mirror/eigen
2
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-01-06 14:14:46 +08:00
Code Issues Packages Projects Releases Wiki Activity
10,978 Commits 11 Branches 89 Tags 168 MiB
cf7adf3a5d
Commit Graph

2845 Commits

Author SHA1 Message Date
Rasmus Munk Larsen
591a554c68 Add TODO to cleanup FMA cost modelling. 2019-09-24 16:39:25 -07:00
Eugene Zhulenev
c64396b4c6 Choose TensorBlock StridedLinearCopy type statically 2019-09-24 16:04:29 -07:00
Eugene Zhulenev
c97b208468 Add new TensorBlock api implementation + tests 2019-09-24 15:17:35 -07:00
Eugene Zhulenev
ef9dfee7bd Tensor block evaluation V2 support for unary/binary/broadcsting 2019-09-24 12:52:45 -07:00
Christoph Hertzberg
e4c1b3c1d2 Fix implicit conversion warnings and use pnegate to negate packets 2019-09-23 16:07:43 +02:00
Christoph Hertzberg
ba0736fa8e Fix (or mask away) conversion warnings introduced in 553caeb6a3 
.
 2019-09-23 15:58:05 +02:00
Rasmus Munk Larsen
1d5af0693c Add support for asynchronous evaluation of tensor casting expressions. 2019-09-19 13:54:49 -07:00
Rasmus Munk Larsen
6de5ed08d8 Add generic PacketMath implementation of the Error Function (erf). 2019-09-19 12:48:30 -07:00
Srinivas Vasudevan
df0816b71f Merging eigen/eigen. 2019-09-16 19:33:29 -04:00
Srinivas Vasudevan
6e215cf109 Add Bessel functions to SpecialFunctions. 
- Split SpecialFunctions files in to a separate BesselFunctions file.

In particular add:
    - Modified bessel functions of the second kind k0, k1, k0e, k1e
    - Bessel functions of the first kind j0, j1
    - Bessel functions of the second kind y0, y1
 2019-09-14 12:16:47 -04:00
Eugene Zhulenev
bf8866b466 Fix maybe-unitialized warnings in TensorContractionThreadPool 2019-09-13 14:29:55 -07:00
Eugene Zhulenev
553caeb6a3 Use ThreadLocal container in TensorContractionThreadPool 2019-09-13 12:14:44 -07:00
Srinivas Vasudevan
facdec5aa7 Add packetized versions of i0e and i1e special functions. 
- In particular refactor the i0e and i1e code so scalar and vectorized path share code.
  - Move chebevl to GenericPacketMathFunctions.


A brief benchmark with building Eigen with FMA, AVX and AVX2 flags

Before:

CPU: Intel Haswell with HyperThreading (6 cores)
Benchmark                  Time(ns)        CPU(ns)     Iterations
-----------------------------------------------------------------
BM_eigen_i0e_double/1            57.3           57.3     10000000
BM_eigen_i0e_double/8           398            398        1748554
BM_eigen_i0e_double/64         3184           3184         218961
BM_eigen_i0e_double/512       25579          25579          27330
BM_eigen_i0e_double/4k       205043         205042           3418
BM_eigen_i0e_double/32k     1646038        1646176            422
BM_eigen_i0e_double/256k   13180959       13182613             53
BM_eigen_i0e_double/1M     52684617       52706132             10
BM_eigen_i0e_float/1             28.4           28.4     24636711
BM_eigen_i0e_float/8             75.7           75.7      9207634
BM_eigen_i0e_float/64           512            512        1000000
BM_eigen_i0e_float/512         4194           4194         166359
BM_eigen_i0e_float/4k         32756          32761          21373
BM_eigen_i0e_float/32k       261133         261153           2678
BM_eigen_i0e_float/256k     2087938        2088231            333
BM_eigen_i0e_float/1M       8380409        8381234             84
BM_eigen_i1e_double/1            56.3           56.3     10000000
BM_eigen_i1e_double/8           397            397        1772376
BM_eigen_i1e_double/64         3114           3115         223881
BM_eigen_i1e_double/512       25358          25361          27761
BM_eigen_i1e_double/4k       203543         203593           3462
BM_eigen_i1e_double/32k     1613649        1613803            428
BM_eigen_i1e_double/256k   12910625       12910374             54
BM_eigen_i1e_double/1M     51723824       51723991             10
BM_eigen_i1e_float/1             28.3           28.3     24683049
BM_eigen_i1e_float/8             74.8           74.9      9366216
BM_eigen_i1e_float/64           505            505        1000000
BM_eigen_i1e_float/512         4068           4068         171690
BM_eigen_i1e_float/4k         31803          31806          21948
BM_eigen_i1e_float/32k       253637         253692           2763
BM_eigen_i1e_float/256k     2019711        2019918            346
BM_eigen_i1e_float/1M       8238681        8238713             86


After:

