UserBenchmark Adjusts Algorthym to Favor Single-Thread Performance, Causing Strange Results

[TechyBen]

20 minutes ago, leadeater said:

I'm sure when Star Citizen comes out all these game engine threading issues will be solved.

Min Requirements: Supercluster

Star Citizen: Hold my space beer.

 

PS, wow, Moose ninja'd me. XD

[NeuesTestament]

2 hours ago, leadeater said:

I'm sure when Star Citizen comes out all these game engine threading issues will be solved.

So you are saying they will never be solved

[Ergroilnin]

I do not really care to be honest.

 

Though if the percentages showed in the OP's post are true, it is a good thing in a way that enthusiasts do know what to expect and look for. It would be way worse if they kept their scoring formula completely secret.

 

I am not an enthusiast per se, but I do like building computers, poking around inside them, reading about new and old stuff etc. And if I want to compare two CPUs (or GPUs), I first look at userbenchmark for a rough estimate and then go for proper benchmarks if I am interested enough. And that usecase of the site won't really change much by this shift.

[vanished]

At first this seems like a bad thing, but then you realize that that arbitrary and meaningless number was always garbage and if you're looking at it, you're a fool, and a change to the algorithm behind it hasn't changed that.  This number never mattered. The fact it's different now doesn't either.  Just look at the actual tests, problem solved.  It's like getting upset over someone's opinion about what product is best, because that's literally what this is.  Do you want an objective measure or an opinion?  This number is an opinion, just ignore it.

[Humbug]

The issue is that people always google stuff such as i7 7700k vs R7 3700x.

The first google result you get when comparing CPUs head to head such as this is userbenchmark, so it is actively misleading thousands of people every day.

 

Remember most people only google this stuff when they need to purchase something, so they are not tech savvy like we are to drill down through the numbers and make sense of it all.

 

Is it possible to contact google and have them consider if they want to have a misleading source as their first search result?

 

I myself have often come across this website when searching for comparisons on various laptop CPUs, because most of the tech channels that we follow do not test laptop CPUs.

[JoostinOnline]

5 minutes ago, Humbug said:

Remember most people only google this stuff when they need to purchase something, so they are not tech savvy like we are to drill down through the numbers and make sense of it all. 

A disappointing number of members on LTT forum have "announced" that they'll no longer be using Userbenchmark. And I'm just like, why was anyone here using that in the first place? These ranking sites are pretty garbage.

 

The only one I will occasionally use is Passmark, which doesn't give weighted scores. Even then I don't consider it too reliable. It's just slightly helpful for getting a point of reference for models I'm not familiar with.

[Humbug]

5 hours ago, JoostinOnline said:

A disappointing number of members on LTT forum have "announced" that they'll no longer be using Userbenchmark. And I'm just like, why was anyone here using that in the first place? These ranking sites are pretty garbage.

 

The only one I will occasionally use is Passmark, which doesn't give weighted scores. Even then I don't consider it too reliable. It's just slightly helpful for getting a point of reference for models I'm not familiar with.

Sometimes you just don't find the CPU comparisons you are looking for elsewhere. With enthusiast and desktop gaming products it's easy, but for less flashy and more obscure laptop CPUs when you want to compare two parts you don't always have hardware unboxed etc running benchmarks. So people google stuff and find that this is one of the few sources...

[AngryBeaver]

I personally feel that they aren't giving a good approximation of the cpu as a whole. Single thread performance matters, but not any more so than multithreaded now. So yes in some games it can make a pretty big difference, but outside of that multi-thread is king.

[Anomnomnomaly]

This thread is funny... so many people thinking Star Citizen will ever actually be finished and released.  

[thechinchinsong]

Its a shame since Userbenchmark is so high on Google's search list, any newcomer to the PC arena is gonna have their data skewed by this website. Hopefully people are gonna be smart enough to do more research on their own, but I'm not getting my hopes up.

[JoostinOnline]

6 hours ago, Humbug said:

With enthusiast and desktop gaming products it's easy, but for less flashy and more obscure laptop CPUs when you want to compare two parts you don't always have hardware unboxed etc running benchmarks.

That's why I mentioned Passmark.

[spartaman64]

5 hours ago, Anomnomnomaly said:

This thread is funny... so many people thinking Star Citizen will ever actually be finished and released.  

it was supposed to be released in 2014 and its horribly mismanaged

[SupremeGOAT]

 

New logo for UserBenchmark free of charge.

