Jump to content
3DCoat Forums

I can do almost 20,000,000 polys (VOX MODE)


ggaliens
 Share

Recommended Posts

  • Advanced Member

I can do almost 20,000,000 polys (VOX MODE)

I have a 1.5GB Ram machine with swap-space set as per Andrews

recommendations.

How many can you do ???

Can someone with a 6GB or larger RAM and 64 bit machine ... can someone like that

comment on their personal experience with 3DCoat ???? How many polys could

you do with 6GB Ram, 8GB Ram ???

Some collected stats would be interesting.

Andrew, have you tried this VOXEL sculpt on a machine with BIG BIG MEMORY 64BIT ???

Is the VOXEL mode coded to take advantage of 64bit architecture/memory spaces ????

Or would that just tend to blow out the graphics card ????

Link to comment
Share on other sites

  • Advanced Member

I don't think a youtube video would add anything to this thread.

Let me respecify what it is I'm interested in hearing about.

I want to know how a user figures out what system resource is the real

bottleneck when 3DCoat starts behaving badly.

Bad behavior is slow and painful and not something that lends itself to good video, LOL.

I have a feeling that for a lot of users it could be the graphics card.

For example, I just sat at a modern DUAL CORE PC with Vista 64bit, 8GB Ram,

256MB NVidia card (bought in fall), and a complex 3DCoat scene. While that 65 bit PC

seemed my more powerful than my usual 32bit,WinXP 1.5GB Ram , Single core,

HP Pavilion a730n ....

That 64bit sculpting session was only marginally more successful than on my HP.

The 64bit session ended in a horrible freeze at 30+ million polygons in VOX MOde.

3.4 MB Swap was in use. Maybe 2.x GB Ram. The unknown seems to be what was happening

at the interface between the graphics card and the PC. I bet the graphics card was overwhelmed.

Andrew ... what can you add to help inform a user when the graphics card is maxed out ???

Anything ???? And I monitor my NVidia card perfomance with an extrnal tool ???

Let's talk about how to discover where the bottlenecks are.

Andrew ... you already said Memory is cheap as dust, well ... the dust won't buy me any perfomance if my graphics card is the real problem.

I don't want to spend $200.00 on 3DCoat, $200-$300 on a top notch graphics card, and $100.00 on Ram. I might be doing better to buy Zbrush in such a case.

So I'm going to have to continue to watch myself so I don't crash catastophically, and figure out the real bottlenecks.

My buddy is going to throw a 1GB Grphics card in his 64bit machine so I can try that to see if it will break the 30million poly barrier.

WE NEED SOME RECOMMENDED HARDWARE along with predictions of performance with such hardware.

Link to comment
Share on other sites

  • Advanced Member

The point of a real time video is to see how responsive X number of polygons will be in a natural situation. A bit pointless having a 60 million polygon model if the brush strokes only update every other second!! Oo

You realize your asking for specs of a product that's still in development!! It's a given that, when it's complete it will come with recommend specs like all software.

Link to comment
Share on other sites

  • Advanced Member

I'm not having any problems with sufficient frame rate with high polygon count.

If I was ... I would have talked about that.

Turn off shadows ... and I can quite easily VOX sculpt on 10,000,000 polygon

model.

My problem is with instability, crashes ... all seeming to happen when the

program have to do a VOX merge, smooth, combine, ... a object level

vox transformation , tweek of some kind. I'm very suspicious that I need

a new graphics card ... but there are some folks who say my card should be

borrowing from my system memory. If that's the case ... then why do I still have problems

on a system with 8GB Ram (testing on such system the other day, 64bit Vista).

I'm trying to gather information on how other folks systems are behaving.

Andrew has an aggressive schdule for delivery of V 3.0 with VOX as far as I

can tell ... and in another thread, he is talking about getting pretty down

and dirty with stats and testing on more hardware. No time like the present

to start formulating a test plan.

Link to comment
Share on other sites

  • Advanced Member

Again, I think you need to remember that your testing a "Alpha" version of this program. Alpha software and bugs, go hand in hand. I'm not sure why you think a more powerful graphics card will make the alpha version of your 3dcoat more stable?

Wouldn't it be likely that the reason it's called alpha is that it's not stable?! :blink:

I have a 4gb + 9600gt dualcore machine and I find anything above 5-7 million not user friendly...

It all depends on what brushes u use, some seem to work 20+ times faster than others. I try and keep away from the surface tools as they lag in comparison. Though it's freaking hard to stay away from the "Move" tool, it really does lag hard, but it's such a essential tool.

