Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > sci.physics > #532129 > unrolled thread

Floating Point Test Suite

Started byFabian Russell <root@localhost.localdomain>
First post2015-11-11 18:01 +0000
Last post2015-11-11 15:28 -0500
Articles 20 on this page of 82 — 14 participants

Back to article view | Back to sci.physics


Contents

  Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-11 18:01 +0000
    Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-11 18:09 +0000
    Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-11 19:58 +0000
      Re: Floating Point Test Suite Chris Ahlstrom <OFeem1987@teleworm.us> - 2015-11-11 15:19 -0500
        Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-11 20:28 +0000
        Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-11 20:37 +0000
          Re: Floating Point Test Suite Chris Ahlstrom <OFeem1987@teleworm.us> - 2015-11-11 16:33 -0500
        Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-11 21:13 +0000
          Re: Floating Point Test Suite Odd Bodkin <bodkinodd@gmail.com> - 2015-11-11 15:16 -0600
            Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-11 21:42 +0000
            Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-11 21:54 +0000
              Re: Floating Point Test Suite Odd Bodkin <bodkinodd@gmail.com> - 2015-11-11 16:43 -0600
                Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-11 23:15 +0000
                  Re: Floating Point Test Suite Odd Bodkin <bodkinodd@gmail.com> - 2015-11-11 17:34 -0600
                  Re: Floating Point Test Suite Odd Bodkin <bodkinodd@gmail.com> - 2015-11-11 17:37 -0600
                    Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-11 23:47 +0000
                      Re: Floating Point Test Suite dvus <dven1@adelphia.net> - 2015-11-12 07:56 -0500
                        Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-12 19:07 +0000
                          Re: Floating Point Test Suite Odd Bodkin <bodkinodd@gmail.com> - 2015-11-12 19:57 -0600
                    Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-12 00:45 +0000
                    Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-12 04:55 +0000
                      Re: Floating Point Test Suite Odd Bodkin <bodkinodd@gmail.com> - 2015-11-12 09:26 -0600
                        Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-12 20:56 +0000
                          Re: Floating Point Test Suite Odd Bodkin <bodkinodd@gmail.com> - 2015-11-12 20:00 -0600
              Re: Floating Point Test Suite Odd Bodkin <bodkinodd@gmail.com> - 2015-11-11 16:48 -0600
                Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-11 23:08 +0000
                  Re: Floating Point Test Suite Odd Bodkin <bodkinodd@gmail.com> - 2015-11-11 17:12 -0600
              Re: Floating Point Test Suite Odd Bodkin <bodkinodd@gmail.com> - 2015-11-11 17:06 -0600
              Re: Floating Point Test Suite dvus <dven1@adelphia.net> - 2015-11-12 07:35 -0500
                Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-12 19:29 +0000
      Re: Floating Point Test Suite Monkey Man <manm0nk3y@gmail.com> - 2015-11-11 21:33 -0600
        Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-12 04:47 +0000
          Re: Floating Point Test Suite Monkey Man <manm0nk3y@gmail.com> - 2015-11-12 20:19 -0600
            Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-13 03:28 +0000
              Re: Floating Point Test Suite Big Fish in a Small Crotch <bigfishinasmallcrotch@myself.com> - 2015-11-12 22:40 -0500
                Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-13 03:59 +0000
                  Re: Floating Point Test Suite benj <none@gmail.com> - 2015-11-13 00:16 -0500
                  Re: Floating Point Test Suite GreyCloud <cumulus@mist.com> - 2015-11-13 15:00 -0700
              Re: Floating Point Test Suite Monkey Man <manm0nk3y@gmail.com> - 2015-11-13 01:40 -0600
                Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-13 08:22 +0000
