT60T74T75T76.binpack
Tests:
STC: https://tests.stockfishchess.org/tests/view/6203fb85d71106ed12a407b7
LLR: 2.94 (-2.94,2.94) <0.00,2.50>
Total: 16952 W: 4775 L: 4521 D: 7656
Ptnml(0-2): 133, 1818, 4318, 2076, 131
LTC: https://tests.stockfishchess.org/tests/view/62041e68d71106ed12a40e85
LLR: 2.94 (-2.94,2.94) <0.50,3.00>
Total: 14944 W: 4138 L: 3907 D: 6899
Ptnml(0-2): 21, 1499, 4202, 1728, 22
closes https://github.com/official-stockfish/Stockfish/pull/3927
Bench: 4919707
The new network caused some issues initially due to the very narrow neuron set between the first two FC layers. Necessary changes were hacked together to make it work. This patch is a mature approach to make the affine transform code faster, more readable, and easier to maintain should the layer sizes change again.
The following changes were made:
* ClippedReLU always produces a multiple of 32 outputs. This is about as good of a solution for AffineTransform's SIMD requirements as it can get without a bigger rewrite.
* All self-contained simd helpers are moved to a separate file (simd.h). Inline asm is utilized to work around GCC's issues with code generation and register assignment. See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101693, https://godbolt.org/z/da76fY1n7
* AffineTransform has 2 specializations. While it's more lines of code due to the boilerplate, the logic in both is significantly reduced, as these two are impossible to nicely combine into one.
1) The first specialization is for cases when there's >=128 inputs. It uses a different approach to perform the affine transform and can make full use of AVX512 without any edge cases. Furthermore, it has higher theoretical throughput because less loads are needed in the hot path, requiring only a fixed amount of instructions for horizontal additions at the end, which are amortized by the large number of inputs.
2) The second specialization is made to handle smaller layers where performance is still necessary but edge cases need to be handled. AVX512 implementation for this was ommited by mistake, a remnant from the temporary implementation for the new... This could be easily reintroduced if needed. A slightly more detailed description of both implementations is in the code.
Overall it should be a minor speedup, as shown on fishtest:
passed STC:
LLR: 2.96 (-2.94,2.94) <-0.50,2.50>
Total: 51520 W: 4074 L: 3888 D: 43558
Ptnml(0-2): 111, 3136, 19097, 3288, 128
and various tests shown in the pull request
closes https://github.com/official-stockfish/Stockfish/pull/3663
No functional change
This PR adds an ability to export any currently loaded network.
The export_net command now takes an optional filename parameter.
If the loaded net is not the embedded net the filename parameter is required.
Two changes were required to support this:
* the "architecture" string, which is really just a some kind of description in the net, is now saved into netDescription on load and correctly saved on export.
* the AffineTransform scrambles weights for some architectures and sparsifies them, such that retrieving the index is hard. This is solved by having a temporary scrambled<->unscrambled index lookup table when loading the network, and the actual index is saved for each individual weight that makes it to canSaturate16. This increases the size of the canSaturate16 entries by 6 bytes.
closes https://github.com/official-stockfish/Stockfish/pull/3456
No functional change
This patch allows old x86 CPUs, from AMD K8 (which the x86-64 baseline
targets) all the way down to the Pentium MMX, to benefit from NNUE with
comparable performance hit versus hand-written eval as on more modern
processors.
NPS of the bench with NNUE enabled on a Pentium III 1.13 GHz (using the
MMX code):
master: 38951
this patch: 80586
NPS of the bench with NNUE enabled using baseline x86-64 arch, which is
how linux distros are likely to package stockfish, on a modern CPU
(using the SSE2 code):
master: 882584
this patch: 1203945
closes https://github.com/official-stockfish/Stockfish/pull/2956
No functional change.
despite usage of alignas, the generated (avx2/avx512) code with older compilers needs to use
unaligned loads with older gcc (e.g. confirmed crash with gcc 7.3/mingw on abrok).
