The Problem With GPU & CPU Gaming Benchmarks - What They Don't Tell You!

In this video, we delve into the intricate world of PC hardware benchmarking. We discuss how the multitude of variables involved in benchmarking can lead to skewed perceptions of a component’s performance. From in-game settings, overclocking/tuning, and software features ie: upscaling technologies, we explore how these factors can significantly impact benchmark results. We also discuss the importance of understanding the context in which benchmarks are conducted. For instance, are the components being tested in their out-of-the-box state, or have they been pushed to their limits through overclocking? We highlight why it’s crucial for consumers, PC gamers, and hardware enthusiasts to approach PC hardware reviews with caution. It’s in your best interest as a consumer to attain as much information as possible from multiple sources to find what works for your specific situation.

***** Check out Epomaker and theTH80 Pro Here: *****

Check out my 13900K Hyperthreading 40 Game Benchmark Vid

Watch my 13700K E Core ON vs OFF 40 Game Benchmark video

Watch my RTX 4090 vs 3090 42 Game Benchmark video

Watch my RTX 4090 at 2GHz video

Support the channel by donating

Want to build a Gaming PC? Use my Amazon Affiliate link below!

00:00 Intro
00:27 Sponsor
01:45 Why this matters
04:31 Test methodologies
07:19 GPU testing
08:48 software features enable/disable
13:28 conclusion

#PCGaming #benchmarks #nvidia #GPUs

[DannyzReviews]

I want to be clear I'm not undermining anyone else's work. I wanted to put this video out there because I often see far too many times someone will watch or read 1 review which will show product X is faster than product Y and declare that as a definitive WIN. But looking around at other reviews with different testing methodologies, will have a different conclusion.

[oscar]

I love comparing my tuned/OC'd parts to the stock benchmarks and there's so much free performance left on the table.

Super happy with my 14700K at 5.7P/5.0R/4.4E, water cooled 3080 12G and tuned DDR5 7000C32

[michaelthompson]

I understand that some people (majority) rely on the ‘out of the box’ settings from reviewers as they know the minimum performance they can expect out of their hardware by the manufacturer. I also understand that some people (minority) want to tap into the best performance out of their system they can get and will tweak their settings in bios/software accordingly. 
However, ‘out of the box’ settings tell you what to expect regardless of the situation ….. whereas tweaking your pc is dependant upon your knowledge, how far you’re willing to push your pc hardware and the silicon quality has soooo many variables affecting what performance it can potentially provide.
I personally tweak some things on the cpu and GPU front minimally for quick and easy performance gain and not put in the hours / days to gain the extra few percentage points of performance and hyper tweak my pc to within an inch of its life. Having a baseline of the product lets you know what you’re minimally going to achieve and imo is a valid thing.

[nossy]

The biggest change in performance for me was turning of Core Isolation in Win 11. Sure my system is now a bit less secure, but it does make a difference in gaming performance.

[HardwareAccent]

Yet another good video. Thanks for sharing. I love this kind of videos with personal take on subjects.

[gstormcz]

I see many benchmarks of cpus with 4090 in test setup, but I would like to see how different would those tested cpus place when with top Radeon (7900xtx).
Either there would be no different results, or the reason is just because 4090 is simply currently fastest gpu and also Nvidia has still bigger market share.
IMHO its also kind of Nvidia marketing..

The most interesting thing would be AI calculated benchmark to tell you expected performance for certain setup, game and settings.

As I play only pair of online games ( higher specs World of warships) and none of them is included on benchmarks, I have to guess only the results depending on old custom pc build tests, benchmarked games running on same engine and finally guess how it would run with rest of my hardware.

Fun fact:
I tried old Quake 1 (gl Quake) on my i3-10100f 6650xt, it is 20+y after game release, 4K "ultra" 600or 800 fps, 1280x720 or even FHD 2500fps. Well, you don't need 4090, you need "optimized game"
Who cares HD textures and raytracing when you get frag?

[Pillokun]

If a tech tuber runs an automated bench run they are doing it wrong, u should always run it yourself and more than 30sec. For instance when it comes to mp games u will have to run several runs, to get to know how low the fps can go during and how the system perform on avg in a normal session.

