that v2 was re-written from scratch, etcI never claimed that for the simple reason that I don't have the slightest idea whether this is true or false and if I have to guess I wouldn't believe this is true.
that v2 was re-written from scratch, etc
I never claimed that for the simple reason
that I don't have the slightest idea whether
this is true or false and if I have to guess I
wouldn't believe this is true.
On July 31, 2023 at 10:44:40 AM UTC-6, Frank Berger wrote:o.k. no problem at all.
Sorry, Frank. I didn't mean that you made both
claims. I have a bad habit of referring to many
people and many things said in one sentence,
which is fine if I don't name anyone but if I do,
then I should delineate which comments were
made by him also.
What may make your claim interesting is that one
hardly ever starts coding new software, especially
eXtremely complex and graphical one, in assembly
language, while disassembling executable files do
generate assembly language code.
Does anybody actually *like* Hungarian notation? It's ugly, and would
seem to be useless given today's IDEs where you can right click on a
variable and bring up the definition immediately.
Code that's full of names like lpszcfbbrFoobar to indicate a long
pointer to a string terminated by zero coded on a friday by someone
who's initials are br serve mostly to get in the way of understanding
the code.
But I'd rather deal with that than assembler. YMMV.
Way back when (ok, late 70s), when Fortran/Pascal/
Cobol were emerging, they were considered horribly
slow. For 'real' applications, Assembler was the only
way. It was fast and small in a time when computers
had virtually no processing power, tiny memory and
limited or no external storage.
Now it's all moot because slow, inefficient software
can be easily hidden by fast machines. Hence
bloatware everywhere.
Sorry, no bg content :-)
MK schrieb am 1. August 2023 um 00:59:48 UTC+2:
What may make your claim interesting is that one
hardly ever starts coding new software, especially
eXtremely complex and graphical one, in assembly
language, while disassembling executable files do
generate assembly language code.
Back in 2005 or 6 Chiva mentioned to me that he
met an programmer in the US who were working
on an incredible good and fast BG AI written in
Assembly. I was very sceptical, because Assembly
would not be my first choice given it's pros and
cons and for obvious other reasons. I haven't asked
Chiva later but it was very clear that it was Xavier.
In personal communication Xavier confirmed that the
AI was written in Assembly
and there is simply nothing was could otherwise explain
the performance advantage.
I regard it as very unlikely that Xavier has taken over
some code. As I wrote: different languages, ....
different architecture, probably different learning
mechanism and different playing strength.
I feel GnuBG code very hard to read (one reason: I hate
Hungarian notation).
MK schrieb am 2. August 2023 um 14:44:20 UTC+2:
"the assembler code will be between 5% and 20% faster."
this is probably the wrong sign "%" should be
replaced with "times" that would make sense
and matches the reality. Would be stupid to
invest 20 times the time to gain 20%.
"You were
one of the first ones to warn about possible violations
of Gnubg's GPL. See this post of from you in May 2009:"
Funny. I don't find the article on gammon village where
my assumption was/might be based upon, but I simply
was wrong by miles.
With the current information I don't see slightest hint
supporting that.
they are as similiar as: https://en.wikipedia.org/wiki/
Twins_(1988_film)
On August 2, 2023 at 4:09:43 PM UTC-6, Frank Berger wrote:
MK schrieb am 2. August 2023 um 14:44:20 UTC+2:
"the assembler code will be between 5% and 20% faster."
this is probably the wrong sign "%" should be
replaced with "times" that would make sense
and matches the reality. Would be stupid to
invest 20 times the time to gain 20%.
You need to learn to read carefully. The word just
before the sentence you re-quoted is "however"!
So, he is indeed contrasting the time invested vs
speed gained. Apparently he is stupid... ;)
I don't think that advices like "You need to learn to read carefully" are particularly helpful, but if you use it you should be sure that you read it correctly yourself. As Tim correctly pointed out I believe it should be times not %. An assemblerthis is probably the wrong sign "%" should beYou need to learn to read carefully. The word just
replaced with "times" that would make sense
and matches the reality. Would be stupid to
invest 20 times the time to gain 20%.
before the sentence you re-quoted is "however"!
So, he is indeed contrasting the time invested vs
speed gained. Apparently he is stupid... ;)
http://www.gammonvillage.com/backgammon/news/article_display.cfm?resourceid=5790Ah ok. I tried to go back in GammonVillage but was stuck earlier. This is a typical Simborg article and he say only on XG that it is in beta and he helps and that it will come out soon.
As I told it. Different implementation language, much higher speed, different architecture (unless 3 is many for Xavier), better playing strength, probably different learning algo (this a bit of interpretation of Xaviers posting where I conclude that heWith the current information I don't see slightest hintCan you tell briefly why do you think you were wrong
supporting that.
"by miles" and what current information do you have
that made you change your mind?
Fortunately, it still exists at archive.org at this url:have you seen the date of the article?
https://web.archive.org/web/20090224005827/http://www.gammonsite.com/extremebg.asp
You all should read and print it for your records. The
second sentence there says "The Neural network is
derived from GNU BG 0.00" exactly as you had quoted
in your Bug-gnubg post and the very last sentence says
"Credit is given to Gary Wong for his work on GnuBg".