                  Re: Floating Point Test Suite benj <none@gmail.com> - 2015-11-13 12:41 -0500
                    Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-13 19:16 +0000
                      Re: Floating Point Test Suite Apollyon <adravirgo@gmail.com> - 2015-11-13 13:04 -0800
                Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-14 22:58 +0000
                  Re: Floating Point Test Suite Big Fish in a Small Crotch <bigfishinasmallcrotch@myself.com> - 2015-11-14 18:08 -0500
                    Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-14 23:39 +0000
                  Re: Floating Point Test Suite Timo <timo@physics.uq.edu.au> - 2015-11-14 18:26 -0800
                    Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-15 07:07 +0000
                      Re: Floating Point Test Suite Timo <timo@physics.uq.edu.au> - 2015-11-14 23:42 -0800
                        Re: Floating Point Test Suite Poutnik <poutnik4nntp@gmail.com> - 2015-11-15 09:49 +0100
                        Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-15 21:01 +0000
                          Re: Floating Point Test Suite jimp@specsol.spam.sux.com - 2015-11-15 21:33 +0000
                          Re: Floating Point Test Suite Timo <timo@physics.uq.edu.au> - 2015-11-15 14:24 -0800
                            Re: Floating Point Test Suite jimp@specsol.spam.sux.com - 2015-11-15 22:55 +0000
                        Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-15 22:25 +0000
                          Re: Floating Point Test Suite Timo <timo@physics.uq.edu.au> - 2015-11-15 14:59 -0800
                      Re: Floating Point Test Suite Poutnik <poutnik4nntp@gmail.com> - 2015-11-15 09:41 +0100
        Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-12 06:24 +0000
          Re: Floating Point Test Suite Monkey Man <manm0nk3y@gmail.com> - 2015-11-12 20:29 -0600
            Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-13 03:49 +0000
              Re: Floating Point Test Suite Monkey Man <manm0nk3y@gmail.com> - 2015-11-12 22:49 -0600
                Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-13 04:59 +0000
                  Re: Floating Point Test Suite Monkey Man <manm0nk3y@gmail.com> - 2015-11-13 01:56 -0600
                    Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-13 08:28 +0000
                      Re: Floating Point Test Suite Monkey Man <manm0nk3y@gmail.com> - 2015-11-13 22:24 -0600
                        Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-14 04:45 +0000
            Re: June 1986 ? ! Fabian Russell <root@localhost.localdomain> - 2015-11-13 06:19 +0000
              Re: June 1986 ? ! GreyCloud <cumulus@mist.com> - 2015-11-13 15:03 -0700
        Re: MonkeyMan is FabianRussell. Fabian Russell <root@localhost.localdomain> - 2015-11-12 19:44 +0000
      Re: Floating Point Test Suite Yousuf Khan <bbbl67@spammenot.yahoo.com> - 2015-11-13 01:59 -0500
        Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-13 07:43 +0000
          Re: Floating Point Test Suite Timo <timo@physics.uq.edu.au> - 2015-11-13 15:11 -0800
            Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-14 00:00 +0000
              Re: Floating Point Test Suite Timo <timo@physics.uq.edu.au> - 2015-11-13 16:31 -0800
                Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-14 02:26 +0000
                  Re: Floating Point Test Suite jimp@specsol.spam.sux.com - 2015-11-14 03:52 +0000
                Re: Floating Point Test Suite Fabian Russell <root@localhost.localdomain> - 2015-11-14 02:46 +0000
                  Re: Floating Point Test Suite Timo <timo@physics.uq.edu.au> - 2015-11-13 19:26 -0800
                    Re: Floating Point Test Suite benj <none@gmail.com> - 2015-11-14 02:11 -0500
                Re: Floating Point Test Suite noTthaTguY <abu.kuanysh05@gmail.com> - 2015-11-14 15:20 -0800
              Re: Floating Point Test Suite jimp@specsol.spam.sux.com - 2015-11-14 00:47 +0000
    Re: Floating Point Test Suite Big Fish in a Small Crotch <bigfishinasmallcrotch@myself.com> - 2015-11-11 15:28 -0500

