Benoit Jacob
02babb9c0f
Provide a empirical lookup table for blocking sizes measured on a Nexus 5. Only for float, only for Android on ARM 32bit for now.
2015-03-15 18:13:12 -04:00
Benoit Jacob
3589a9c115
actual_panel_rows computation should always be resilient to parameters not consistent with the known L1 cache size, see comment
2015-03-15 18:12:18 -04:00
Benoit Jacob
1dd3d89818
Fix a unused-var warning
2015-03-15 18:07:19 -04:00
Benoit Jacob
e56aabf205
Refactor computeProductBlockingSizes to make room for the possibility of using lookup tables
2015-03-15 18:05:12 -04:00
Benoit Jacob
488c15615a
organize a little our default cache sizes, and use a saner default L1 outside of x86 (10% faster on Nexus 5)
2015-03-13 14:51:26 -07:00
Gael Guennebaud
1330f8bbd1
bug #973 , improve AVX support by enabling vectorization of Vector4i-like types, and enforcing alignement of Vector4f/Vector2d-like types to preserve compatibility with SSE and future Eigen versions that will vectorize them with AVX enabled.
2015-03-13 21:15:50 +01:00
Gael Guennebaud
d99ab35f9e
Fix internal::random(x,y) for integer types. The previous implementation could return y+1. The new implementation uses rejection sampling to get an unbiased behabior.
2015-03-13 21:12:46 +01:00
Doug Kwan
657407227e
Fix bug in pdiv<Packet1cd> which swaps 32-bit halves of a pair of
...
doubles instead of swapping the doubles.
2015-03-11 15:13:37 -07:00
Deanna Hood
f89fcefa79
Add hyperbolic trigonometric functions from std array support
2015-03-11 13:13:30 +10:00
Deanna Hood
a5e49976f5
Add log10 array support
2015-03-11 08:56:42 +10:00
Deanna Hood
19a71056ae
Allow calling of square(array) in addition to array.square()
2015-03-11 06:59:28 +10:00
Deanna Hood
31fdd67756
Additional unary coeff-wise functors (isnan, round, arg, e.g.)
2015-03-11 06:39:23 +10:00
Gael Guennebaud
0ee391863e
Avoid undeflow when blocking size are tuned manually.
2015-03-06 21:51:09 +01:00
Gael Guennebaud
14a5f135a3
bug #969 : workaround abiguous calls to Ref using enable_if.
2015-03-06 17:51:31 +01:00
Gael Guennebaud
87681e508f
bug #978 : early return for vanishing products
2015-03-06 16:11:22 +01:00
Gael Guennebaud
cd3bbffa73
Improve blocking heuristic: if the lhs fit within L1, then block on the rhs in L1 (allows to keep packed rhs in L1)
2015-03-06 14:31:39 +01:00
Gael Guennebaud
58740ce4c6
Improve product kernel: replace the previous dynamic loop swaping strategy by a more general one:
...
It consists in increasing the actual number of rows of lhs's micro horizontal panel for small depth such that L1 cache is fully exploited.
2015-03-06 10:30:35 +01:00
Gael Guennebaud
7550107028
Product optimization: implement a dynamic loop-swapping startegy to improve memory accesses to the destination matrix in the case of K-rank-update like products, i.e., for products of the kind: "large x small" * "small x large"
2015-03-05 10:03:46 +01:00
Benoit Steiner
0196141938
Fixed the optimized AVX implementation of the fast rsqrt function
2015-03-02 13:49:39 -08:00
Benoit Steiner
4fd7f47692
Added an optimized version of rsqrt for SSE and AVX that is used when EIGEN_FAST_MATH is defined.
2015-03-02 09:38:47 -08:00
Benoit Steiner
fb53384b0f
Improved the default implementation of prsqrt
2015-02-28 01:51:26 -08:00
Benoit Steiner
306fceccbe
Pulled latest updates from trunk
2015-02-27 13:05:26 -08:00
Benoit Steiner
2386fc8528
Added support for 32bit index on a per tensor/tensor expression. This enables us to use 32bit indices to evaluate expressions on GPU faster while keeping the ability to use 64 bit indices to manipulate large tensors on CPU in the same binary.
2015-02-27 12:57:13 -08:00
Benoit Jacob
6466fa63be
Reimplement the selection between rotating and non-rotating kernels
...
using templates instead of macros and if()'s.
That was needed to fix the build of unit tests on ARM, which I had
broken. My bad for not testing earlier.
2015-02-27 15:30:10 -05:00
Benoit Steiner
05089aba75
Switch to truncated casting when converting floating point types to integer. This ensures that vectorized casts are consistent with scalar casts
2015-02-27 09:27:30 -08:00
Benoit Steiner
573b377110
Added support for vectorized type casting of tensors
2015-02-27 08:46:04 -08:00
Benoit Jacob
2fc3b484d7
remove trailing comma
2015-02-27 11:37:45 -05:00
Benoit Jacob
33669348c4
Disable Packet2f/2i halfpacket support in NEON.
...
