Discussion The Benchmarks we need

[nmb255]

Whenever there is a new CPU or GPU range the benchmarkers rush to benchmark that family of chips against the family of chips from the other camp. Using as many benchmarks as they can. Whilst essential, I think they are missing a trick.

A frequent question in these here forums is - Should I upgrade. Therefore something that would be really useful would be benchmarks of new generation vs older generations to help demonstrate when it's worth making the leap of not from 3 or 4 or more generations back.

At best what we get the situation between now vs previous generation which is not far enough back for most.

It would be interesting to see what difference is in games the using: 9900k, 8700k, 6700k,4770k processors and 2080, 1080, 980 and 780 graphics cards for example. Then someone with a 6700k and a 980 could see if they would benefit most from a CPU or GPU upgrade path. Which is a commonly asked question.

I'm sure the data is out there that each gen is a % increase on the previous one and it can be pieced together, but doing it in one test suite would be using the latest titles would give a clearer picture of where and how much the improvement to gain are.

 

[PC Tailor]

K actually think a step further back, it depends what benchmarks you are running.

Frankly, I personally could care less about Furmark, Userbench, PassMark results, because they don't live in real world application.

NVMe SSDs for example, mind blowing on benchmarks, in real world application, in most applications, negligible difference to a SATA interface SSD. If however we're talking actual in application benchmarks, then yes, I would love to see whether it truly is worth upgrading from a generation some years prior.

I still run a 4690K overclocked and rarely see this come up in comparisons in real world application anymore, but it still serves me excellently.

 

[nmb255]

K actually think a step further back, it depends what benchmarks you are running.

Frankly, I personally could care less about Furmark, Userbench, PassMark results, because they don't live in real world application.

NVMe SSDs for example, mind blowing on benchmarks, in real world application, in most applications, negligible difference to a SATA interface SSD. If however we're talking actual in application benchmarks, then yes, I would love to see whether it truly is worth upgrading from a generation some years prior.

I still run a 4690K overclocked and rarely see this come up in comparisons in real world application anymore, but it still serves me excellently.

For sure, benchmarking under typical conditions is missing too. Another often asked question is "I want to game and stream" and the always replied answer is that a better mutitasking processor is the way to go. Based on the performance under extreme multitasking benchmark applications.

It would be interesting to see the real benchmarks of such a situation with different processors and graphic card combinations to properly iron out the sweet spot and fall off for this niche requirement. Often the answers read that without the latest and most core count you won't be able to stream and game at the same time. Which is probably stretching a point just a little.