When u run at high fps either because of low settings at low res and/or with high end cpu/ram with tweaked settings u will always experience more uneven fps perf so to speak. for instance if you look at the sky or similar(ie a scene in the game where the game system is not gpu bound) and then when u look down or the scene becomes more gpu bound because of stuff happening on the screen the fps can be very high and feel like a stutter.

Anyway remember that gn uses 12700k with 6000xmp sticks for the gpu benchmarks, ie they limit the perf of the high end nvidia gpus compared to amd ones.

Hardware unboxed are better as when they do cpu comparisons they use 7200mt/s but yeah.. they are also not that great when they do more indepth testing like ddr4 vs ddr5 or when they tested the ipc of 12, 13 and 13 gen as they gimped the cpus with 3ghz ringbus instead of letting it go to at least 4.5ghz which all lga1700 "k-skus" can hit with ease.

[jettangeles]

i just discovered your channel & you're so right about people not wanting to get into game settings anymore & optimizing it themselves. also, you should start game optimizing videos like hardware unboxed. i think that'll be great!

[jjlw]

The main difference I see is the areas in games where TechTubers benchmark. Some use built in benchmarks while most use custom runs. Providing the locations of the benchmarks would help immensely for those looking to compare.

[curie]

i undervolt mine, whats the lowest power consumption while being almost the same performance is the best for me

[slimal]

Do reviewers not include details of their hardware status and test settings?
Or is this video for folks who are new to PC gaming?
Please enlighten me, because I feel I'm missing the point of this video.

[astreakaito]

I dislike how tech influencers really skew expectation and makes consumers chase for the wrong tree or believe straight up bs. I particularly dislike TV/monitor reviewers and how incredibly stupid their pixel response time race is when it really doesn't matter that much when it comes to motion performance at all .vs other variables, and this affected the monitor industry for the worse, to the point where TVs are now by far objectively the best displays for gaming and not the "gaming" monitors. Ppl are wasting thousands on literal plastic garbage, it hurts.

[Itsyoungsaturn]

Should I upgrade my 4070 ti to to a 4090 with a 5900x? Or should I wait, and upgrade my mother board and cpu to a ryzen 7000 series and prepare for the 5000 series nvida gpus ?

[Greez]

It's funny how DDR4 beyond 3600 C16/C18 was basically never in a techtube benchmark. Yet now, HUB is benchmarking with tuned 7200 DDR5. Which I think is a nice change... but imagine if normies saw what some 3600C14 on Zen, and at least 3800C14 on intel could do.. there'd be a bit more nuance to simply moving to DDR5. It's a nightmare to get stable, so it really pays to be an X3D chip. AMD's saving grace....

[JamesSmith-swnk]

This is why people buy PS5's & Xbox's. No tuning required! lol!

[eggsdooley]

Can you come over to my place and make my Microsoft Flight Simulator VR settings work? kthxbye

[thisisashan]

I think a lot should be said that some benchmarks were or are, sponsored in one way or another which does indeed cater favortism.
Apple. They lie about everything.

[alexarose]

I got scammed listening to that 6 cores is enough for gaming bs for that one guy now in 12 cores lol

[impuls]

HUB benchmarked at 3Ghz ring and I tried running Pubg with that, it sucked. It gave the feeling of playing against 100ms players. Put it back to 4.5Ghz and it was playing on a 40ms ping server again(ofc ping stayed the same). Arcitecture internal delays is way outside what even HUB guys tests and they even need Buildzoid to get 7200Mhz tuned ram setup to work. HUB is making so many wierd desicions now just to stay relewant with their huge outdated benchmark routine. Atleast they are transparant with it but its not enough when 90% of their audience dont understand what ring speed even is. They just go around saying the Intel processors suck when infact they still can be better. To top it all off all the benchmarks of the new 14th gen had no overclocking segments even though its an K part. They just dont have the skill to tune all the voltages involves now to get it working in time for when a review have to be published.

[NatrajChaturvedi]

My problem is how they have given the impression to the unitiated or the unthinking that the FPS counter is everything and that you should strive to squeeze every last drop of FPS in every game. I at least am not the kind of guy that can feel a significant difference between something like 100 - 120 fps and 160 - 240 fps. If its a slower singleplayer game or especially a top down game, I don't care about running that game at 200 fps! 60 fps is just fine.