Hey, this may well be a simple question, then again.
I would just like to know what you think on graphics cards when it comes to rendering in Bryce. How much of a difference does it make to render if your using, say an older g-force4 64MB, to one of the newer cards on the market today? Will it effect the quality of the render?

Graphics cards have absolutely no effect on the render. Only thing it does is determine how many polys you can move around while setting up the scene. CPU speed and RAM determine how fast a scene renders.

Cheers Aldaron. Now I now why I cant have too much in a scene. Bryce crashes and I throw my monitor at the wall, bloody crappy G-Force card I use!

Well RAM may play a part in that too. I have a G-force 4 64 mb card and 1 gig RAM and can put quite a bit in scene but no crashes.

yea RAM is the main issue for making big scenes. The graphics card kicks in when you're trying to view it in Direct3D or OpgenGL modes and such I think bryce 5 has a 3rd mode also with vidcard which I never use.. scree3d or such? or whatever it's spelled? anyways, that one uses vid-card too. But during rendering it's 100% up to cpu and Ram, and also up to your HardDrive at the moment you run out of RAM, but you probably would prefer not to run out of ram as it renders faster when there's no HD needed :)

Aye, Sans. Your graphics card is not what's crasing Bryce, at all. It's a RAM issue, or swap-file issue. Bryce hasn't been programmed to use any hardware acceleration during rendering. The reason for this is so that a broader array of PC's are supported...

The reason for this is so that a broader array of PC's are supported... That and graphics cards aren't made with raytracing in mind anyways so it'll be doubtfull of how much use they would be anyways.

Aye, I disagree kindly on that matter, RayRaz. CPU's were certainly NOT made with Raytracing in mind, either! GPU's and CPU's both do what they do because software tells them to. Considering the power of nowaday GPU's, it would be very easy to tell them to do raytracing, from a software standpoint. And GPU's have higher memory clocks and plenty of shader technology to allow such a thing. There is simply no standard for such things... Yet. With the PS3 and Cell processors coming out shortly, all kinds of things are going to be possible. We'll be seeing realtime raytracing for the first time, ever, in the next year or so. And it will be very, very powerful...

Thanks for all the replies. But my main question was about the quality of the end render, the picture quality. I have 512 dual DDR3200, a 2.6 Barton with a 7n400pro2 mother-board by Gigabyte and of course a Ge-force 4 64MB. Would it be a good idea to increase my page file?

Cheers;)

Message edited on: 05/29/2005 09:20

Hey Rayraz/LSD- Pure3D cards. (but who can afford them if it is not a business investment!)

You needn't go beyond 768 MB or possibly 1.5 GB with the page file. Microsoft says: "For best performance, do not set the initial size to less than the minimum recommended size under Total paging file size for all drives. The recommended size is equivalent to 1.5 times the RAM on your computer. It is good practice to leave the paging file at its recommended size. However, you may increase its size if you frequently use programs that use much memory." Then, neither the amount of RAM nor the speed of the CPU affect the quality of the render. RAM doesn't play any role in rendering except making it possible for you to open bigger files and/or to edit in them smoother. CPU speed is the only thing that affects the speed of rendering. Not quality, speed. The end result will always be the same, cause the processor will always calculate the same things. With a faster processor you'll "only" arrive to that result sooner.

RAM affects the speed too because once you run out of ram you will need to use the swapfile on your harddrive and your processor WILL have to wait for your harddrive to keep up. LSD, while cpu's are build more for all-round number crunching, GPU's these days are optimized for very specific calculations needed for real-time rendering using openGl or DirectX. Ofcourse a GPU's processing abilities far exceed that of a CPU when it comes to openGL and DirectX, but it only does so because it was build specifically for this purpose and this purpose only. I doubt that, even with the faster memory, a GPU would be any match for a CPU when it comes to raytracing, simply because the CPU is optimized for basic number crunching, which is in essence all there is to raytracing at the moment. I do agree that IF programmed to use features at which GPU's can use their full potential, a powerfull combination of cpu and gpu rendering is possible, but it would still be a hybrid render system a lot unlike bryces pure raytracing engine.

"RAM affects the speed too because once you run out of ram you will need to use the swapfile on your harddrive and your processor WILL have to wait for your harddrive to keep up." I wouldn't be so certain. If you take a look at your scene and the appropriate BMP file, you'll see that the size of the BMP is always the same, regardless whether you saved just an empty BMP or a full rendered picture. Once you opened a file and it occupied a certain amount of memory, rendering uses just CPU. Check with Task Manager. I did right now. Memory use fell down by a tiny number of kilobytes, but the CPU occupancy jumped up to 98%. Swap File comes into use if you add models, textures and so on to an existing scene.