CPU: Intel Haswell with HyperThreading (6 cores)
Benchmark                  Time(ns)        CPU(ns)     Iterations
-----------------------------------------------------------------
BM_eigen_i0e_double/1            15.8           15.8     44097476
BM_eigen_i0e_double/8            99.3           99.3      7014884
BM_eigen_i0e_double/64          777            777         886612
BM_eigen_i0e_double/512        6180           6181         100000
BM_eigen_i0e_double/4k        48136          48140          14678
BM_eigen_i0e_double/32k      385936         385943           1801
BM_eigen_i0e_double/256k    3293324        3293551            228
BM_eigen_i0e_double/1M     12423600       12424458             57
BM_eigen_i0e_float/1             16.3           16.3     43038042
BM_eigen_i0e_float/8             30.1           30.1     23456931
BM_eigen_i0e_float/64           169            169        4132875
BM_eigen_i0e_float/512         1338           1339         516860
BM_eigen_i0e_float/4k         10191          10191          68513
BM_eigen_i0e_float/32k        81338          81337           8531
BM_eigen_i0e_float/256k      651807         651984           1000
BM_eigen_i0e_float/1M       2633821        2634187            268
BM_eigen_i1e_double/1            16.2           16.2     42352499
BM_eigen_i1e_double/8           110            110        6316524
BM_eigen_i1e_double/64          822            822         851065
BM_eigen_i1e_double/512        6480           6481         100000
BM_eigen_i1e_double/4k        51843          51843          10000
BM_eigen_i1e_double/32k      414854         414852           1680
BM_eigen_i1e_double/256k    3320001        3320568            212
BM_eigen_i1e_double/1M     13442795       13442391             53
BM_eigen_i1e_float/1             17.6           17.6     41025735
BM_eigen_i1e_float/8             35.5           35.5     19597891
BM_eigen_i1e_float/64           240            240        2924237
BM_eigen_i1e_float/512         1424           1424         485953
BM_eigen_i1e_float/4k         10722          10723          65162
BM_eigen_i1e_float/32k        86286          86297           8048
BM_eigen_i1e_float/256k      691821         691868           1000
BM_eigen_i1e_float/1M       2777336        2777747            256


This shows anywhere from a 50% to 75% improvement on these operations.

I've also benchmarked without any of these flags turned on, and got similar
performance to before (if not better).

Also tested packetmath.cpp + special_functions to ensure no regressions.
 2019-09-11 18:34:02 -07:00
Deven Desai
cdb377d0cb Fix for the HIP build+test errors introduced by the ndtri support. 
The fixes needed are
 * adding EIGEN_DEVICE_FUNC attribute to a couple of funcs (else HIPCC will error out when non-device funcs are called from global/device funcs)
 * switching to using ::<math_func> instead std::<math_func> (only for HIPCC) in cases where the std::<math_func> is not recognized as a device func by HIPCC
 * removing an errant "j" from a testcase (don't know how that made it in to begin with!)
 2019-09-06 16:03:49 +00:00
Eugene Zhulenev
d918bd9a8b Update ThreadLocal to use separate Initialize/Release callables 2019-09-10 16:13:32 -07:00
Eugene Zhulenev
e3dec4dcc1 ThreadLocal container that does not rely on thread local storage 2019-09-09 15:18:14 -07:00
Srinivas Vasudevan
e38dd48a27 PR 681: Add ndtri function, the inverse of the normal distribution function. 2019-08-12 19:26:29 -04:00
Eugene Zhulenev
47fefa235f Allow move-only done callback in TensorAsyncDevice 2019-09-03 17:20:56 -07:00
Eugene Zhulenev
a8d264fa9c Add test for const TensorMap underlying data mutation 2019-09-03 11:38:39 -07:00
Eugene Zhulenev
f68f2bba09 TensorMap constness should not change underlying storage constness 2019-09-03 11:08:09 -07:00
Alberto Luaces
c694be1214 Fixed Tensor documentation formatting. 2019-07-23 09:24:06 +00:00
Eugene Zhulenev
79c402e40e Fix shadow warnings in TensorContractionThreadPool 2019-08-30 15:38:31 -07:00
Eugene Zhulenev
edf2ec28d8 Fix block mapper type name in TensorExecutor 2019-08-30 15:29:25 -07:00
Eugene Zhulenev
f0b36fb9a4 evalSubExprsIfNeededAsync + async TensorContractionThreadPool 2019-08-30 15:13:38 -07:00
Eugene Zhulenev
619cea9491 Revert accidentally removed <memory> header from ThreadPool 2019-08-30 14:51:17 -07:00
Eugene Zhulenev
66665e7e76 Asynchronous expression evaluation with TensorAsyncDevice 2019-08-30 14:49:40 -07:00
Eugene Zhulenev
bc40d4522c Const correctness in TensorMap<const Tensor<T, ...>> expressions 2019-08-28 17:46:05 -07:00
Eugene Zhulenev
6e77f9bef3 Remove shadow warnings in TensorDeviceThreadPool 2019-08-28 10:32:19 -07:00
Rasmus Larsen
84fefdf321 Merged in ezhulenev/eigen-01 (pull request PR-683) 
Asynchronous parallelFor in Eigen ThreadPoolDevice
 2019-08-26 21:49:17 +00:00
maratek
8b5ab0e4dd Fix get_random_seed on Native Client 
Newlib in Native Client SDK does not provide ::random function.
Implement get_random_seed for NaCl using ::rand, similarly to Windows version.
 2019-08-23 15:25:56 -07:00
Eugene Zhulenev
6901788013 Asynchronous parallelFor in Eigen ThreadPoolDevice 2019-08-22 10:50:51 -07:00
Eugene Zhulenev
071311821e Remove XSMM support from Tensor module 2019-08-19 11:44:25 -07:00
Rasmus Munk Larsen
facc4e4536 Disable tests for contraction with output kernels when using libxsmm, which does not support this. 2019-08-07 14:11:15 -07:00
Rasmus Munk Larsen
eab7e52db2 [Eigen] Vectorize evaluation of coefficient-wise functions over tensor blocks if the strides are known to be 1. Provides up to 20-25% speedup of the TF cross entropy op with AVX. 
A few benchmark numbers:

name                              old time/op             new time/op             delta
BM_Xent_16_10000_cpu              448µs ± 3%              389µs ± 2%  -13.21%
(p=0.008 n=5+5)
BM_Xent_32_10000_cpu              575µs ± 6%              454µs ± 3%  -21.00%          (p=0.008 n=5+5)
BM_Xent_64_10000_cpu              933µs ± 4%              712µs ± 1%  -23.71%          (p=0.008 n=5+5)
 2019-08-07 12:57:42 -07:00
Rasmus Munk Larsen
0987126165 Clean up unnecessary namespace specifiers in TensorBlock.h. 2019-08-07 12:12:52 -07:00
Rasmus Munk Larsen
e2999d4c38 Fix performance regressions due to https://bitbucket.org/eigen/eigen/pull-requests/662. 
The change caused the device struct to be copied for each expression evaluation, and caused, e.g., a 10% regression in the TensorFlow multinomial op on GPU:


