jasonp_sf committed [r1074] on Code

begin transforming GPU stage 1 to use structures of arrays and decouple p/q values from their roots (WIP)

jasonp_sf committed [r1073] on Code

minor cleanup

jasonp_sf committed [r1072] on Code

switch size optimization to arbitrary precision throughout; separate ODE solver state; accurately track sign change of Jacobian determinant during homotopy path tracking

jasonp_sf committed [r1071] on Code

add 1 to degree of rotations; generalize objective function to optimize subsets of parameters; add support for degree 4 and 5

jasonp_sf committed [r1070] on Code

remove rectangular sizeopt norm; it was only to compare with poly51opt

jasonp_sf committed [r1069] on Code

adjust parameterization; add Gerbicz collision search; modify final stages of collision search to add radix sorting

jasonp_sf committed [r1068] on Code

add test program for fast collision search

jasonp_sf committed [r1067] on Code

streamline arithmetic

jasonp_sf committed [r1066] on Code

do hashtable-based collision search instead of the last few radix sort passes (25% speedup)

jasonp_sf committed [r1065] on Code

allow CPU stage 1 to batch multiple special-q together; moderately optimize radix sort; refactor control flow

jasonp_sf committed [r1064] on Code

merge control flow between CPU and GPU polyselect stage1, initialize both codebases identically, and make the CPU code use radix sorting instead of a hashtable

jasonp_sf committed [r1063] on Code

fold in changes from specialq64 branch; overhaul CPU poly select stage 1 to use many of the same tools that the GPU code has developed

jasonp_sf committed [r1062] on Code

Add ODE with adaptive stepsize for homotopy predictor; experiment with automatic problem scaling; adjust curve tracking to account for determinant of Jacobian; put initial values in homotopy callback

jasonp_sf committed [r1061] on Code

update include

jasonp_sf committed [r1060] on Code

generalize solver to arbitrary homotopies; add better Newton sequencing, backtracking line search and 2nd order predictor; remove dead code

jasonp_sf committed [r1059] on Code

overhaul degree 6 size optimization (in progress)

jasonp_sf committed [r1058] on Code

refine least squares

jasonp_sf committed [r1057] on Code

size optimization research

BRG committed [r1056] on Code

Update NVIDIA architecture capability code

BRG committed [r1055] on Code

Remove unnecessary CUDA library references

BRG committed [r1054] on Code

NVIDIA CUB is now a part of the 12.5 CUDA toolkit and no longer needs to be added separately

BRG committed [r1053] on Code

Update NVIDIA CUB to the current version 9which is NOW a part of NVIDIA CCCL)

BRG committed [r1052] on Code

update to CUDA 12.5

vegamink created a blog post

Msieve help

vegamink committed [r1051] on Code

Add VBITS info to output.

BRG committed [r1050] on Code

update to Nvidia Cuda 12.3

jasonp_sf committed [r1049] on Code

fix bug in singleton removal

BRG committed [r1048] on Code

BRG committed [r1047] on Code

Update to current version of Nvidia CUB

BRG committed [r1046] on Code

update to Nvidia Cuda 12.0

BRG committed [r1045]

update to CUDA 11.8

BRG committed [r1044]

Update Visual Studio build to Visual Studio 2022 and CUDA 11.6

BRG committed [r1043]

Update GPU kernel loading code to find compute capabilities 5.0 .. 8.6

BRG committed [r1042]

update to CUDA 11.4

BRG committed [r1041]

update to CUDA 11.3

BRG committed [r1040]

correct error in the non-gpu Visual Studio build

BRG committed [r1039]

update to CUDA 11.2

jasonp_sf committed [r1038]

fix buffer overflows caused by array offsets exceeding 2^32

BRG committed [r1037]

update to CUDA 11.2 and Visual Studio 2019

jasonp_sf committed [r1036]

fix MPI block Lanczos

BRG committed [r1035]

