Search found 234 matches

by pauli
Wed Mar 13, 2024 2:55 am
Forum: C47/C43, WP43, WP34S
Topic: Inquiry: C47 Bigint Capabilities
Replies: 26
Views: 604

Re: Inquiry: C47 Bigint Capabilities

I don't remember the C47 supporting Euler's ϕ. We could expose more of the number theoretic integer functions is there was sufficient interest.
by pauli
Mon Mar 11, 2024 4:54 am
Forum: DM32 BETA Bugs
Topic: DM32 Numerical Precision for Transcendental Functions
Replies: 14
Views: 673

Re: DM32 Numerical Precision for Transcendental Functions

The WP 34S calculates internal results to 39 digits (mostly) and then rounds to 34. I'm not yet aware of a case where it rounds incorrectly, at least for the fundamental functions. The WP 34S doesn't carry any hidden digits between the user's operations. TI used to do this and IMO it's quite mislead...
by pauli
Sun Mar 10, 2024 9:30 pm
Forum: C47/C43, WP43, WP34S
Topic: C47: Integration performance and accuracy
Replies: 42
Views: 3817

Re: C47: Integration performance and accuracy

wgr wrote:
Sun Mar 10, 2024 4:29 pm
Further, if the calculator line is advertised as "World's most precise RPN calculators", little differences like this would be nice to work on.
It's never been this IMO. The WP 34S is more accurate due to rounding correctly more often.

Still it's only been 1 ULP out that I've seen.
by pauli
Sun Mar 10, 2024 11:50 am
Forum: C47/C43, WP43, WP34S
Topic: C47: Integration performance and accuracy
Replies: 42
Views: 3817

Re: C47: Integration performance and accuracy

It isn't a problem inherent to using decimal128 -- both Intel's library and the C47 use decimal128 for registers.

I doubt the Intel mathematics library will be "improved".
by pauli
Wed Feb 28, 2024 8:15 am
Forum: C47/C43, WP43, WP34S
Topic: C47 bezel available
Replies: 63
Views: 12440

Re: C47 bezel available

I suspect the goal is to encourage wide spread adoption of the C47 software. To that end, we need to encourage the use of the spiff72 overlays. He is distributing in the US. They only cost $5. The spiff72 adhesive overlay looks as good, or maybe even better than the "other" ;) overlay. He...
by pauli
Wed Feb 28, 2024 8:13 am
Forum: C47/C43, WP43, WP34S
Topic: C47 bezel available
Replies: 63
Views: 12440

Re: C47 bezel available

BigEd wrote:
Tue Feb 27, 2024 2:42 pm
Hmm, you've baffled me there. ...
They are all too easily interpreted by the areas not angles. The areas are necessarily quadratic.
by pauli
Tue Feb 27, 2024 5:01 am
Forum: C47/C43, WP43, WP34S
Topic: C47 bezel available
Replies: 63
Views: 12440

Re: C47 bezel available

IMO pie charts are rubbish. Converting a linear value into a quadratic area greatly enhances any large value contributions. That it is visual is great but it is quite misleading. I have a similar objection to least squares fits [outliers contribute quadratically to the outcome]. To avoid knee jerk r...
by pauli
Mon Feb 26, 2024 1:25 am
Forum: C47/C43, WP43, WP34S
Topic: C47 bezel available
Replies: 63
Views: 12440

Re: C47 bezel available

114 more overlays there.

It seems like quite a few devices are being converted to the C47.
by pauli
Sun Feb 25, 2024 3:36 am
Forum: C47/C43, WP43, WP34S
Topic: C47: Integration performance and accuracy
Replies: 42
Views: 3817

Re: C47: Integration performance and accuracy

Am I doing something really wrong? I don't think so. Even using the 34S in double precision isn't going to be comparable to the C47. The 34S doesn't use more accurate algorithms when in double precision, rather it does the same thing as it would in single precision mode. The C47 has had some algori...
by pauli
Fri Feb 23, 2024 3:00 am
Forum: C47/C43, WP43, WP34S
Topic: C47: Integration performance and accuracy
Replies: 42
Views: 3817

Re: C47: Integration performance and accuracy

The C47 jumps to longer internal reals moderately often. The WP 34S was designed so that 16 digit results were accurate. The C47 produces 34 digit results which will often take longer. The 34S in double precision mode also produces 34 digit results but only limited updates to the mathematical routin...