Benchmark                       Time(ns)        CPU(ns)     Iterations
----------------------------------------------------------------------
BM_Multinomial_gpu_1_100000_4     128173         231326           2922  1.610G items/s

VS

Benchmark                       Time(ns)        CPU(ns)     Iterations
----------------------------------------------------------------------
BM_Multinomial_gpu_1_100000_4     146683         246914           2719  1.509G items/s
 2019-08-02 11:18:13 -07:00
Eugene Zhulenev
3cd148f983 Fix expression evaluation heuristic for TensorSliceOp 2019-07-09 12:10:26 -07:00
Eugene Zhulenev
6083014594 Add outer/inner chipping optimization for chipping dimension specified at runtime 2019-07-03 11:35:25 -07:00
Deven Desai
7eb2e0a95b adding the EIGEN_DEVICE_FUNC attribute to the constCast routine. 
Not having this attribute results in the following failures in the `--config=rocm` TF build.

```
In file included from tensorflow/core/kernels/cross_op_gpu.cu.cc:20:
In file included from ./tensorflow/core/framework/register_types.h:20:
In file included from ./tensorflow/core/framework/numeric_types.h:20:
In file included from ./third_party/eigen3/unsupported/Eigen/CXX11/Tensor:1:
In file included from external/eigen_archive/unsupported/Eigen/CXX11/Tensor:140:
external/eigen_archive/unsupported/Eigen/CXX11/src/Tensor/TensorChipping.h:356:37: error:  'Eigen::constCast':  no overloaded function has restriction specifiers that are compatible with the ambient context 'data'
    typename Storage::Type result = constCast(m_impl.data());
                                    ^
external/eigen_archive/unsupported/Eigen/CXX11/src/Tensor/TensorChipping.h:356:37: error:  'Eigen::constCast':  no overloaded function has restriction specifiers that are compatible with the ambient context 'data'
external/eigen_archive/unsupported/Eigen/CXX11/src/Tensor/TensorAssign.h:148:56: note: in instantiation of member function 'Eigen::TensorEvaluator<const Eigen::TensorChippingOp<1, Eigen::TensorMap<Eigen::Tensor<int, 2, 1, long>, 16, MakePointer> >, Eigen::Gpu\
Device>::data' requested here
    return m_rightImpl.evalSubExprsIfNeeded(m_leftImpl.data());

```

Adding the EIGEN_DEVICE_FUNC attribute resolves those errors
 2019-07-02 20:02:46 +00:00
Gael Guennebaud
ef8aca6a89 Merged in codeplaysoftware/eigen (pull request PR-667) 
[SYCL] :

Approved-by: Gael Guennebaud <g.gael@free.fr>
Approved-by: Rasmus Larsen <rmlarsen@google.com>
 2019-07-02 12:45:23 +00:00
Eugene Zhulenev
4ac93f8edc Allocate non-const scalar buffer for block evaluation with DefaultDevice 2019-07-01 10:55:19 -07:00
Mehdi Goli
9ea490c82c [SYCL] : 
* Modifying TensorDeviceSYCL to use `EIGEN_THROW_X`.
  * Modifying TensorMacro to use `EIGEN_TRY/CATCH(X)` macro.
  * Modifying TensorReverse.h to use `EIGEN_DEVICE_REF` instead of `&`.
  * Fixing the SYCL device macro in SpecialFunctionsImpl.h.
 2019-07-01 16:27:28 +01:00
Eugene Zhulenev
81a03bec75 Fix TensorReverse on GPU with m_stride[i]==0 2019-06-28 15:50:39 -07:00
Rasmus Munk Larsen
74a9dd1102 Fix preprocessor condition to only generate a warning when calling eigen::GpuDevice::synchronize() from device code, but not when calling from a non-GPU compilation unit. 2019-06-28 11:56:21 -07:00
Rasmus Munk Larsen
70d4020ad9 Remove comma causing warning in c++03 mode. 2019-06-28 11:39:45 -07:00
Eugene Zhulenev
6e7c76481a Merge with Eigen head 2019-06-28 11:22:46 -07:00
Eugene Zhulenev
878845cb25 Add block access to TensorReverseOp and make sure that TensorForcedEval uses block access when preferred 2019-06-28 11:13:44 -07:00
Mehdi Goli
7d08fa805a [SYCL] This PR adds the minimum modifications to the Eigen unsupported module required to run it on devices supporting SYCL. 
* Abstracting the pointer type so that both SYCL memory and pointer can be captured.
* Converting SYCL virtual pointer to SYCL device memory in Eigen evaluator class.
* Binding SYCL placeholder accessor to command group handler by using bind method in Eigen evaluator node.
* Adding SYCL macro for controlling loop unrolling.
* Modifying the TensorDeviceSycl.h and SYCL executor method to adopt the above changes.
 2019-06-28 10:08:23 +01:00
Christoph Hertzberg
adec097c61 Remove extra comma (causes warnings in C++03) 2019-06-26 16:14:28 +02:00
Eugene Zhulenev
229db81572 Optimize evaluation strategy for TensorSlicingOp and TensorChippingOp 2019-06-25 15:41:37 -07:00
Rasmus Larsen
c1b0aea653 Merged in Artem-B/eigen (pull request PR-654) 
Minor build improvements