Update Visual Studio 2019 build to use Nvidia CUDA 11.1

jasonp_sf committed [r1034]

update changelog

jasonp_sf committed [r1033]

avoid crash with GMP-6.2.0

BRG committed [r1032]

remove Visual Studio *.user file committed in error

BRG committed [r1031]

update to CUDA v11.0

BRG committed [r1030]

add Visual Studio 2019 build files

BRG committed [r1029]

add Visual Studio 2019 build files

BRG committed [r1028]

update for CUDA 10

BRG committed [r1027]

Update the Visual Studio 2017 build to CUDA 10

BRG committed [r1026]

Update Visual Studio 2017 build files to accommodate latest source changes

jasonp_sf committed [r1025]

merge CPU code of msieve-lacuda branch

jasonp_sf committed [r1024]

merge trunk changes

jasonp_sf committed [r1023]

add several polynomial selection tweaks

BRG committed [r1022]

update to CUDA 9.2

jasonp_sf committed [r1021]

make initial maximum merge size configurable

jasonp_sf committed [r1020]

remove exit() call from NFS filtering error path

jasonp_sf committed [r1019]

combine positive and negative intervals; make sieve size variable; add profiling hooks

jasonp_sf committed [r1018]

add fixes for MPI builds

BRG committed [r1017]

add Visual Studio 2017 build (with CUDA 9.0)

BRG committed [r1016]

update to CUDA 9.0

jasonp_sf committed [r1015]

add first wave of MPI changes for large vectors; specify vector size from makefile

jasonp_sf committed [r1014]

make the choice of superblock size account for wider vectors

jasonp_sf committed [r1013]

fix dump interval computations to account for large vectors

jasonp_sf committed [r1012]

add support for wider (128-bit or 256-bit) vectors in the linear algebra

jasonp_sf committed [r1011]

upate CUB library

jasonp_sf committed [r1010]

merge trunk

jasonp_sf committed [r1009]

update makefile to build GPU code for more rece...

jasonp_sf committed [r1008]

start next version

jasonp_sf created a blog post

Msieve v1.53 released

Msieve released /msieve/Msieve v1.53/Readme

Msieve released /msieve/Msieve v1.53/msieve153_win64_gpu.zip

Msieve released /msieve/Msieve v1.53/msieve153_win64.zip

Msieve released /msieve/Msieve v1.53/msieve153_gpu.zip

Msieve released /msieve/Msieve v1.53/msieve153_src.tar.gz

Msieve released /msieve/Msieve v1.53/msieve153.zip

jasonp_sf committed [r1007]

release 1.53

jasonp_sf committed [r1006]

finish v1.53

jasonp_sf committed [r1005]

correct previous commit

jasonp_sf committed [r1004]

fix preprocessor use to avoid gcc complaints

BRG committed [r1003]

improve windows diagnostic messages when GPU so...

BRG committed [r1002]

correct Visual STudio 2015 build (build sort en...

BRG committed [r1001]

test access

BRG committed [r1000]

test access

BRG committed [r999]

test access

BRG committed [r998]

further correction to the Visual Studio 2015 build

BRG committed [r997]

correct error in the output of the subversion n...

BRG committed [r996]

BRG committed [r995]

Remove string subroutines available in Visual S...

BRG committed [r994]

add Visual Studio 2015 build files

BRG committed [r993]

update Visual Studio build for sort engine changes

BRG committed [r992]

update Visual Studio build for sort engine changes

jasonp_sf committed [r991]

cap the number of poly select GPU stage 1 threa...

jasonp_sf committed [r990]

remove debug print

jasonp_sf committed [r989]

replace b40c GPU sort cde with CUB library

jasonp_sf committed [r988]

extend the fix in r981

jasonp_sf committed [r987]

fix buffer overflow in NFS poly scoring for deg...

jasonp_sf committed [r986]

fix r984

jasonp_sf committed [r985]

allow -march=native

jasonp_sf committed [r984]

add an option to go straight to NFS with no oth...

BRG committed [r983]

Further update Visual Studio 2013 build to CUDA...