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

Groups > comp.arch > #3265 > unrolled thread

Re: Solving the Floating-Point Goldilocks Problem!

Back to article view | Back to comp.arch

Contents

  Re: Solving the Floating-Point Goldilocks Problem! jmfbahciv <See.above@aol.com> - 2011-08-30 13:32 +0000
    Re: Solving the Floating-Point Goldilocks Problem! "Andy \"Krazy\" Glew" <andy@SPAM.comp-arch.net> - 2011-08-30 07:02 -0700
      Re: Solving the Floating-Point Goldilocks Problem! Morten Reistad <first@last.name> - 2011-08-30 17:09 +0200
      Re: Solving the Floating-Point Goldilocks Problem! jmfbahciv <See.above@aol.com> - 2011-08-31 13:21 +0000
        Re: Solving the Floating-Point Goldilocks Problem! "Andy \"Krazy\" Glew" <andy@SPAM.comp-arch.net> - 2011-08-31 07:55 -0700
          Re: Solving the Floating-Point Goldilocks Problem! jmfbahciv <See.above@aol.com> - 2011-09-01 13:39 +0000
            Re: Solving the Floating-Point Goldilocks Problem! "Rod Speed" <rod.speed.aaa@gmail.com> - 2011-09-02 01:24 +1000
            Re: Solving the Floating-Point Goldilocks Problem! "Andy \"Krazy\" Glew" <andy@SPAM.comp-arch.net> - 2011-09-01 08:28 -0700
              Re: Solving the Floating-Point Goldilocks Problem! jmfbahciv <See.above@aol.com> - 2011-09-02 13:44 +0000
        Re: Solving the Floating-Point Goldilocks Problem! "Rod Speed" <rod.speed.aaa@gmail.com> - 2011-09-01 04:30 +1000
          Re: Solving the Floating-Point Goldilocks Problem! jmfbahciv <See.above@aol.com> - 2011-09-01 13:39 +0000
            Re: Solving the Floating-Point Goldilocks Problem! "Rod Speed" <rod.speed.aaa@gmail.com> - 2011-09-02 01:26 +1000

#3265 — Re: Solving the Floating-Point Goldilocks Problem!

Andy "Krazy" Glew wrote:
> On 8/29/2011 9:56 AM, nmm1@cam.ac.uk wrote:
>> In article<4E5BB93F.2030406@SPAM.comp-arch.net>,
>> Andy \"Krazy\" Glew<andy@SPAM.comp-arch.net>  wrote:
>>> On 8/29/2011 4:56 AM, Terje Mathisen wrote:

<snip interesting discussion>

>
> I am collecting these against the next time I am involved in an effort
> like MMX.  Where shortsighted people will say "all we need to support is
> 4:2:2 YUV, and we should add a special instruction just for that",
> rather than more general support.  Or vice versa.

Please consider putting your collections on bitsavers, or somewhere,
so that the knowledge doesn't get lost.

/BAH

[toc] | [next] | [standalone]

#3269

On 8/30/2011 6:32 AM, jmfbahciv wrote:
> Andy "Krazy" Glew wrote:
>> On 8/29/2011 9:56 AM, nmm1@cam.ac.uk wrote:
>>> In article<4E5BB93F.2030406@SPAM.comp-arch.net>,
>>> Andy \"Krazy\" Glew<andy@SPAM.comp-arch.net>   wrote:
>>>> On 8/29/2011 4:56 AM, Terje Mathisen wrote:
>
> <snip interesting discussion>
>
>>
>> I am collecting these against the next time I am involved in an effort
>> like MMX.  Where shortsighted people will say "all we need to support is
>> 4:2:2 YUV, and we should add a special instruction just for that",
>> rather than more general support.  Or vice versa.
>
> Please consider putting your collections on bitsavers, or somewhere,
> so that the knowledge doesn't get lost.
>
> /BAH

I am putting my collection on comp-arch.net --- my own personal 
wikipedia.  I had not considered it as Bitsavers, but fair enough.

Why "my own personal" - well, I started putting it on wikipedia, but got 
tired of stuff being deleted by the wikipedia trolls.

Why not "bitsavers"?  Well, my impression is that bitsavers is about 
saving manuals, etc.

