In J80x, there's no deriv_jcalculus_ because deriv_jcalculus_ is a
replacement for the d. primitive. So, in J807, you can just use that
primitive.
I think the idea was that by making it a library routine instead of a
primitive it would be easier for motivated J programmers to enhance it
with addit
For those interested in Benchmarks I have split my original benchmark routine
into 2 scripts. One the Kernighan/Van Wyk benchmarks and an advanced benchmarks
script. The advanced benchmarks were taken from some of the functionality that
the Scheme language benchmarks for which I found J implemen
… I’d never had thought things like that would be that quickly found
looks like it was quite tricky to get to the root cause
kudos to Henry and whoever else looked into this
Am 24.05.21 um 21:30 schrieb Henry Rich:
> This crash happened when a recursive explicit function was calling
> itself usin
a) No apologies needed - thanks for all the work you do!!
b) You give me incentive to move to 64 bit OS on the Pi -- Thanks!
- joey
> On 2021May 24, at 14:22, bill lam wrote:
>
> a. I'm lazy, sorry about that. I'll update when I have time.
> b. only arm64 has decent support of simd.
>
> On
a. I'm lazy, sorry about that. I'll update when I have time.
b. only arm64 has decent support of simd.
On Tue, May 25, 2021, 5:15 AM Joey K Tuttle wrote:
> Good segue into a couple of questions that have been on my mind.
>
> a) Any thoughts comments on a current Beta release for Raspbian?
>
> b)
Good segue into a couple of questions that have been on my mind.
a) Any thoughts comments on a current Beta release for Raspbian?
b) Are there strong reasons to move to a 64 bit OS on the Raspberry Pi?
- joey
> On 2021May 24, at 13:54, bill lam wrote:
>
> Matrix multiplication on arm64 andr
Matrix multiplication on arm64 android should already be fully optimized,
including
Blas routine with arm64 asimd kernel
Openmp multithreading
Optimized on desktop too, J runs as fast as other multithreaded optimized
blas lapack such as openblas.
On Mon, May 24, 2021, 3:53 PM Ric Sherlock wr
This crash happened when a recursive explicit function was calling
itself using a virtual argument, and the recursion gave an error. In
that case a memory block was not timely freed, and was freed later after
it had gone out of scope. I don't know why it didn't show up on jconsole.
Good to ha
I can reproduce it! Great, thanks.
Henry Rich
On 5/24/2021 10:24 AM, Raul Miller wrote:
On OSX 10.14.6 (Mojave), I load the attached in j903 beta-j then run
0 dv ast
and it crashes.
Interestingly, if I include 0 dv ast in the script, that does not crash.
I hope this helps,
---
On OSX 10.14.6 (Mojave), I load the attached in j903 beta-j then run
0 dv ast
and it crashes.
Interestingly, if I include 0 dv ast in the script, that does not crash.
I hope this helps,
--
Raul
Engine: j903/j64avx2/darwin
Beta-j: commercial/2021-05-18T17:06:59
Library: 9.03.03
Qt IDE: 1.9.
I can't get this to crash under JQt. Can you provide detailed
instructions (I know you did once, I just want to make sure I'm doing
what you're doing)?
And can others reproduce this crash?
Henry Rich
On 5/23/2021 10:37 PM, Raul Miller wrote:
This does not crash for me under jconsole.
This
Oops
dv=: 0, 1+ [:; [:dv&.> 1}.]
Thanks,
--
Raul
On Mon, May 24, 2021 at 8:40 AM Raoul Schorer
wrote:
> Hi and thanks to all for helpting!
>
> Raul, I think you forgot to copy your working verb in your last.
> Ethejiesa, good catch and great insight! Your verb yields a domain error on
>
Ethejiesa, I stand corrected! Your solution works.
Thanks!
On Mon, May 24, 2021 at 2:40 PM Raoul Schorer
wrote:
> Hi and thanks to all for helpting!
>
> Raul, I think you forgot to copy your working verb in your last.
> Ethejiesa, good catch and great insight! Your verb yields a domain error
>
Just adding the definition of your (was f in one of your replies below
Raul)…
dv=: {{(],(>:x) dv >@[)/ |. x ;`(<@[)@.(0=#@]) }.y}}
Works fine, glad you found and resolved the tree construction to match, nice.
> On 24 May 2021, at 10:35 pm, Raul Miller wrote:
>
> Thank you, that makes sen
Hi and thanks to all for helpting!
Raul, I think you forgot to copy your working verb in your last.
Ethejiesa, good catch and great insight! Your verb yields a domain error on
J902, though.
Cheers,
Raoul
On Mon, May 24, 2021 at 2:35 PM Raul Miller wrote:
> Thank you, that makes sense.
>
> And,
Thank you, that makes sense.
And, I also see that I built a faulty value for ast.
Here's a fixed ast and a working depth vector verb:
t=:'.'
ast=:t;(t;t);(t;(t;t);t);<(t;(t;t;t);(t;t;t);t)
ast
+-+-+---+-+
|.|+-+-+|+-+-+-+|+-+---+---+-+|
|
Just to provide some context to Henry's statement that things have changed
a bit since J8.05, below are the timings I get on my phone (Pixel 4a) using
J902.
,.f"0]2^>:i.13
0.024127
1e_5
2e_6
3e_6
3.4e_5
0.000909
0.000425
0.012697
0.020461
0.139175
1.00075
6.6658
56.7179
On Mon,
17 matches
Mail list logo
