aboutsummaryrefslogtreecommitdiff
path: root/src/core/NEON/kernels/arm_gemm
AgeCommit message (Collapse)Author
2021-02-25Update int8 A55 kernelsGeorgios Pinitas
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I15a443806a9773bd18b212fe23d9e4d126948be4 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5191 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-02-24Account Ksections correctly for interleaved and hybrid_indirect gemm kernelsGeorgios Pinitas
Resolves: COMPMID-4131 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I6ab15d5dd2e89ccda6acf06e1bef7fd85e31ce54 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5169 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-02-24Use vector resizing instead of temporary assign in gemm_hybrid_indirectGeorgios Pinitas
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: Ic782e37ad50a54a3e94f30b5e826dda1b3e23c7e Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5167 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-02-23Remove unreachable code in gemm_hybrid_quantized for computing k_blockGeorgios Pinitas
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I24981652b93f15d5c17c253b3e2bb0b29c1532ff Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5148 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-02-22Comply with Trademark rules for use of Neon, Arm and MaliSheri Zhang
Full trademarks available in README.md Resolves: COMPMID-4257 Signed-off-by: Sheri Zhang <sheri.zhang@arm.com> Change-Id: Ibfba2adf2eef3449433f467464ebd87d7198474d Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5116 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-02-17Regenerate kernels and update A55 versionsGeorgios Pinitas
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I9eae76c77db03b8806af65729da34ab2d77f95f2 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4965 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-01-18Cycle estimate-based kernel selection for dot product quantized s8/u8 kernelsGeorgios Pinitas
Resolves: COMPMID-3990 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: If840c79209940535450f4ea1cbf6b0ec646a168e Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4866 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-01-17Add missing std headersSang-Hoon Park
Missing std headers - limits, algorithm, cstddef - are added where they have to be. Partially implements: COMPMID-3808 Change-Id: Ia31f75370f8440dcb753e5ac6eb2eac18e9c63f3 Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4861 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-12-23Fix baremetal arm_compute_validation build errorsSiCongLi
* Add -C flag to instruct preprocessor not to strip comments. This is to prevent marker comments like '// fall through' that suppresses certain warnings from being removed. * Fix unused variable warnings. * Add M_PI definition that's missing from certain toolchain standard libraries. Resolves COMPMID-4054 Change-Id: I1d641db668685d4b678f3d0efed84bfe9e630b4b Signed-off-by: SiCongLi <sicong.li@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4692 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2020-12-07COMPMID-3869: Update Sconstruct to support SVE/SVE2Manuel Bottini
Modifying scons to build with SVE/SVE2 Updating the documentation with examples Change-Id: I80875206599d5444b9c21ac75c4a8e4efd30d8b5 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4629 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2020-12-03Update GEMV heuristics for quantized types for A53Georgios Pinitas
Switch assembly kernels to dispatch a 4x4 blocked GEMM kernel for A53 when M <= 4 instead of the 8x12 u16 based one. Resolves: COMPMID-3983 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: Ic46a1b51a7c075e46dcb5cd578c75260ded0540c Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4640 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-11-23COMPMID-3987: Nightly failure - Android builds failing in dataset and validationManuel Bottini
Removing warnings from vector library in GCC 7.1+ Removing warning in wanted switch cases fall throughs GCAccessor moving constructor removed Removing parentheses equality checks in stb_image Small fixes in GEMM test suite Change-Id: I8ba8e3fa20b45c32e5b6219473e0f33ab787ca30 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4483 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-11-23Update tuning numbers for A55 for both fp16 and fp32Georgios Pinitas
Resolves: COMPMID-3974 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I6d5189e44ebeda1575a80dd14ec3a09c75f68e03 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4521 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-11-17COMPMID-3970: Failure when building with GCC < 6Georgios Pinitas
