EXTREMES in testing hardware, normal budget combinations never tested

[cvaldes]

Real world situation: let's say you read five reviews (as they are commonly written now). You play five games frequently and another 10-15 games less so. Some games perform better on Team Green, others better on Team Red. Team Blue's GPUs are rarely tested.

So how do you pick which GPU to buy? Maybe one company's GPU works way better when RT is turned on. There's also the notion of graphics quality output. So maybe one company's super-sampling resolution upscaling technology looks way better than the competition's? What is that worth to you? To the guy next door? To someone with way deeper pockets?

I don't think buyers are lazy. I just think that some people reading PC product reviews don't actually have a clear understanding of what is being measured, what's not being measured and how it ultimately comes down to the reader opening up their wallet anyhow.

And even crazier, there's nothing preventing someone from owning multiple computers (gasp, I know, I know) or multiple components.

In the end, every decision you make is some sort of compromise, not just PC component purchase decisions. There's no perfect solution for everyone all the time which is why people aren't just driving Henry Ford's Model T in 2024.

 

[LabRat 891]

I just checked for you. I await your first set of reviews.

View attachment 340621

You're giving me ideas...

What hosting service, etc. did ya use to look up domain availability?

 

[Super Firm Tofu]

You're giving me ideas...

What hosting service, etc. did ya use to look up domain availability?

That was from name.com

 

[Bill_Bright]

I don't think buyers are lazy.

Sadly, I think most are. Most buyers pick their choice without doing proper research.

Many pick their choice simply because of the packaging. If that were not true, there would not be entire sciences and marketing degrees based on the science (and phycology) of packaging and advertising designed to attract sales.

Many others pick their choice solely on price. Those either buy the cheapest to save on the budget, or they buy the most expensive thinking it costs more, therefore it must be better.

Oh, as to the team blue comment that their GPUs are rarely used - I do note that more and more users are gaming with integrated graphics, and many games are coded to provide good game play with lessor graphics solutions.

I will also note that contrary to the beliefs of many, in particular, many on this site, computers are not just about playing games - even among computer enthusiasts - many of whom may not even game.

 

[LabRat 891]

I've tested with realistic hardware for like a decade .. then people kept giving me shit .. "you're bottlenecking your CPU", "you are bottlenecking the GPU" ..

*both* is really needed, IMO. But, it's unrealistic to expect 'both perspectives' being tested by 1 reviewer.
Having a 'bottlenecked' and 'maxed out' comparison, can be very alluding towards how a GPU, etc. might perform in someone's oddball use-case.

SOMEBODY should do 'oddball combo/lopsided' reviews, yes.
It's just entirely impractical to have 'all in one' from a single reviewer-author.

 

[nomdeplume]

You don't test a Bugatti on city streets with an amateur driver and where there are red lights, pedestrians and traffic jams. Or on the freeway where there is a 70mph speed limit.

Actually, this is exactly how they test them. On public roads around the world.

Same with testing GPUs and CPUs.

Are you sure this time?

I've tested with realistic hardware for like a decade .. then people kept giving me shit .. "you're bottlenecking your CPU", "you are bottlenecking the GPU" ..

/thread

 

[iameatingjam]

GPUs are tested with the best CPU available.
CPUs are tested with the best GPU available.

Well that is the best way to do it... If you want to find performance on random component combinations you usually can just using google.

 

[cvaldes]

Sadly, I think most are. Most buyers pick their choice without doing proper research.

Many pick their choice simply because of the packaging. If that were not true, there would not be entire sciences and marketing degrees based on the science (and phycology) of packaging and advertising designed to attract sales.

Many others pick their choice solely on price. Those either buy the cheapest to save on the budget, or they buy the most expensive thinking it costs more, therefore it must be better.

Oh, as to the team blue comment that their GPUs are rarely used - I do note that more and more users are gaming with integrated graphics, and many games are coded to provide good game play with lessor graphics solutions.

I will also note that contrary to the beliefs of many, in particular, many on this site, computers are not just about playing games - even among computer enthusiasts - many of whom may not even game.

