CPU-Z changes benchmark code (Ryzen suffers)

[Humbug]

this saga sums up everything wrong with synthetic benchmarks.

 

neither the new code nor the old code has been written with the objective of solving a computational problem.

[Mihle]

[Tinfoil Hat]

 

Intel paid them do do so!

[samiscool51]

10 hours ago, porina said:

Does anyone actually use the CPU-Z bench? I don't recall them relating it to any particular real world scenario so I've not really paid attention to it.

it's no-longer used due to its workload being top easiy to do

plus, newer benchmarking software comes out and it's even more punishing than the last

in fact the most CPU intensive program is not something like adobe premiere pro

it's Universe Sandbox 2, it WREAKS CPU's more than Po from kung fu panda trying to defeat his biggest enemy..... stairs.....

sericery, i don't know why reviewers aren't using them in their reviews, it shows how good a CPU is with BIG workloads

this guy shows how it does on a 6700K overclocked (i don't remember what to) and the CPU is not having the time of it's life

 

[Doobeedoo]

Well as long as scores show to be in line what would be expected and be precise in comparison all good. 

[DocSwag]

7 hours ago, ApolloFury said:

Joke > your head

It's hard for me to tell when people are joking... I was on the fence as to whether or not it was a joke or not.

 

Sarcasm doesn't carry well on the internet.

[Sauron]

As with any synthetic benchmark, I wouldn't base my purchase decision based solely on it - this is a perfect example of why it could be unreliable, since some parts may or may not reflect real world applications.

[Sauron]

13 hours ago, porina said:

Does anyone actually use the CPU-Z bench? I don't recall them relating it to any particular real world scenario so I've not really paid attention to it.

I used it to stress test my cpu when I overclocked it, it works fine when it's about comparing a cpu to itself.

[Humbug]

8 minutes ago, valdyrgramr said:

This is why I find synthetic benchmarks useless.  It's nothing more than a childish, "My dad can beat up your dad!" fight.  It doesn't show any real world benefit, is generally useless in real world application, and you should really compare your buying options on real world scenarios like how fast it unzips, renders, and so on with tasks that you actually plan to do.  Not, how fast it moves through unrealistic graphics or other tasks for the sake of "ooooo shiny!"

Exactly, they are not trying to really solve any computational problem.

They are trying to create something which is equally demanding compared to an 'average' of real world application. Which is crazy because every real world application is different. And then when some architecture comes out which performs their algorithm faster then they have to go back and change their algorithm... If the next Intel CPU architecture is too slow in this bench but fast in real world apps are they going to then revise the bench again?

 

I understand why they do it. Because they are trying to perform the same as the average of other apps. Rather than get the best performance out of their app...

 

Which is why we should refer real world testing. By the same token give me a benchmark suite of real games like Doom 4 and GTA 5 anyday over 3D mark...

[K0MP4CT]

13 hours ago, Darth Revan said:

Looks like I pissed off AMD fanboys.

I don't own any AMD products. You didn't piss anyone off. Your statement was incorrect, and you were corrected. Grow up.

[Taf the Ghost]

There's the definite meta-level of "well, Intel gets toasted on this, so let's change it!" that can't be avoided.  CPU-Z could have dealt with that by being much more clear that they botched up originally with the Benchmark. They simply used a workout that was inefficient on Intel as some sort of Benchmark.  Ryzen's architecture, with is insanely good at certain workloads, simply blew up what was previously thought "hard".  Which it was, on Intel's architecture.

[mr moose]

On 5/7/2017 at 10:51 AM, done12many2 said:

 

How did you reach this determination? 

 

 

I'm pretty sure you just pulled the Intel pay off theory completely out of your back side.   Additionally, multi core smartphones and desktop parts have nothing in common.  

 

You know someone has an irrational emotional attachment with zero argument when they resort to the shill gambit.