Page 4 of 5 — ← Prev page 1 2 3 [4] 5  Next page →


#532644

FromMonkey Man <manm0nk3y@gmail.com>
Date2015-11-12 22:49 -0600
Message-ID<QLidnaMXKqNH9tjLnZ2dnUU7-d2dnZ2d@giganews.com>
In reply to#532636
On Fri, 13 Nov 2015 03:49:33 +0000, Fabian Russell wrote:

>> Checking rounding on multiply, divide and add/subtract.
>> * is neither chopped nor correctly rounded.
>> Division appears to round correctly.
>> Addition/Subtraction appears to round correctly.
>> Sticky bit used incorrectly or not at all.
>> FLAW:  lack(s) of guard digits or failure(s) to correctly round or chop
>> (noted above) count as one flaw in the final tally below.
>> 
>> 
> It seems that multiplication does not meet ideal rounding.
> 
> This is a FLAW, which that computations are correct but rounding is not
> ideal.
> 
> I don't know what could be causing this.
> 
> What processor do you have?

Hey Fabian,

I read through your other reply and am working on composing my response.

I've got a:

8<
  *-cpu                   
       product: AMD FX(tm)-8350 Eight-Core Processor
       vendor: Advanced Micro Devices [AMD]
       physical id: 1
       bus info: cpu@0
       size: 1400MHz
       capacity: 4GHz
       width: 64 bits
       capabilities: fpu fpu_exception wp vme de pse tsc msr pae mce cx8 
apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall 
nx mmxext fxsr_opt pdpe1gb rdtscp x86-64 constant_tsc rep_good nopl 
nonstop_tsc extd_apicid aperfmperf pni pclmulqdq monitor ssse3 fma cx16 
sse4_1 sse4_2 popcnt aes xsave avx f16c lahf_lm cmp_legacy svm extapic 
cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs xop skinit wdt 
lwp fma4 tce nodeid_msr tbm topoext perfctr_core perfctr_nb arat cpb 
hw_pstate npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid 
decodeassists pausefilter pfthreshold vmmcall bmi1 cpufreq
WARNING: output may be incomplete or inaccurate, you should run this 
program as super-user.
>8

[toc] | [prev] | [next] | [standalone]


#532645

FromFabian Russell <root@localhost.localdomain>
Date2015-11-13 04:59 +0000
Message-ID<pan.2015.11.13.04.58.11@localhost.localdomain>
In reply to#532644
On Thu, 12 Nov 2015 22:49:30 -0600, Monkey Man wrote:

> 
> I've got a:
> 
>        product: AMD FX(tm)-8350 Eight-Core Processor
>

Well, the FLAW is not going to be of any concern except if you are
a stickler for perfection.

I can't see what could possibly be causing it.

On a lark, you may want to remove the "-O2" compiler optimization
flag from linux.sh and see if that makes a difference.

Or you just may want to consider this whole thing to be entertainment
with no real consequence.

[toc] | [prev] | [next] | [standalone]


#532665

FromMonkey Man <manm0nk3y@gmail.com>
Date2015-11-13 01:56 -0600
Message-ID<P7idnVFGAvMGCtjLnZ2dnUU7-cWdnZ2d@giganews.com>
In reply to#532645
Hi Fabian,

On Fri, 13 Nov 2015 04:59:51 +0000, Fabian Russell wrote:

> On Thu, 12 Nov 2015 22:49:30 -0600, Monkey Man wrote:
> 
> 
>> I've got a:
>> 
>>        product: AMD FX(tm)-8350 Eight-Core Processor
>>
>>
> Well, the FLAW is not going to be of any concern except if you are a
> stickler for perfection.
>

Which, I'm not since this is my home computer. If I can get the test 
suite to run at work and I find issues, then I'll probably dig into it.

> 
> I can't see what could possibly be causing it.
> 
> On a lark, you may want to remove the "-O2" compiler optimization flag
> from linux.sh and see if that makes a difference.
>

Sure, I'll give that a shot and diff the results to see if anything jumps 
out. I'll post the results up here as well in case your interested.

> 
> Or you just may want to consider this whole thing to be entertainment
> with no real consequence.

I am 100% certain you're right. However, I'm into entertainment with no 
real consequence, and apparently so is a large swath of the youth these 
days (ex. Minecraft, Call of Duty, YouTube, etc.). Full disclosure, I'm a 
bit of a Minecraft junkie.

Thanks for the help Fabian,

Monkey Man

[toc] | [prev] | [next] | [standalone]


#532667

FromFabian Russell <root@localhost.localdomain>
Date2015-11-13 08:28 +0000
Message-ID<pan.2015.11.13.08.27.08@localhost.localdomain>
In reply to#532665
On Fri, 13 Nov 2015 01:56:11 -0600, Monkey Man wrote:

> 
> If I can get the test 
> suite to run at work and I find issues, then I'll probably dig into it.
> 

Well, if you do find issues, then be sure to run some other independent
tests.

