Newton stability tests

A place to discuss everything related to Newton Dynamics.

Moderators: Sascha Willems, walaber

Newton stability tests

Postby Leadwerks » Fri Apr 28, 2017 12:32 pm

Is the source code for your Newton / Bullet / PhysX tests available anywhere? I've got people on my forum who insist "PhysX is the best because Nvidia" even when I am showing them proof:
http://www.leadwerks.com/werkspace/topi ... wton-sails
User avatar
Leadwerks
 
Posts: 515
Joined: Fri Oct 27, 2006 2:54 pm

Re: Newton stability tests

Postby Julio Jerez » Fri Apr 28, 2017 12:56 pm

those are the nvida and bullet test written and set up by nvidia and bullet. I just adjusted newton because they try to set newton the way the set PhysX.
you can read the conversation with mr Pierre here.
I can list here the emails form nvidea and why the do not want to include newton as integrated by my
viewtopic.php?f=9&t=8836

it is nVidia that changes the test of the competitor because they is that tool to the develop to use their engine. so they refuse to integrate other engine properly.

but like I said, you are not going to convince a user of the contrary, most use think that polarity, or running fast, is equal to correctness, and when a company spend the on million of dollar a year to maintain a narrative, there is no way that can be changed.


if they want to compared and remove bias,
there is a Unity Newton Plugin that the can use to compare Newton and PhysX side by side.
you should probably integrate PhysX and bullet that way you can get rid of that constant war.

let them shoes what they think is bet for them.
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Newton stability tests

Postby Leadwerks » Fri Apr 28, 2017 1:11 pm

Thanks for the link. This is the proof I wanted.
Julio Jerez wrote:there is a Unity Newton Plugin that the can use to compare Newton and PhysX side by side.

Dude, you're killing me. Do you realize who you're talking to? :roll:
User avatar
Leadwerks
 
Posts: 515
Joined: Fri Oct 27, 2006 2:54 pm

Re: Newton stability tests

Postby Leadwerks » Fri Apr 28, 2017 1:20 pm

How do I play a demo? The test.script.txt plays a bunch of different scenarios but switches every five seconds. The other scripts don't seem to do anything.
User avatar
Leadwerks
 
Posts: 515
Joined: Fri Oct 27, 2006 2:54 pm

Re: Newton stability tests

Postby Julio Jerez » Fri Apr 28, 2017 1:25 pm

Oh I am sorry, I just was trying to say that trying to explain a person a fact is very difficult when the person do not want to understand.
the only way you will get your users away from comparison war and grass is greener of the other side of the fence discussions is if you integrate PhysX in your engine.

here is the email I got from nvidea when I submitted my integration and despite the test show that Netwon was in fact competitive and in many cases better engine, they decided not to use my integration and went with their own bogus newton version.

From: John Ratcliff [mailto:jratcliffscarab@gmail.com]
Sent: Thursday, June 06, 2013 10:48 PM
To: Julio Jerez; Jerez, Julio
Subject: Re: PEEL tests

Tomorrow I will add the Bullet peel plugin to the depot so we can add it to the data set.
On Jun 7, 2013 12:34 AM, "John Ratcliff" <jratcliffscarab@gmail.com> wrote:
Julio,

I'm about to head to bed now; but I've been going through some of the tests; which I assume you have done as well.

Here are some things I have noticed

#1 : Newton seems to be a lot more efficient with compound shapes than PhysX. This is especially noticeable in test #59 and #60

#2 : Newton handles interpenetration much better than PhysX.

#3 : Newton seems to have a much more aggressive sleep algorithm than PhysX; as can be seen in large box stack tests where PhysX falls over but Newton puts them to sleep fairly quickly. This isn't necessarily a good or bad thing; it's usually just a tuning parameter.

#4 : All of the demos involving constraints where there are long chains or, especially the 'nets', the behavior of Newton looks *much* better. The performance is lower, but that's almost for sure simply because the solver-iteration count on the PhysX constraint is too low. I'm sure if we increased the solver-iteration count on the PhysX constraint to make the behavior match more closely to the quality of the behavior in Newton the performance would be much more close.

