[Proposal] Modern Benchmark that would have mem + cpu focus!

[Kona64]

Hey guys, i had an idea for a benchmark and i don't know where to pitch it, so i'll write my proposal here and y'know, see where it goes. So IMO there is a need for a modern benchmark that is more utilitarian in focus and 2D in nature, and i though of something i do on a regular basis which i use as a "benchmark" and "stability test" which could be awesome for a benchmark which focuses greatly on CORE COUNT, FREQUENCY, AND MEMORY OPTIMIZATION AND FREQUENCY.

My proposal is: making a program which uses a standalone version of google's open source chromium browser to load and render a window of chrome with 60~120 tabs of websites containing all sorts of html5 and javascript and whatever.

Why is it good? Because all of this stuff has to be processed by the cpu and memory (if you disable hardware acceleration) and it's a LOT of data going to-and-fro from cpu to mem, i have used this in the past as a memory overclock stability-test, because it uses a lot of cpu power to process everything, and it uses both a lot of mem in terms of raw gb of ram and in terms of bandwidth and latency. It also scales well with higher core-count cpu's!

Why would it interest users? Because hardcore browsing is something that a lot of people do, and is becoming more and more popular as people learn the value of being able to manage data and tabs and communications, an example of this is how much the popularity of dual or triple monitor setups has exploded in the past 4~6 years. So this would be a meaningful and practical way to measure how "good" your PC is for hardcore web browsing. Why would it interest XOCers? Because superpi is old and new users don't care about it anymore; it has lost relevance and doesn't give a "practical" idea of real-world PC use. This way we'd have a cool bench which can scale well with mem overclocking and optimization BUT has more tangible benefit for modern PCs, this would bring more people to become interested and appreciate overclocking! Also you'd need a few GB's of ram so no maxmem  closer to regular use-case! This might add new meta's to benching such as maybe using DR DDR4 for rank interleaving, who knows, it would be a bench that would strain the PC for a use case that MOST USERS USE THEIR PC FOR (most people use PC's for web-browsing(social networks research etc...) AND gaming, the vast majority of the benches so far focus a lot on gaming, or productivity in the form of creation(video editing, 3d rendering...)!).

Practical Considerations for implementation:
idk, i'm a noob, but i imagine we could like package a version of chromium and all of the content of the "tabs" in a file (like the full web-pages with banners and everything packaged with the bench so no internet connection is required), when the user runs the benchmark there's a "delay" as everything gets copied to the ram to remove storage speed from the equation, then the script executes the browser opening and loading all the tabs and gives us a time of how long it took (or maybe a score based off time)
then this version of chrome could be pre-configured to not use any hardware acceleration when "rendering" images and banners and whatnot directly off cpu and mem
so it's 100% software, adding more strain to mem and cpu, and removing "gpu drivers and gpu in general" from the equation. **Chromium would also need to be modded to render all the tabs at once instead of how chrome seems to do some of the work only when a tab gets opened!

Personal experiences:
i use a 9980XE and all the threads and a lot of mem gets used when i open my bookmark folders, i think it would be a cool bench, and it's a fast bench like spi1m, wouldn't take over 25s to run on a 6+ core if you do sub-60 tabs, and there could be like 2 options like spi1m VS spi32m, one option loads 80 tabs, and the second loads 200! and it scales well with core count as well!  Google chrome and firefox are based off of chromium so this bench would use the same resources people use daily.
I think this would be a bench that would have practical value for evaluating a PC for scholarly/research/web-browsing potential, it would fun for OCers cause it stresses mem a lot (something that's lacking in most popular benches today) and it would help maybe companies that work with making drivers, software, and hardware, possibly optimize things for the improvement of modern computing! I've had a few memory overclocks be "modern gaming-stable" but crash as soon as i try to open a 40+ bookmark folder!

How could this be made into a reality?
We'd need people to be willing to help organize something, i'm somewhat of a noob at programming and i'm quite busy and inexperienced, we could for example make a kickstarter or something to pay someone to work on it, i don't know, we could make a discord server where we coordinate efforts to implement this! so if people want this to happen they need to comment and get in touch so that we can form a group to get working on this!

Credit:
I would like to thank Ground, Ozzie, and Mythical for having helped brainstorming to write this poposal!
I would like to thank all the XOCers and people that make up this community for helping newcomers and stimulating people to get into this amazing hobby!

 

Edited June 21, 2020 by Kona64
Small Fixes

[Kona64]

We could also, for example, make this benchmark be Windows 10 native to get points, forcing the community to get a little more familiar with windows 10 which is more widely used by modern users and more "pratical" for newcomers, as well as make the results more "useful" for the average user!! (also to motivate the geniuses out there to help make a good benching W10 distro i can use ??)

Edited June 21, 2020 by Kona64
Small Fixes

[TerraRaptor]

I have flashback of pcmark "tweaks", there were some for web page rendering there ?

[Tyllo]

Sounds like a fun benchmark, I love optimizing mem for daily so this is right up my alley if its aimed at non-maxmem.

[Leeghoofd]

Better make the coding rock solid, as many will challenge your programming skills, looking forward to it

[Kona64]

1 hour ago, Leeghoofd said:

Better make the coding rock solid, as many will challenge your programming skills, looking forward to it

Ah i posted here cause i know my coding skills aren't up to the task hahaha, my hope was that someone (some people) might find this cool and help me get the the project up and running