The last time I did a search for "floating point tests" there wasn't
much available.  I may have a few links if you want.

[toc] | [prev] | [next] | [standalone]


#532971

FromMonkey Man <manm0nk3y@gmail.com>
Date2015-11-13 22:24 -0600
Message-ID<W9Gdnf5QWrYYKtvLnZ2dnUU7-audnZ2d@giganews.com>
In reply to#532667
Hi Fabian,

On Fri, 13 Nov 2015 08:28:48 +0000, Fabian Russell wrote:

> On Fri, 13 Nov 2015 01:56:11 -0600, Monkey Man wrote:
> 
> 
>> If I can get the test suite to run at work and I find issues, then I'll
>> probably dig into it.
>> 
>> 
> Well, if you do find issues, then be sure to run some other independent
> tests.
> 
> The last time I did a search for "floating point tests" there wasn't
> much available.  I may have a few links if you want.

That'd be great if you have any links! I did run the test again last 
night and got the exact same results. Later on, I did a bit of searching 
for other suites and the few links I found pointed at URL's that have 
probably been down for quite some time.

I'm getting my computer ready to throw LFS back on so I can try tuning 
the binaries for the best performance on my CPU.

Oh, and I forgot to mention that I tried running the test suite on a 
computer at work today and it looks like the binaries were compiled for 
32-bit processors and will not work in their current state. I'll try 
recompiling them when I've got some time.

What a tragedy in Paris today :/

Have a good one Fabian,

Monkey Man

[toc] | [prev] | [next] | [standalone]


#532975

FromFabian Russell <root@localhost.localdomain>
Date2015-11-14 04:45 +0000
Message-ID<pan.2015.11.14.04.43.30@localhost.localdomain>
In reply to#532971
On Fri, 13 Nov 2015 22:24:37 -0600, Monkey Man wrote:

> 
> That'd be great if you have any links!
>

Here's one that should keep you busy for a while:

https://www.research.ibm.com/haifa/projects/verification/fpgen/ieeets.html


I was supposed to go on vacation to the Caribbean for a few weeks,
my associate is lending me his private mansion on the beach,
but when I discovered this suite I decided to cancel everything
and hole myself up with it instead.

No, I'm just kidding. (I think.)

Anyway, take care, and happy computing.

[toc] | [prev] | [next] | [standalone]


#532654 — Re: June 1986 ? !

FromFabian Russell <root@localhost.localdomain>
Date2015-11-13 06:19 +0000
SubjectRe: June 1986 ? !
Message-ID<pan.2015.11.13.06.17.51@localhost.localdomain>
In reply to#532614
On Thu, 12 Nov 2015 21:17:45 -0800, Jeff-Relf.Me wrote:

> 
> June 1986 ? !
> 
> A _Lot_ has changed since then.
>

Jeffie, do not pollute my thread with your thoroughly ignorant
and uninformed comments.

Find a place that is more appropriate to your emotional and
intellectual level --> a nut house.

[toc] | [prev] | [next] | [standalone]


#532853 — Re: June 1986 ? !

FromGreyCloud <cumulus@mist.com>
Date2015-11-13 15:03 -0700
SubjectRe: June 1986 ? !
Message-ID<f5-dnTREx_XRw9vLnZ2dnUU7-VWdnZ2d@bresnan.com>
In reply to#532654
On 11/12/15 23:19, Fabian Russell wrote:
> On Thu, 12 Nov 2015 21:17:45 -0800, Jeff-Relf.Me wrote:
>
>>
>> June 1986 ? !
>>
>> A _Lot_ has changed since then.
>>
>
> Jeffie, do not pollute my thread with your thoroughly ignorant
> and uninformed comments.
>
> Find a place that is more appropriate to your emotional and
> intellectual level -->  a nut house.
>
>
OOh.  Like he's shaking in his shoes... from laughter.
LOL!!!

-- 
My problem is that I don't have enough middle fingers.

