GTS vs MX for 3D Studio Max?

[Faskis2]

I'm planning to buy a new video card (I own an old TNT) and I´m not sure if it´s worth to pay for a GeForce 2 GTS or a GeForce 2 MX. I own a 19" monitor and work at 1280x1024x32

I want to use 3DS Max, and I´m not sure if the extra bandwith and texture pipelines will really affect the performance in displaying the viewports. I plan to work with as many textures as I can displayed on the viewport, often displaing at least 2 textures per mesh.

It´s obvious that for games the extra hardware makes a difference, but anyone knows if it is a difference for 3D applications? Any experiences? Keep in mind that I´m not talking about rendering time (no video card acceleration involved there), just about modelling and viewport drawing.

Thanks

 

[biXen]

I'm not positive, but I think you may notice a difference in the viewports too. They can be heavy enough when you have a big scene, and a lot of RAM, and a good video card will help.

 

[cool]

CPU especially raw FPU power is still king in 3dsmax. Sure, if working with lots of textures more video memory will help. I've worked with 3dsmax some time ago and the transition from V3-3000 to GF2MX didn't bring me as much performance as the transition from a Cel400 to a P3-650e but then what I did in 3dsmax was more playing around than useful stuff...

 

[BenSkywalker]

The GTS has a ~11% edge in clock speed, that does help in high geometry situations(hard T&L can make a very big difference as long as it is set up right for use with 3DSM). Nothing to do with the memory bandwith at all, you can forget that completely. IIRC the TNT2 still has more fillrate then the latest and greatest from Intergraph(though it was close if I'm not getting senile), the GF2MX has more then enough. I assume that you use or will be using high enough polygon complexity to keep you geometry limited? I always do, and pretty much everyone else I know does also no matter what visualization package they are using.

Additional RAM could be quite useful if you work with large texture loads. If you can get a GF2MX clocked to 200MHZ with 64MB of RAM you are probably looking at the best price/performance ratio you are going to find for 3DSM(well, if you are handy with a soldering iron you can mod it to a Quadro2MX and use the Elsa drivers which kick @ss in 3DSM).

I'd say go with a 64MB GF2MX if you want the best overall bang for your buck(the added on board RAM for storing textures, AGP swapping is a lot better then PCI, but if you are dealing with high geometry loads already you don't want to kill all your bandwith). Which version of 3DSM are you running/planning on running?

 

[Faskis2]

Thanks for your answers, quite complete. The version of 3D Studio is v.4.

I´ve seen some benchmarks in Tom´s hardware at (sorry Anand for the free banner ) using the SPECviewperf test and there´s not much difference upgrading the processor in most cases, (usually less than 15% for a 400Mhz increment) but there´s quite a difference changing from a MX to a GTS (sometime double performance for the same CPU). However sometimes there´s absolutely no difference between the two cards (CPU limit) or even no difference at all no matter what CPU/graphics card combination (perhaps system memory bottleneck or something). Just forget about the Ultra and the higher end CPUs (poor quality/price balance).

If someone can explain this benchmarks, and why the GTS almost double the MX sometimes and others there´s no difference at all. How would this translate to MAX workflow? Any ideas?.

Thanks

 

[EvilDonnyboy]

uh, sometimes, 3DSMAX needs a lot of texture power so the GTS shows it's muscle. Sometime 3DSMAX needs a lot of FPU power fromt eh CPU, so the video cards don't have a chance to flex muscle.

 

[BenSkywalker]

If someone can explain this benchmarks, and why the GTS almost double the MX sometimes and others there´s no difference at all. How would this translate to MAX workflow? Any ideas?."

AWAdvs is the closest test to 3DSM. That test it based on AliasWavefront software(makers of Maya), the rest of the tests are aimed more towards CAD/MCAD performance(versus visualization). Frankly I'm a bit surprised by the scores using any of the boards he has listed, my GeForce DDR paired with a Duron 650@~700 is scoring higher on some of the test then any board save the Ultra(which is 64MB), odd(yes, I'm running the same version of ViewPerf).

Overall I doubt you would be able to tell the difference in workflow without a stopwatch, though if you can afford a GF2 64MB card it would be better(faster T&L at the very least).

 

[Ausm]

I would go with the GTS

Ausm