What's also eXtremely intrigueing is the one sentenceHave you read the article you refered to? "eXtreme Gammon started when Oasya (Snowie) told us they would start charging money to have their program running on GammonSite. As we did not want to (and couldn't) pay the requested price, I started to work on
before the last: "Master Gammon/Master Backgammon
computer program (in cooperation with Oasya)". What!?
Does anyone has a clue as to what this is referring to?
I would expect from you better than trying to distractI tend to get ironic when confronted with theories I regard weird.
from a serious subject with silly shit like that... :(
On 8/4/2023 4:21 AM, MK wrote:
On August 2, 2023 at 4:09:43 PM UTC-6, Frank Berger wrote:
MK schrieb am 2. August 2023 um 14:44:20 UTC+2:
"the assembler code will be between 5% and 20% faster."
this is probably the wrong sign "%" should be
replaced with "times" that would make sense
and matches the reality. Would be stupid to
invest 20 times the time to gain 20%.
You need to learn to read carefully. The word just
before the sentence you re-quoted is "however"!
So, he is indeed contrasting the time invested vs
speed gained. Apparently he is stupid... ;)
The article at bkgm.com indeed says "%" rather
than "times" but Frank is saying that he thinks
that whoever transcribed the interview made a
mistake, and that Xavier actually said "times".
As Tim correctly pointed out I believe it should
be times not %.
An assembler programmer that is only able to
get 20% against Delphi code would be highly
incompetent.
Given that XG is a lot faster than GnuBG or
BGBlitz supports my point.
Can you tell briefly why do you think you were wrong
"by miles" and what current information do you have
that made you change your mind?
As I told it. Different implementation language, much
higher speed, different architecture (unless 3 is many
for Xavier), better playing strength, probably different
learning algo (this a bit of interpretation of Xaviers
posting where I conclude that he uses some kind of
reinforcement lerning but I maybe wrong wheras
GnuBG uses supervized learning).
Fortunately, it still exists at archive.org at this url:
https://web.archive.org/web/20090224005827/http://www.gammonsite.com/extremebg.asp
You all should read and print it for your records. The
second sentence there says "The Neural network is
derived from GNU BG 0.00" exactly as you had quoted
in your Bug-gnubg post and the very last sentence says
"Credit is given to Gary Wong for his work on GnuBg".
have you seen the date of the article?
What's also eXtremely intrigueing is the one sentence
before the last: "Master Gammon/Master Backgammon
computer program (in cooperation with Oasya)". What!?
Does anyone has a clue as to what this is referring to?
Have you read the article you refered to? "eXtreme
Gammon started when Oasya (Snowie) told us they
would start charging money to have their program
running on GammonSite....
So GammonSite replaced Snowie with GnuBG in 2004.
2009 XG came out. That the 2009 code is still based on
GnuBG code?
I regard the probability as very low.
Unless the XG code will be available there will be no
proof in any kind.
I would expect from you better than trying to distract
from a serious subject with silly shit like that... :(
I tend to get ironic when confronted with theories I
regard weird.
But even not assuming anything and going only by
the text, Xavier makes two related statements with
the word "however" linking the two. If you look up
the definition of "however", you'll see that it means
"but", "yet", "on the other hand", "despite that", etc.
indicating a negative correlation...
Simborg is a very experienced interviewer. There
is no reason to think that he made a transcribing
mistake that nobody, including himself and Xavier,
caught in an interview that was published in many
places including his own web site. Are you two the
only smarties to catch it now 10+ years later?
On 8/4/2023 7:58 PM, MK wrote:
..... indicating a negative correlation...
The word "however" makes sense either way.
In fact, it makes slightly more sense with "times."
Consider the following analogy.
"I spent a lot of money installing solar panels on my
house; however, they have paid for themselves in
energy savings over the last five years."
... "however" makes perfect sense in this context,
... was a benefit that justified the investment.
This is entirely analogous to saying that you spent
20x as much time writing code in the first part of a
sentence, and then explaining that there was a
benefit that justified that investment.
With "%" instead of times, there is still a contrast
between expenditure with payoff, but the payoff
is small.
contrast between a large expenditure and a small
payoff is *less* than the contrast between a large
expenditure and a large payoff, so the 'however' is
less convincing.
On August 1, 2023 at 4:55:25 PM UTC-6, Simon Woodhead wrote:
Way back when (ok, late 70s), when Fortran/Pascal/
Cobol were emerging,
Daily backups, weekly payroll, etc. took hours to run
and required changing tape reels, hard disk platters,
etc. That's why there was a breed called "computer
operators" and they mostly worked night shifts...
On Wednesday, August 2, 2023 at 7:02:16 AM UTC-3, MK wrote:
On August 1, 2023 at 4:55:25 PM UTC-6, Simon Woodhead wrote:You two either are kids or are senile.
Way back when (ok, late 70s), when Fortran/Pascal/
Cobol were emerging,
In early and mid 70s Fortran was already well established, almost a standard. It came out in the 60s, or even before.