[Animal901]

On 7/25/2019 at 3:33 AM, Dabombinable said:

Even fucking Skyrim can use 4 cores...so 8 years later, how is single threaded performance that important?

That old game again.....that game was made way before quad core. I think dual core just came out. New vegas ran naturally on dual. So nervernind what you said.  

 

 

 

 

On another note.  This forum has a big problem with running cinebench 24/7. It's not always about that multicore. People are still using phenom ii, bulldozer and core 2 quad machines.  Ryzen hasn't been the killer everyone expected.  I still remember the pain of owning an amd athlon xp 2400+ with the fancy radeon 9800pro. They consistently relesed bad drivers that crashed on battlefield 1942. everyone forgets how often you would have to reinstall xp back then. Thanks amd/ati.

[Bananasplit_00]

9 minutes ago, Animal901 said:

That old game again.....that game was made way before quad core. I think dual core just came out. New vegas ran naturally on dual. So nervernind what you said.  

I am fairly sure we had quad core CPUs before Skyrim lmao

[Animal901]

21 minutes ago, Bananasplit_00 said:

I am fairly sure we had quad core CPUs before Skyrim lmao

I keep forgeting skyrim wasn't elder scrolls 4. Cause when you played 3 why would you play 5?

 

The real improvements y'all are seeing is the extra money you're putting into your video cards. Also the amount of Chrome tabs you can keep opening to test your 384gb of ram. 

 

I played half life 2 on a potato. 

[Emberstone]

3 hours ago, Animal901 said:

That old game again.....that game was made way before quad core.

 

3 hours ago, Animal901 said:

I keep forgeting skyrim wasn't elder scrolls 4.

Technically the first quad core CPU released the same year as ES IV Oblivion (2006), though of course games of that era were a long way from being meaningfully privy to multiple processors.

[Animal901]

Skyrim runs just fine on cheap consumer hardware.  Like an amd athlon64x2 at 2.6ghz with 2mb cache, 4gb of ram and a 20 dollar 1gb radeon. I just decommissioned an opteron 1218he and it could play minecraft. Could not watch 4k though. 

[mr moose]

8 hours ago, Animal901 said:

 This forum has a big problem with running cinebench 24/7.

I think some of the problem with this forum is that they see things that use more than 4 cores and just assume  80% of consumers demands need more than 4 cores.  Majority of consumers still only do light office work, emails, facebook and the odd game.  A far cry from needing more than 4 cores, and if we're honest a far cry from there having been a significant consumer demand to push hard in that direction.  This is why we are only starting to see real growth in core counts in recent times.  And even that still only panders to entry level workstations and DIY video/render buffs. 

 

Remember using more than 4 cores and needing to use more than 4 cores are two different things.  There are other factors like the speed of the cores, HT or real cores, RAM amount and speed and is the game GPU limited.

[thorhammerz]

41 minutes ago, mr moose said:

I think some of the problem with this forum is that they see things that use more than 4 cores and just assume  80% of consumers demands need more than 4 cores. 
...
Remember using more than 4 cores and needing to use more than 4 cores are two different things.

Those extra 15% FPS aren't going to buy themselves .

[mr moose]

4 hours ago, thorhammerz said:

Those extra 15% FPS aren't going to buy themselves .

Those extra 15 FPS while your working on a spreadsheet or emailing the boss last weeks accounts?

[Animal901]

To think someone created vertical sync to be used.  Lmfao.....

[thorhammerz]

44 minutes ago, mr moose said:

Those extra 15 FPS while your working on a spreadsheet or emailing the boss last weeks accounts?

While streaming that game while editing that video of you streaming that game while running the latest-and-greatest MC benchmark....

 

[mr moose]

3 minutes ago, thorhammerz said:

While streaming that game while editing that video of you streaming that game while running the latest-and-greatest MC benchmark....

 

That's not the 80% of consumer demand though.

 

EDIT: I'm not having a good day,. leave me alone

[Animal901]

If you know the processor and how it stands then there is no reason to google it, or look at userbenchmark. I've been running my older stuff on Userbenchmark. There's nothing about it that says Intel is better than ryzen in any way. The group leaders tend to have some really cool hardware. That happens to be ryzen.  Userbenchmark still hates my phenom II x4 955, 8gb samsung ram, hp quadro 2000 1gb digital audio workstation just like you. Nothing about core 2 quad is confused to be better than ryzen. I7 isn't confused either.  Cinebench rendering a lame photo isn't exactly proving anything. If you know so much about the hardware, then why does anyone benchmark at all?