My renders on complex scenes increased SIGNIFICANYLY when I added more RAM to my pc. And I mean like in the order of up to 3 times at fast for scenes that used to have my harddrive swapping non-stop before the upgrade and needed no swapping after the upgrade. The bmp file is only a fraction of the memory usage, the real memory is in stored 3D data and calculations and textures and such things. If you render a scene of 500Mbyte there is no way it will render as fast on a 256mb system as it would on a 512mb system.

Aye, I completely agree with Rayraz on this one. Physical RAM really speeds things up, when every scene you have is more than half of what you're using. Many of my scenes would simply not work at all, not even load in, if I didn't have 512MB. I remember working on scenes back when I had 256, and just stopping, because Bryce generally needs twice as much RAM as your scene's total data size... Meaning, that at 512MB RAM, I can comfortable work with scenes under 256MB in size. Currently, I have two projects which will require another 512MB before I can do anything else to them. My fortress and my little house scene both have grown beyond the 512MB arena, and I can't even save changes to either file... So, I await another stick of DDR, and then I can move forward and finish those scenes...

I've worked scenes with more then 3 GB of total data on a 512mb system.. Just split them up into bitesize scenes, then merge later. It won't render as efficiently because of the above stated reasons, but it will work.

Message edited on: 05/29/2005 14:48

LSD, be patient when saving a file. Bryce will go on till it finishes. Just step outside, take a walk, mow the grass, take a nap, shoot some hoops, shower, sooner or later Bryce will come back. The last pic I worked on took twenty minutes to save, if that's all the computer was doing. You remember Brycetech's CrystalPalace, twentysome hours to save. Sans2012, you're answer is up there. Bryce doesn't care about your videocard except for when useing openGl modes.

The trick to overcoming long save times : save as a different file name. But, that's not the issue in my little house scene. It's more of an issue, again, of Bryce's internal naming database. Too many parts, too many pieces, and it just stops. Especially when dealing with highly detailed imported meshes. Bryce 3 and 4 were even worse. If you had more than 256 separate (mesh) parts, it would crash. They fixed it in Bryce 5, one of the ONLY things they fixed, but I think they merely increased the limit to 1024 or so...

Aaaaaaaa, you two, you are not listening! RENDERING ITSELF DOES NOT ADD MEMORY OVERHEAD. PERIOD. You are talking about a different thing. You are talking about opening a scene. Of course your drive will swap if you try to open a 500 MB scene on a 256 MB computer. There's no space in memory for the complete scene. So when you start rendering a scene bigger than the RAM you have, computer has to load different parts of the scene one after the another. But it's not a matter of rendering, it's a matter of LOADING, because the complete scene is not in memory. Repeat, the rendering itself does not depend on memory. The calculations are always done at the same speed regardless how much memory you might have. There's no speedup in rendering if you add more RAM, there's a speedup in loading. If, for instance, we are talking about a 50 MB scene, it will render at the same speed with 256 MB RAM or with 1 GB RAM. If rendering was RAM-dependent, the computer with more RAM would render the same scene faster. It's not the case.

RENDERING ITSELF DOES NOT ADD MEMORY OVERHEAD. PERIOD. Then where does it store the not previously opened textures? And what about the various variables and numbers it needs to remember during calculations? All the lightrays that are not there when you open the file because they only start being simulated during render. Where does it cache it's calculations from the previous render pass which it might use during the next one?

Aye, I understand what you and Xenix are saying, Erlik! And if you were working with a 50 MB scene alone, at render-time it wouldn't matter if you had 256 or 1024 MB of RAM. You're right... But you're also wrong. The swap file ALWAYS kicks in, because my PC is never just rendering. There is always, always, 30 or more other processes going on. And Bryce is rarely, if ever, "On Top". So in the case of normal-sized scenes, which to me are over 200MB, there is even MORE swapping going on. If I had more physical RAM, this would be minimized. And in the case of my two scens that are well over 500MB, which I can open and render but not ADD anything to without instabilities on my next save, having more physical RAM is the only way to overcome this barrier. Like I said, I've dealt with this before with multiple, dozens of scenes really. When I moved up to 512 DDR on this PC, it opened up all kinds of worlds. So, we're all correct on this issue, in certain ways. But in the other ways, I have to say I agree with Rayraz on the issue of Physical RAM, and how it really does affect render speed in large scenes. If your scene is paging (swap file, virtual memory) during rendering, it's hit a bottleneck... And since all of my scenes page at render-time, it's a bottleneck for me!

