Jump to content
Search In
  • More options...
Find results that contain...
Find results in...
BlueScope819

PSA about Userbenchmark

Recommended Posts

Posted · Original PosterOP

General background:

 

Upon the release of the Ryzen 3000 series, Userbenchmark adjusted their ranking system to a different one that isn't any longer representative of real performance. When called out, they accused their critics of being "anonymous call center shills". Rigged or falsified benchmarks are destructive to the PC ecosystem: builders (and sometimes even entire companies) fall for these fake benchmarks, build worse PCs, and starve out the creators of good hardware. No industry ever got better by giving all the funding to the loser.

 

This can influence the decisions of PC builders negatively and should be mentioned here. This type of behavior is destructive to the core of the community on LTT forum, because LTT an educational resource, it attracts novice PC builders and educates them about computer hardware. Userbenchmark is a company that targets new users with SEO (just google any two processor names and scroll down), while providing a source of, at best, misinformation, and at worst, intentionally attempting to mislead users into buying computer hardware that is not suited to their workload.

 

Here are some examples of this.

 

Spoiler

U1.thumb.PNG.95d1a4d9255b661ba834c8c012986cc2.PNG

Spoiler

U2.thumb.PNG.a5aaa9493180679571eab7695bdd005b.PNG

 

Okay, so what should I do?

Well, the solution is pretty simple. Don't use Userbenchmark. Below is a list of some much better solutions that reputable benchmarkers use.

Spoiler
On 8/23/2018 at 7:37 PM, Crunchy Dragon said:

More of a benchmark than stress test, this lets you test your CPU’s multicore speed by using your CPU to render an image. I see Cinebench commonly used to get an idea of thermal performance, as well as multithreaded performance in general. Usually if a system isn’t performing as normal, it can be recommended that a user runs Cinebench to get an idea of how their CPU is performing compared to how it should be performing, since Cinebench scores are easy to find online.

 

Unigine Benchmarks -- https://unigine.com/en/products/benchmarks

 

There are a few Unigine programs used for benchmarking, I personally recommend Unigine Heaven and Superposition. These are used usually to test GPU performance and overclocks. I’ve used Unigine programs combined with Cinebench to identify bottlenecks in systems in the past, as well as my personal GPU overclock.

 

Aida64 and Prime95 -- https://www.aida64.com/ https://www.guru3d.com/files-details/prime95-download.html

These are in the same boat, as they are programs designed for extended stress testing of your CPU. The workloads they put a CPU under are usually not close to real world stress levels. These are usually used to measure temperatures at full load over time since they run longer than Cinebench, as well as to test overclock stability over time.

 

Prime95 may experience problems when using it on Haswell and newer Intel CPUs with automatic voltage settings. It should also be noted that with these stress tests that you either need proper cooling or to set voltages to manual to prevent any overheating issues while testing.

 

Aida64 is also useful for its ability to show temperatures per core while running, as well as thermal throttling.

 

 

Furmark -- https://geeks3d.com/furmark/

Furmark is a extended stress test for your GPU. The workloads your GPU is under are definitely higher than pretty much anything you’ll likely use it for. Furmark is commonly used to test overclock stability over time.

BE WARNED: Furmark has been known to kill Nvidia Kepler(GTX 6xx-7xx)series and lower GPUs. Maxwell(GTX 9xx) and higher GPUs are not known to be killed by Furmark, and I’ve personally never heard of it happening.

General guidance for comparing computer components

 

Try and compare with as similar a configuration as possible

  • For example, if you are trying to find the difference between a Ryzen 3 and a Ryzen 5, make sure all other aspects of the system are the same. (Graphics card, memory, etc).

 

Benchmark your workload.

  • If you are building a Gaming PC, don't use benchmarks that aren't games. Large, workstation workloads don't allow you to compare performance in your game.

 

Use multiple benchmarks

  • Some games take advantage of more of the CPU or GPU than others. All games are not created equal, so make sure you benchmark a range of games.

