Hello.
I have isolated the issue: it is the fused-multiply-add instruction
set (FMA on Intel processors). Running -march=skylake -mno-fma not
only does not hang, but passes make check-all (using R's native BLAS).
My intuition remains that something in the new more precise ebd0 code
used in dpois_r
This seems entirely avoidable, given that there is a relatively simple
formula for converting 2-ary indices [i,j] of S to 1-ary indices k of
S[lower.tri(S, TRUE)]:
k <- i + round(0.5 * (2L * n - j) * (j - 1L)) # for i >= j
I ought to be slightly more precise here: _coercion_ is avoidable,
be
Hello, Martin et. al.
I apologize for top posting, but I believe I have tracked down the
difference why last time my build worked and now it hangs on
`dchisq(c(Inf, 1e80, 1e50, 1e40), df=10, ncp=1)`. and it's NOT the
BLAS. I built against both 3.15 AND R's vanilla and it hung both
times. The issue
Hi all,
AFAICT https://stat.ethz.ch/R/daily/R-patched.tar.gz is still R 4.0.5 patched.
Probably needs a branch bump. FYI,
Gabor
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Thanks. 'by' is implemented by tapply, and it seems to behave rather
erratically for empty inputs:
> FUN = function(x) x[1,]
> FUNx <- function(x) FUN(df[x, , drop = FALSE])
> tapply(seq_len(10), df$b, FUNx) %>% storage.mode # array
> tapply(seq_len(0), c(), FUNx) %>% storage.mode # logical
> ta
> Gábor Csárdi writes:
> While trying to reproduce a NOTE for
> * checking for new files in some other directories ... NOTE
> I noticed that the check code uses
> Sys.getenv("LOGNAME")
> to query the name of the current user. However on many systems this is
> not set, so this is the empty s
Currently, the class for dense symmetric matrices in packed storage,
"dspMatrix", inherits its subset (i.e., `[`) methods from the "Matrix"
class. As a result, subsetting "dspMatrix" requires coercion to
"matrix". If memory cannot be allocated for this "matrix", then an error
results.
n <- 30