Address pre-N4387 tuple usage Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: Iefe6e08e27b8fe1e688d2ff9db8cb7e172b568f3 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4429 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2020-11-12COMPMID-3776: Indirect GEMMGeorgios Pinitas
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I51a1b0f098bc3a8c408c50c92221e4df3061e12c Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4343 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-09-22COMPMID-3757: (u)int8: Don't select the 16-bit route on A53 for cases with ↵David Mansell
very few rows Also added 2D version of the 16-bit route, and altered the selection heuristic so that 2D mode will be used in cases where 1D mode won't thread well. Change-Id: I0057fde08456771dc0090ac51f50d82f8bb86044 Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3903 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: SiCong Li <sicong.li@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-09-09MLCE-229: Fixed requantization per channel in asm kernelmorgolock
Change-Id: Iaf1465f3144371e153ce123ac00da5cc092f77df Signed-off-by: morgolock <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3939 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2020-08-19MLCE-229: Support for negative shifts in asm kernelsmorgolock
Change-Id: I2c5e98aae7698963f106d7423df0e65cd00ee2a9 Signed-off-by: morgolock <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3710 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-07-24[ONCPUML-120]: Tweak of the launch heuristics for hybrid_u8u32_dot_16x4 kernelAleksandr Nikolaev
Hybrid kernel turns to be faster for qasymm8 than quantized_wrapper with interleaved. Signed-off-by: Aleksandr Nikolaev <aleksandr.nikolaev@arm.com> Change-Id: I200646aee6cdcabfe125b746c7d87bfa7d06e0fc Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3585 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-07-23COMPMID-3578: Update FP32/int8 kernel selection.David Mansell
Upgrade the current 'is_preferred()' mechanism with a new framework, where kernels instead provide an estimated cycle count figure. Compatibility with old mechanism is achieved via a wrapper which replaces a "true" result with an estimate of 0, and a "false" result with UINT64_MAX. This mechanism is then used to select between 'interleaved' and 'hybrid' FP32 NEON kernels. This uses a simple system based on counting MACs performed and bytes of data transferred (for rearrange/merge operations) and dividing by fixed performance figures, which are provided for A53, A55, A73 and 'default' figures (based on A76). Separately, a new route for performing int8 GEMMs by using the int16 kernel is provided. This performs significantly (for uint8) or slightly (for int8) better on A53 than the existing int8 route. Optimized 8-to-16 bit transforms are also included. Change-Id: I53b2e59eb9368793c78c2081e17d2445361bcc47 Signed-off-by: David Mansell <David.Mansell@arm.com> Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/250120 Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: bsgcomp <bsgcomp@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3609 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2020-07-16[ONCPUML-97]: Implement "int8" support for 2D decomposition at high core countsAleksandr Nikolaev
Interleaved2d functionality was extended to uint8 and int8 kernels. Change-Id: If78facbce56e9ec7b2f4c23436af0bd5db7f7b69 Signed-off-by: Aleksandr Nikolaev <aleksandr.nikolaev@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3467 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-07-15[ONCPUML-96] FP16 support for 2D decomposition at high core counts.cfRod
Added changes to gemm_fp16 to pick gemm interleaved pretransposed 2D for hgemm_24x8 and sgemm_12x8. Also added the change for scheduling hints based on datatype F16. Signed-off-by: cfRod <crefeda.rodrigues@arm.com> Change-Id: Idd754cf14b47d00a70eab79bbb5ee3ecaf77450f Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3477 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-07-09COMPMID-3324: Adjusting capitalization of Arm copyright claim to reflect Arm ↵Michele Di Giorgio
preferred presentation Change-Id: Ib7dcfcbb24b408999dfae366b9da396485aacf78 Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3525 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-07-07COMPMID-3324: Remove pretransposed support from NEON backendGeorgios Pinitas
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I394c6c539969940e0119cbc14174909d47e65de6 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3519 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-07-06COMPID-3324: Clean GEMM kernelsGeorgios Pinitas
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I170de1671e061a78740caee31fb4a1b8642c1369 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3505 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2020-07-02COMPMID-3324: Fix per-channel quantization on N blockingGeorgios Pinitas