Link to comment
Share on other sites

  • Advanced Member

"Alpha software and bugs, go hand in hand. "

Right ... and there has to be a ROOT cause for every bug.

THE ROOT cause is not ... "It's alpha software".

And in this case ... since the bug only appears when system

resources are very demanding on both the main RAM and the graphics card.

And since I can't "MEASURE" the demand on the graphics card in as direct a manner

as I can mesure the PAGING files and RAM usage ... I have to be at least

a little suspicious of having an under-performing graphics card where ULTIMATE

capacity is concerned. I am still of the opinion that a graphics card borrowing resources

from the system is probably a very bad thing ... and this may very well be what I have

going on on my machine.

I have no complaints about brush speed at all. I understand that they

all have inherent speed limitations of their own.

I'm definately getting mostly OUT OF MEMORY ERRORS. But I would not

trust 3DCOat to be categorizing these OUT OF MEMORY ERRORS with anything

but large "sweeping generalized buckets". I know my self that I'd be hard pressed

to write code that always proeduces a most meaningful error message that cannot

be misconstrued. So I'm happy to continue investigating the likelihood of this being

a case of lame graphics card.

Link to comment
Share on other sites

  • Member
And in this case ... since the bug only appears when system

resources are very demanding on both the main RAM and the graphics card.

So I'm happy to continue investigating the likelihood of this being

a case of lame graphics card.

In another thread, Andrew posted the following specs for 3dCoat:

Minimal requirements are the same - Radeon 9200/FX 5500/GMA945, 128 Video RAM, 512 RAM, 1 GGz CPU

Optimal by price - NV 8600 (512) + 2 core CPU +2GB RAM

Optimal - NV 8800 and better, 4GB RAM, 4 core CPU

Did you say earlier that you have a 256mb graphics card?

If so, Andrew mentions the need for a 512mb graphics card for optimal performance. Can you try a better graphics card?

Cheers

Doug

Link to comment
Share on other sites

  • Advanced Member

What kind of polycount would you consider adequate for you?

My quarms about brush performance at high densitis are of the way the brush some times jumps and I have to try and recreate the line. Though, I think this is just down to the way 3dcoat handles lag and it'll be tweaked to avoid such spikes in lag.

I have a 9600gt 512 card

Link to comment
Share on other sites

  • Advanced Member

Well ... if polygons can be "STREAMED" through a graphics card into a 2D framebuffer as some folks have at least alluded to (and would make some sense for Cards to work this way in a perfect world ) maybe I should accept no polygon limit ????

Seriously. Why not have "Stand-Ins" for some Objects (SEE MIPMAP in wikipedia and thing 3D analog for MIPMAP) ???

Sure this is not an existing feature ... but it would make a great test for how graphics cards work.

If polygons can't be streamed ... and we can't have my Billions of polygons by streaming them from DISK in a not so tricky way (should be extremely simple), then the fallout is that Cards do inherently have a MAXIMUM MAXIMUM polygon limit in addition to a MAXIMUM framerate with a given polygon set.

As it turns out ... I was probably mistaken about my assertion that my card has 256MB Ram. I think in fact I was biased by someone elses suggestion that an NVIDIA 8500GT has 256MB. NVIDIA 8500 may come in at least two variants 256MB and 512MB). Shame on my for not finding that control panal a long time ago.

My NVIDIA Control panel does say 512MB. So I'm now thinking that when I went to my buddies house to test on 64bit machine ... he may well have had less memory than me on Graphics card. So maybe explains a bit of why 64bit 8GB main RAM didn't make the experience that much better. There was another bottleneck that got tighter.

I'm a programmer 9-5 in multimedia and video compression ... and I know that it's not always easy to diagnose root causes of bugs in a complex system such as 3DCoat. I don't have the source code. But I do see some of the BUG-REPORTS that pop up in the Web-Browser after a crash. I try to envision what they might mean ... and what I can't deduce from them.

If Andrew says a 1GB Graphics card is cheap ( I think he did ), then I do hold out some hope for more polygons in the near future.

I WANT 100 Million polygons in my scenes. I want to be able to work with 20 Million in memory at once. That should be enough for me. I want to be able to off-load some to DISK by keeping aorund only PROXIES in memory since in VOXEL space a proxy should be super easy to calculate.

I'm dreaming big dreams, I know. But I think Andrew can get us lots more polygons ... with some clever tricks.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...