 

Sources:

 

  • TweakTown - "Userbenchmark changes their CPU rating algorithm to push huge focus on single-threaded performance while nerfing multi-threaded performance because 'Ryzen is too fast'"
  • Techspot- "Enthusiasts were taken back by the website's questionable decision to adjust its scoring algorithm in favor of single-threaded performance, inflating scores in such a way that doesn't appear to be entirely representative of the processor in question."
  • TomsHardware - "It doesn't take a degree in computer science to see the obvious disconnect here."
  • Techordeal - "This change penalizes processors which focus on multi-core performance over single and quad-core performance. The result is certainly handy for Intel, who were seeing their rivals offer better performance scores for much lower prices."
  • AMD subreddit - "PSA: Use Benchmark.com have updated their CPU ranking algorithm and it majorly disadvantages AMD Ryzen CPUs"
  • AMD subreddit
  • LTT - "UserBenchmark Adjusts Algorithm to Favor Single-Thread Performance, Causing Strange Results."https://www.reddit.com/r/Amd/comments/gk4kzv/more_proof_that_userbenchmark_is_run_by_12yearolds/

 


@BlueScope819 so I can see your post

#MuricaParrotGang

"My name is Legion 'Murica Parrot Gang, for we are many."

Mentioned in 8/5/2020 TechLinked

Link to post
Share on other sites
29 minutes ago, BlueScope819 said:

Sources

So where does UB benefit?

 

IN other words, what do they get from lying about performance? What's the angle here?


So rise up, all ye lost ones, as one, we'll claw the clouds

Link to post
Share on other sites
1 minute ago, Radium_Angel said:

IN other words, what do they get from lying about performance? What's the angle here?

Marketing by you-know-who? /Conspirary


Remember to quote or @mention others, so they are notified of your reply

Link to post
Share on other sites
2 minutes ago, Radium_Angel said:

So where does UB benefit?

 

IN other words, what do they get from lying about performance? What's the angle here?

Perhaps they don't need profit, they just have something to prove. Whether or not they're getting paid, their testing methodology is pretty obtuse, and they are very prideful of their own figures and derisive of other outlets.


I WILL find your ITX build thread, and I WILL recommend the SIlverstone Sugo SG13B

 

Primary PC:

i7 8086k (won) - EVGA Z370 Classified K - G.Kill Trident Z RGB - Force MP500 - Jedi Order Titan Xp - Hyper 212 Black (with RGB Riing flair) - EVGA G2 650W - Black and green theme, Razer branwashed me.

Draws 400 watts under max load, for reference.

 

Linux Proliant ML150 G6:

Dual Xeon X5560 - 24GB ECC DDR3 - GTX 750 TI - old Seagate 1.5TB HDD - Dark moded Ubuntu (and Win7, cuz why not)

 

How many watts do I need? Seasonic Focus threadUserbenchmark (Et al.) is trash explained, PSU misconceptions, protections explainedgroup reg is bad

Link to post
Share on other sites
5 minutes ago, Fasauceome said:

Perhaps they don't need profit, they just have something to prove.

That's pretty bizarre. I can't imagine their hosting costs are modest.

What about the site cpuboss.com are they any more accurate?


So rise up, all ye lost ones, as one, we'll claw the clouds

Link to post
Share on other sites
10 minutes ago, Radium_Angel said:

That's pretty bizarre. I can't imagine their hosting costs are modest.

They run hella ads though, the only thing that matters is that they manage to keep their SEO up high, aside from that I don't think the actual integrity of the data is important, and they probably don't need to take bribes from Intel.


I WILL find your ITX build thread, and I WILL recommend the SIlverstone Sugo SG13B

 

Primary PC:

i7 8086k (won) - EVGA Z370 Classified K - G.Kill Trident Z RGB - Force MP500 - Jedi Order Titan Xp - Hyper 212 Black (with RGB Riing flair) - EVGA G2 650W - Black and green theme, Razer branwashed me.

Draws 400 watts under max load, for reference.

 

Linux Proliant ML150 G6:

Dual Xeon X5560 - 24GB ECC DDR3 - GTX 750 TI - old Seagate 1.5TB HDD - Dark moded Ubuntu (and Win7, cuz why not)

 

How many watts do I need? Seasonic Focus threadUserbenchmark (Et al.) is trash explained, PSU misconceptions, protections explainedgroup reg is bad

Link to post
Share on other sites
Posted · Original PosterOP
2 hours ago, Radium_Angel said:

That's pretty bizarre. I can't imagine their hosting costs are modest.

What about the site cpuboss.com are they any more accurate?

Really it's important to not look at CPU boss or any other website, but benchmark for your workload. As I said:

 

General guidance for comparing computer components

 

Try and compare with as similar a configuration as possible

  • For example, if you are trying to find the difference between a Ryzen 3 and a Ryzen 5, make sure all other aspects of the system are the same. (Graphics card, memory, etc).

 

Benchmark your workload.

  • If you are building a Gaming PC, don't use benchmarks that aren't games. Large, workstation workloads don't allow you to compare performance in your game.

