How to choose the machine configuration for maximum performance WebGL?

Since the application needed to render a large number of 3D scenes using WebGL, our company decided to purchase a high-performance desktop machine. I wondered how to choose the machine? Multi-core CPU or high-frequency CPU? Game graphics or professional graphics? Gaming computer or professional graphics workstation。
I can’t find this solution online, I hope to get your’s advice.

Budget: 4300$

You won’t benefit much from multiple cores as much as a single fast one. Graphics wise I’d always go for a high end gaming one. It’s not like you can use all the features that the gpu offers when working with webgl, but you will be able to crunch more stuff.

2 Likes

Could you explain some more background to this? Is the application only supposed to work on the machine?

Definetly more of a gaming class graphics card, basically workstation class cards like Quadros mainly benefit from more precision (64bit) you can’t use with WebGL anyway. When i got my Titan X it was cheaper and more powerful than the Quadros, in terms of vram it is still overkill.

If you’re interested im releasing a IndexedVolume plugin soon, it boosts large scenes massively and with auto-instancing it can push the boundaries where the process would crash without.

1 Like

Yeah, the way I choose is similar to your advice.

Yes, only supposed to work on the machine. My application has two version: high quality and low quality.The low quality version has been used for Poor performance computer. This high quality version will be used for high quality displays and animations, then used in exhibitions, conferences, screencasts and then on PPT.

RTX 2080Ti was choiced

That’s so great and coincidentally helpful for me , my application’s core tech is instancing and need to be optimized further. I has watched this.

minimal if any

RAM is always a thing, the more the better. The kite demo of unreal for example requires about 24 or 26 GB RAM already to edit.

For browser this can be different, if you run the application in a container (or set some flags in chrome) you can get more available ram for your process, since it is limited around 2-4 (it probably scales to how much is available)

1 Like
  • RAM can help, but difference in throughput between ram is fairly minimal.
  • Disk speed is almost irrelevant.
  • CPU is important, if you have a lot of draw calls especially. I noticed around 20% performance boost in object throughput when going from an old i7 to an i9.

If you have a specific application that you are running - further improvements are probably to be gained from software side.

1 Like

At this point though, i’d start to question whether the browser + WebGL is the right platform for this :slight_smile:

Has purchased this machine:

Dell XPS8930
GPU: RTX2080 Super
CPU: core-i7 9700K
RAM: 16G DDR4 2666MHZ no-ECC
SSD: 512G M.2 NVME 固态
Price: 2600$ (in china)

Why is RTX2080 Super?
RTX2080 Super is much cheaper than RTX2080 Ti, However, their performance gap is not very large.

Why is core-i7 9700K?
core-i7 9700K has high clock frequency and 8 cores. Single browser tab can be benefit from this high clock frequency. And WebStorm and chrome will also be heavily used, this two application can be benefit from this 8 cores cpu.

I will not use this machine for Large 3D editing program

1 Like