In the file. Bryce stores everything in the file. And once it starts rendering it just draws the things on the screen. Or in the separate picture file, as the case might be. If a texture had not been opened, it was not in the file. Now, if Bryce supported instancing and displacement, the situation might be quite different, I suppose. Also, I suppose that the interim calculaations are stored in the TMP file.

I expirienced tremendous rendering slowdown at the scenes which exceed amount of RAM (talking about 320 MB scene file on PC with 256 Mb memory only), from 5 hours on 512 Mb memory one, to 28 hours on the one with only 256 Mb of RAM... RAM is crucial to rendering as it's speed is too :D

Now, if Bryce supported instancing and displacement, the situation might be quite different, I suppose. In this case it would be even more apparent yes. But I would be just as right as I am now ;-) Are you suggesting that bryce loads all your bitmap textures, non of which you really use in ANY way in your wireview (except when using daz's new 5.5 textured opengl views probably) still get loaded into your RAM? If this is the case then I would have to say bryce's memory management is quite disappointing in this aspect.

Disappointing is an understatement, Rayraz. After playing with Maya for a few months now and Lightwave last year for a bit, and keeping a close eye on my RAM and CPU usage, I have to say that you really get what you pay for. I love Bryce to death, but it's got a lot of room to grow. Also, I'm using a now-outdated system in many aspects. The only thing GOOD about my system is all the stupid little proggies that add up to it being a graphics powerhouse. Bryce, Rhino, Maya, Photoshop... Riptide... I can't wait to pull my HD's and slap them in a new system someday soon! (Recently Ser AgentSmith notified me that my Athlon 2000 was clocked down, not at it's best setting, and I got a quick 400MHz jump out of the deal!

Ray, I'm not suggesting it, I'm saying it. :-) Bryce saves everything into the file: textures, models, lights, camera positions, animation paths... When you open a file, everything loads into RAM. OTOH, Bryce, as Darlisa quoted Doc Mojo, does not have a poly import limit programmed in. In my experience, it does not have a scene limit either. Both are dependent on your RAM. One guy with XSI said Mental Ray has a limit of about 2 million polys, cause all of them have to be in RAM to render (2 GB of usable memory on Windows). Compare that with the fact that Kemal could render a 320 MB scene on a 256 MB computer. I'd say that Bryce has quite good memory management. Not to mention that my scene with 68 million polys was livelier in Bryce than a 2.6 mil poly scene in Cinema.

Aye, but did that 68 million polys include Terrain polys? They aren't the same, RAM-wise, by any means... Try pulling in even 4 or 5 million mesh-object polys into Bryce and watch it crumble under the onslaught. And I don't know what that XSI guy's problem was, but I've crammed well over 50 million polys through Mental Ray before. I suppose he forgot about swap files, too? I agree, a Bryce file saves all of it's data, and has no references or "instancing". But don't force me to post screenshots of Swap File and Physical Ram usage, Erlik! As a scene renders, it can swap out of Physical RAM completely, depending on what you're doing. I'm not saying this is a good idea, but it can't be helped under certain circumstances.

Okay, I certainly don't understand everything that you guys are saying, but I have just managed to transfer almost all I need to my new laptop, so I've just started using it for a few little things. Now what I'm wondering about is that I seem to see it behaving SLOWER that the PC did, I'm thinking here about moving things around in wireframe. The PC has just 512 of ram, but the laptop has 2gigs of ram - shouldn't it be faster?

No, Fran. Not necessarily. What are the clock speeds and RAM speeds of your two machines? RAM doesn't actually make anyone's CPU faster. It's just that it can help you RENDER faster, in certain cases. I've done test renders with 256 and 512 (256x2 modules) and if you have just one strip of 256 it will render faster than having 2, IF and only if your scene size is still contained by your RAM limit... This is due to paging size discrepencies in the Front-Side Bus... It gets techy, but to sum it up : more is better. Laptops are always slower than their desktop counterparts, at the same clock speed. Just part of the deal.

Uh... I dunno how to find out what the speeds of either clock or RAM are. If you could tell me where to look?