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

#3277

In article <4E5CED7B.5040209@SPAM.comp-arch.net>,
Andy \"Krazy\" Glew <andy@SPAM.comp-arch.net> wrote:
>On 8/30/2011 6:32 AM, jmfbahciv wrote:
>> Andy "Krazy" Glew wrote:
>>> On 8/29/2011 9:56 AM, nmm1@cam.ac.uk wrote:
>>>> In article<4E5BB93F.2030406@SPAM.comp-arch.net>,
>>>> Andy \"Krazy\" Glew<andy@SPAM.comp-arch.net>   wrote:

>> Please consider putting your collections on bitsavers, or somewhere,
>> so that the knowledge doesn't get lost.
>>
>> /BAH
>
>I am putting my collection on comp-arch.net --- my own personal 
>wikipedia.  I had not considered it as Bitsavers, but fair enough.
>
>Why "my own personal" - well, I started putting it on wikipedia, but got 
>tired of stuff being deleted by the wikipedia trolls.
>
>Why not "bitsavers"?  Well, my impression is that bitsavers is about 
>saving manuals, etc.

We should secure sites like bitsavers, multicians, trailing-edge and
the various private sites for posterity.

I suggest we publish them in a quantity that forces the national
libraries of good archiving states to store them.

The US, Canada, UK, Norway, Sweden and Germany all have very good
archives. (I am sure there are more). 

Norway has a limit of 50 published copies where you can voluntarily
give up copies for the arhives, the mandatory limit is at 500. 

The archival package is 6 copies, and it qualifies as a tax payment, 
deductible on a pro-rata sales volume. 

Three of the copies goes into deep-time storage, chilled, packed in
argon etc. They are getting pretty good at preserving stuff. 

-- mrr

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

#3298

Andy "Krazy" Glew wrote:
> On 8/30/2011 6:32 AM, jmfbahciv wrote:
>> Andy "Krazy" Glew wrote:
>>> On 8/29/2011 9:56 AM, nmm1@cam.ac.uk wrote:
>>>> In article<4E5BB93F.2030406@SPAM.comp-arch.net>,
>>>> Andy \"Krazy\" Glew<andy@SPAM.comp-arch.net>   wrote:
>>>>> On 8/29/2011 4:56 AM, Terje Mathisen wrote:
>>
>> <snip interesting discussion>
>>
>>>
>>> I am collecting these against the next time I am involved in an effort
>>> like MMX.  Where shortsighted people will say "all we need to support is
>>> 4:2:2 YUV, and we should add a special instruction just for that",
>>> rather than more general support.  Or vice versa.
>>
>> Please consider putting your collections on bitsavers, or somewhere,
>> so that the knowledge doesn't get lost.
>>
>> /BAH
>
> I am putting my collection on comp-arch.net --- my own personal
> wikipedia.  I had not considered it as Bitsavers, but fair enough.

Not wiki.   Definitely not wiki.  I don't care where as long as all
your tradeoffs, which is the most important piece of designing but never
gets published.  In our biz, our mistakes and initial desing reasonings
rarely got put down on paper.

>
> Why "my own personal" - well, I started putting it on wikipedia, but got
> tired of stuff being deleted by the wikipedia trolls.


I have yet to believe anything on wiki.  I think of it as a non-feature
and eventually dangerous.
>
> Why not "bitsavers"?  Well, my impression is that bitsavers is about
> saving manuals, etc.

Well, I don't care where as long as the thinking processes oof those
designs are preserved.  :-)  Making computers represent numbers
and do mere arithmetic is hard and most people, even those who
become bit gods in other areas, have no idea about the problems.
Training new people to design these things is in the startosphere
of techniques, IMO.  Every kiddie is going to have a wonderful
idea and won't discover the pitfalls until too late.