I believe that it was erroneously turned on, since Packet2f/2i intrinsics are unimplemented,
and code trying to use halfpackets just fails to compile on NEON, as it tries to use the
default implementation of pload/pstore and the types don't match.
2015-02-27 11:35:37 -05:00
Benoit Jacob
b7fc8746e0
Replace a static assert by a runtime one, fixes the build of unit tests on ARM
...
Also safely assert in the non-implemented path that should never be taken in practice,
and would return wrong results.
2015-02-27 10:01:59 -05:00
Benoit Steiner
f41b1f1666
Added support for fast reciprocal square root computation.
2015-02-26 09:42:41 -08:00
Gael Guennebaud
bcf9bb5c1f
Avoid packing rhs multiple-times when blocking on the lhs only.
2015-02-26 17:01:33 +01:00
Gael Guennebaud
4ec3f04b3a
Make sure that the block size computation is tested by our unit test.
2015-02-26 17:00:36 +01:00
Gael Guennebaud
a8ad8887bf
Implement a more generic blocking-size selection algorithm. See explanations inlines.
...
It performs extremely well on Haswell. The main issue is to reliably and quickly find the
actual cache size to be used for our 2nd level of blocking, that is: max(l2,l3/nb_core_sharing_l3)
2015-02-26 16:04:35 +01:00
Gael Guennebaud
400becc591
Fix typos in block-size testing code, and set peeling on k to 8.
2015-02-26 15:57:06 +01:00
Benoit Jacob
692136350b
So I extensively measured the impact of the offset in this prefetch. I tried offset values from 0 to 128 (on this float* pointer, so implicitly times 4 bytes).
...
On x86, I tested a Sandy Bridge with AVX with 12M cache and a Haswell with AVX+FMA with 6M cache on MatrixXf sizes up to 2400.
I could not see any significant impact of this offset.
On Nexus 5, the offset has a slight effect: values around 32 (times sizeof float) are worst. Anything else is the same: the current 64 (8*pk), or... 0.
So let's just go with 0!
Note that we needed a fix anyway for not accounting for the value of RhsProgress. 0 nicely avoids the issue altogether!
2015-02-25 12:37:14 -05:00
Christoph Hertzberg
531fa9de77
bug #970 : Add EIGEN_DEVICE_FUNC to RValue functions, in case Cuda supports RValue-references.
2015-02-24 21:03:28 +01:00
Benoit Jacob
26275b250a
Fix my recent prefetch changes:
...
- the first prefetch is actually harmful on Haswell with FMA,
but it is the most beneficial on ARM.
- the second prefetch... I was very stupid and multiplied by sizeof(scalar)
and offset of a scalar* pointer. The old offset was 64; pk = 8, so 64=pk*8.
So this effectively restores the older offset. Actually, there were
two prefetches here, one with offset 48 and one with offset 64. I could not
confirm any benefit from this strange 48 offset on either the haswell or
my ARM device.
2015-02-23 16:55:17 -05:00
Christoph Hertzberg
ecbf2a6656
log1p is defined only for real Scalars in C++11
2015-02-21 19:58:24 +01:00
Gael Guennebaud
3cf642baa3
Fix compilation of unit tests disabling assertion cheking
2015-02-21 14:13:48 +01:00
Gael Guennebaud
2da1594750
Fix doc of Ref<>
2015-02-20 11:52:22 +01:00
Gael Guennebaud
b192e29eae
In C++11 destructors do not throw by default (fix CommaInitializer unit test)
2015-02-20 09:28:34 +01:00
Benoit Steiner
ab41652d81
Pulled latest changes from trunk
2015-02-19 21:23:37 -08:00
Benoit Steiner
7765039f1c
Marked the CUDA packet primitives as EIGEN_DEVICE_FUNC since they'll end up being executed on the GPU device.
2015-02-19 21:22:51 -08:00
Gael Guennebaud
a66f5fc2fd
Fix regression with C++11 support of lambda: now internal::result_of falls back to std::result_of in C++11.
2015-02-19 23:32:12 +01:00
Gael Guennebaud
1b7e12847d
Fix some calls to result_of on binary functors as unary ones.
2015-02-19 23:30:41 +01:00
Gael Guennebaud
0f4dd15dfc
Declare const some const variables
2015-02-19 23:28:57 +01:00
Gael Guennebaud
829dddd0fd
Add support for C++11 result_of/lambdas
2015-02-19 15:18:37 +01:00
Benoit Jacob
db05f2d01e
rotating kernel: avoid compiling anything outside of ARM
2015-02-18 15:43:52 -05:00
Benoit Jacob
0ed00d5438
remove a newly introduced redundant typedef - sorry.
2015-02-18 15:05:01 -05:00
Benoit Jacob
9bd8a4bab5
bug #955 - Implement a rotating kernel alternative in the 3px4 gebp path
...
This is substantially faster on ARM, where it's important to minimize the number of loads.
This is specific to the case where all packet types are of size 4. I made my best attempt to minimize how dirty this is... opinions welcome.
Eventually one could have a generic rotated kernel, but it would take some work to get there. Also, on sandy bridge, in my experience, it's not beneficial (even about 1% slower).
2015-02-18 15:03:35 -05:00
