CPU/GPU review methods ( time to include 1% lows please )

[TheLostSwede]

So majority agrees on this forum, most thrustworty review sites have already added 1% lows.

26 people are not a majority of this forum... But whatever makes you feel good.

 

[oxrufiioxo]

26 people are not a majority of this forum... But whatever makes you feel good.

I don't even understand why they are not included W1zzard is one of the best reviewers period so I give him the benefit of the doubt as to why he doesn't include them in cpu reviews but it still forces me to seek out other publications to make a purchasing decision.

I guess in some ways it's a good thing because people should be looking at 4-5 different reviews to get a general baseline of the hardware anyways.

 

[JRMBelgium]

26 people are not a majority of this forum... But whatever makes you feel good.

Look. I am only asking because
A
There is no additional work, the benchmark has to run anyway, so why not include the data that's already been gathered while benchmarking?
B,
It would make TPU articles more valuable. TPU not only leaves out 1% lows but also streaming + gaming performance, they also recycle old benchmark data ( latest GPU review includes 3 AMD drivers and 2 Nvidia drivers ) and they claim "All games are tested using the same game version."wich doesn't really makes sence for titles like Battlefield V wich are still getting patches to improve performance and stability. They also don't include the most played games like Tom Clancy's Rainbow Six: Sieg or Call of Duty Warzone and don't compare medium/ultra framerates.

And I get that most of the things I mentioned take extra work, more manpower. But including 1% lows simply doesn't because the data is already there. I upgraded from 2700x to 3900x and after that I lost the trust in TPU simply because their AVG fps results don't represent actual gameplay experience at all. Going from 80-90fps drops to 120-130fps drops makes a world of difference in Battlefield but in the review it's like there is no gain at all.

 

[EarthDog]

But 1% only tells the story with THAT CPU only. As you said, going from a 2700x to 3900x made a difference in 1% lows... but GPU tests are with ONE CPU (most anyway). So anyone with a different CPU wouldn't see these results either. You can't cover every single configuration. That said, I agree to have 1% in there...but for different reasons.

It's also hard to test on the same driver for long without retesting everything as well. Sometimes a new GPU comes out that is only good from XXX.zz version forward so testing with that and newer cards needs to be done on the release driver. That's a shit load of effort every time something new comes out (note, not talking major arch changes, but look at, for example, RTX 2080/2080 Ti. Then lower tier released months later, then non RTX tier.... all new drivers......Oh, then AMD gets mixed in there. Navi big boys... then the 5600 xt, then the 5500xt..less than 4 weeks apart! oh and let's not forget Adrenalin 2020.... so, seven (at least) COMPLETE retests in 1.5 years sounds like a lot of effort for little reason. I don't mind if it is mixed that way... there isn't much choice without throwing a ton more hours at it.

 

[moproblems99]

streaming + gaming performance,

Let's not waste W1zzards time with stuff like this.

 

[John Naylor]

It is certainly a relevant stat ... but before that Id like to be able to click a box and she all the out of box performance numbers, shift to overclocked