/BAH

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

#3301

On 8/31/2011 6:21 AM, jmfbahciv wrote:
> Andy "Krazy" Glew wrote:
>> On 8/30/2011 6:32 AM, jmfbahciv wrote:
>>> Andy "Krazy" Glew wrote:
>>>> On 8/29/2011 9:56 AM, nmm1@cam.ac.uk wrote:
>>>>> In article<4E5BB93F.2030406@SPAM.comp-arch.net>,
>>>>> Andy \"Krazy\" Glew<andy@SPAM.comp-arch.net>    wrote:
>>>>>> On 8/29/2011 4:56 AM, Terje Mathisen wrote:
>>>
>>> <snip interesting discussion>
>>>
>>>>
>>>> I am collecting these against the next time I am involved in an effort
>>>> like MMX.  Where shortsighted people will say "all we need to support is
>>>> 4:2:2 YUV, and we should add a special instruction just for that",
>>>> rather than more general support.  Or vice versa.
>>>
>>> Please consider putting your collections on bitsavers, or somewhere,
>>> so that the knowledge doesn't get lost.
>>>
>>> /BAH
>>
>> I am putting my collection on comp-arch.net --- my own personal
>> wikipedia.  I had not considered it as Bitsavers, but fair enough.
>
> Not wiki.   Definitely not wiki.  I don't care where as long as all
> your tradeoffs, which is the most important piece of designing but never
> gets published.  In our biz, our mistakes and initial desing reasonings
> rarely got put down on paper.
>
>>
>> Why "my own personal" - well, I started putting it on wikipedia, but got
>> tired of stuff being deleted by the wikipedia trolls.
>
>
> I have yet to believe anything on wiki.  I think of it as a non-feature
> and eventually dangerous.

Recording the design rationales, rejected alternatives, etc., is what I 
am trying to do with comp-arch.net.  Slowly. Very slowly. Proprietary 
FUD always gets in the way, usually due to self-censorship (e.g. I won't 
post this now, because that would give you a clue as to what I am 
working on. Even though everything is public domain.)

Are you against wikis overall, or just wikipedia?

Like I said above, wikipedia is a loss for this sort of thing (1) 
because stuff gets deleted, and (2) because wikipedia wants 
authoritative references. I am trying to record folklore, less formal stuff.

But wikis - easy to edit hypertext - are my preferred way of recording 
such stuff.

There is lossage because I don't have good diagram support on any wiki I 
know well.  And I need better security / delayed publication / 
distributed version control / offline access, etc.

But hypertext is the best way I know of to do this.  And wikis are the 
easiest form of hypertext to CREATE - perhaps not to read, but at the 
moment I am content creation limited. Must make composing the content as 
easy as possible.

So, therefore, comp-arch.net is largely a private and semipublic wiki. 
Mostly readable by the world; writable by only me and a few others.

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

#3323

Andy "Krazy" Glew wrote:
> On 8/31/2011 6:21 AM, jmfbahciv wrote:
>> Andy "Krazy" Glew wrote:
>>> On 8/30/2011 6:32 AM, jmfbahciv wrote:
>>>> Andy "Krazy" Glew wrote:
>>>>> On 8/29/2011 9:56 AM, nmm1@cam.ac.uk wrote:
>>>>>> In article<4E5BB93F.2030406@SPAM.comp-arch.net>,
>>>>>> Andy \"Krazy\" Glew<andy@SPAM.comp-arch.net>    wrote:
>>>>>>> On 8/29/2011 4:56 AM, Terje Mathisen wrote:
>>>>
>>>> <snip interesting discussion>
>>>>
>>>>>
>>>>> I am collecting these against the next time I am involved in an effort
>>>>> like MMX.  Where shortsighted people will say "all we need to support is
>>>>> 4:2:2 YUV, and we should add a special instruction just for that",
>>>>> rather than more general support.  Or vice versa.
>>>>
>>>> Please consider putting your collections on bitsavers, or somewhere,
>>>> so that the knowledge doesn't get lost.
>>>>
>>>> /BAH
>>>
>>> I am putting my collection on comp-arch.net --- my own personal
>>> wikipedia.  I had not considered it as Bitsavers, but fair enough.
>>
>> Not wiki.   Definitely not wiki.  I don't care where as long as all
>> your tradeoffs, which is the most important piece of designing but never
>> gets published.  In our biz, our mistakes and initial desing reasonings
>> rarely got put down on paper.
>>
>>>
>>> Why "my own personal" - well, I started putting it on wikipedia, but got
>>> tired of stuff being deleted by the wikipedia trolls.
>>
>>
>> I have yet to believe anything on wiki.  I think of it as a non-feature
>> and eventually dangerous.
>
> Recording the design rationales, rejected alternatives, etc., is what I
> am trying to do with comp-arch.net.  Slowly. Very slowly. Proprietary
> FUD always gets in the way,

I understood that :-).

