[fairliearrow2]

I want to build my own desktop PC for Maxwell Render 3.2.

So what areas should my build specialize in, if any, to make Maxwell Render 3.2 work at its best? For instance would more RAM be the best thing to spend more cash on? Or a better graphics card? Or something else such the CPU? In short how does Maxwell render so that I can make my computer reflect its favorite aspect/s? I will probably use a Linux OS if that matters.

Thank you. [ if this is the wrong thread for this question then please put it elsewhere.]

Accidental double post. Firefox crashed etc so it all went a bit pear shaped. Sigh.

[jojojoj]

I would go for a fast processor like i7 5820k hexacore and overclock it. It's pretty easy, just make sure to get a decent cooler. Get a sufficient amount of RAM, depending on what you plan to do. Bhe minimum is 16GB, and try to get a mobo with 8 slots for RAM so you can add more later on.

[Mihai]

A few of the forums hardware experts might chime in to help you further, but the general idea is:

- graphics card won't speed up the rendering. But since V3, with the FastML feature, it's an advantage to get a card with as much VRAM as possible since FastML will use the cards RAM. Not enough RAM on the card for a given render resolution = it switches to the "old" way of viewing Multilight changes ie, you have to click the little Preview button and wait for the main render view to update. Cards with 4GB of VRAM are pretty common these days.

- CPU is all that matters, I have no idea what is the best these days but there are a few sites that use Maxwell as a benchmark, or also Cinebench etc.

- as mentioned above, 16GB would be the minimum these days, 24GB would be decent.

- after you've assembled everything, run a few torture tests for CPU and RAM, for at least a few hours (using specialized apps for this), to check that you have no overheating issues or faulty RAM modules.

[jojojoj]

Yeah cpu matters most. You could get something faster than what mentioned above, i.e. i75960x or double xeons. Fact is that comes to a price point that's much higher. You generally want a machine with highest possible single core clock speed and high core count. Clock speed is for all the operations that are not multithreaded and often makes much more for viewport perfomance. Take the very best video card and a crappy proc and you'll get a terrible viewport lag.

[fairliearrow2]

Thanks. That helps a lot. That's what I need to know!