Approved-by: Rasmus Larsen <rmlarsen@google.com>
 2019-05-31 22:27:04 +00:00
Rasmus Munk Larsen
b08527b0c1 Clean up CUDA/NVCC version macros and their use in Eigen, and a few other CUDA build failures. 2019-05-31 15:26:06 -07:00
tra
b4c49bf00e Minor build improvements 
* Allow specifying multiple GPU architectures. E.g.:
  cmake -DEIGEN_CUDA_COMPUTE_ARCH="60;70"
* Pass CUDA SDK path to clang. Without it it will default to /usr/local/cuda
which may not be the right location, if cmake was invoked with
-DCUDA_TOOLKIT_ROOT_DIR=/some/other/CUDA/path
 2019-05-31 14:08:34 -07:00
Michael Tesch
c5019f722b Use pade for matrix exponential also for complex values. 2019-05-08 17:04:55 +02:00
Rasmus Larsen
e92486b8c3 Merged in rmlarsen/eigen (pull request PR-643) 
Make Eigen build with cuda 10 and clang.

Approved-by: Justin Lebar <justin.lebar@gmail.com>
 2019-05-20 17:02:39 +00:00
Eugene Zhulenev
01654d97fa Prevent potential division by zero in TensorExecutor 2019-05-17 14:02:25 -07:00
Eugene Zhulenev
96a276803c Always evaluate Tensor expressions with broadcasting via tiled evaluation code path 2019-05-16 16:15:45 -07:00
Rasmus Munk Larsen
ab0a30e429 Make Eigen build with cuda 10 and clang. 2019-05-15 13:32:15 -07:00
Rasmus Larsen
c8d8d5c0fc Merged in rmlarsen/eigen_threadpool (pull request PR-640) 
Fix deadlocks in thread pool.

Approved-by: Eugene Zhulenev <ezhulenev@google.com>
 2019-05-13 20:04:35 +00:00
Christoph Hertzberg
4ccd1ece92 bug #1707: Fix deprecation warnings, or disable warnings when testing deprecated functions 2019-05-10 14:57:05 +02:00
Rasmus Munk Larsen
e5ac8cbd7a A) fix deadlocks in thread pool caused by EventCount 
This fixed 2 deadlocks caused by sloppiness in the EventCount logic.
Both most likely were introduced by cl/236729920 which includes the new EventCount algorithm:
01da8caf00

bug #1 (Prewait):
Prewait must not consume existing signals.
Consider the following scenario.
There are 2 thread pool threads (1 and 2) and 1 external thread (3). RunQueue is empty.
Thread 1 checks the queue, calls Prewait, checks RunQueue again and now is going to call CommitWait.
Thread 2 checks the queue and now is going to call Prewait.
Thread 3 submits 2 tasks, EventCount signals is set to 1 because only 1 waiter is registered the second signal is discarded).
Now thread 2 resumes and calls Prewait and takes away the signal.
Thread 1 resumes and calls CommitWait, there are no pending signals anymore, so it blocks.
As the result we have 2 tasks, but only 1 thread is running.

bug #2 (CancelWait):
CancelWait must not take away a signal if it's not sure that the signal was meant for this thread.
When one thread blocks and another submits a new task concurrently, the EventCount protocol guarantees only the following properties (similar to the Dekker's algorithm):
(a) the registered waiter notices presence of the new task and does not block
(b) the signaler notices presence of the waiters and wakes it
(c) both the waiter notices presence of the new task and signaler notices presence of the waiter
[it's only that both of them do not notice each other must not be possible, because it would lead to a deadlock]
CancelWait is called for cases (a) and (c). For case (c) it is OK to take the notification signal away, but it's not OK for (a) because nobody queued a signals for us and we take away a signal meant for somebody else.
Consider:
Thread 1 calls Prewait, checks RunQueue, it's empty, now it's going to call CommitWait.
Thread 3 submits 2 tasks, EventCount signals is set to 1 because only 1 waiter is registered the second signal is discarded).
Thread 2 calls Prewait, checks RunQueue, discovers the tasks, calls CancelWait and consumes the pending signal (meant for thread 1).
Now Thread 1 resumes and calls CommitWait, since there are no signals it blocks.
As the result we have 2 tasks, but only 1 thread is running.

Both deadlocks are only a problem if the tasks require parallelism. Most computational tasks do not require parallelism, i.e. a single thread will run task 1, finish it and then dequeue and run task 2.

This fix undoes some of the sloppiness in the EventCount that was meant to reduce CPU consumption by idle threads, because we now have more threads running in these corner cases. But we still don't have pthread_yield's and maybe the strictness introduced by this change will actually help to reduce tail latency because we will have threads running when we actually need them running.



B) fix deadlock in thread pool caused by RunQueue

This fixed a deadlock caused by sloppiness in the RunQueue logic.
Most likely this was introduced with the non-blocking thread pool.
The deadlock only affects workloads that require parallelism.
Most computational tasks don't require parallelism.

PopBack must not fail spuriously. If it does, it can effectively lead to single thread consuming several wake up signals.
Consider 2 worker threads are blocked.
External thread submits a task. One of the threads is woken.
It tries to steal the task, but fails due to a spurious failure in PopBack (external thread submits another task and holds the lock).
The thread executes blocking protocol again (it won't block because NonEmptyQueueIndex is precise and the thread will discover pending work, but it has called PrepareWait).
Now external thread submits another task and signals EventCount again.
The signal is consumed by the first thread again. But now we have 2 tasks pending but only 1 worker thread running.