 

Use multiple benchmarks

  • Some games take advantage of more of the CPU or GPU than others. All games are not created equal, so make sure you benchmark a range of games.

@BlueScope819 so I can see your post

#MuricaParrotGang

"My name is Legion 'Murica Parrot Gang, for we are many."

Mentioned in 8/5/2020 TechLinked

Link to post
Share on other sites

-= Topic moved to Member Reviews =-


Tech News Posting Guidelines - READ BEFORE POSTING | Community Standards | Forum Staff

LTT Folding Users Tips, Tricks and FAQ | F@H Contribution | My Rig | Project Steamroller

 

Spoiler

 †  

 

Character is like a Tree and Reputation like its Shadow. The Shadow is what we think of it; The Tree is the Real thing.  ~ Abraham Lincoln

You have enemies? Good. That means you've stood up for something, sometime in your life.  ~ Winston Churchill

Docendo discimus - "the best way to learn is to teach" ~ Benjamin Jantz

 

I am a StarCitizen are you? My ships: Aegis Eclipse, Aegis Sabre, Aegis Gladius, Aopoa Nox, KI P52 Merlin, KI P72 Archimedes and the RSI Constellation Aquila.

 

My Phones are a Nokia Lumia 925 with WM10 and a Microsoft Lumia 950 XL with WM10 running the Fast Ring insider updates. Broke :(

Samsung Note 9 and a Samsung S9+

 

About Myself:   https://linustechtips.com/main/profile/229093-sansvarnic/?tab=field_core_pfield_46

 

 CHRISTIAN MEMBER 

 

 

Link to post
Share on other sites
On 7/7/2020 at 12:00 AM, BlueScope819 said:

Really it's important to not look at CPU boss or any other website, but benchmark for your workload. As I said:

 

General guidance for comparing computer components

 

Try and compare with as similar a configuration as possible

  • For example, if you are trying to find the difference between a Ryzen 3 and a Ryzen 5, make sure all other aspects of the system are the same. (Graphics card, memory, etc).

 

Benchmark your workload.

  • If you are building a Gaming PC, don't use benchmarks that aren't games. Large, workstation workloads don't allow you to compare performance in your game.

 

Use multiple benchmarks

  • Some games take advantage of more of the CPU or GPU than others. All games are not created equal, so make sure you benchmark a range of games.

Usefull advice!.... for those who start their own bemchmark resource site or channel.

 

But useless for newcomers thay buy 1st time, or upgraders that want to compare their current setup to newer things. They dont have the 1 or 2 sets of hardware available for comparison let alone multiple sets to see the relative differences.

 

But agree wholeheartetly on that kind of poor reference resources. But its not isolated and not new either, just look at all the psu wattage calculators that "overprovisiom" by 200-300% or the infamous bottleneck calculators that always find some level of fault in any combination (and when they dont they are wrong too because of different workload types).

 

Basically these are not informative resources, they are advertisement billboards and/or sales funnels. And just like the good old fashion door to door salesman selling whatever people already had or didnt need, there is always some audience they have success with. 

Link to post
Share on other sites
Posted · Original PosterOP
17 minutes ago, Bartholomew said:

But useless for newcomers thay buy 1st time, or upgraders that want to compare their current setup to newer things. They dont have the 1 or 2 sets of hardware available for comparison let alone multiple sets to see the relative differences.

Youtube is a great source for those who wish to determine the differences between hardware types, as so many people have put their results out there you are bound to find something similar to what you are looking to benchmark.


@BlueScope819 so I can see your post

#MuricaParrotGang

"My name is Legion 'Murica Parrot Gang, for we are many."

Mentioned in 8/5/2020 TechLinked

Link to post
Share on other sites
24 minutes ago, BlueScope819 said:

Youtube is a great source for those who wish to determine the differences between hardware types, as so many people have put their results out there you are bound to find something similar to what you are looking to benchmark.

 

On 7/7/2020 at 12:00 AM, BlueScope819 said:

to find the difference between a Ryzen 3 and a Ryzen 5, make sure all other aspects of the system are the same. (Graphics card, memory, etc).

Simmilar or same? But agree 100% so dont read me wrong :)

And overall i agree your advice is sound and a much better option than relying on 1 or 2 "commercial" sources, especially on comparing the workloads you need and not just some pointless hypothetical stuff (or things youll never need).