#5 : The ragdoll demo looks better in PhysX than in Newton. This may be just because you haven't fully hooked up all of the constraint properties yet? The PhysX constraints have a lot of spring/softness to them that tend to make ragdolls look a little nicer.

#6 : In general Newton seems to be very performance competitive with PhysX until you start hitting extreme situations of massive numbers of dynamic objects all in a giant pile. I'm going to assume this is what you meant by an artificial demo. I agree this isn't necessarily that big of a deal.

As we discussed earlier, you have to be careful about drawing too many conclusions too quickly, since many differences could be due to iteration counts or other kinds of tuning parameters.

These are just some things that I have observed so far.

You are putting a lot of work into getting Newton to work with PEEL; I think we should discuss a plan to present the data when we are finished. Perhaps something on either your website or my blog. Something which is diplomatic and complimentary.

In my personal view, Newton is a very impressive high quality physics engine with excellent behavior and performance. The fact that it is even competitive with commercial physics engines that have had dozens of engineers working on them for years is really an incredible testament to your talent and dedication.

I think you have done an amazing job, and we should communicate that to the public as well.

Great work.

John
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Newton stability tests

Postby Julio Jerez » Fri Apr 28, 2017 1:27 pm

you just run the executable and select the engine you want to compare and then select different demos for the menu with the space bar.
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Newton stability tests

Postby Julio Jerez » Fri Apr 28, 2017 1:32 pm

there is a newer version of that tool that was release later but as usual the did not integrate any of the new changes so the continue to straw manning the competitors.
There is not much I can do because NVidia is extremely dishonest and has lot of power and money to silence the competitors in almost every area.
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Newton stability tests

Postby Leadwerks » Fri Apr 28, 2017 1:33 pm

Dude, you have a quote from Nvidia saying Newton is better. I love it! :lol:
User avatar
Leadwerks
 
Posts: 515
Joined: Fri Oct 27, 2006 2:54 pm

Re: Newton stability tests

Postby Julio Jerez » Fri Apr 28, 2017 1:57 pm

Not from NVidia, but from one of the people from NVidia who asked me to integrate Newton to that tool, you will be surprised of the thing these people think about newton in private.

this is being going on for almost five years now and is very difficult to fight with such a dishonest player like the PhysX folks. To make is clear, that tool was an internal tool NVidia uses to compare their stuff vs the competitors. NVidea is very rudeless, their main competitor was havoc for many years. With the increase popularity of Bullet they decide is was time to address them as well so the added engines like Bullet and later Newton.

Their Newton integration was really bad, so there never bothered to make correction until one day while working at the same company, that man showed me that tool since they were not going to be able to hide since I was the physics programmer so the offered me to integrate Newton the best I could. They never though that newton was going to even work.

I had to sign an NDA and after about few weeks of work they sent a letter basically sayin newton was good but that the will not publish it, instead they will make a some blog or whatever which they never did. I was removed from source control and I could not say anything for more than a year until they decided to make it public.

When they made it public they did not incorporated a single line of code of my integration of Newton
and when with the bogus one the show to every developer when the go to make their presentations.
after that the release more version and is more of the same, basically NVidia straw man the competitors.
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Newton stability tests

Postby Leadwerks » Fri Apr 28, 2017 1:59 pm

Well, your demo converted one of my users:
http://www.leadwerks.com/werkspace/topi ... ntry106438
User avatar
Leadwerks
 
Posts: 515
Joined: Fri Oct 27, 2006 2:54 pm

Re: Newton stability tests

Postby Julio Jerez » Sat Apr 29, 2017 1:39 pm

I read the thread, Isn't funny how your user was so 100% with that the test was bias, base solely on the big studio uses PhysX, and how was he sure there were hundreds of test that will show otherwise but he could not produce a single one.
The proof isn't right infront of my eyes, his test (which is bias just like every other test that claims one physics engine trumps the rest) shows that for his implementation of each physics engine to do the same thing, he produced the best results with Newton. Who's to say I wouldn't be capable of doing that with any other physics engine?


and then the same user end up saying this.
I've got to say I never thought Newton was THAT good of a physics engine, very underrated i must admit. You do tend to pay in most demos in performance for Newtons accuracy though.