It may be possible to fix this in a different way: make EventCount::CancelWait forward wakeup signal to a blocked thread rather then consuming it. But this looks more complex and I am not 100% that it will fix the bug.
It's also possible to have 2 versions of PopBack: one will do try_to_lock and another won't. Then worker threads could first opportunistically check all queues with try_to_lock, and only use the blocking version before blocking. But let's first fix the bug with the simpler change.
 2019-05-08 10:16:46 -07:00
Christoph Hertzberg
e6667a7060 Fix stupid shadow-warnings (with old clang versions) 2019-05-07 18:32:19 +02:00
Christoph Hertzberg
e54dc24d62 Restore C++03 compatibility 2019-05-07 18:30:44 +02:00
Rasmus Larsen
ac50afaffa Merged in ezhulenev/eigen-01 (pull request PR-633) 
Check if gpu_assert was overridden in TensorGpuHipCudaDefines
 2019-04-29 16:29:35 +00:00
Eugene Zhulenev
01d7e6ee9b Check if gpu_assert was overridden in TensorGpuHipCudaDefines 2019-04-25 11:19:17 -07:00
Eugene Zhulenev
8ead5bb3d8 Fix doxygen warnings to enable statis code analysis 2019-04-24 12:42:28 -07:00
Rasmus Munk Larsen
144ca33321 Remove deprecation annotation from typedef Eigen::Index Index, as it would generate too many build warnings. 2019-04-24 08:50:07 -07:00
Eugene Zhulenev
a7b7f3ca8a Add missing EIGEN_DEPRECATED annotations to deprecated functions and fix few other doxygen warnings 2019-04-23 17:23:19 -07:00
Anuj Rawat
8c7a6feb8e Adding lowlevel APIs for optimized RHS packet load in TensorFlow 
SpatialConvolution

Low-level APIs are added in order to optimized packet load in gemm_pack_rhs
in TensorFlow SpatialConvolution. The optimization is for scenario when a
packet is split across 2 adjacent columns. In this case we read it as two
'partial' packets and then merge these into 1. Currently this only works for
Packet16f (AVX512) and Packet8f (AVX2). We plan to add this for other
packet types (such as Packet8d) also.

This optimization shows significant speedup in SpatialConvolution with
certain parameters. Some examples are below.

Benchmark parameters are specified as:
Batch size, Input dim, Depth, Num of filters, Filter dim

Speedup numbers are specified for number of threads 1, 2, 4, 8, 16.

AVX512:

Parameters                  | Speedup (Num of threads: 1, 2, 4, 8, 16)
----------------------------|------------------------------------------
128,   24x24,  3, 64,   5x5 |2.18X, 2.13X, 1.73X, 1.64X, 1.66X
128,   24x24,  1, 64,   8x8 |2.00X, 1.98X, 1.93X, 1.91X, 1.91X
 32,   24x24,  3, 64,   5x5 |2.26X, 2.14X, 2.17X, 2.22X, 2.33X
128,   24x24,  3, 64,   3x3 |1.51X, 1.45X, 1.45X, 1.67X, 1.57X
 32,   14x14, 24, 64,   5x5 |1.21X, 1.19X, 1.16X, 1.70X, 1.17X
128, 128x128,  3, 96, 11x11 |2.17X, 2.18X, 2.19X, 2.20X, 2.18X

AVX2:

Parameters                  | Speedup (Num of threads: 1, 2, 4, 8, 16)
----------------------------|------------------------------------------
128,   24x24,  3, 64,   5x5 | 1.66X, 1.65X, 1.61X, 1.56X, 1.49X
 32,   24x24,  3, 64,   5x5 | 1.71X, 1.63X, 1.77X, 1.58X, 1.68X
128,   24x24,  1, 64,   5x5 | 1.44X, 1.40X, 1.38X, 1.37X, 1.33X
128,   24x24,  3, 64,   3x3 | 1.68X, 1.63X, 1.58X, 1.56X, 1.62X
128, 128x128,  3, 96, 11x11 | 1.36X, 1.36X, 1.37X, 1.37X, 1.37X

In the higher level benchmark cifar10, we observe a runtime improvement
of around 6% for AVX512 on Intel Skylake server (8 cores).

On lower level PackRhs micro-benchmarks specified in TensorFlow
tensorflow/core/kernels/eigen_spatial_convolutions_test.cc, we observe
the following runtime numbers:

AVX512:

Parameters                                                     | Runtime without patch (ns) | Runtime with patch (ns) | Speedup
---------------------------------------------------------------|----------------------------|-------------------------|---------
BM_RHS_NAME(PackRhs, 128, 24, 24, 3, 64, 5, 5, 1, 1, 256, 56)  |  41350                     | 15073                   | 2.74X
BM_RHS_NAME(PackRhs, 32, 64, 64, 32, 64, 5, 5, 1, 1, 256, 56)  |   7277                     |  7341                   | 0.99X
BM_RHS_NAME(PackRhs, 32, 64, 64, 32, 64, 5, 5, 2, 2, 256, 56)  |   8675                     |  8681                   | 1.00X
BM_RHS_NAME(PackRhs, 32, 64, 64, 30, 64, 5, 5, 1, 1, 256, 56)  |  24155                     | 16079                   | 1.50X
BM_RHS_NAME(PackRhs, 32, 64, 64, 30, 64, 5, 5, 2, 2, 256, 56)  |  25052                     | 17152                   | 1.46X
BM_RHS_NAME(PackRhs, 32, 256, 256, 4, 16, 8, 8, 1, 1, 256, 56) |  18269                     | 18345                   | 1.00X
BM_RHS_NAME(PackRhs, 32, 256, 256, 4, 16, 8, 8, 2, 4, 256, 56) |  19468                     | 19872                   | 0.98X
BM_RHS_NAME(PackRhs, 32, 64, 64, 4, 16, 3, 3, 1, 1, 36, 432)   | 156060                     | 42432                   | 3.68X
BM_RHS_NAME(PackRhs, 32, 64, 64, 4, 16, 3, 3, 2, 2, 36, 432)   | 132701                     | 36944                   | 3.59X

AVX2:

