10900k vs 5950x overclocked for gaming

[Robert896r1]

https://kingfaris.co.uk/cpu/battle

I've known Faris for a couple of years and he's always very thoughtful in his testing and just sticks to the facts. Unlike mainstream viewers, he'll also list all his timings and the performance he was getting out of the platform for testing so you can directly compare your setup vs his.

Please note all the setups here are stress tested for daily stability. Only on an AIO thus no 'high end' cooling.

 

[Robert896r1]

10900k was probably sucking over 300w and it still couldn't bear the 5950x in most cases

In games, 10900k’s pull 400w+

Everyone knows that.

 

[Robert896r1]

You wouldn't use a 5950X for gaming, 5800X is or even 5600X is just as good if not better. The best one is the 5800X.

Couple of things about this 'review? You may notice the difference between Stock and PBO on these charts is nothing to marginal, Just enabling PBO does little or nothing, all it does is set higher power limits and most motherboard are feeding the CPU plenty out of the box, to overclock Zen 3 you add (insert Mhz here) to the boost and play with the curve optimiser, usually setting negative 10, 15 or 20 and the CPU will run 5Ghz + on all cores in games.

Its what i did, Negative 15, +200Mhz boost and boom..... 5.05Ghz all cores, its so simple, i don't know why these 'apparent? Tech reviewers still haven't learned how to do that.

And 5.4Ghz? either that is a very 'cherrypicked' sample or he's ramming a ton of volts into it, the kind of volts no one in their right mind would run 24/7.

It's in the review: This profile is a 5950X running with PBO2, FMax at +150MHz, PPT/TDC/EDC at 270/190/190, and a stable all-core negative offset of 14. The RAM is overclocked to 3733MHz C14.

Along with all his timings, aida, what it boosted to etc.

 

[Robert896r1]

Just wanted to say thanks for posting this, I checked out the other article he did on RAM latency/bandwidth. Loads of real data, very interesting. I tend to ignore conclusions and learn from the data (or try to).
It seems to impact framerates to a very significant degree, perhaps even as much as IPC. In particular it seems to affect the 1% lows. It's kind of relevant to the modern intel/amd argument as it seems that latency and bandwidth are actually incredibly important for gaming.

I'm currently still stuck on a 6700 but I bought some expensive b-die a while ago, because it seemed a good upgrade from a price performance point of view. That fact that he's showed the actual timings is very helpful.

Timings on ram matter so much for performance. Being transparent with the data overall was a key goal. It was also an interesting experiment to see how many people actually review the data vs accept or reject the results based on how it suits their bias.