Hey guys, I'm buying a new PC and i have budget around $2000 give or take, what do you suggest?
I'm more concerned with Simulations in Houdini and rendering in Mantra mainly (for now, i may learn Redshift in the future).
New PC for Houdini (2000$ Budget)
5091 7 2- OmarWanis
- Member
- 1 posts
- Joined: May 2017
- Offline
- BabaJ
- Member
- 2127 posts
- Joined: Sept. 2015
- Offline
I use Mantra, but one of the advantages of using a gpu renderer like Redshift is that your going to invest in a good graphics card/s.
And that has the side benefit of better viewport response(please correct me guys if I am wrong).
But if using Mantra, instead of spending your money on a gpu I would say get as much RAM as possible.
And since simulations are your priority, go with number of cores/threads over cpu speed.
Of course, I'm sure there are some elements/processes of certain types of simulations which are necessarily single-threaded, so sometimes cpu spead can help.
Perhaps the more experienced guys could comment on that point.
And on the matter of RAM. Dont' go less than 32G….64G or 128G is better but, your on a budget.
I'm using 16G…and although it's good for learning, I'm always having to scale back on complexity of my projects, because 16GB just doesn't cut it.
And in a certain way, that cut's back on learning a bit, because I get no ‘feedback’ to look at and learn for higher ‘quality’ image sequences to tweak.
And that has the side benefit of better viewport response(please correct me guys if I am wrong).
But if using Mantra, instead of spending your money on a gpu I would say get as much RAM as possible.
And since simulations are your priority, go with number of cores/threads over cpu speed.
Of course, I'm sure there are some elements/processes of certain types of simulations which are necessarily single-threaded, so sometimes cpu spead can help.
Perhaps the more experienced guys could comment on that point.
And on the matter of RAM. Dont' go less than 32G….64G or 128G is better but, your on a budget.
I'm using 16G…and although it's good for learning, I'm always having to scale back on complexity of my projects, because 16GB just doesn't cut it.
And in a certain way, that cut's back on learning a bit, because I get no ‘feedback’ to look at and learn for higher ‘quality’ image sequences to tweak.
Edited by BabaJ - Oct. 16, 2018 09:04:52
- GOgraphR
- Member
- 72 posts
- Joined: Oct. 2018
- Offline
BabaJ
I'm using 16G…and although it's good for learning, I'm always having to scale back on complexity of my projects, because 16GB just doesn't cut it.
And in a certain way, that cut's back on learning a bit, because I get no ‘feedback’ to look at and learn for higher ‘quality’ image sequences to tweak.
The biggest drawback is “interactivity”, if you are learning, you loose your focus, if you think, you are “tweaking” a parameter, just to see what happens, and it turns out, your RAM is “exploding”. Had that a few times… yu change a parameter (0.1 -> 0.01) and then… Result available the next day !! But that's OK. in Some cases, this change multiplies to the power of 3, and that means all of a sudden (not really suddenly) your setup has to deal with 1000 times more geometry… *hihi* good to have a secondary access to your machine (In my case SSH from another laptop) to issue a “kill -9” and you are back in the cycle…
So yes, get as much RAM as you can afford, and probably a little bit more !
And have a look a entagma.com, they had a VLOG lately on this topic.
I am looking to find a refurbed dual socket with 128GB RAM and spend on a more recent GPU (GTX1070 or RX580)
- BabaJ
- Member
- 2127 posts
- Joined: Sept. 2015
- Offline
…The biggest drawback is “interactivity”…
I think we are talking about different sides of the same ‘issues’.
When I mentioned about no feedback, I'm talking about not getting any results at all because the system simply does not have enough RAM to begin with.
Even if the system does have enough RAM or even lots of RAM, getting the results can still take time, especially if it's single threaded…from that perspective you still ‘…lose your focus…’ because of the time it takes, even though you can still get the results.
In that spirit people use renderers like RedShift because as a gpu renderer you can get results faster, other pros/cons (cpu vs. gpu) not considered.
- GOgraphR
- Member
- 72 posts
- Joined: Oct. 2018
- Offline
@BabaJ: I Agree!
But I was unclear, I did not talk about rendering. I am an Apprentice, and as such, have no _real_need_ for fast renders. I was talking about setup.
And again I agree, RedShift might speed up significantly, but to me, it is a 3rd party renderer I can not use.
CPU/GPU tradeoff - Well, the whole thing we all do call “Life” is a single heap of compromises…
Sometimes I wish I had a licence to a Greenback printer and enough green ink in my attic…:(
But I was unclear, I did not talk about rendering. I am an Apprentice, and as such, have no _real_need_ for fast renders. I was talking about setup.
And again I agree, RedShift might speed up significantly, but to me, it is a 3rd party renderer I can not use.
CPU/GPU tradeoff - Well, the whole thing we all do call “Life” is a single heap of compromises…
Sometimes I wish I had a licence to a Greenback printer and enough green ink in my attic…:(
- atracksler
- Member
- 41 posts
- Joined: July 2005
- Offline
https://vimeo.com/286199410 [vimeo.com]
I just went to amazon and this whole thing is @$2600. You could get less RAM and a couple different parts and probably hit your goal, you could also get some parts from Ebay….
Inasmuch as I'm considering something like this….
I just went to amazon and this whole thing is @$2600. You could get less RAM and a couple different parts and probably hit your goal, you could also get some parts from Ebay….
Inasmuch as I'm considering something like this….
- Thingy1972
- Member
- 9 posts
- Joined: June 2017
- Offline
Hello
I just wondered whether you could use some of the budget to get Indie and then you can use the free test demo version of Redshift. I found Mantra just waaaaaay too slow on my machine - even on my new Ryzen 7 2700x - to get anywhere sensible and it drove me up the wall. Redshift is very very very very fast in comparison.
I just wondered whether you could use some of the budget to get Indie and then you can use the free test demo version of Redshift. I found Mantra just waaaaaay too slow on my machine - even on my new Ryzen 7 2700x - to get anywhere sensible and it drove me up the wall. Redshift is very very very very fast in comparison.
- anon_user_40689665
- Member
- 648 posts
- Joined: July 2005
- Offline
-
- Quick Links