Parameters                                                     | Runtime without patch (ns) | Runtime with patch (ns) | Speedup
---------------------------------------------------------------|----------------------------|-------------------------|---------
BM_RHS_NAME(PackRhs, 128, 24, 24, 3, 64, 5, 5, 1, 1, 256, 56)  | 26233                      | 12393                   | 2.12X
BM_RHS_NAME(PackRhs, 32, 64, 64, 32, 64, 5, 5, 1, 1, 256, 56)  |  6091                      |  6062                   | 1.00X
BM_RHS_NAME(PackRhs, 32, 64, 64, 32, 64, 5, 5, 2, 2, 256, 56)  |  7427                      |  7408                   | 1.00X
BM_RHS_NAME(PackRhs, 32, 64, 64, 30, 64, 5, 5, 1, 1, 256, 56)  | 23453                      | 20826                   | 1.13X
BM_RHS_NAME(PackRhs, 32, 64, 64, 30, 64, 5, 5, 2, 2, 256, 56)  | 23167                      | 22091                   | 1.09X
BM_RHS_NAME(PackRhs, 32, 256, 256, 4, 16, 8, 8, 1, 1, 256, 56) | 23422                      | 23682                   | 0.99X
BM_RHS_NAME(PackRhs, 32, 256, 256, 4, 16, 8, 8, 2, 4, 256, 56) | 23165                      | 23663                   | 0.98X
BM_RHS_NAME(PackRhs, 32, 64, 64, 4, 16, 3, 3, 1, 1, 36, 432)   | 72689                      | 44969                   | 1.62X
BM_RHS_NAME(PackRhs, 32, 64, 64, 4, 16, 3, 3, 2, 2, 36, 432)   | 61732                      | 39779                   | 1.55X

All benchmarks on Intel Skylake server with 8 cores.
 2019-04-20 06:46:43 +00:00
Rasmus Munk Larsen
039ee52125 Tweak cost model for tensor contraction when parallelizing over the inner dimension. 
https://bitbucket.org/snippets/rmlarsen/MexxLo
 2019-04-12 13:35:10 -07:00
Jonathon Koyle
9a3f06d836 Update TheadPoolDevice example to include ThreadPool creation and passing pointer into constructor. 2019-04-10 10:02:33 -06:00
Deven Desai
66a885b61e adding EIGEN_DEVICE_FUNC to the recently added TensorContractionKernel constructor. Not having the EIGEN_DEVICE_FUNC attribute on it was leading to compiler errors when compiling Eigen in the ROCm/HIP path 2019-04-08 13:45:08 +00:00
Eugene Zhulenev
629ddebd15 Add missing semicolon 2019-04-02 15:04:26 -07:00
Eugene Zhulenev
4e2f6de1a8 Add support for custom packed Lhs/Rhs blocks in tensor contractions 2019-04-01 11:47:31 -07:00
Deven Desai
2dbea5510f Merged eigen/eigen into default 2019-03-19 16:52:38 -04:00
David Tellenbach
bd9c2ae3fd Fix include guard comments 2019-03-15 15:29:17 +01:00
Eugene Zhulenev
001f10e3c9 Fix segfaults with cuda compilation 2019-03-11 09:43:33 -07:00
Eugene Zhulenev
899c16fa2c Fix a bug in TensorGenerator for 1d tensors 2019-03-11 09:42:01 -07:00
Eugene Zhulenev
0f8bfff23d Fix a data race in NonBlockingThreadPool 2019-03-11 09:38:44 -07:00
Gael Guennebaud
2df4f00246 Change license from LGPL to MPL2 with agreement from David Harmon. 2019-03-07 18:17:10 +01:00
Rasmus Munk Larsen
3c3f639fe2 Merge. 2019-03-06 11:54:30 -08:00
Rasmus Munk Larsen
f4ec8edea8 Add macro EIGEN_AVOID_THREAD_LOCAL to make it possible to manually disable the use of thread_local. 2019-03-06 11:52:04 -08:00
Rasmus Munk Larsen
41cdc370d0 Fix placement of "#if defined(EIGEN_GPUCC)" guard region. 
Found with -Wundefined-func-template.

Author: tkoeppe@google.com
 2019-03-06 11:42:22 -08:00
Rasmus Munk Larsen
cc407c9d4d Fix placement of "#if defined(EIGEN_GPUCC)" guard region. 
Found with -Wundefined-func-template.

