-
Hi,
Currently with master (commit ID: 9ee68ffaac9af986...) using the tunecache.tsv file from one SM version can cause difficulties with another SM version (e.g using a tuning file from sm20 when run…
bjoo updated
10 years ago
-
Hi guys,
while reviewing your code for a better integration of qdp-jit I notice that you define a bunch of preprocessor macros in 'quda_internal.h', e.g. 'InverterType' etc. Are these necessary? One …
-
mpicxx -Wall -O3 -D__COMPUTE_CAPABILITY__=200 -DMULTI_GPU -DGPU_STAGGERED_DIRAC -DGPU_FATLINK -DGPU_UNITARIZE -DGPU_GAUGE_FORCE -DGPU_DIRECT -DBUILD_QDP_INTERFACE -DBUILD_MILC_INTERF…
-
I tries to run 'su3_leapfrog' and 'su3_rhmc_hisq' using quad 0.7 (136a4ca8d74f6f87f17a286a17a29e5fc6d0130c) and milc from lattice/milc.
It seems the multishift inverter does not converge.
For the hi…
-
Methods as e.g. plsDA, geoDA. Perhaps take a look if it's in general an interesting package. Last update was in November 2013.
Also linDA and quDA are available, but I don't know the difference towar…
-
(paraphrasing message from Alejandro)
Multi-gpu clover is broken for link reconstruction different than 18 in quda-0.7 and tifr-redux branches (therefore in tmclover-quda as well). I don't know why, …
-
Add support for chirally-projected fields, i.e. a full four-component field \psi, stored as two two-component fields (1+\gamma_5)/2 \* \psi and (1-\gamma_5)/2 \* \psi. Or, in QDP notation:
void
QCDP…
-
A technique used by Mathias Wagner is to keep the solution vector in CG always in high precision, this is supposed to help convergence, and may help especially with half precision:
e.g., x += alpha …
-
The lib/Makefile compilation command on line 135 for blas_magma.cpp needs fixing.
detar updated
10 years ago
-
I have noticed that the staggered CG routine in the test suite doesn't always converge monotonically. I've run some full double-precision tests with reliable updating switched off, and I see the resid…