 

You also got to love the graphs where a lot of sources just chose scale and offset for graphs so that a difference between a score of 10000.0 and 10000.1 seems like some huge deal 😂

Link to post
Share on other sites

UB has a very odd hard-on for quad-core CPUs for some reason


The Workhorse

R7 3700X | RTX 2070 Super | 32GB DDR4-3200 | 512GB SX8200P + 2TB 7200RPM Barracuda Compute | Windows 10 Pro

 

The Portable Station

Core i7 7700H | GTX 1060 | 8GB DDR4-2400 | 128GB SSD + 1TB HGST | Windows 10

 

Samsung Galaxy Note8 SM-N950F

Exynos 8895 ARM Mali G71 MP20 | 6GB LPDDR4 | 64GB internal + 128GB microSD | 6.3" 1440p "Infinity Display" AMOLED | Android Pie 9.0 w/ OneUI

Link to post
Share on other sites
1 hour ago, D13H4RD said:

UB has a very odd hard-on for quad-core CPUs for some reason

something about the i5 2400- i5 7600 just gets them off


Good luck, Have fun, Build PC, and have a last gen console for use once a year. I should answer most of the time between 9 to 3 PST

NightHawk 2.0: R7 2700 @4.0ghz, B450m Steel Legends, H105, 4x8gb Gell EVO 2866, XFX RX 580 8GB, Corsair RM750X, 500 gb 850 evo, 500gb 850 pro and 5tb Toshiba x300

Skunkworks: R5 3500U, 16gb, 250 intel 730, 500gb Adata XPG 6000 lite, Vega 8. HP probook G455R G6

Condor (MC server): 6600K, z170m plus, 16gb corsair vengeance LPX, samsung 750 evo, EVGA BR 450.

Bearcat (F@H box) core 2 duo, 1x4gb EEC DDR2, 250gb WD blue, 9800GTX+, STRIX 660ti, supermicro PSU, dell T3400.

Rappter(unfinished compute server) HP DL380G6 2xE5520 24GB ram with 4x146gb 10k drives and 4x300gb 10K drives, running NOTHING can't get anything to work

Spirt  (unfinished NAS) Cisco Security Multiservices Platform server e5420 12gb ram, 1x6 1tb raid 6 for plex + Need funding 16+1 2tb raid 6 for mass storage.

PSU Tier List      Motherboard Tier List      How to get PC parts cheap    HP probook 445R G6 review

 

"Stupidity is like trying to find a limit of a constant. You are never truly smart in something, just less stupid."  @CircleTech

 

Link to post
Share on other sites
Posted · Original PosterOP
On 7/9/2020 at 2:07 AM, D13H4RD said:

UB has a very odd hard-on for quad-core CPUs for some reason

You should read their write up of the 3300x, it's really funny.


@BlueScope819 so I can see your post

#MuricaParrotGang

"My name is Legion 'Murica Parrot Gang, for we are many."

Mentioned in 8/5/2020 TechLinked

Link to post
Share on other sites
3 hours ago, BlueScope819 said:

You should read their write up of the 3300x, it's really funny.

I have. It reaffirms my beliefs that they have a weird fetish for quad cores. 


The Workhorse

R7 3700X | RTX 2070 Super | 32GB DDR4-3200 | 512GB SX8200P + 2TB 7200RPM Barracuda Compute | Windows 10 Pro

 

The Portable Station

Core i7 7700H | GTX 1060 | 8GB DDR4-2400 | 128GB SSD + 1TB HGST | Windows 10

 

Samsung Galaxy Note8 SM-N950F

Exynos 8895 ARM Mali G71 MP20 | 6GB LPDDR4 | 64GB internal + 128GB microSD | 6.3" 1440p "Infinity Display" AMOLED | Android Pie 9.0 w/ OneUI

Link to post
Share on other sites

Is UB still good for comparing GPUs though? I'm glad I came across this thread, cause I had a feeling that UB CPU rankings and comparisons were a bit.. off.

Link to post
Share on other sites
4 minutes ago, Hyper_V said:

Is UB still good for comparing GPUs though? I'm glad I came across this thread, cause I had a feeling that UB CPU rankings and comparisons were a bit.. off.

It's fine to use CPU and GPU UserBenchmark, if you just use it as a sort of guide to roughly estimate the difference between multiple components.

You just need to know what to look for and take it with a grain of salt.

For CPU: looks at the single core speed difference and then the amount of cores/threads (and ignore the whole 'score' thing).

For GPU: look at the game FPS numbers, if you want to compare gaming performance

 

Again, use it as somewhat of guidance. Just know there are people running great overclocked systems and people running who have a whole bunch of background programs during the benchmark. Both scores get submitted 'equally', so the scores on UB are not 100% correct.


"We're all in this together, might as well be friends" Tom, Toonami.