If you're using XP : Control Panel, System. It will tell you there the CPU clock speed and how much RAM you are using. (smiles, "How did you buy a computer without knowing what was in it, Fran!?!?")

Oh... I already looked there. Loads of numbers and stuff. Basically on the laptop it says: Pentium 4CPU 3.60GHz 3.60GHz 2.00GB of RAM and I think the Graphic card is: NVIDIA GeForce GO 6800 That's what it says under "display adapters" anyway The PC is... AMD Athlon (TM) XP 2700+ 2.17GHz 512MB of RAM and the graphics card - again found under "Display Adapters" is: Inno GeForce FX 5200 Is any of that the clock speed or RAM speed you are talking about? And to answer your other question, I didn't buy it, my husband bought it for me - and he won't tell me how much. It's supposed to be specifically for me to do rendering and other graphics stuff on. But the Laptop is way slower than the PC when you open Bryce 5.5, then "Import Object" Orbital's Tower and Balloon and try moving them around in wireframe mode. Is there some setting I need to change for the laptop? I would like to get it moving faster.

Thanks again every one your help is appreciated. Everyone seems to have their own belief on this topic, glad we could stir up so much information. One other question, wouldn't your hard drive speed (access speed) also play a part in the render time or performance? Cheers all.

Yes, the speed of hard disk would play a role in the speed of performance. Faster the disk, less time to access the swap file. Fragmentation will play a certain role, too. Whoa, Fran, I don't know how much your husband paid, but it wasn't peanuts. That's one mean machine - in CPU, in graphics and especially in RAM. Can you send it to me if you don't want it? :-) I played with a very similar laptop quite recently, immediately after I got 5.5 and the speed was normal. It had a weaker graphics card, Go 6600, with 128 MB video RAM and when I used a proper mouse instead of the damn touchpad, it all went perfectly normally. (To see the amount of video RAM: right-click on the desktop -> Settings -> Advanced -> Adapter). What you could do is to go to System Properties and then to Advanced and click on Settings under Performance, and then again on Advanced. Put the mark beside Programs in the upper two cattegories, and in Virtual Memory check Let Windows manage... or System managed size, whatever the case might be. BTW, you won't be able to use it on batteries much, cause it probably drains them quite quickly. The one I mention above emptied its batteries in about an hour and a half.

Thanks Erlik, The only thing that wasn't as you said was "System managed size" so I've just changed that. Hopefully it'll now speed up a little, and no, I'm not sending it to you. ggg Husband would kill me, and I kinda like my head where it is, on my shoulders. I don't really use it on batteries... so far. I really just wanted it for Bryce-ing, Winging and Photography. Glad you like it, I was beginning to wonder if it was worth its un-named price. I rather like the screen, it's lovely and clear. (but I dunno what it's called so don't ask me)

Yup, that IS quicker. Thanks Erlik.

One other question, wouldn't your hard drive speed (access speed) also play a part in the render time or performance? As soon as bryce runs out of RAM and is forced to (significantly) use the harddrive, yes, your harddrive's reading and writing speeds and such will play part in the render time. The way thigns work with RAM and different RAM speeds and sizes is: A: More RAM storage space allows more data to be stored in the RAM. B: The faster your RAM, the faster it can read and write data and the faster it can feed any data to the CPU when needed. During rendering your CPU does calculations, nothing more, nothing less. But a CPU is of no use if there is no way of feeding the CPU these calculations. If you don't tell the CPU it's calculation, then it "won't know what it should calculate", and it won't do a thing. The CPU receives it's calculations/instructions whatever you wanna call it from other sources. The RAM is the fastest external (outside of the CPU) communicating source in the system to send the CPU it's instructions. In the most ideal circumstances the CPU would receive it's instructions at such a speed that it does not ever need to wait. Now I think you can imagine, that if your computer calculates things that need so much data that it simply runs out of storage space when trying to put all this data in the RAM, it needs to put this data somewhere else. This "somewhere else" is the swapfile on your harddrive. Since your harddrive is (by far) not capable of reading and writing data as fast as your RAM the CPU can't receive it's instructions as fast anymore thus the whole process will slow down. So in conclusion, if at any moment bryce is forced to go use the harddrive during rendering, yes, the harddrive's speed will have influence. But since a harddrive (even the fastest one) is so much slower then RAM it's way more effective to just get more ram rather then to buy a faster harddrive.

Your knowledge on this subject has been amazing, please take my thanks, all of you. Cheers. Michael.