UserBenchmark vs Hardware Unboxed

[Asteliks]

Hi everyone!

Today I've came across a YT video from Hardware Unboxed “Discussing UserBenchmark's Dodgy CPU Weighting Changes”. I got there due to a link under “Incompetent (coar core) smearers” on the about UserBenchmark page in the "Why we do it section". Thus I open the case of UserBenchmark vs Hardware Unboxed. I would like to hear from you guys what is your take on this topic are HU clams baseless or do they have some merit? I’m asking this question to settle this matter ones and for all. And especially for those that are new to these topics and/or unexperienced. The LinusTech court is in session.

[LogicalDrm]

Maybe use our search function. This isn't new thing, and I remember it being a topic at WAN Show or Level1 News (the only tech news shows I watch).

[minibois]

This has been discussed quite a bit in a previous - now locked - thread:

 

The reason UserBenchmark changed the weighting was because they felt like single core was still very important. I guess that is true to some extent and I was okay  with what they did at the time (I didn't agree with it, but I was indifferent on the change).

Now at this point they have kind of buried the 64-core score at the bottom of the page, which I think is a shitty move from them. That means everything that has over 8 threads will not be shown its true potential in one quick overview. In my opinion, that is what UB is all about; seeing a quick and dirty comparison between 2 CPU's.

 

Not that UB has ever been super accurate; but it's just kind of a simple metric to compare. But I don't know what their end-goal is in not presenting the full picture of a CPU in one go. I don't like it though!

 

(Welcome to the forum, by the way! )

[Zando_]

Hardware Unboxed > random shit out of the bottom of your jacket that you haven't worn since last winter > trash > toe gunge from rubber gumboots after getting water in them and not wearing socks > actual shit > UserBenchmark. For accurate numbers anyways (especially on CPU), for GPUs they usually have deeeecent-ish ballpark numbers. 

[Princess Luna]

 

They changed it later though... but any ways Userbenchmark is fine if you know it's only good for a rough idea where hardware lands in relation to other hardware... that's it.

[amdorintel]

3 minutes ago, minibois said:

UserBenchmark

i used to go to that website a lot, but some people dont seem to give userbenchmark that high regard.

[Jurrunio]

I dont like the idea of giving a set performance number in the first place. The balance between faster cores individually and more cores overall are seen differently between different tasks. Currently their scaling works well for gaming, but only gaming. Otherwise pros with unlimited budget (but only a single CPU) will be buying 9900KS for rendering instead of a TR 3970X or 10980XE, the CPU they should be buying.

[Mira Yurizaki]

I treat aggregate sites like these only to get a very high level overview of how two parts compare, but nothing more than that. It's if I want something quick and dirty, but nothing as an absolute.

[Asteliks]

1 minute ago, Mira Yurizaki said:

I treat aggregate sites like these only to get a very high level overview of how two parts compare, but nothing more than that. It's if I want something quick and dirty, but nothing as an absolute.

could you be more Specific? Any sites you would recommend for an in-depth searcher?

[Mira Yurizaki]

2 minutes ago, Asteliks said:

could you be more Specific? Any sites you would recommend for an in-depth searcher?

AnandTech has a section where they listed the parts they've tested on their benchmarking suite and allows you to compare two parts. It doesn't cover everything, but it covers enough that you can fill in the gaps.

 

I don't know of any other site that has something like this.

[Otto_iii]

Userbenchmark is not considered that reputable, at best its a decent place to look if you are just getting into hardware and trying to get a very rough reference of components ability and you know literally 0% about what anything is, at all, just trying to learn names of PC Hardware and relative (very loosely relative) performance.

The whole website is based off a outdated 5mb or so app that just benches a users own hardware, in very quick and not at all deep ways to get a very basic metric, then they repost overall results on the website through millions of users results. Time and time again hardware and how it functions in real world apps, games etc doesn't line up with it, so it should be taken with a huge grain of salt, or like i said before only used as a very loose metric for performance when leaning about hardware for the first time. 

If you are at the point you are looking to spend real money its much more advised to look at actual benchmarks, whether its channels like Hardware Unboxed(Techspot), GamersNexus, OptimumTech, JayzTwoCents, UFDTech, LTT, Pauls Hardwar, BitWit, TechYesCity etc or likewise reputable websites like AnandTech, Guru3D, TechPowerUp, Techspot(Hardware Unboxed) etc and plenty more.  

I'd even take low viewer count, 10-50k subscriber techtubers so long as they appear to be testing correctly, over UserBenchMark.  

Real world performance>A outdated 5mb benchmark app