Latest Version

Questions and answers for COW Macintosh and Windows
MikeAtBookup
Posts: 187
Joined: Mon Aug 01, 2016 2:02 pm

Re: Latest Version

Post by MikeAtBookup »

Bogg wrote:As I mentioned before 10B n/s making a limit of 100 million nodes instead of 100 thousand nodes no improvement and hence still broken.
The ridiculous candidate font size hasn't been fixed. Please remember to place the backsolve info side by side when this is fixed.
Fixing the piece siding speed bug is pleasant but its not like you fixed anything usefull.
I'm on a laptop at my in-laws through the holidays so I cannot check this to be certain but you should be able to set EPD analysis to 10,000 or more which equates to ten billion nodes.

You'd be surprised at how many of the "older" customers are praising the "ridiculous" candidate font size. :) It's on my list to allow adjusting it. What did you mean by placing the "backsolve info side by side?" I intend to try reorienting the Pedigree ribbon left to right rather than vertical so that they are more easily compared.

You mentioned wanting your money back. Did you purchase recently? These upgrades were all free to everyone who purchased over the last four years.
-------------------
Mike Leahy
:geek: Head Geek at bookup.com
Bogg
Posts: 26
Joined: Mon Sep 19, 2016 8:06 pm

Re: Latest Version

Post by Bogg »

1) 10B nodes is useless. I state again my 2 year old rig does 10B nodes per second. That is a limit of 1 second. Who do you think is satisfied with that? Who do you think even wants this parameter? or max ply for that matter. Perhaps you could clue me in on some use for either of them.
2) If the font size for the candidates is lowered and you keep the backsolve information stacked then as they are 1/2 the font size, probably around 8, they will be unreadable.
3) Yes I have purchased recently. I don't see where that matters. This release and all future releases are not free, they were paid for with our qualifying payments. If I had purchased earlier I would already have been yelling for my money back. No update for 4 years, I'm taking your word for this, and then this piece of ... is what you present.
MikeAtBookup
Posts: 187
Joined: Mon Aug 01, 2016 2:02 pm

Re: Latest Version

Post by MikeAtBookup »

Bogg wrote:1) 10B nodes is useless. I state again my 2 year old rig does 10B nodes per second. That is a limit of 1 second. Who do you think is satisfied with that? Who do you think even wants this parameter? or max ply for that matter. Perhaps you could clue me in on some use for either of them.

There are quite a few theorists who are interested in comparing depth ply analysis for different engines for their own reasons. Similarly they like to compare analysis done with equal amounts of time.

On my two year old single core HP desktop, I get around 10 million (not billion) nodes in about three seconds if memory serves. You might be projecting that every chess student is using a rig comparable to your own. Most COW users are reasonably satisfied with 100 million nodes of analysis which, with the more recent engines, would best most grandmasters. I would call that fast and far from useless.

The program still uses EPD files precisely because those with powerful engines and powerful computers are likely to do their own analysis outside of the COW program and then import the results after.

I'm still not clear why changing the setting to 1000000 (one trillion nodes) wouldn't work for you.

The impetus behind the new version is to migrate from a rather brittle database schema that is limited to a dozen GB per tree to a single file model that can handle 140TB per tree. The move also allows using ones data on portable devices and on Macintosh. It required an almost complete re-write of the program.
-------------------
Mike Leahy
:geek: Head Geek at bookup.com
Bogg
Posts: 26
Joined: Mon Sep 19, 2016 8:06 pm

Re: Latest Version

Post by Bogg »

1) I didn't ask why anyone would want time, time is obvious, it is the only limit that matters. If someone wants to use ply or even nodes for some arcane reason fine, I am only doing analysis and those limits are worthless and get in the way.
2) I would set the number of million nodes per second to a large number if the software allowed, but it doesn't. It is only possible to change one of the parameters if that. If you change a second it resets the other to its default. I am unable to do further testing to verify that they aren't all like the slide speed was (only staying changed while you looked at it).