Direct the column to start from in the quantized code Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I8231e0b541c6b1b76becf349a1d6ddf973ade9e2 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3488 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-06-30COMPMID-3324: Handle unused variable in SVE based GEMM kernels.Georgios Pinitas
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: Ic201433d6c2191c1498390d97dd371e578a081fe Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3480 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-06-17COMPMID-3520: Move ndrange.hpp header from arm_gemm to assemblyMichele Di Giorgio
Change-Id: I6352a520ce38230cdfbad346b176cb659ab242a7 Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3327 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-06-10COMPMID-3324: Remove extra bracketGeorgios Pinitas
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I9e4dedf2c216ee6229674cd4b71fa665afce3097 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3316 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: SiCong Li <sicong.li@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-05-28move ndrange.hpp header into arm_compute directoryVincent ABRIOU
ndrange.hpp file is needed by arm_gemm_compute_iface.hpp when building armnn. In a cross compilation environment like Yocto Openembedded, the installed include files comes from either /arm_compute, /include or /support directories. So to avoid armnn build failure, ndrange.hpp need to be moved in arm_compute/core/NEON/kernels/arm_gemm directory. Other files that need ndrange.hpp to build are updated accordingly. Signed-off-by: Vincent ABRIOU <vincent.abriou@st.com> Change-Id: Icc8e18b23dd06c56e0c49a84d4a51bfd85fe2290 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3271 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2020-05-26COMPMID-3499: Fix integer overflow for large GEMM on NEONSang-Hoon Park
Change-Id: Id9eef3abc8a902b52ba61772f716f2ba2b97f7d4 Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3245 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-05-20COMPMID-3496: Fix for GEMM_INTERLEAVE_2DGian Marco Iodice
GEMM_INTERLEAVE_2D was wrongly selected by the heuristic also in case of maxthreads < 8 Change-Id: If531d44c6f00ae6f8e3a4bf22428829b252bc3d6 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3225 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-05-11COMPMID-3467: Zero-initialize std::array in ndrange.hppGeorgios Pinitas
Fix uninitialized variable warning. Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: Ifeabc68e2ed3bc35b6cb68890716cceea6b519db Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3170 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2020-05-10[ONCPUML-7] arm_compute support for ND parallelismJoseph Dobson
Currently 1D ranges of work are specified by the scheduler via two integers, start and end. This limit opportunities for advance parallelism and scheduling This patch expands the interfaces to allow for ND parallism. `GemmCommon::get_window_size` now returns an `NDRange` specifying the work in N-dimensions rather than with the single integer it used prior (1D) Execute now takes an `NDCoordinate` which specifies an `NDRange` with a start position for that work along with an `NDCoordinate` to specify the thread location In addition to expanding the interface to enable this functionality, we have added the capability to SGEMM when the number of threads is high this has the effective of allowing a much greater degree of parallelism where te problem dimension would previously have limited the number of threads. Change-Id: I3e1a8b7276216627bec4ff6f24ac2147552ea9fb Signed-off-by: Joseph Dobson <joseph.dobson@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2962 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-03-12COMPMID-3160: Add Bfloat16 support in NEGEMMConvolutionLayerGeorgios Pinitas
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I0e449306c138a562ffc1455e76ec44b2fd059d85 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2860 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-01-31COMPMID-3003: Integrate assembly kernels utilizing MMLA instruction.Georgios Pinitas
MMLA is a matrix-multiply instruction introduced on armv8.6-A Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I572a54981d48f5a1e0e9e51102cb7ae28ad87806 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2663 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2020-01-23COMPMID-2923 Integrate arm_gemm per channel quantizationMichalis Spyrou
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Change-Id: I8667e75843fdd6ac75bd8272a86a348b830da28d Reviewed-on: https://review.mlplatform.org/c/2548 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-12-11COMPMID-2841: Enable aarch32 buildsGeorgios Pinitas
