5 ms·

Dave, doesn’t that video show more like “50% faster”? Here’s the timecode (&t=360) [1] for the “production difficulty” result (which really doesn’t seem to be,

by boulos 6y ago

Dave, doesn’t that video show more like “50% faster”? Here’s the timecode (&t=360) [1] for the “production difficulty” result (which really doesn’t seem to be, but whatever). Isn’t there a better Vray or Arnold comparison somewhere? As in my summary comment, an A100 can now run real scenes, but will cost you ~$10k per card. For $10k, you get a lot more threads from AMD. [1] https://m.youtube.com/watch?v=ZlmRuR5MKmU&t=360 https://m.youtube.com/watch?v=ZlmRuR5MKmU&t=360

shaklee3 6y ago
What do you mean by a lot more threads? Are you comparing an epyc?
- boulos 6y ago
  Yeah, that came off clumsily (I’d lost part of my comment while switching tabs on my phone). An AMD Rome/Milan part will give you 256 decent threads on a 2S box with a ton of RAM for say $20-25k at list price (e.g., a Dell power edge without any of their premium support or lots of flash). By comparison, the list price of just an A100 is $15k (and you still need a server to drive the thing). So for shops shoving these into a data center they still need to do a cost/benefit tradeoff of “how much faster is this for our shows, can anyone else make use of it, how much power do these draw...”. If anything, the note about more and more software using CUDA is probably as important as “ray tracing is now sufficiently faster” since the lack of reuse has held them back (similar things for video encoding historically: if you’ve got a lot of cpus around, it was historically hard to beat for $/transcode).
  - erosenbe0 6y ago
    Is the high priced 256 thread part that interesting for rendering? You can get 4 of the 64 thread parts on separate boards and each one will have its own 8 channel ddr instead of having to share that bandwidth. Total performance will be higher for less or same money. Power budget will be higher but only a couple dollars a day, at most. But I haven't been involved in a cluster for some time, so not really sure what is done these days.
  - shaklee3 6y ago
    The reason I asked is I did a performance trade-off with a v100 and dual epyc rome with 64 cores, and the v100 won handily for my tasks. That obviously won't always be the case, but in terms of threads you're now comparing 256 to 5000+, but obviously not apples to apples.
dahart 6y ago
Yes, this example isn’t quite as high as the 2-10x range I claimed, but I still liked it as an example because the CPU is very beefy, and it’s newer and roughly the same list price as the GPU being compared. I like that they compare power consumption too, and ultimately the GPU comes out well ahead. There are lots of other comparisons that show huge x-factors, this one seemed less likely to get called out for cherry picking, and @berkut’s critique of texture memory consumption for large production scenes is fair... we’re not all the way there yet. But, 50% faster is still “worth it”. In the video, Sam mentions that if you compare lower end components on both sides, the x-factor will be higher.

by boulos 6y ago

shaklee3 6y ago

What do you mean by a lot more threads? Are you comparing an epyc?

boulos 6y ago
Yeah, that came off clumsily (I’d lost part of my comment while switching tabs on my phone). An AMD Rome/Milan part will give you 256 decent threads on a 2S box with a ton of RAM for say $20-25k at list price (e.g., a Dell power edge without any of their premium support or lots of flash). By comparison, the list price of just an A100 is $15k (and you still need a server to drive the thing). So for shops shoving these into a data center they still need to do a cost/benefit tradeoff of “how much faster is this for our shows, can anyone else make use of it, how much power do these draw...”. If anything, the note about more and more software using CUDA is probably as important as “ray tracing is now sufficiently faster” since the lack of reuse has held them back (similar things for video encoding historically: if you’ve got a lot of cpus around, it was historically hard to beat for $/transcode).
- erosenbe0 6y ago
  Is the high priced 256 thread part that interesting for rendering? You can get 4 of the 64 thread parts on separate boards and each one will have its own 8 channel ddr instead of having to share that bandwidth. Total performance will be higher for less or same money. Power budget will be higher but only a couple dollars a day, at most. But I haven't been involved in a cluster for some time, so not really sure what is done these days.
- shaklee3 6y ago
  The reason I asked is I did a performance trade-off with a v100 and dual epyc rome with 64 cores, and the v100 won handily for my tasks. That obviously won't always be the case, but in terms of threads you're now comparing 256 to 5000+, but obviously not apples to apples.

dahart 6y ago

Yes, this example isn’t quite as high as the 2-10x range I claimed, but I still liked it as an example because the CPU is very beefy, and it’s newer and roughly the same list price as the GPU being compared. I like that they compare power consumption too, and ultimately the GPU comes out well ahead. There are lots of other comparisons that show huge x-factors, this one seemed less likely to get called out for cherry picking, and @berkut’s critique of texture memory consumption for large production scenes is fair... we’re not all the way there yet. But, 50% faster is still “worth it”. In the video, Sam mentions that if you compare lower end components on both sides, the x-factor will be higher.