> usually due to self-censorship (e.g. I won't
> post this now, because that would give you a clue as to what I am
> working on. Even though everything is public domain.)

Which is why I asked that you put it somewhere, some time, so your
work doesn't get "lost" if you ever stop working.  I don't know,
nor have any suggestion, w.r.t. how you do this.  It's real tricky.
There are lots of things I didn't get preserved because of similar
reasons.  There are times I wish I had broken security policies
but it just wasn't possible back then.

>
> Are you against wikis overall, or just wikipedia?


Anything which allows assholes to destory knowledge and/or
productivity.

>
> Like I said above, wikipedia is a loss for this sort of thing (1)
> because stuff gets deleted, and (2) because wikipedia wants
> authoritative references. I am trying to record folklore, less formal stuff.
>
> But wikis - easy to edit hypertext - are my preferred way of recording
> such stuff.
>
> There is lossage because I don't have good diagram support on any wiki I
> know well.  And I need better security / delayed publication /
> distributed version control / offline access, etc.
>
> But hypertext is the best way I know of to do this.  And wikis are the
> easiest form of hypertext to CREATE - perhaps not to read, but at the
> moment I am content creation limited. Must make composing the content as
> easy as possible.
>
> So, therefore, comp-arch.net is largely a private and semipublic wiki.
> Mostly readable by the world; writable by only me and a few others.

Well, at least there are a "few others".  I'm an auld fart and do
not have any valid suggestions about how to go about doing this stuff.
However, I am very experienced in watching knowledge get destroyed--
some was because of old-age but most was because of a determined effort
to eradicate that knowledge.

/BAH

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

#3327

jmfbahciv wrote
> Andy "Krazy" Glew wrote
>> jmfbahciv wrote
>>> Andy "Krazy" Glew wrote
>>>> jmfbahciv wrote
>>>>> Andy "Krazy" Glew wrote
>>>>>> nmm1@cam.ac.uk wrote
>>>>>>> Andy \"Krazy\" Glew<andy@SPAM.comp-arch.net> wrote
>>>>>>>> Terje Mathisen wrote

>>>>>> I am collecting these against the next time I am involved in an
>>>>>> effort like MMX.  Where shortsighted people will say "all we
>>>>>> need to support is 4:2:2 YUV, and we should add a special
>>>>>> instruction just for that", rather than more general support.
>>>>>> Or vice versa.

>>>>> Please consider putting your collections on bitsavers,
>>>>> or somewhere, so that the knowledge doesn't get lost.

>>>> I am putting my collection on comp-arch.net --- my own personal
>>>> wikipedia.  I had not considered it as Bitsavers, but fair enough.

>>> Not wiki.   Definitely not wiki.  I don't care where as long as all
>>> your tradeoffs, which is the most important piece of designing but
>>> never gets published.  In our biz, our mistakes and initial desing
>>> reasonings rarely got put down on paper.

>>>> Why "my own personal" - well, I started putting it on wikipedia,
>>>> but got tired of stuff being deleted by the wikipedia trolls.

>>> I have yet to believe anything on wiki.  I think of it as a
>>> non-feature and eventually dangerous.

>> Recording the design rationales, rejected alternatives, etc., is
>> what I am trying to do with comp-arch.net.  Slowly. Very slowly.
>> Proprietary FUD always gets in the way,

> I understood that :-).

>> usually due to self-censorship (e.g. I won't post this now,
>> because that would give you a clue as to what I am
>> working on. Even though everything is public domain.)

> Which is why I asked that you put it somewhere, some time, so
> your work doesn't get "lost" if you ever stop working.  I don't know,
> nor have any suggestion, w.r.t. how you do this.  It's real tricky.

Not really. Even gmail alone provides a pretty decent place to dump it
with some distribution of the password to allow others access to that.

Synergy is one example of the sort of preservation of that sort of design notes stuff.

> There are lots of things I didn't get preserved because of similar
> reasons.  There are times I wish I had broken security policies
> but it just wasn't possible back then.

>> Are you against wikis overall, or just wikipedia?