I wouldn't be complaining if it was possible to set the two limits that are useless to some ridiculous number to get them out of the way. A 1 trillion node limit would work fine for me as long as I can also change the time and max ply limits. If the upper limit is strictly 1 trillion nodes I would guess that there would be some other users with machines that run 5B+ n/s that might want to analyze longer than 3 minutes but again that is only a guess.

My point is you shouldn't be placing upper limits other than perhaps 2^63 (10^21). Why do you hamstring your product making features usable only for some? If even one user is dissatisfied you made a mistake. What most users are satisfied with is irrelevant. Thinking you are doing a good job because your random, arbitrary limits satisfy most of your customers is idiotic. Good programmers don't set random, arbitrary limits. My guess is that back in the old days you were stuck with 2^31 and for some reason don't see fit to move into the 64-bit world. It can't be because you are using an out of date computer or compiler. I seem to remember you saying that you had 40K users. That would mean that you made over 6 million on COW sales alone.

P.S.
a 100M node limit using Komodo or stockfish probably won't beat an A player. That limit would be hit in .1 seconds on even slow computers.
MikeCrockett
Posts: 21
Joined: Sun Aug 28, 2016 3:15 pm

Re: Latest Version

Post by MikeCrockett »

It seems Mr. Bogg's won't be satisfied no matter what explanation he is given. Unfortunately, trollish, whining behavior such as this seems to be a common issue with BB forums. It wouldn't surprise me at all if Mr. Boggs happened to be a 3rd party software provider attempting to denigrate one of his competitors. Even assuming he is a legitimate customer, he should know that no power user, trying to squeak the tiniest bit of extra computational performance from his computer would perceive much difference when the end result is measured in centipawns.
Bogg
Posts: 26
Joined: Mon Sep 19, 2016 8:06 pm

Re: Latest Version

Post by Bogg »

Mr. Crockett:

I am not certain of you qualifications so please forgive the insult if you have any. I have been a master for ~30 years and a highly sought after software guru for ~35. Like many people accusing others of being trolls I doubt that you have even the faintest understanding of the discussion elsewise you would have kept silent. Mike gave no reason for the unnecessary limitations that he placed on his customers' use of his software. He only said that most users are more or less satisfied. His job is to, when possible, satisfy all of his customers and here it was possible. As I said, good software designers do not imbed unnecessary limits in their software, period.

What third party software are you referring to? Mike's software fills a niche that I don't think anyone has bothered to compete in. Additionally, I have told Mike more than once that his old software, pre-COW, was a tough act to follow. I am complaining because these particular design flaws have existed for a long time and I can't think of any reason that I would find acceptable for not fixing them, especially during a rewrite.

If you were a stronger chess player you would realize that even stockfish and komodo, the two best chess engines by miles, sometimes need to do very deep analysis, 30 minutes or more on my computer and it sounds like 6+ days on Mike's, to arrive at a stable answer. I am not talking about 1 or 2 centipawns but 20 to 50 and on rare occasions more. Sometimes the engines just miss the mark, just like we humans.

The only reason this has been brought up in this forum is because Mike asked me to. (The reason Mike gave may answer some complaints. Mike said he gets to much email and sometimes a message gets deleted without being read. So if you are being ignored and were trying to communicate via email try here before ranting about Mike being a jerk.) I don't like pointing out mistakes in front of others. That should be done in private, IMO. Plus, it is bad enough looking like a donkey in a private conversation if I am wrrr-wr--not right.

Lastly, it is not Boggs and there is no mr. it is just Bogg. :lol:
MikeCrockett
Posts: 21
Joined: Sun Aug 28, 2016 3:15 pm

Re: Latest Version

Post by MikeCrockett »

Bogg wrote: Mike's software fills a niche that I don't think anyone has bothered to compete in. Additionally, I have told Mike more than once that his old software, pre-COW, was a tough act to follow.
Bogg wrote: I'm taking your word for this, and then this piece of ... is what you present
Bogg wrote: Make the fixes ... by 2016/12/31 or return my money. My next step is BBB and small claims court.
You clearly have been a long time user. You have benefitted from his initial product or you wouldn't have bought it. He has provided updates as promised. You do not own the software nor have you purchased a license to acquire the rights to the modify the software yourself. You were only granted a right to use it, as-is.

