Forum: Poser - OFFICIAL


Subject: What would be the specs for the ultimate CG workstation?

Paloth opened this issue on Sep 18, 2007 ยท 86 posts


Dale B posted Wed, 19 September 2007 at 6:25 AM

If you aren't quite ready to upgrade just yet, you might want to keep an eye on the AMD sites; the chip wars are about to get hot once again. The Phenom processors are about to hit the channel (the official end of the Athlon series cores), with 2,3, and 4 core variations (the triple core is a quad that didn't pass on one core, so they plan on burning the fuse to the bad one, and having the triple core as a value slated offering to get the most out of their yields.). There is supposed to be a list of existing AM2 mobo's out there that -will- support the Phenom.....and frankly, the underlying tech is still much better than Intel's. One group unaffiliated with either side has already come out with numbers that show the AMD chips have a better power consumption curve when idling than the Intel duals and dual duals (which is what their quad cores actually are; another tidbit that points to the differences in base tech is that Intel is busy trying to adapt AMD's memory controller on the chip to their product, and copying the HyperTransport scheme and naming something else so they can claim they have this great new idea....not playing catchup with the littler fellow). Now the power consumption is more relevant for the server market....or people who leave the system on all the time. For rendering appliances, it will make a bit of difference in cost to run (as in your power bill, and the ambient heat the thing generates). As for memory; do some research before you choose a memory speed. Bigger numbers do not a cool speed up mean. It's been a rule of thumb (there might have been a change with the very last revision) since DDR2 came out that the actual performance difference between the DDR2-667 and DDR2-800 has been less than 5%....with the prices not exactly reflecting this truth. Most of the testers have found a benchmarking difference, but could notice no actual performance difference. So you get into the 'Do I get the geekiest fastest, or the slower, cheaper memory and get more of it for the same cost?' question. If you are rendering, the latter is the answer, natch. Video is almost totally up in the air at the moment. With uber cards affordable by the masses (not the pro cards, just uber cards), SLI and Crossfire pretty much -not- being supported by anyone save the game companies, M$ dinking around with OpenGL support yet again, it's a coin toss. But if you have the desk space, dual monitors can be soooo helpful. If you go with RAID, then I'd recommend a 0+1 array; that way you get the speed of data striping, and a backup in case your main array go boom. Ideally you'd want a not so large 0+1 for OS and apps, then a second, much larger 0+1 for content and workfiles. But whether you go for RAID or not, plan from the beginning to keep your content and work files on separate physical drives from the applications and OS. That way if one either of them tanks, you don't lose your work. And you might want to invest in one of the networked drives, or put together a 500 gig USB2 safety drive to backup all your perishables onto. And a battery backup. Don't forget the backup. Even if it lasts just long enough for the OS to save the work and shut down, it will save you no end of grief....