[toc] | [prev] | [next] | [standalone]


#532421 — Re: MonkeyMan is FabianRussell.

FromFabian Russell <root@localhost.localdomain>
Date2015-11-12 19:44 +0000
SubjectRe: MonkeyMan is FabianRussell.
Message-ID<pan.2015.11.12.19.42.26@localhost.localdomain>
In reply to#532262
On Thu, 12 Nov 2015 11:27:56 -0800, Jeff-Relf.Me wrote:

> 
> [snip drool and drivel]
>

Shaddup, you pathetic little gadfly.

Somewhere there is a big pile of dog shit with your name on it.

No go and find it.

[toc] | [prev] | [next] | [standalone]


#532659

FromYousuf Khan <bbbl67@spammenot.yahoo.com>
Date2015-11-13 01:59 -0500
Message-ID<hYWdnctGXfjGF9jLnZ2dnUU7-amdnZ2d@giganews.com>
In reply to#532168
On 11/11/2015 2:58 PM, Fabian Russell wrote:
> What?  No takers?

What? You're surprised?

First of all, this is completely off-topic in the sci.physics newsgroup. 
It may be okay for the comp.os.linux.advocacy newsgroup, but I won't 
presume to speak for them, it might be off-topic there too.

Second, insulting Windows users is so passe, and so counterproductive, 
as it's 96% of the userbase. If you wanted real numbers and to get real 
statistics, you can't ignore Windows.

Third, even on Linux, the days of extracting a tarball and compiling 
some source code are so over on Linux too, for the vast majority of 
Linux users. Get familiar with a package manager.

Fourth, there's a reason why the source code hasn't been updated since 
1995: nobody has worried about compliance since then. The days of 
worrying about being IEEE 754 or 854 compliant are over. These days, 
what's of more importance is what hardware is being used to calculate 
the FP numbers now. On x86, you have the traditional x87 floating point 
unit, which is now been obsoleted by the multimedia floating point 
units. Then at the same time, you have GPU floating units which make the 
x86 multimedia units look slow. These days the issues are which is the 
fastest way to calculate FP numbers, not how compliant they are.

> What's the matter?  Floating point numbers are not impressive enough to
> your digital savoir-faire?

Frankly, no.

> Floating point numbers make the computing world go around.  Nothing could
> be done without them, from calculating a flight to Mars to balancing
> grandma's check book.

And that's because floating point is so ubiquitous and trouble-free 
these days that it's boring. It's like worrying whether your excretory 
system is working, you know the shit is coming out whether you worry 
about it or not. 1995 is calling, it wants its computing challenges back.

	Yousuf Khan

[toc] | [prev] | [next] | [standalone]


#532664

FromFabian Russell <root@localhost.localdomain>
Date2015-11-13 07:43 +0000
Message-ID<pan.2015.11.13.07.43.29@localhost.localdomain>
In reply to#532659
On Fri, 13 Nov 2015 01:59:44 -0500, Yousuf Khan wrote:

> 
> First of all, this is completely off-topic in the sci.physics newsgroup.
>

No, it is not.  Is very relevant.

But nothing I will say could ever penetrate your prejudice --
a prejudice which is based on much misinformation.

Computational physics is quite a fundamental area both of
application and of theoretical research, and the thorough
understanding of FP is crucial.

You evidently have never performed numerical analysis
of any kind beyond the use of a "canned" package.
Therefore, I can only conclude that you are unqualified
to form a definitive opinion on the matter.


> 
> Second, insulting Windows users is so passe, and so counterproductive,
>

Maybe, but it sure is fun.

You should try it sometime.  Yeah, put away your stuffed shirt
and try it.


> 
> Third, even on Linux, the days of extracting a tarball and compiling 
> some source code are so over on Linux too,
>

Are they?  Funny, I don't seem to have noticed.

Let me enlighten you, sir, to the fact that many important Linux software
is available ONLY as a tarball and MUST be compiled.

Let me enlighten you, sir, to the fact that the very popular Gentoo Linux
distribution is based entirely on tarball compilation.

Also, there never were any golden days of extracting tarballs in Linux.
From the very beginning there have always been package managers AND
tarball compilation.  In fact, not much has really changed at all.