Better performance thus requires gcc >= 9 on hardware supporting avx2/avx512
closes https://github.com/official-stockfish/Stockfish/pull/2969
No functional change
This patch ports the efficiently updatable neural network (NNUE) evaluation to Stockfish.
Both the NNUE and the classical evaluations are available, and can be used to
assign a value to a position that is later used in alpha-beta (PVS) search to find the
best move. The classical evaluation computes this value as a function of various chess
concepts, handcrafted by experts, tested and tuned using fishtest. The NNUE evaluation
computes this value with a neural network based on basic inputs. The network is optimized
and trained on the evalutions of millions of positions at moderate search depth.
The NNUE evaluation was first introduced in shogi, and ported to Stockfish afterward.
It can be evaluated efficiently on CPUs, and exploits the fact that only parts
of the neural network need to be updated after a typical chess move.
[The nodchip repository](https://github.com/nodchip/Stockfish) provides additional
tools to train and develop the NNUE networks.
This patch is the result of contributions of various authors, from various communities,
including: nodchip, ynasu87, yaneurao (initial port and NNUE authors), domschl, FireFather,
rqs, xXH4CKST3RXx, tttak, zz4032, joergoster, mstembera, nguyenpham, erbsenzaehler,
dorzechowski, and vondele.
This new evaluation needed various changes to fishtest and the corresponding infrastructure,
for which tomtor, ppigazzini, noobpwnftw, daylen, and vondele are gratefully acknowledged.
The first networks have been provided by gekkehenker and sergiovieri, with the latter
net (nn-97f742aaefcd.nnue) being the current default.
The evaluation function can be selected at run time with the `Use NNUE` (true/false) UCI option,
provided the `EvalFile` option points the the network file (depending on the GUI, with full path).
The performance of the NNUE evaluation relative to the classical evaluation depends somewhat on
the hardware, and is expected to improve quickly, but is currently on > 80 Elo on fishtest:
60000 @ 10+0.1 th 1
https://tests.stockfishchess.org/tests/view/5f28fe6ea5abc164f05e4c4c
ELO: 92.77 +-2.1 (95%) LOS: 100.0%
Total: 60000 W: 24193 L: 8543 D: 27264
Ptnml(0-2): 609, 3850, 9708, 10948, 4885
40000 @ 20+0.2 th 8
https://tests.stockfishchess.org/tests/view/5f290229a5abc164f05e4c58
ELO: 89.47 +-2.0 (95%) LOS: 100.0%
Total: 40000 W: 12756 L: 2677 D: 24567
Ptnml(0-2): 74, 1583, 8550, 7776, 2017
At the same time, the impact on the classical evaluation remains minimal, causing no significant
regression:
sprt @ 10+0.1 th 1
https://tests.stockfishchess.org/tests/view/5f2906a2a5abc164f05e4c5b
LLR: 2.94 (-2.94,2.94) {-6.00,-4.00}
Total: 34936 W: 6502 L: 6825 D: 21609
Ptnml(0-2): 571, 4082, 8434, 3861, 520
sprt @ 60+0.6 th 1
https://tests.stockfishchess.org/tests/view/5f2906cfa5abc164f05e4c5d
LLR: 2.93 (-2.94,2.94) {-6.00,-4.00}
Total: 10088 W: 1232 L: 1265 D: 7591
Ptnml(0-2): 49, 914, 3170, 843, 68
The needed networks can be found at https://tests.stockfishchess.org/nns
It is recommended to use the default one as indicated by the `EvalFile` UCI option.
Guidelines for testing new nets can be found at
https://github.com/glinscott/fishtest/wiki/Creating-my-first-test#nnue-net-tests
Integration has been discussed in various issues:
https://github.com/official-stockfish/Stockfish/issues/2823https://github.com/official-stockfish/Stockfish/issues/2728
The integration branch will be closed after the merge:
https://github.com/official-stockfish/Stockfish/pull/2825https://github.com/official-stockfish/Stockfish/tree/nnue-player-wip
closes https://github.com/official-stockfish/Stockfish/pull/2912
This will be an exciting time for computer chess, looking forward to seeing the evolution of
this approach.
Bench: 4746616