Decouples the execution state from the architecture. Now architectures can be set as (armv7a, armv8a, etc) and execution state using the `estate` flag with the following options (auto, 32, 64). Change-Id: Ie7f757b3565495a39c7e20fb350a72fd9c5a2a4f Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2438 Reviewed-by: Pablo Marquez <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-10-28MLCE-146: VTS failures A55Pablo Tello
Change-Id: I924449e9eafc622d6457d2808300ae7b66ce449b Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/2163 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-10-23COMPMID-2577: Fuse bias addition and activation in gemm assembly kernelsGeorgios Pinitas
Change-Id: I7f52112d2d05b1ea3d3f3d4b19b8eafab05d6c44 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2141 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
2019-10-11COMPMID-2697: Dynamic Tensors Type2 test caseGeorgios Pinitas
Change-Id: I8b8ac272e6c49dc51570fb0612d74f29772371e7 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2072 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2019-10-08COMPMID-2486: Remove disabled compiler warningsMichalis Spyrou
Removed -Wno-unused-parameter and -Wno-deprecated-declarations compilation flags. Plus, 3RDPARTY_UPDATE. Change-Id: I43098c7af527d5651aad3c597b508a56f8813dda Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/2041 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-09-30COMPMID-2452: Dot product optimizations on merge/transformsGeorgios Pinitas
-Adds optimized gemm transforms for AArch64. -Optimized gemm merger to only support alpha==1 and (beta==0 || beta=1) cases Change-Id: I55793b12a0381f4fd53f521d0e57416809904d96 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2003 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
2019-09-26COMPMID-2452: Replace DOT assemble macro with the corresponding word directives.Georgios Pinitas
Change-Id: Id602cfe2d00deed6d994ba4c90cdc5914a8e6016 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1987 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
2019-09-25COMPMID-2674: Buffer overrun in GEMM assembly tranform kernels.Georgios Pinitas
Block transforms could access elements out of bounds in case of input sizes smaller than the transform blocks. Fixes the edge cases by short-circuiting the accesses in such cases. Change-Id: I11d172ecd80b4dde46496e9d4b446de7fb9d5dc7 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1976 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
2019-08-23COMPMID-2336: Update GEMM codeMichalis Spyrou
Fix incorrect stride Change-Id: I5bb9660bd73c40d587fa869d2488b7ab5f5b1ea7 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/1807 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2019-08-21COMPMID-2596 MobilenetSSD produce wrong resultsMichalis Spyrou
Update GEMM assembly code. Change-Id: Id315c51a11aa89915727c4d388e9335982216a2d Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/1774 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-07-29COMPMID-2336: Guard quint8 gemm files for aarch64 only.Georgios Pinitas
gemm_quint8 is only supported for 64-bit thus guarding to avoid any build related issues. Change-Id: Id8784dbacc467780318bd340f895a5abbd383182 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1638 Reviewed-by: Pablo Marquez <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-07-26COMPMID-2178: Update GEMM assembly code.Georgios Pinitas
Perform offset reduction and requantization within the assembly wrapper. Change-Id: I5d5b3e1f6f9ef4c71805362c57f88ff199c027a3 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1541 Comments-Addressed: Pablo Marquez <pablo.tello@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-07-23Use .inst directives instead of .int directives.Ramana Radhakrishnan
Has a couple of benefits, one is a disassembler that actually understands dot product will start showing the dot product instruction for what it is rather than just a random .word. For the interested parties in actually why compilers and toolchains manage to disassemble this , please go and look up mapping symbols from toolchains. Secondly .word is a data directive and if you ever have a customer run Arm compute library on big endian, on AArch64 this will not work. This is because data on big endian is well, big endian but the code section is not big endian but just little endian. Admittedly there will be many other things that need to be fixed for big endian to work reliably. Eyeballed satisfactorily with a simple case. If someone could run this through a test run with the CI that would be great. Thanks, Ramana Change-Id: I0b9573ecbed298afc967d675b0542a6fe72b4c52 Signed-off-by: Ramana Radhakrishnan <ramana.radhakrishnan@arm.com> Reviewed-on: https://review.mlplatform.org/c/1588 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
generated by cgit v1.2.1 (git 2.23.0) at 2024-05-28 06:29:40 +0000