> 
> Fourth, there's a reason why the source code hasn't been updated since 
> 1995: nobody has worried about compliance since then.
>

Please be accurate in your reporting.

No one has worries because few CAN worry, that is, few have the competence
to even know that worrying may be necessary.


>
> the traditional x87 floating point 
> unit, which is now been obsoleted by the multimedia floating point 
> units.
>

The 387 FP unit is NOT obsolete.

You call the "new" FP unit the "multimedia" FP unit.  This only
demonstrates your lack of technical understanding.

The "new" FP unit embodies the SIMD instruction set, which means
Single Instruction Multiple Data.  The fact that SIMD can be
used for multimedia purposes does not make the unit a "multimedia unit."

But most likely all that you do with your computer is watch YouTube
videos, and because of that you claim an authority.


>
> Then at the same time, you have GPU floating units which make the 
> x86 multimedia units look slow.
>

GPU FP is not standardized.  Use it at your own risk.


>
> These days the issues are which is the 
> fastest way to calculate FP numbers, not how compliant they are.
> 

For your YouTube Mickey Mouse videos is does not make a difference.

But for serious scientific computation it sure as hell does.


> 
> And that's because floating point is so ubiquitous and trouble-free
>

It is that way because there are professional who DO take an
active interest.  Their work, behind the scenes, insures that,
among other things, your YouTube Mickey Mouse is jitter free.


>
> It's like worrying whether your excretory 
> system is working,
>

Until you start pissing blood, then you run to the specialist
and beg for the knowledge and expertise that you could never attain.

[toc] | [prev] | [next] | [standalone]


#532881

FromTimo <timo@physics.uq.edu.au>
Date2015-11-13 15:11 -0800
Message-ID<1a0ddc23-4b7b-4931-ae68-60e8b5f68518@googlegroups.com>
In reply to#532664
On Friday, November 13, 2015 at 5:43:48 PM UTC+10, Fabian Russell wrote:
> On Fri, 13 Nov 2015 01:59:44 -0500, Yousuf Khan wrote:
> > 
> > First of all, this is completely off-topic in the sci.physics newsgroup.
> 
> No, it is not.  Is very relevant.
> 
> But nothing I will say could ever penetrate your prejudice --
> a prejudice which is based on much misinformation.
> 
> Computational physics is quite a fundamental area both of
> application and of theoretical research, and the thorough
> understanding of FP is crucial.

Some understanding of FP is crucial. Thorough understanding of FP is almost never needed for computational physics.

> > These days the issues are which is the 
> > fastest way to calculate FP numbers, not how compliant they are.
> 
> For your YouTube Mickey Mouse videos is does not make a difference.
> 
> But for serious scientific computation it sure as hell does.

Almost never. In computational physics, if your result depends significantly on whether FP numbers are rounded according to a particular standard, you've done it wrong.

[toc] | [prev] | [next] | [standalone]


#532898

FromFabian Russell <root@localhost.localdomain>
Date2015-11-14 00:00 +0000
Message-ID<pan.2015.11.14.00.01.43@localhost.localdomain>
In reply to#532881
On Fri, 13 Nov 2015 15:11:05 -0800, Timo wrote:

> 
> Thorough understanding of FP is almost never needed for
> computational physics.
>

Anyone who limits his desires and interests to only what is
absolutely necessary is a fool

Your attitude is poor and not becoming of a true professional. 

> 
> In computational physics, if your result depends significantly 
> on whether FP numbers are rounded according to a particular standard,
> you've done it wrong.
>

This is highly debatable.  But since your conclusion is essentially
foregone, it would be futile to pursue this further.

[toc] | [prev] | [next] | [standalone]


#532909

FromTimo <timo@physics.uq.edu.au>
Date2015-11-13 16:31 -0800
Message-ID<9c420671-d8a6-44da-8382-170f9300410e@googlegroups.com>
In reply to#532898
On Saturday, November 14, 2015 at 10:02:01 AM UTC+10, Fabian Russell wrote:
> On Fri, 13 Nov 2015 15:11:05 -0800, Timo wrote:
> 
> > Thorough understanding of FP is almost never needed for
> > computational physics.
> 
> Anyone who limits his desires and interests to only what is
> absolutely necessary is a fool

I certainly don't recommend doing so.

