F5F Stay Refreshed Hardware Desktop 3DS Render Server Build AMD version 3DS Render Server for AMD platforms

3DS Render Server Build AMD version 3DS Render Server for AMD platforms

3DS Render Server Build AMD version 3DS Render Server for AMD platforms

9
905xA
Senior Member
667
01-18-2022, 02:44 PM
#1
We aim to upgrade our PowerEdge FC430 units with two Intel Xeon E5-2680 v3 rendering nodes for our Max 3ds render farm. This involves replacing our current setup that handles VRAY, Phoenix, and Forest pack plugins. We also have several AS-2022TG-HTRF systems running dual AMD Opteron processors. Our performance metrics are heavily influenced by core count, so we're focusing on cost per core improvements. We've been exploring racking options like Threadripper 2990WX and Ryzen 9 3900X, as well as Epyc dual CPU servers (7401, 7551). We're gathering insights from others who have experience with similar setups or benchmarks. We're currently testing the Ryzen 9 3900X and dual Xeon machines in our office, finding them slightly less efficient than an i9-7900X on a VRay test. Upcoming benchmarks will compare actual 3DS Max renders over the weekend to evaluate performance differences. Our research phase is just starting with limited downtime, so we're reaching out for advice, feedback, or suggestions to help others in this situation.
9
905xA
01-18-2022, 02:44 PM #1

We aim to upgrade our PowerEdge FC430 units with two Intel Xeon E5-2680 v3 rendering nodes for our Max 3ds render farm. This involves replacing our current setup that handles VRAY, Phoenix, and Forest pack plugins. We also have several AS-2022TG-HTRF systems running dual AMD Opteron processors. Our performance metrics are heavily influenced by core count, so we're focusing on cost per core improvements. We've been exploring racking options like Threadripper 2990WX and Ryzen 9 3900X, as well as Epyc dual CPU servers (7401, 7551). We're gathering insights from others who have experience with similar setups or benchmarks. We're currently testing the Ryzen 9 3900X and dual Xeon machines in our office, finding them slightly less efficient than an i9-7900X on a VRay test. Upcoming benchmarks will compare actual 3DS Max renders over the weekend to evaluate performance differences. Our research phase is just starting with limited downtime, so we're reaching out for advice, feedback, or suggestions to help others in this situation.

A
Anskar
Member
59
01-25-2022, 11:06 AM
#2
Check servethehome.com for top ratings. The 7702p currently outperforms dual 8280 models at a much lower cost. https://www.servethehome.com/amd-epyc-77...er-socket/ They mention c-ray as a ray tracing test, though v-ray isn't ideal. Focus on price per core—Epyc 7282 shines, along with 7552 and 7452 for dual sockets. Single-socket blade/1U units seem to deliver the best results. Also note Epyc's efficiency is impressive, drawing just 270W peak. Be sure to request quotes from both options before comparing specs, as pricing varies widely between vendors.
A
Anskar
01-25-2022, 11:06 AM #2

Check servethehome.com for top ratings. The 7702p currently outperforms dual 8280 models at a much lower cost. https://www.servethehome.com/amd-epyc-77...er-socket/ They mention c-ray as a ray tracing test, though v-ray isn't ideal. Focus on price per core—Epyc 7282 shines, along with 7552 and 7452 for dual sockets. Single-socket blade/1U units seem to deliver the best results. Also note Epyc's efficiency is impressive, drawing just 270W peak. Be sure to request quotes from both options before comparing specs, as pricing varies widely between vendors.

D
DeenBurbigo
Member
61
01-31-2022, 05:56 AM
#3
Very useful post/info. We've been talking about the depreciation value of the high-end products we bought earlier, and how quickly newer, cheaper alternatives can surpass them in 1-3 years—this is just part of our business. Our team tends to choose the more affordable choices regardless of performance, so I'm unsure if we'll be able to justify a server purchase with a $7,500 CPU. For us, it makes sense because the last upgrade cost about $40,000 for average equipment, and a dual 7742 node would more than double our current processing or rendering capacity at half that price, while also reducing our footprint. The main downsides are the upfront expense and depreciation. We'll keep exploring this but appreciate the insights you shared as they helped us evaluate the "high-end" option.
D
DeenBurbigo
01-31-2022, 05:56 AM #3

Very useful post/info. We've been talking about the depreciation value of the high-end products we bought earlier, and how quickly newer, cheaper alternatives can surpass them in 1-3 years—this is just part of our business. Our team tends to choose the more affordable choices regardless of performance, so I'm unsure if we'll be able to justify a server purchase with a $7,500 CPU. For us, it makes sense because the last upgrade cost about $40,000 for average equipment, and a dual 7742 node would more than double our current processing or rendering capacity at half that price, while also reducing our footprint. The main downsides are the upfront expense and depreciation. We'll keep exploring this but appreciate the insights you shared as they helped us evaluate the "high-end" option.