> Anything which allows assholes to destory knowledge and/or productivity.

Trouble is that misses the point on it being one hell of a resource as well.

>> Like I said above, wikipedia is a loss for this sort of thing (1) because
>> stuff gets deleted, and (2) because wikipedia wants authoritative
>> references. I am trying to record folklore, less formal stuff.

>> But wikis - easy to edit hypertext - are my preferred way of recording such stuff.

>> There is lossage because I don't have good diagram support on any
>> wiki I know well.  And I need better security / delayed publication /
>> distributed version control / offline access, etc.

>> But hypertext is the best way I know of to do this.  And wikis are
>> the easiest form of hypertext to CREATE - perhaps not to read, but
>> at the moment I am content creation limited. Must make composing the
>> content as easy as possible.

>> So, therefore, comp-arch.net is largely a private and semipublic wiki.
>> Mostly readable by the world; writable by only me and a few others.

> Well, at least there are a "few others".  I'm an auld fart and do
> not have any valid suggestions about how to go about doing this stuff.
> However, I am very experienced in watching knowledge get destroyed--
> some was because of old-age but most was because of a determined
> effort to eradicate that knowledge.

Not even possible anymore. In spades with wikileaks. 

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

#3329

On 9/1/2011 6:39 AM, jmfbahciv wrote:
> Andy "Krazy" Glew wrote:

>> Recording the design rationales, rejected alternatives, etc., is what I
>> am trying to do with comp-arch.net.  Slowly. Very slowly. Proprietary
>> FUD always gets in the way,
>
> I understood that :-).
>
>> usually due to self-censorship (e.g. I won't
>> post this now, because that would give you a clue as to what I am
>> working on. Even though everything is public domain.)
>
> Which is why I asked that you put it somewhere, some time, so your
> work doesn't get "lost" if you ever stop working.  I don't know,
> nor have any suggestion, w.r.t. how you do this.  It's real tricky.
> There are lots of things I didn't get preserved because of similar
> reasons.  There are times I wish I had broken security policies
> but it just wasn't possible back then.

You should be able to seee that putting stuff on Bitsavers doesn't save 
the sensitivity problem.

In my copious spare time I am rewriting the wiki software to have a 
delayed publication model: I write stuff that is public but sensitive 
because of time, and arrange for it to get published later.   A year 
later, three years later, randomized, all subject to the appropriate NDA.

In the meantime, I write it up, put it in my private area, and try to 
remember to move it to the public area after enough time has gone by.


>> Are you against wikis overall, or just wikipedia?
>
> Anything which allows assholes to destory knowledge and/or
> productivity.

Then you are not against wikis. Just wikipedia.

Wiki/hypertext make it easier to create knowledge.  Hypertext makes it 
easy to see links;  IMHO the biggest thing Ward Cunningham invented with 
wiki was a way of creating links to stuff that hasn't been written yet, 
making it easier to create hypertext.   The multiuser aspect if wiki is 
just icing on the cake.  I have a number of single user wikis (more than 
one of them because of security issues, usually; if wikis had good 
security, I might need only one).

By the way: at both Intel and AMD, considerable computer architecture 
knowledge is created and managed using wikis.   One of the best things I 
did for AMD was introduce them to wikis.

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

#3374

Andy "Krazy" Glew wrote:
> On 9/1/2011 6:39 AM, jmfbahciv wrote:
>> Andy "Krazy" Glew wrote:
>
>>> Recording the design rationales, rejected alternatives, etc., is what I
>>> am trying to do with comp-arch.net.  Slowly. Very slowly. Proprietary
>>> FUD always gets in the way,
>>
>> I understood that :-).
>>
>>> usually due to self-censorship (e.g. I won't
>>> post this now, because that would give you a clue as to what I am
>>> working on. Even though everything is public domain.)
>>
>> Which is why I asked that you put it somewhere, some time, so your
>> work doesn't get "lost" if you ever stop working.  I don't know,
>> nor have any suggestion, w.r.t. how you do this.  It's real tricky.
>> There are lots of things I didn't get preserved because of similar
>> reasons.  There are times I wish I had broken security policies
>> but it just wasn't possible back then.
>
> You should be able to seee that putting stuff on Bitsavers doesn't save
> the sensitivity problem.