> Your attitude is poor and not becoming of a true professional. 

Your reading comprehension is poor and not becoming of a true professional. Correcting your claim that thorough understanding of FP is crucial for computational physics is not the same as saying that a thorough understanding of FP is not desirable for those working in computational physics.

(While far from crucial, it's clearly desirable. However, I wouldn't even call it important.)

> > In computational physics, if your result depends significantly 
> > on whether FP numbers are rounded according to a particular standard,
> > you've done it wrong.
> 
> This is highly debatable.

You have a counter-example?

> But since your conclusion is essentially
> foregone, it would be futile to pursue this further.

Especially if you don't have a counter-example.

On how many decades of work in computational physics do you base your claims?

[toc] | [prev] | [next] | [standalone]


#532949

FromFabian Russell <root@localhost.localdomain>
Date2015-11-14 02:26 +0000
Message-ID<pan.2015.11.14.02.27.03@localhost.localdomain>
In reply to#532909
On Fri, 13 Nov 2015 16:31:31 -0800, Timo wrote:

> 
>> > In computational physics, if your result depends significantly 
>> > on whether FP numbers are rounded according to a particular standard,
>> > you've done it wrong.
>> 
>> This is highly debatable.
> 
> You have a counter-example?
> 

No one is talking about accuracy; the whole point is about
REPRODUCABILITY.

[toc] | [prev] | [next] | [standalone]


#532966

Fromjimp@specsol.spam.sux.com
Date2015-11-14 03:52 +0000
Message-ID<fonihc-n7s.ln1@mail.specsol.com>
In reply to#532949
Fabian Russell <root@localhost.localdomain> wrote:
> On Fri, 13 Nov 2015 16:31:31 -0800, Timo wrote:
> 
>> 
>>> > In computational physics, if your result depends significantly 
>>> > on whether FP numbers are rounded according to a particular standard,
>>> > you've done it wrong.
>>> 
>>> This is highly debatable.
>> 
>> You have a counter-example?
>> 
> 
> No one is talking about accuracy; the whole point is about
> REPRODUCABILITY.
 
So reproducible inaccuracy is a good thing?


-- 
Jim Pennino

[toc] | [prev] | [next] | [standalone]


#532953

FromFabian Russell <root@localhost.localdomain>
Date2015-11-14 02:46 +0000
Message-ID<pan.2015.11.14.02.46.09@localhost.localdomain>
In reply to#532909
On Fri, 13 Nov 2015 16:31:31 -0800, Timo wrote:

> 
> Especially if you don't have a counter-example.
> 

A cheap tactic.

If I cannot produce a satisfactory counter-example then that somehow
will discredit all that I have said.

Well, I am not going to provide a SPECIFIC counter-example.  It may
take a while for me to procure one, and I have other fish to fry as
well.

But here is an excerpt from the Intel ICC manual:

"Using the default option keyword -fp-model fast or /fp:fast,
you may get significant differences in your result depending
on whether the compiler uses x87 or SSE2 instructions to implement
floating-point operations. Results are more consistent when the
other option keywords are used."

http://www.cs.kent.edu/~farrell/cc14/intel/Documentation/en_US/compiler_c/main_cls/index.htm

Who am I going to believe?

Hint: It's not you.

[toc] | [prev] | [next] | [standalone]


#532958

FromTimo <timo@physics.uq.edu.au>
Date2015-11-13 19:26 -0800
Message-ID<9a7ff5a7-a988-4e47-b982-395ebe7e7886@googlegroups.com>
In reply to#532953
On Saturday, November 14, 2015 at 12:47:04 PM UTC+10, Fabian Russell wrote:
> On Fri, 13 Nov 2015 16:31:31 -0800, Timo wrote:
> 
> > Especially if you don't have a counter-example.
> 
> A cheap tactic.
> 
> If I cannot produce a satisfactory counter-example then that somehow
> will discredit all that I have said.

Evidence for claims is often useful for supporting claims. If there is no evidence, what is the basis for your claim?