Now you want to get ugly with him, threaten him with a small claim lawsuit as if you are somehow owed something? You may be a good chess player but your interpersonal skills leave a lot to be desired. The best you can hope to achieve is harassment. That doesn't help anyone or anything.

Are there issues that need to be fixed? Yes. No argument. Mike's quality control leaves much to be desired, but he does try to address them as he can. For a one-man programming shop, he has done remarkably well. Normally products of this nature require a team of programmers and business management professionals and that type of overhead drives up the cost. If you take the time to google Mike's business address you'll learn he is operating out of his home, and yet he manages to provide free software updates which no one else can reasonably afford to do.

As far as Quality Control goes, I do wish Mike would focus on correcting the existing issues in his program code before cloning them to another operating system. Copying the existing flaws to another platform simply multiplies the maintenance problems. With that said, fixing little bugs doesn't bring in sales revenue either and it's a tricky balancing act that all business owners need to work out for themselves. With that said, Mike shouldn't simply assume that all his customers are happy simply because they are being complacent.

Does Bogg's issue deserve attention? Yes. Does that mean it's the most important issue? No. It's just the squeaky wheel syndrome.
MikeAtBookup
Posts: 187
Joined: Mon Aug 01, 2016 2:02 pm

Re: Latest Version

Post by MikeAtBookup »

Bogg is correct that the EPD setting reverts to 100 million if you set it to anything higher than 100. Mea culpa. It's on the list to fix...

http://www.bookup.com/home/chess-openin ... am-status/

As for posting conduct, I am loathe to delete anyone's comments. I do ask that folks remember that kids use this software as well as this forum. I'd refer you to https://www.chromium.org/conduct but the most important tenet is to have every discussion be respectful and kind.
-------------------
Mike Leahy
:geek: Head Geek at bookup.com
Bogg
Posts: 26
Joined: Mon Sep 19, 2016 8:06 pm

Re: Latest Version

Post by Bogg »

Mike:

I wrote a lengthy reply earlier but evidently forgot to save it. I will make it very short, for me :lol: , this time.
1) I made the error that I make over and over, my rig does ~10M n/s not B. That would make being able to enter 100,000M in the epd sufficient for about five years i guess. A general note to readers: nodes are meaningless on a multicore rig. Make sure to turn off hyperthreading if you want the best results as x cores have less overhead (wasted work) than 2x half cores. On my windows 7 pro rig setting the thread count to the physical core count with hyperthreading enabled only uses half of the processor. Perhaps someone else has different results.
2) I am complaining primarily because I am annoyed. Unless you add the ability to complete the current ply I have very little use for the analyze epd feature, not none but very little.
3) I still want the ability to change the font size for the candidates window. Having a huge clear font is nice but not at the expense of the comments window. It may only be me that hates needing to scroll to see some candidates. If you decide to make this change remember to not stack the backsolve info else it will be unreadable at just about any font/2 size.
4) I really hate the new scroll bar functionality in the comments window. Clicking just jumps back and forth to the top or bottom. I have yet to find a convenient way to edit a pasted comment that goes to much over one window size.
MikeAtBookup
Posts: 187
Joined: Mon Aug 01, 2016 2:02 pm

Re: Latest Version

Post by MikeAtBookup »

Bogg wrote:Mike:
1) I made the error that I make over and over, my rig does ~10M n/s not B. That would make being able to enter 100,000M in the epd sufficient for about five years i guess.
That's what I thought. The version released today allows up to 2,000 nodes (in millions) just in case you do some day get a rig that can handle a billion nodes. :)
Bogg wrote:3) I still want the ability to change the font size for the candidates window. Having a huge clear font is nice but not at the expense of the comments window. It may only be me that hates needing to scroll to see some candidates. If you decide to make this change remember to not stack the backsolve info else it will be unreadable at just about any font/2 size.
Done in today's build 68 release for Windows (and 67 for Macintosh).
-------------------
Mike Leahy
:geek: Head Geek at bookup.com
Post Reply