Bias?

[bipolarsuxs]

Some people on facebook were irritated because I posted this, https://www.userbenchmark.com/UserRun/29216223. Said they are bias FFS. Comments?

 

Thanks

[5x5]

Userbenchmark is a known fraudulent site. They are correct that it should never be trusted about anything as the results are fake and manipulated

[Slottr]

Userbenchmark is notoriously shitty

 

They have an extreme bias towards intel chips, and they stick AMD chips in the gutter for it.

 

Don't use their site for anything. There's so many alternatives that will provide accurate information

[Bombastinator]

33 minutes ago, 5x5 said:

Userbenchmark is a known fraudulent site. They are correct that it should never be trusted about anything as the results are fake and manipulated

Fraudulent is strong.  “A whole lot less accurate than they claim to be” perhaps. There’s a much wider width to their accuracy than they say, and they use metrics of questionable value.  
 

I still don’t think they’re totally useless, just less useful than they claim to be.

[Bombastinator]

I suspect the issue is the site that was picked.  They have severe reputation problems atm.

[5x5]

30 minutes ago, Bombastinator said:

Fraudulent is strong.  “A whole lot less accurate than they claim to be” perhaps. There’s a much wider width to their accuracy than they say, and they use metrics of questionable value.  
 

I still don’t think they’re totally useless, just less useful than they claim to be.

I've caught them changing my own results and fudging my numbers months apart. Definitely fraudulent

[Bombastinator]

3 minutes ago, 5x5 said:

I've caught them changing my own results and fudging my numbers months apart. Definitely fraudulent

Strong argument

[mr moose]

The testing itself is not biased,  in order for it to be biased it would have to report a different result depending on the name on the product rather than how it performed in said test.  If the same test is used on multiple samples then the result is not biased,  you just have to be careful how you interpret the results.   That said their written articles do tend to have less than desired wording and conclusions.

 

 

[Bombastinator]

3 minutes ago, mr moose said:

The testing itself is not biased,  in order for it to be biased it would have to report a different result depending on the name on the product rather than how it performed in said test.  If the same test is used on multiple samples then the result is not biased,  you just have to be careful how you interpret the results.   That said their written articles do tend to have less than desired wording and conclusions.

 

 

@5x5 is claiming actual data manipulation.  
 

I did hear very bad things about their algorithms and the complaints about their sampling make sense.  Misleading rather that fraudulent perhaps though it could be argued that the functional difference is small.  This is excepting the data manipulation claim.  That bit potentially changes things.

[Craftyawesome]

34 minutes ago, 5x5 said:

I've caught them changing my own results and fudging my numbers months apart. Definitely fraudulent

Do you have screenshots? Sounds interesting.

[mr moose]

Just now, Bombastinator said:

@5x5 is claiming actual data manipulation.  
 

I did hear very bad things about their algorithms and the complaints about their sampling make sense.  Misleading rather that fraudulent perhaps though it could be argued that the functional difference is small.  This is excepting the data manipulation claim.  That bit potentially changes things.

Yer,  I will wait for actually evidence to that before I believe it.  Anyone can claim that.  Lots of claims regarding the manipulation and algorithm's go around, but when you filter through the results they are not far from any other benchmark.  AMD score well in multi core, Intel in single thread.  That is starting to change now, but we have to remind ourselves that zen has only just overtaken Intel on single thread performance and they have  really low market saturation still so it will take a while for the benchmarks to catch up.

 

I wouldn't read their marketing guff and articles.  You can clearly hear a bias in written form (there is no need to talk about moar cores and forcus on single thread as much as they do in articles).

 

 

[5x5]

3 hours ago, Bombastinator said:

@5x5 is claiming actual data manipulation.  
 

I did hear very bad things about their algorithms and the complaints about their sampling make sense.  Misleading rather that fraudulent perhaps though it could be argued that the functional difference is small.  This is excepting the data manipulation claim.  That bit potentially changes things.

 

3 hours ago, Craftyawesome said:

Do you have screenshots? Sounds interesting.

Yes and no. Cannot post anything visual due to work NDA having a clause regarding using tire party tools meaning if lose my job if I were to post anything proving I ran unlicenced software on my system

 

But my i7s score kept increasing every few months despite nothing about the system changing. Ita point score got better despite it getting older. Not normal inho

[MonkeyPants]

Which website do you guys use instead of https://www.userbenchmark.com ?

[5x5]

7 minutes ago, MonkeyPants said:

Which website do you guys use instead of https://www.userbenchmark.com ?

GamersNexus, OC3D and Anandtech. By far the most reliable.

[minibois]

9 hours ago, bipolarsuxs said:

Some people on facebook were irritated because I posted this, https://www.userbenchmark.com/UserRun/29216223. Said they are bias FFS. Comments?

In my opinion, their bias is 'best' shown in their reviews and the way they choose to represent their 'overall score'.

Example: https://cpu.userbenchmark.com/Compare/Intel-Core-i3-9100F-vs-AMD-Ryzen-7-2700X/4054vs3958

9100F vs. 2700X.

4c4t vs. 8c16t.

 

They have the same 'Overall score'.

But you have to dive a bit deeper to actually see how the CPU's differ.

Simply put, they put a large emphasis on single-core speed and not as much on multi-core speeds, which in my opinion is equally important nowadays.

But according to them 8-core (and more) speeds are not important, as telling gamers it's important is like selling ice to Inuit. According to Userbenchmark of course.

 

Plus, another criticism you could give them is something that is a pro and con.

Pro: Userbenchmark has a TON of users that have run their benchmarks, so a LOT of data to compare.

Con: not all users know 'how to benchmark', or more simply put; not everyone benchmarks in the same way.

There is a performance difference between running a benchmark with no background tasks active and running a benchmark while watching YouTube, etc. etc. Because of the very open nature of Userbenchmark, it's difficult to know if all users have run a 'correct' benchmark.

9 minutes ago, MonkeyPants said:

Which website do you guys use instead of https://www.userbenchmark.com ?

You can still use userbenchmark to get a rough idea of performance, just know what you should be looking at for your purpose.

Their overall score is something I would ignore altogether, as that is based on this formula:

Quote

Single-core scoring now accounts for 40% of the result, and quad-core score takes a hit down to providing 57% of the score. Finally, the multicore score (specified to be 64 threads) only makes up 2% of the final score.

You can look at overall score, but it's much better to look at those individual numbers (which are present on the comparison pages too), to actually gauge how well a CPU would work for you.

But only use Userbenchmark for a rough idea, because you don't know about the consistency of all the benchmarks.

 

If you want an actual idea of the performance difference between CPU's, you would need to find a review that goes over all the CPU's you want to compare, as then you know the parameters the CPU's have ran under (what overclock (if any), what cooler, what videocard, etc. etc.).

Of course it's not always possible to find a review comparing the just the two you need to compare, which is when Userbenchmark can be used to get rough idea of performance difference.

[5x5]

@miniboisthere's also the fact that they openly attack established tech outlets such as Hardware Unboxed, Gamers Nexus and Anandtech which itself is insanity.