Since you're not willing/able to provide a counterexample, perhaps you could briefly explain. Performing some numerical calculation in computational physics, one will have some target level of accuracy. Suppose one aims for a relative error of 1e-6. If the final error due to FP rounding is larger than that, I'd say that the calculation is done incorrectly. You appear to disagree. Can you explain why? 

> Well, I am not going to provide a SPECIFIC counter-example.  It may
> take a while for me to procure one, and I have other fish to fry as
> well.
> 
> But here is an excerpt from the Intel ICC manual:
> 
> "Using the default option keyword -fp-model fast or /fp:fast,
> you may get significant differences in your result depending
> on whether the compiler uses x87 or SSE2 instructions to implement
> floating-point operations. Results are more consistent when the
> other option keywords are used."
> http://www.cs.kent.edu/~farrell/cc14/intel/Documentation/en_US/compiler_c/main_cls/index.htm

Different meaning of "significant". Not relevant to the point.

> Who am I going to believe?
> 
> Hint: It's not you.

Fine, go ahead and ignore what computational physicists say about what's important in computational physics.

[toc] | [prev] | [next] | [standalone]


#532982

Frombenj <none@gmail.com>
Date2015-11-14 02:11 -0500
Message-ID<v8B1y.57533$gK.37131@fx26.iad>
In reply to#532958
On 11/13/2015 10:26 PM, Timo wrote:
> On Saturday, November 14, 2015 at 12:47:04 PM UTC+10, Fabian Russell
> wrote:
>> On Fri, 13 Nov 2015 16:31:31 -0800, Timo wrote:
>>
>>> Especially if you don't have a counter-example.
>>
>> A cheap tactic.
>>
>> If I cannot produce a satisfactory counter-example then that
>> somehow will discredit all that I have said.
>
> Evidence for claims is often useful for supporting claims. If there
> is no evidence, what is the basis for your claim?
>
> Since you're not willing/able to provide a counterexample, perhaps
> you could briefly explain. Performing some numerical calculation in
> computational physics, one will have some target level of accuracy.
> Suppose one aims for a relative error of 1e-6. If the final error due
> to FP rounding is larger than that, I'd say that the calculation is
> done incorrectly. You appear to disagree. Can you explain why?

Of course he can't explain why. Because Fabio is off chasing his own 
tail. Fact is obviously, if your calculation needs an accuracy such that 
the rounding standard of the floating point unit comes into play it's 
clear that calculation is in BIG trouble and needs to be reworked.

OF course it's also obvious that a floating point rounding standard 
SHOULD do so correctly, but if it does not, that would be of little 
consequence to any properly done calculation.

You know how it is around here Timo. Truth is just so hard to believe 
when there so many better sounding other stories out there!
-- 

       ___           ___           ___            ___
      /\  \         /\  \         /\__\          /\  \
     /::\  \       /::\  \       /::|  |         \:\  \
    /:/\:\  \     /:/\:\  \     /:|:|  |     ___ /::\__\
   /::\~\:\__\   /::\~\:\  \   /:/|:|  |__  /\  /:/\/__/
  /:/\:\ \:|__| /:/\:\ \:\__\ /:/ |:| /\__\ \:\/:/  /
  \:\~\:\/:/  / \:\~\:\ \/__/ \/__|:|/:/  /  \::/  /
   \:\ \::/  /   \:\ \:\__\       |:/:/  /    \/__/
    \:\/:/  /     \:\ \/__/       |::/  /
     \::/__/       \:\__\         /:/  /
      ~~            \/__/         \/__/

[toc] | [prev] | [next] | [standalone]


#533151

FromnoTthaTguY <abu.kuanysh05@gmail.com>
Date2015-11-14 15:20 -0800
Message-ID<be226774-f3d1-4795-9672-5f32553fc497@googlegroups.com>
In reply to#532909
IEEE-758 is inherently chaotic, due to rounding, as is clear
from the spec (which is an article in *Computer, mid-'80s

> On how many decades of work in computational physics do you base your claims?

[toc] | [prev] | [next] | [standalone]


Page 4 of 5 — ← Prev page 1 2 3 [4] 5  Next page →

Back to top | Article view | sci.physics


csiph-web