Yes, I know.  If I had a good idea about how to do it, I'd have told you :-).

>
> In my copious spare time

<GRIN>

> I am rewriting the wiki software to have a
> delayed publication model: I write stuff that is public but sensitive
> because of time, and arrange for it to get published later.   A year
> later, three years later, randomized, all subject to the appropriate NDA.
>
> In the meantime, I write it up, put it in my private area, and try to
> remember to move it to the public area after enough time has gone by.

Yes, and this has a single-point failure potential.  These problems
aren't unique to just your stuff.  The computing biz was an
engineering corporate effort so the usual papers documenting mistakes,
trade-offs, and ideas dropped because the tech isn't available wasn't
done.  The science effort has this incorporated as part of their
procedures.  Even writing about the most mundane mistakes is
rewarded.

>
>
>>> Are you against wikis overall, or just wikipedia?
>>
>> Anything which allows assholes to destory knowledge and/or
>> productivity.
>
> Then you are not against wikis. Just wikipedia.


Or anything else that shows up which is used as anti-production.

>
> Wiki/hypertext make it easier to create knowledge.  Hypertext makes it
> easy to see links;  IMHO the biggest thing Ward Cunningham invented with
> wiki was a way of creating links to stuff that hasn't been written yet,
> making it easier to create hypertext.   The multiuser aspect if wiki is
> just icing on the cake.  I have a number of single user wikis (more than
> one of them because of security issues, usually; if wikis had good
> security, I might need only one).

Heh.  sounds like 10 useful projects there :-)).

>
> By the way: at both Intel and AMD, considerable computer architecture
> knowledge is created and managed using wikis.   One of the best things I
> did for AMD was introduce them to wikis.

Can infomartion be purged if it's not been accessed for some
pre-determined period of time?  Sources have been lost in the
past because of inactivity.  For instance, a site can have throrough
BACKUP procedures.  However, if a file hasn't been "touched" by a human,
for various definitions of touching, then noone notices that a file
has been deleted.  If the BACKUPs have a "hole" in time and the deletion
occurred during that hole, the file cannot be restored.  Note that
I'm using terminolgoy based on my old work days.

Somebody may have to translate for me using terms which are more
modern.

/BAH

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

#3308

jmfbahciv wrote
> Andy "Krazy" Glew wrote
>> jmfbahciv wrote
>>> Andy "Krazy" Glew wrote
>>>> nmm1@cam.ac.uk wrote
>>>>> Andy \"Krazy\" Glew<andy@SPAM.comp-arch.net> wrote
>>>>>> Terje Mathisen wrote

>>>> I am collecting these against the next time I am involved in an
>>>> effort like MMX.  Where shortsighted people will say "all we need
>>>> to support is 4:2:2 YUV, and we should add a special instruction
>>>> just for that", rather than more general support.  Or vice versa.

>>> Please consider putting your collections on bitsavers, or somewhere,
>>> so that the knowledge doesn't get lost.

>> I am putting my collection on comp-arch.net --- my own personal
>> wikipedia.  I had not considered it as Bitsavers, but fair enough.

> Not wiki.   Definitely not wiki.  I don't care where as long as all
> your tradeoffs, which is the most important piece of designing
> but never gets published.  In our biz, our mistakes and initial
> desing reasonings rarely got put down on paper.

>> Why "my own personal" - well, I started putting it on wikipedia,
>> but got tired of stuff being deleted by the wikipedia trolls.

> I have yet to believe anything on wiki.

More fool you.

> I think of it as a non-feature and eventually dangerous.

More fool you.

>> Why not "bitsavers"?  Well, my impression is that bitsavers is about
>> saving manuals, etc.

> Well, I don't care where as long as the thinking processes oof those
> designs are preserved.  :-)  Making computers represent numbers
> and do mere arithmetic is hard and most people, even those who
> become bit gods in other areas, have no idea about the problems.
> Training new people to design these things is in the startosphere
> of techniques, IMO.  Every kiddie is going to have a wonderful
> idea and won't discover the pitfalls until too late.

Kiddies dont design computers. 

