Groups | Search | Server Info | Keyboard shortcuts | Login | Register
Groups > comp.os.linux.misc > #59411
| Subject | Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints |
|---|---|
| Newsgroups | comp.os.linux.misc |
| References | <YPKdnaTfaLzAq5b6nZ2dnZfqnPidnZ2d@earthlink.com> <veg8cq$k36i$1@dont-email.me> |
| From | "186282@ud0s4.net" <186283@ud0s4.net> |
| Organization | wokiesux |
| Date | 2024-10-15 02:43 -0400 |
| Message-ID | <LpScnb7e54pgk5P6nZ2dnZfqn_qdnZ2d@earthlink.com> (permalink) |
On 10/13/24 6:45 AM, Pancho wrote: > On 10/13/24 03:54, 186282@ud0s4.net wrote: > >> The new technique is basic—instead of using complex >> floating-point multiplication (FPM), the method uses integer >> addition. Apps use FPM to handle extremely large or small >> numbers, allowing applications to carry out calculations >> using them with extreme precision. It is also the most >> energy-intensive part of AI number crunching. >> > > That isn't really true. Floats can handle big and small, but the reason > people use them is for simplicity. "Simple", usually. Energy/time-efficient ... not so much. > The problem is that typical integer calculations are not closed, the > result is not an integer. Addition is fine, but the result of division > is typically not an integer. So if you use integers to model a problem > every time you do a division (or exp, log, sin, etc) you need to make a > decision about how to force the result into an integer. The question is how EXACT the precision HAS to be for most "AI" uses. Might be safe to throw away a few decimal points at the bottom. > Floats actually use integral values for exponent and mantissa, but they > automatically make ballpark reasonable decisions about how to force the > results into integral values for mantissa and exponent, meaning > operations are effectively closed (ignoring exceptions). So the > programmer doesn't have to worry, so much. > > Floating point ops are actually quite efficient, much less of a concern > than something like a branch misprediction. A 20x speed up (energy > saving) sounds close to a theoretical maximum. I would be surprised if > it can be achieved in anything but a few cases. Well ... the article insists they are NOT energy-efficient, esp when performed en-masse. I think their prelim tests suggested an almost 95% savings (sometimes). Anyway, at least the IDEA is back out there again. We old guys, oft dealing with microcontrollers, knew the advantages of wider integers over even 'small' FP. Math processors disguised the amount of processing required for FP ... but it was STILL there.
Back to comp.os.linux.misc | Previous | Next — Previous in thread | Next in thread | Find similar
Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints "186282@ud0s4.net" <186283@ud0s4.net> - 2024-10-12 22:54 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Richard Kettlewell <invalid@invalid.invalid> - 2024-10-13 10:15 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints The Natural Philosopher <tnp@invalid.invalid> - 2024-10-13 13:25 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Pancho <Pancho.Jones@proton.me> - 2024-10-13 14:23 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints The Natural Philosopher <tnp@invalid.invalid> - 2024-10-14 11:16 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints not@telling.you.invalid (Computer Nerd Kev) - 2024-10-15 07:10 +1000
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints The Natural Philosopher <tnp@invalid.invalid> - 2024-10-14 23:47 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints "186282@ud0s4.net" <186283@ud0s4.net> - 2024-10-15 02:31 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints The Natural Philosopher <tnp@invalid.invalid> - 2024-10-15 12:03 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints "186282@ud0s4.net" <186283@ud0s4.net> - 2024-10-16 02:54 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints "186282@ud0s4.net" <186283@ud0s4.net> - 2024-10-15 02:30 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Mike Scott <usenet.16@scottsonline.org.uk.invalid> - 2024-10-15 08:52 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Richard Kettlewell <invalid@invalid.invalid> - 2024-10-15 09:14 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Pancho <Pancho.Jones@proton.me> - 2024-10-13 11:45 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints "186282@ud0s4.net" <186283@ud0s4.net> - 2024-10-15 02:43 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints The Natural Philosopher <tnp@invalid.invalid> - 2024-10-15 12:06 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints "186282@ud0s4.net" <186283@ud0s4.net> - 2024-10-16 02:38 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Pancho <Pancho.Jones@proton.me> - 2024-10-16 08:23 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Richard Kettlewell <invalid@invalid.invalid> - 2024-10-16 11:56 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints "186282@ud0s4.net" <186283@ud0s4.net> - 2024-10-18 00:20 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Richard Kettlewell <invalid@invalid.invalid> - 2024-10-18 17:34 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints 186282ud0s3 <186283@ud0s4.net> - 2024-10-18 14:12 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Chris Ahlstrom <OFeem1987@teleworm.us> - 2024-10-18 15:56 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints The Natural Philosopher <tnp@invalid.invalid> - 2024-10-18 21:06 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints rbowman <bowman@montana.com> - 2024-10-15 19:20 +0000
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Chris Ahlstrom <OFeem1987@teleworm.us> - 2024-10-15 15:46 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints rbowman <bowman@montana.com> - 2024-10-16 02:35 +0000
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints "186282@ud0s4.net" <186283@ud0s4.net> - 2024-10-16 03:13 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Chris Ahlstrom <OFeem1987@teleworm.us> - 2024-10-16 07:40 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2024-10-16 16:16 +0000
Less politics, more Linux! (was: Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints) vallor <vallor@cultnix.org> - 2024-10-16 16:25 +0000
Re: Less politics, more Linux! Chris Ahlstrom <OFeem1987@teleworm.us> - 2024-10-16 14:24 -0400
Re: Less politics, more Linux! vallor <vallor@cultnix.org> - 2024-10-16 18:29 +0000
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints The Natural Philosopher <tnp@invalid.invalid> - 2024-10-16 20:02 +0100
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints rbowman <bowman@montana.com> - 2024-10-16 21:07 +0000
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints Chris Ahlstrom <OFeem1987@teleworm.us> - 2024-10-17 06:54 -0400
Re: Well DUH ! AI People Finally Realize They Can Ditch Most Floating-Point for Big Ints The Natural Philosopher <tnp@invalid.invalid> - 2024-10-17 13:38 +0100
csiph-web