Recommendation for new system
Recommendation for new system
What would you guys consider a good workstation?
What cpu? Still a dual Xeon or dual AMD with 96 threads? Ryzen?. What graphics card?
Any good EU deals at the moment?
We have a budget of let's say around €7000
What cpu? Still a dual Xeon or dual AMD with 96 threads? Ryzen?. What graphics card?
Any good EU deals at the moment?
We have a budget of let's say around €7000
- Janusz Biela
- Posts: 3265
- Joined: Mon Mar 13, 2006 10:39 am
- Location: Finland
- Contact:
Re: Recommendation for new system
You know me ![Razz :P](./images/smilies/icon_razz.gif)
I do not like GPU (Power Plant system render) and limited Path Tracing.
So Dual Xeon.
![Razz :P](./images/smilies/icon_razz.gif)
I do not like GPU (Power Plant system render) and limited Path Tracing.
So Dual Xeon.
Re: Recommendation for new system
What about AMD Threadripper 2990WX vs i9 9980XE?
-
- Posts: 5
- Joined: Wed Feb 29, 2012 4:14 pm
- Janusz Biela
- Posts: 3265
- Joined: Mon Mar 13, 2006 10:39 am
- Location: Finland
- Contact:
Re: Recommendation for new system
These benchmarks sucks. Most of them use CPU instruction without going via BUS system to memory or very rare use it.
Real benchmark should stress CPU and RAM via BUS system. Idea is to swap big data and calculate Global Illumination - so it means a lot memory blocks must be proceed by CPU back and forth. Famous Cinebench use raytracing which is very bad for testing CPU. I suggest use our Kray Benchmark which stress completely whole system and gives high accuracy because use GI procedures include shooting photons.
You can notice in all these benchmark that use render, scores are different: they have spread with a couple percent.
..its my opinion. Render speed test should use CPU in complex situation not only pure calculation.
Real benchmark should stress CPU and RAM via BUS system. Idea is to swap big data and calculate Global Illumination - so it means a lot memory blocks must be proceed by CPU back and forth. Famous Cinebench use raytracing which is very bad for testing CPU. I suggest use our Kray Benchmark which stress completely whole system and gives high accuracy because use GI procedures include shooting photons.
You can notice in all these benchmark that use render, scores are different: they have spread with a couple percent.
..its my opinion. Render speed test should use CPU in complex situation not only pure calculation.
Re: Recommendation for new system
Problem of course I can't test a system I don't have yet....
Would you still recommend dual xenon 72TH over Threadripper 64TH or i9 36TH?
Would you still recommend dual xenon 72TH over Threadripper 64TH or i9 36TH?
Re: Recommendation for new system
They are equivalent.
- Janusz Biela
- Posts: 3265
- Joined: Mon Mar 13, 2006 10:39 am
- Location: Finland
- Contact:
Re: Recommendation for new system
If price is similar I would choose Intel....even, if Intel CPU will be 10-20% more expensive.
Re: Recommendation for new system
Thank you, all depends on the boss his Goodwill and how much he wants to spend. At least I know what to fight forJanusz Biela wrote:If price is similar I would choose Intel....even, if Intel CPU will be 10-20% more expensive.
![Very Happy :D](./images/smilies/icon_biggrin.gif)
![Very Happy :D](./images/smilies/icon_biggrin.gif)
- Janusz Biela
- Posts: 3265
- Joined: Mon Mar 13, 2006 10:39 am
- Location: Finland
- Contact:
Re: Recommendation for new system
In my opinion: Stick with Intel.