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

#3322

Rod Speed wrote:
> jmfbahciv wrote
>> Andy "Krazy" Glew wrote
>>> jmfbahciv wrote
>>>> Andy "Krazy" Glew wrote
>>>>> nmm1@cam.ac.uk wrote
>>>>>> Andy \"Krazy\" Glew<andy@SPAM.comp-arch.net> wrote
>>>>>>> Terje Mathisen wrote
>
>>>>> I am collecting these against the next time I am involved in an
>>>>> effort like MMX.  Where shortsighted people will say "all we need
>>>>> to support is 4:2:2 YUV, and we should add a special instruction
>>>>> just for that", rather than more general support.  Or vice versa.
>
>>>> Please consider putting your collections on bitsavers, or somewhere,
>>>> so that the knowledge doesn't get lost.
>
>>> I am putting my collection on comp-arch.net --- my own personal
>>> wikipedia.  I had not considered it as Bitsavers, but fair enough.
>
>> Not wiki.   Definitely not wiki.  I don't care where as long as all
>> your tradeoffs, which is the most important piece of designing
>> but never gets published.  In our biz, our mistakes and initial
>> desing reasonings rarely got put down on paper.
>
>>> Why "my own personal" - well, I started putting it on wikipedia,
>>> but got tired of stuff being deleted by the wikipedia trolls.
>
>> I have yet to believe anything on wiki.
>
> More fool you.
>
>> I think of it as a non-feature and eventually dangerous.
>
> More fool you.
>
>>> Why not "bitsavers"?  Well, my impression is that bitsavers is about
>>> saving manuals, etc.
>
>> Well, I don't care where as long as the thinking processes oof those
>> designs are preserved.  :-)  Making computers represent numbers
>> and do mere arithmetic is hard and most people, even those who
>> become bit gods in other areas, have no idea about the problems.
>> Training new people to design these things is in the startosphere
>> of techniques, IMO.  Every kiddie is going to have a wonderful
>> idea and won't discover the pitfalls until too late.
>
> Kiddies dont design computers.

They do.  Getting it to market requires experience.

/BAH

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

#3328

jmfbahciv wrote
> Rod Speed wrote
>> jmfbahciv wrote
>>> Andy "Krazy" Glew wrote
>>>> jmfbahciv wrote
>>>>> Andy "Krazy" Glew wrote
>>>>>> nmm1@cam.ac.uk wrote
>>>>>>> Andy \"Krazy\" Glew<andy@SPAM.comp-arch.net> wrote
>>>>>>>> Terje Mathisen wrote

>>>>>> I am collecting these against the next time I am involved in an
>>>>>> effort like MMX.  Where shortsighted people will say "all we need
>>>>>> to support is 4:2:2 YUV, and we should add a special instruction
>>>>>> just for that", rather than more general support.  Or vice versa.

>>>>> Please consider putting your collections on bitsavers,
>>>>> or somewhere, so that the knowledge doesn't get lost.

>>>> I am putting my collection on comp-arch.net --- my own personal
>>>> wikipedia.  I had not considered it as Bitsavers, but fair enough.

>>> Not wiki.   Definitely not wiki.  I don't care where as long as all
>>> your tradeoffs, which is the most important piece of designing
>>> but never gets published.  In our biz, our mistakes and initial
>>> desing reasonings rarely got put down on paper.

>>>> Why "my own personal" - well, I started putting it on wikipedia,
>>>> but got tired of stuff being deleted by the wikipedia trolls.

>>> I have yet to believe anything on wiki.

>> More fool you.

>>> I think of it as a non-feature and eventually dangerous.

>> More fool you.

>>>> Why not "bitsavers"?  Well, my impression is that bitsavers is
>>>> about saving manuals, etc.

>>> Well, I don't care where as long as the thinking processes oof those
>>> designs are preserved.  :-)  Making computers represent numbers
>>> and do mere arithmetic is hard and most people, even those who
>>> become bit gods in other areas, have no idea about the problems.
>>> Training new people to design these things is in the startosphere
>>> of techniques, IMO.  Every kiddie is going to have a wonderful
>>> idea and won't discover the pitfalls until too late.

>> Kiddies dont design computers.

> They do.

Not anymore.

> Getting it to market requires experience.

There arent any done by kiddies anymore. 

[toc] | [prev] | [standalone]

Back to top | Article view | comp.arch

csiph-web