Author: tkoeppe@google.com
 2019-03-06 11:40:06 -08:00
Eugene Zhulenev
1bc2a0a57c Add missing return to NonBlockingThreadPool::LocalSteal 2019-03-06 10:49:49 -08:00
Eugene Zhulenev
4e4dcd9026 Remove redundant steal loop 2019-03-06 10:39:07 -08:00
Eugene Zhulenev
25abaa2e41 Check that inner block dimension is continuous 2019-03-05 17:34:35 -08:00
Eugene Zhulenev
5d9a6686ed Block evaluation for TensorGeneratorOp 2019-03-05 16:35:21 -08:00
Eugene Zhulenev
a407e022e6 Tune tensor contraction threadpool heuristics 2019-03-05 14:19:59 -08:00
Eugene Zhulenev
56c6373f82 Add an extra check for the RunQueue size estimate 2019-03-05 11:51:26 -08:00
Eugene Zhulenev
b1a8627493 Do not create Tensor<const T> in cxx11_tensor_forced_eval test 2019-03-05 11:19:25 -08:00
Eugene Zhulenev
efb5080d31 Do not initialize invalid fast_strides in TensorGeneratorOp 2019-03-04 16:58:49 -08:00
Eugene Zhulenev
b95941e5c2 Add tiled evaluation for TensorForcedEvalOp 2019-03-04 16:02:22 -08:00
Eugene Zhulenev
694084ecbd Use fast divisors in TensorGeneratorOp 2019-03-04 11:10:21 -08:00
Rasmus Munk Larsen
cf4a1c81fa Fix specialization for conjugate on non-complex types in TensorBase.h. 2019-03-01 14:21:09 -08:00
Rasmus Munk Larsen
6560692c67 Improve EventCount used by the non-blocking threadpool. 
The current algorithm requires threads to commit/cancel waiting in order
they called Prewait. Spinning caused by that serialization can consume
lots of CPU time on some workloads. Restructure the algorithm to not
require that serialization and remove spin waits from Commit/CancelWait.
Note: this reduces max number of threads from 2^16 to 2^14 to leave
more space for ABA counter (which is now 22 bits).
Implementation details are explained in comments.
 2019-02-22 13:56:26 -08:00
Gael Guennebaud
9ac1634fdf Fix conversion warnings 2019-02-19 21:59:53 +01:00
Rasmus Munk Larsen
071629a440 Fix incorrect value of NumDimensions in TensorContraction traits. 
Reported here: #1671
 2019-02-19 10:49:54 -08:00
Rasmus Larsen
efeabee445 Merged in ezhulenev/eigen-01 (pull request PR-590) 
Do not generate no-op cast() and conjugate() expressions
 2019-02-14 21:16:12 +00:00
Eugene Zhulenev
7b837559a7 Fix signed-unsigned return in RuqQueue 2019-02-14 10:40:21 -08:00
Eugene Zhulenev
f0d42d2265 Fix signed-unsigned comparison warning in RunQueue 2019-02-14 10:27:28 -08:00
Eugene Zhulenev
106ba7bb1a Do not generate no-op cast() and conjugate() expressions 2019-02-14 09:51:51 -08:00
Eugene Zhulenev
8c2f30c790 Speedup Tensor ThreadPool RunQueu::Empty() 2019-02-13 10:20:53 -08:00
Eugene Zhulenev
21eb97d3e0 Add PacketConv implementation for non-vectorizable src expressions 2019-02-08 15:47:25 -08:00
Eugene Zhulenev
1e36166ed1 Optimize TensorConversion evaluator: do not convert same type 2019-02-08 15:13:24 -08:00
Steven Peters
953ca5ba2f Spline.h: fix spelling "spang" -> "span" 2019-02-08 06:23:24 +00:00
Eugene Zhulenev
59998117bb Don't do parallel_pack if we can use thread_local memory in tensor contractions 2019-02-07 09:21:25 -08:00
Eugene Zhulenev
8491127082 Do not reduce parallelism too much in contractions with small number of threads 2019-02-04 12:59:33 -08:00
Eugene Zhulenev
eb21bab769 Parallelize tensor contraction only by sharding dimension and use 'thread-local' memory for packing 2019-02-04 10:43:16 -08:00
Gael Guennebaud
d586686924 Workaround lack of support for arbitrary packet-type in Tensor by manually loading half/quarter packets in tensor contraction mapper. 2019-01-30 16:48:01 +01:00
Christoph Hertzberg
a7779a9b42 Hide some annoying unused variable warnings in g++8.1 2019-01-29 16:48:21 +01:00
Christoph Hertzberg
c9825b967e Renaming even more I identifiers 2019-01-26 13:22:13 +01:00
Christoph Hertzberg
934b8a1304 Avoid I as an identifier, since it may clash with the C-header complex.h 2019-01-25 14:54:39 +01:00
Rasmus Munk Larsen
ee550a2ac3 Fix flaky test for tensor fft. 2019-01-16 14:03:12 -08:00
Eugene Zhulenev
1e6d15b55b Fix shorten-64-to-32 warning in TensorContractionThreadPool 2019-01-11 11:41:53 -08:00
Eugene Zhulenev
0abe03764c Fix shorten-64-to-32 warning in TensorContractionThreadPool 2019-01-10 10:27:55 -08:00
Gael Guennebaud
d812f411c3 bug #1654: fix compilation with cuda and no c++11 2019-01-09 18:00:05 +01:00
Eugene Zhulenev
e70ffef967 Optimize evalShardedByInnerDim 2019-01-08 16:26:31 -08:00
Rasmus Munk Larsen
dd6d65898a Fix shorten-64-to-32 warning. Use regular memcpy if num_threads==0. 2018-12-12 14:45:31 -08:00
Gael Guennebaud
cf697272e1 Remove debug code. 2018-12-09 23:05:46 +01:00
Gael Guennebaud
450dc97c6b Various fixes in polynomial solver and its unit tests: 
- cleanup noise in imaginary part of real roots
 - take into account the magnitude of the derivative to check roots.
 - use <= instead of < at appropriate places
 2018-12-09 22:54:39 +01:00
Rasmus Munk Larsen
8a02883d58 Merged in markdryan/eigen/avx512-contraction-2 (pull request PR-554) 
Fix tensor contraction on AVX512 builds

Approved-by: Rasmus Munk Larsen <rmlarsen@google.com>
 2018-12-05 18:19:32 +00:00
Mark D Ryan
36f8f6d0be Fix evalShardedByInnerDim for AVX512 builds 
evalShardedByInnerDim ensures that the values it passes for start_k and
end_k to evalGemmPartialWithoutOutputKernel are multiples of 8 as the kernel
does not work correctly when the values of k are not multiples of the
packet_size.  While this precaution works for AVX builds, it is insufficient
for AVX512 builds where the maximum packet size is 16.  The result is slightly
incorrect float32 contractions on AVX512 builds.

