[Rd] Workaround very slow NAN/Infinities arithmetic?

brodie gaslam brod|e@g@@|@m @end|ng |rom y@hoo@com
Thu Sep 30 14:27:47 CEST 2021


I'm not an R core member, but happen to have looked a little bit at this
issue myself.  I've seen similar things on Skylake and Coffee Lake 2
(9700, one generation past your latest) too.  I think it would make sense
to have some handling of this, although I would want to show the trade-off
with performance impacts on CPUs that are not affected by this, and on
vectors that don't actually have NAs and similar.  I think the performance
impact is likely to be small so long as branch prediction is active, but
since branch prediction is involved you might need to check with different
ratios of NAs (not for your NA bailout branch, but for e.g. interaction
of what you add and the existing `na.rm=TRUE` logic).

You'll also need to think of cases such as c(Inf, NA), c(NaN, NA), etc.,
which might complicate the logic a fair bit.

Presumably the x87 FPU will remain common for a long time, but if there
was reason to think otherwise, then the value of this becomes

Either way, I would probably wait to see what R Core says.

For reference this 2012 blog post[1] discusses some aspects of the issue,
including that at least "historically" AMD was not affected.

Since we're on the topic I want to point out that the default NA in R
starts off as a signaling NA:

    example(numToBits)   # for `bitC`
    ## [1] 0 11111111111 | 0000000000000000000000000000000000000000011110100010
    bitC(NA_real_ + 0)
    ## [1] 0 11111111111 | 1000000000000000000000000000000000000000011110100010

Notice the leading bit of the significant starts off as zero, which marks
it as a signaling NA, but becomes 1, i.e. non-signaling, after any

This is meaningful because the mere act of loading a signaling NA into the
x87 FPU is sufficient to trigger the slowdowns, even if the NA is not
actually used in arithmetic operations.  This happens sometimes under some
optimization levels.  I don't now of any benefit of starting off with a
signaling NA, especially since the encoding is lost pretty much as soon as
it is used.  If folks are interested I can provide patch to turn the NA
quiet by default.



[1]: https://randomascii.wordpress.com/2012/05/20/thats-not-normalthe-performance-of-odd-floats/
[2]: https://en.wikipedia.org/wiki/NaN#Encoding

> On Thursday, September 30, 2021, 06:52:59 AM EDT, GILLIBERT, Andre <andre.gillibert using chu-rouen.fr> wrote:
> Dear R developers,
> By default, R uses the "long double" data type to get extra precision for intermediate computations, with a small performance tradeoff.
> Unfortunately, on all Intel x86 computers I have ever seen, long doubles (implemented in the x87 FPU) are extremely slow whenever a special representation (NA, NaN or infinities) is used; probably because it triggers poorly optimized microcode in the CPU firmware. A function such as sum() becomes more than hundred times slower!
> Test code:
> a=runif(1e7);system.time(for(i in 1:100)sum(a))
> b=a;b[1]=NA;system.time(sum(b))
> The slowdown factors are as follows on a few intel CPU:
> 1)      Pentium Gold G5400 (Coffee Lake, 8th generation) with R 64 bits : 140 times slower with NA
> 2)      Pentium G4400 (Skylake, 6th generation) with R 64 bits : 150 times slower with NA
> 3)      Pentium G3220 (Haswell, 4th generation) with R 64 bits : 130 times slower with NA
> 4)      Celeron J1900 (Atom Silvermont) with R 64 bits : 45 times slower with NA
> I do not have access to more recent Intel CPUs, but I doubt that it has improved much.
> Recent AMD CPUs have no significant slowdown.
> There is no significant slowdown on Intel CPUs (more recent than Sandy Bridge) for 64 bits floating point calculations based on SSE2. Therefore, operators using doubles, such as '+' are unaffected.
> I do not know whether recent ARM CPUs have slowdowns on FP64... Maybe somebody can test.
> Since NAs are not rare in real-life, I think that it would worth an extra check in functions based on long doubles, such as sum(). The check for special representations do not necessarily have to be done at each iteration for cumulative functions.
> If you are interested, I can write a bunch of patches to fix the main functions using long doubles: cumsum, cumprod, sum, prod, rowSums, colSums, matrix multiplication (matprod="internal").
> What do you think of that?
> --
> Sincerely
>     [[alternative HTML version deleted]]
> ______________________________________________
> R-devel using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel

More information about the R-devel mailing list