Daily backups, weekly payroll, etc. took hours to runNo, in the data centers they were called "system operators", and there were day and night shifts, non-stop. It was the "time sharing" era, you'd pay for the CPU time and peripherals used.
and required changing tape reels, hard disk platters,
etc. That's why there was a breed called "computer
operators" and they mostly worked night shifts...
You'd rather be in good terms with them, or they'd put your lot of punch-cards the last in the deck of jobs, so your printout would take an eternity to be spitted out.
On August 2, 2023 at 7:02:16 AM UTC-3, MK wrote:
You two either are kids or are senile.
I wrote to Simborg, asking whether "%" versus
"times" was a typo.
He wrote back promptly, saying, "You may as
well be talking Mandarin to me. I have no idea
what those squiggly numbers mean. But I have
copied Xavier on this and maybe he can clarify."
But at least it's clear that Simborg could very
well have made a mistake in transcription, since
by his own admission he has no intuitive feeling
for what the numbers "ought" to be.
On 6/08/2023 5:12 pm, Grunty wrote:
On August 2, 2023 at 7:02:16 AM UTC-3, MK wrote:
You two either are kids or are senile.
Probably both.
... they were called "system operators", and
there were day and night shifts, non-stop.
Yes, I was one of them. I learned to play
backgammon on the night shift.
I learned how to program from that experience,
Mark IV and Assembler.
It also made me curious whether you are and/or
actually speak Mandarin (and whether he knows
the answers)?
Even if he didn't, I would have expected an answer
indicating that such a mistake is unlikely because
he would have had Xavier sign off on the final text
before publishing it, as any responsible, respectful
interviewer would/should have done.
But, why are you guys still dwelling on this? What if
Xavier says he meant "times", not "%"? Will you then
believe him that Assembler code can run "20 times"
faster then Delphi code..?
The typo could also have been introduced by bkgm.com. That is, maybe
Xavier signed off on it, but "between 5x and 20x" got changed to
"between 5% and 20%" in the process of reformatting for the bkgm.com
website.
On 8/7/2023 8:27 AM, I wrote:
The typo could also have been introduced by
bkgm.com. That is, maybe Xavier signed off
on it, but "between 5x and 20x" got changed
to "between 5% and 20%" in the process of
reformatting for the bkgm.com website.
Actually, it seems that this possibility can be
eliminated, because "5% and 20%" is what it
said on Simborg's own website.
Another possibility is that there was a typo,
and Xavier was given a chance to spot it, but
he was in a hurry and missed it.
On 8/6/2023 6:07 PM, MK wrote:
It also made me curious whether you are and/or
actually speak Mandarin (and whether he knows
the answers)?
I do speak Mandarin, but not as well as I speak
Hakka or Cantonese.
But, why are you guys still dwelling on this? What
if Xavier says he meant "times", not "%"? Will you
then believe him that Assembler code can run "20
times" faster then Delphi code..?
Believing him is a separate question. It's of interest
to know what he is claiming, regardless of whether
the claim is credible.
I haven't yet stated my personal opinion of what
Xavier actually said. To be honest, 5% seems low
to me and 20x seems high to me.
It would surprise me to learn that Xavier would spend
20x the effort to achieve only a 5% speedup.
On the other hand, I would also be surprised to learn
that a 20x speedup can be expected.
But I can believe that there might be some situations
where there are data structures in Delphi that make it
very easy to program certain things, but which end up
being very slow because .....
I don't think this would happen often, but it only has
to have happened once for a claim of a 20x speedup
to be true.
Good to know. ;) I always wondered if some Greek
people get offended by the expression "It's Greek
to me" that's been used since Roman times. For a
second, I wondered if what/how he said offended
you?
I had already pointed this out to you, without
giving a link, in replying to one of your posts a
few days ago, when I wrote:
"There is no reason to think that he made a
"transcribing mistake that nobody, including
"himself and Xavier, caught in an interview
"that was published in many places including
"** his own web site **.
On 8/7/2023 8:12 PM, MK wrote:
I had already pointed this out to you, without
giving a link, in replying to one of your posts a
few days ago, when I wrote: ....
"that was published in many places including
"** his own web site **.
Simborg's Backgammon Learning Center website
is dead. That's why I had to go to the Wayback
Machine. Are you referring to some other website
that Simborg maintains?
Most shocking was to read you say that his site
was dead. Sure, the links to the exact page with
the interview weren't resolving but I'm sure the
site itself was accessible while I was looking for
the interview page there.
On 8/9/2023 6:52 AM, MK wrote:
Most shocking was to read you say that his
site was dead.
Your investigations have also been valuable
to me, because I see now that the new site
https://www.backgammonlearningcenter.com
is alive whereas the old site
https://www.thebackgammonlearningcenter.com
is dead.
I had not thought to remove "the" from the URL.
But the interview with Xavier does not seem to be
on the new site.
Sysop: | Keyop |
---|---|
Location: | Huddersfield, West Yorkshire, UK |
Users: | 297 |
Nodes: | 16 (2 / 14) |
Uptime: | 17:04:56 |
Calls: | 6,667 |
Calls today: | 1 |
Files: | 12,216 |
Messages: | 5,336,853 |