This commit fixes the problem by ensuring that k is always a multiple of
the packet_size if the packet_size is > 8.
 2018-12-05 12:29:03 +01:00
Christoph Hertzberg
0ec8afde57 Fixed most conversion warnings in MatrixFunctions module 2018-11-20 16:23:28 +01:00
Deven Desai
e7e6809e6b ROCm/HIP specfic fixes + updates 
1. Eigen/src/Core/arch/GPU/Half.h

   Updating the HIPCC implementation half so that it can declared as a __shared__ variable


2. Eigen/src/Core/util/Macros.h, Eigen/src/Core/util/Memory.h

   introducing a EIGEN_USE_STD(func) macro that calls
   - std::func be default
   - ::func when eigen is being compiled with HIPCC

   This change was requested in the previous HIP PR
   (https://bitbucket.org/eigen/eigen/pull-requests/518/pr-with-hip-specific-fixes-for-the-eigen/diff)


3. unsupported/Eigen/CXX11/src/Tensor/TensorDeviceThreadPool.h

   Removing EIGEN_DEVICE_FUNC attribute from pure virtual methods as it is not supported by HIPCC


4. unsupported/Eigen/CXX11/src/Tensor/TensorReduction.h

   Disabling the template specializations of InnerMostDimReducer as they run into HIPCC link errors
 2018-11-19 18:13:59 +00:00
Rasmus Munk Larsen
72928a2c8a Merged in rmlarsen/eigen2 (pull request PR-543) 
Add parallel memcpy to TensorThreadPoolDevice in Eigen, but limit the number of threads to 4, beyond which we just seem to be wasting CPU cycles as the threads contend for memory bandwidth.

Approved-by: Eugene Zhulenev <ezhulenev@google.com>
 2018-11-13 17:10:30 +00:00
Rasmus Munk Larsen
cda479d626 Remove accidental changes. 2018-11-12 18:34:04 -08:00
Rasmus Munk Larsen
719d9aee65 Add parallel memcpy to TensorThreadPoolDevice in Eigen, but limit the number of threads to 4, beyond which we just seem to be wasting CPU cycles as the threads contend for memory bandwidth. 2018-11-12 17:46:02 -08:00
Rasmus Munk Larsen
93f9988a7e A few small fixes to a) prevent throwing in ctors and dtors of the threading code, and b) supporting matrix exponential on platforms with 113 bits of mantissa for long doubles. 2018-11-09 14:15:32 -08:00
Rasmus Munk Larsen
07fcdd1438 Merged in ezhulenev/eigen-02 (pull request PR-534) 
Fix cxx11_tensor_{block_access, reduction} tests
 2018-10-25 18:34:35 +00:00
Eugene Zhulenev
8a977c1f46 Fix cxx11_tensor_{block_access, reduction} tests 2018-10-25 11:31:29 -07:00
Christoph Hertzberg
449ff74672 Fix most Doxygen warnings. Also add links to stable documentation from unsupported modules (by using the corresponding Doxytags file). 
Manually grafted from d107a371c6
 2018-10-19 21:10:28 +02:00
Christoph Hertzberg
40fa6f98bf bug #1606: Explicitly set the standard before find_package(StandardMathLibrary). Also replace EIGEN_COMPILER_SUPPORT_CXX11 in favor of EIGEN_COMPILER_SUPPORT_CPP11. 
Grafted manually from a4afa90d16
 2018-10-19 17:20:51 +02:00
Rasmus Munk Larsen
dda68f56ec Fix GPU build due to gpu_assert not always being defined. 2018-10-18 16:29:29 -07:00
Eugene Zhulenev
9e96e91936 Move from rvalue arguments in ThreadPool enqueue* methods 2018-10-16 16:48:32 -07:00
Eugene Zhulenev
217d839816 Reduce thread scheduling overhead in parallelFor 2018-10-16 14:53:06 -07:00
Rasmus Munk Larsen
d52763bb4f Merged in ezhulenev/eigen-02 (pull request PR-528) 
[TensorBlockIO] Check if it's allowed to squeeze inner dimensions

Approved-by: Rasmus Munk Larsen <rmlarsen@google.com>
 2018-10-16 15:39:40 +00:00
Eugene Zhulenev
900c7c61bb Check if it's allowed to squueze inner dimensions in TensorBlockIO 2018-10-15 16:52:33 -07:00
Gael Guennebaud
f0fb95135d Iterative solvers: unify and fix handling of multiple rhs. 
m_info was not properly computed and the logic was repeated in several places.
 2018-10-15 23:47:46 +02:00
Gael Guennebaud
2747b98cfc DGMRES: fix null rhs, fix restart, fix m_isDeflInitialized for multiple solve 2018-10-15 23:46:00 +02:00
Gael Guennebaud
d835a0bf53 relax number of iterations checks to avoid false negatives 2018-10-15 10:23:32 +02:00
Gael Guennebaud
8214cf1896 Make sparse_basic includable from sparse_extra, but disable it since sparse_basic(DynamicSparseMatrix) does not compile at all anyways 2018-10-11 10:27:23 +02:00
Christoph Hertzberg
3f2c8b7ff0 Fix a lot of Doxygen warnings in Tensor module 2018-10-09 20:22:47 +02:00
Gael Guennebaud
93a6192e98 fix mpreal for mpfr<4.0.0 2018-10-09 09:15:22 +02:00
Rasmus Munk Larsen
d16634c4d4 Fix out-of bounds access in TensorArgMax.h. 2018-10-08 16:41:36 -07:00
Rasmus Munk Larsen
1a737e1d6a Fix contraction test. 2018-10-08 16:37:07 -07:00
Gael Guennebaud
2eda9783de typo 2018-10-08 21:37:46 +02:00
Gael Guennebaud
6cc9b2c831 fix warning in mpreal.h 2018-10-08 18:25:37 +02:00
Gael Guennebaud
e29bfe8479 Update included mpreal header to 3.6.5 and fix deprecated warnings. 2018-10-08 17:09:23 +02:00
Gael Guennebaud
64b1a15318 Workaround stupid warning 2018-10-08 12:01:18 +02:00