Sorry if my post seemed rude, that is never my intention.

"Why do we suffer a lifetime for a moment of happiness?" - Anonymous

 

Link to post
Share on other sites

Userbenchmark, from what I've found, is popular because it's relatively clean and provides a straightforward comparison between products - even if that comparison is incorrect. It's unfortunate, because when someone new to building PCs wants to find the best hardware for their money, they might not take into account differences in systems and performance weights (single vs quad core, for example). 

 


Tag me @Opencircuit74 or quote me if you want me to respond, message me if you want me to spec out a decent low-budget build. (US) CPU: I7-6700k RAM: 32GB (8X4) DDR4-3000 GPU: PNY RTX 2060 MOBO: ASUS Z170-A PSU: Corsair RM850 SSD: ADATA XPG SX8200 PRO 1TB HDD: Seagate Barracuda 2TB Case: NZXT S340 White Cooler: Cooler Master Lite 240 AIO

Link to post
Share on other sites
Posted · Original PosterOP
41 minutes ago, Opencircuit74 said:

Userbenchmark, from what I've found, is popular because it's relatively clean and provides a straightforward comparison between products - even if that comparison is incorrect. It's unfortunate, because when someone new to building PCs wants to find the best hardware for their money, they might not take into account differences in systems and performance weights (single vs quad core, for example). 

 

Yup, according to UB a 14nm Intel single core CPU clocked at 5.7ghz is the best thing to exist since sliced bread.


@BlueScope819 so I can see your post

#MuricaParrotGang

"My name is Legion 'Murica Parrot Gang, for we are many."

Mentioned in 8/5/2020 TechLinked

Link to post
Share on other sites
5 hours ago, BlueScope819 said:

Yup, according to UB a 14nm Intel single core CPU clocked at 5.7ghz is the best thing to exist since sliced bread.

And that a 9100f (4c/4t 3.7ghz) beats my 2700x (8c 16t 3.8ghz). 


5Head :wine_glass: AH YES

Link to post
Share on other sites

The simple fact that they attack other outlets and openly disparage them is telling enough.

You never see GN, HU, LTT or whoever go around screaming at other youtubers/reviewers/tech media. That's HIGHLY unprofessional and what I'd expect out of a 12-year-old fanboy, not someone running a "benchmarking" website.

Link to post
Share on other sites
Posted · Original PosterOP
4 minutes ago, 5x5 said:

The simple fact that they attack other outlets and openly disparage them is telling enough.

You never see GN, HU, LTT or whoever go around screaming at other youtubers/reviewers/tech media. That's HIGHLY unprofessional and what I'd expect out of a 12-year-old fanboy, not someone running a "benchmarking" website.

I totally agree. It's perfectly fine to point out errors with other outlets testing methodology for example (See: Steve from Gamers Nexus roasting Linus about measuring internal case temperatures with a thermal camera through a tempered glass side panel during the roast), however despairing other reputable outlets as "Call center shills" is grossly unprofessional.


@BlueScope819 so I can see your post

#MuricaParrotGang

"My name is Legion 'Murica Parrot Gang, for we are many."

Mentioned in 8/5/2020 TechLinked

Link to post
Share on other sites
On 7/6/2020 at 1:18 PM, BlueScope819 said:

Okay, so what should I do?

Well, the solution is pretty simple. Don't use Userbenchmark. Below is a list of some much better solutions that reputable benchmarkers use.

  Reveal hidden contents

 

I would add in PassMark as a decent benchmark suite. It's not perfect, but it does use a pretty comprehensive and well programmed set of tests that run long enough to get a decent baseline for your hardware. (they also throw out the 2D benchmark scores for GPUs, despite showing them to you, since they are impossible to use for real-world comparisons) They even have the ability to search through other benchmarks that use the same CPU+GPU as you, and put them on the comparison chart.


CPURyzen 5 3600 Cooler: GAMMAXX 400 with dual Arcic P12 PWM RAM: G.Skill Ripjaws V 16gb 3600 16-16-16 MoboASRock X570M Pro4

Graphics Card: ASRock Reference RX 5700 XT Case: Antec P5 PSU: Rosewill Capstone 750M Monitor: MSI Optix MAG241C Case Fans: 3x Arctic P12 PWM

Storage: Mushkin ECO2 512GB SSD, HP EX900 1TB NVMe, dual Constellation ES 2GB in RAID1

https://hwbot.org/submission/4497882_btgbullseye_gpupi_v3.3___32b_radeon_rx_5700_xt_13min_37sec_848ms

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×