Is user benchmark inaccurate?

[InternetSwag]

I was comparing the 11400f to my cpu, the 8700. And it looks like decent progress about 13% faster in every aspect.

But if I look at gaming benchmarks, they're virtually identical.

I know gaming isn't really an exact science because of gpu and resolutions etc but even like looking at cpu bound games or low res high fps it seems more of a draw.

So idk about this userbenchmark site.
Maybe it's ok as a quick glance, but I wouldn't trust it 100%

 

[Barbarian Conan]

I was comparing the 11400f to my cpu, the 8700. And it looks like decent progress about 13% faster in every aspect.

But if I look at gaming benchmarks, they're virtually identical.

I know gaming isn't really an exact science because of gpu and resolutions etc but even like looking at cpu bound games or low res high fps it seems more of a draw.

So idk about this userbenchmark site.
Maybe it's ok as a quick glance, but I wouldn't trust it 100%

It is a known turd.
Both the AMD and Intel subreddits banned links to it.

When AMD released higher core count CPUs, they assigned more weight to single core performance. Some intel CPUs were also caught in the crossfire:

 

[DrJohnZoidberg]

No. There is a reason it was banned from being used on the Intel subreddit.

It has zero useful information on it as their scoring formula doesn't make any sense.

 

[supersunbird]

Does tomshardware still do the comparison list? Found those useful back in the day.

 

[cguy]

One possible real reason is that the 11400f supports AVX-512, so for benchmarks that use it, they may see up to a 1.8x performance increase (per clock). So it can come out on top for math benchmarks despite a lower clock rate.

Furthermore the 11400f also has a significantly larger L1 and L2 cache, that for certain benchmarks could be enormously beneficial.

Games typically won’t use AVX-512, and will typically consist of many types of workloads so that the benefit from the large cache would be less dramatic than a benchmark consisting of a single workload that wholly just fits into one of the new larger caches.