He still thinks performance is what makes PhysX good, this is like saying you have a race car in say Nasca or Indianapolis and this car is two or even three times faster than any other car but the engine blows up on the 50's lap. To an spectator that only saw the first 25 laps, there is not why you can explain him that to claim victory you need to win the race.

This is what Physx does in those large test, what good it is to be very fast but very wrong? as if going fast somehow is going to make up for the total incorrect and erroneous physics solutions.

When they make a pyramid of 30 row high and say you see we are faster that anyone else, what they do not tell you is that if you wait long enough the pyramid collapses, and this happens to stacks that are as low as 20 rows high, you can't claim to be the fastest if you do not solve the problem.

But that is what propaganda does to people, they can only see they first few second and conclude what the propaganda and misrepresentation is design to convey.
We live and a world where presentation and image is more important than facts and reality. We are seeing in real time now with alternative Facts.
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Newton stability tests

Postby rokups » Thu Dec 06, 2018 3:51 am

they decided not to use my integration and went with their own bogus newton version.


Is your integration available anywhere? Me and some people would love to try your version instead of one at https://github.com/Pierre-Terdiman/PEEL
rokups
 
Posts: 1
Joined: Thu Dec 06, 2018 3:48 am

Re: Newton stability tests

Postby Julio Jerez » Thu Dec 06, 2018 12:44 pm

It is Thursday today, I just compiled and after fix some compiler errors, it seems to work, for most demos.

If you give me over the weekend so that I can test the rest, and remove all the shenanigans Nvidia team think they have to saddle the competitor, just because their stuff does not do well, and if there did not it will make their stuff look bad.

Remember these test are just synthetic nonsense made by Nvidia to bamboozle developer into thinking that physic was better that havoc, it just exploded on them with so many other alternative engines.
The demos are all riddle with hacks that apply only to physx, stuff like having a mass and an inertia, and apply that to all engine, but internally physic multiplying the inertia by 10, but not for the other.

Those hacks only an experienced person in numerical simulation can spot but they were not made for experience people where made for potential client that know jack about physics.
The only thing physic has over newton is that it runs faster. Newton has far more feature, is more accurate and more stable. That I can say because newton simulate the equations of motions honestly, not inventions of brand new law of physics that work in some scenario but not others. the only thing they can do is to approach newton but never surpace it, of course they will never admit that, and the moment they get something better, they make press release with ridiculous claim "we are the only one doing this, the only one doing that and so on" never recognizing that many of us were doing that long before, and the reason they can do it is because the have lot of money to hire an army of people to do what other has done but was never recognized.

Anyway, on Saturday I will check that all demos
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Newton stability tests

Postby Julio Jerez » Mon Dec 10, 2018 1:02 pm

I did some check, but I could not complete it.
Basically since we deprecated visual studio 2010, the engine not longer integrate with that tool, so I have to spend the majority of the weekend making the changes to the tool so that I can be opened and compiled with vs2013.

Vs 2013 is the most that I could get it to compile, that tool is among the worse pieces of code I have ever seen in more than 30 years making a living as a professional programmer.
You find stuff the that is not only inline assembly, no no no, the engeneers at Nvidia decided inline assembly is not enough, they went a step further and used inline machine code, stuff like emit intrinsic, cpp keywords uses as variables, it is absolutely dreadfully.

Because of that, it can't be open with any version of visual studio newer than 2013 and even that requires ton of refactoring.

Anyway after I did that, last night I have it running, and I test the first 125 demos, some of the do not run because of the shenanigans added by Mr Pierre plus and all the demos I added were also removed.
I am adding those back.

If you go to the announce thread you will see the 1.1 version, but it is a work inprogreess, I will go over all the demos this weekend and make sure they all work.
So do not be alarm if some crash or malfunction, chances are is because some hack added to favor physx and not the others.

Pay particular attention to the physics behavior,
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles


Return to General Discussion

Who is online

Users browsing this forum: No registered users and 4 guests