For sure many pick solely on price. Their primary priority is to save money which they presumably have better uses for elsewhere, a perfectly reasonable justification if you're not a billionaire.

And yes there are some people who pick on looks. 175W TDP graphics cards with triple fans and fat 3-slot heatsinks because "it looks badass".

Many online commenters (not just here at TPU) have a very warped sight of how technology hardware is used, by whom and how often. Some people hear "videogaming" and only imagine a Windows desktop PC with a discrete graphics card driving some sort of desktop monitor between 21" and 42".

Well, 130+ million Switch unit sales sort of escaped these people. Same with the millions of PS5 and Xbox Series X|S units. Let alone smartphones and tablets. That's right: Angry Birds made more money than Gollum or Redfall. And then they are shocked to hear that the lowly 2060 and 3060 GPUs are the ones most commonly reported in the regular Steam Hardware Surveys.

And yes they are even more shocked to hear that enterprise PC usage cases drive most design decisions. That's right, there are more people doing work on PCs than gaming. A secretary at a law office, the receptionist at your dentist office, some warehouse worker, an architect, the fine folks at the IRS processing your tax return, et cetera.

That's 250W TDP GPU? Well maybe you don't need to be burning up all those electrons if you're just running Excel, composing e-mails, running some database client, especially if you're in some government office with hundreds of other people doing the same thing.

Anyhow, this whole discussion thread has been most entertaining maybe not all of it intentionally.

 

[Bill_Bright]

Actually, this is exactly how they test them. On public roads around the world.

Okay, You are right. Perhaps that was a bad example/analogy on my part.

HOWEVER - is that the ONLY way they test the Bugatti? And is the results of that "real-world" scenario the ONLY criterial they use to (1) rate the Bugatti and (2) compare with competing cars?

No. Not just no, but hell no! - as you well know.

Are you sure this time?

Yes. If the goal is to determine the maximum capability/performance of the device, there MUST be NO outside limitations or restrictions (bottlenecks) placed on that device. It MUST be allowed to totally run free and not be held back by the limitations of any another device. So yes, I am absolutely sure.

That does NOT mean, however, that scenario is the only scenario to test, or compare a device in a review. As long as the exact same restrictions are placed on ALL the devices under review, then inserting such bottlenecks is fine.

 

[Steevo]

Why would I buy a 7900XTX or 4090 and pair it with a 9700K or 10 series, or a low power APU, or 3800X or old hardware that $100 or 150 more gets me a 7800X3D? I wouldn't

Why would I buy a system and not load it with RAM when its cheap? I wouldn't

Why would I buy a system and not understand that without everything being equal it will be as fast as the slowest component? I wouldn't


I understand if we are only talking about a sub $200 GPU, but those too with some thinking, their performance can be extrapolated.

NVIDIA GeForce RTX 3050 6 GB Review - The Fastest Slot-Powered GPU

Priced at $180, the GeForce RTX 3050 6 GB offers an accessible entry point into PC gaming. Operating solely on PCIe slot power, it's compatible with all PCs, including older OEM models, eliminating the need for power connectors. While noise levels are impressive, the price-to-performance ratio...

www.techpowerup.com

A 3050 that gets beaten by 1660Ti at 1080 which is the only playable resolution for said hardware. At that resolution the CPU does matter, but not that much as long as it isn't an antique. I had and used a 1100T CPU up till last November or so, and a old GPU with it.

How about this review

https://www.techpowerup.com/review/asrock-radeon-rx-7600-xt-steel-legend/ it shows the competition to that card with prices that were current when it was written.

Arc A750 is $200, pair it with a $ 110 CPU and it won't have any bottleneck with those cards

Intel Core i5-10400F Review - Six Cores with HT for Under $200

Intel's new Core i5-10400F offers a large performance jump over the previous generation Core i5-9400F because of its six-core/twelve-thread design. In this Core i5-10400F review we also test the feasibility of overclocking through BCLK, or by relaxing the PL1 and PL2 Turbo Limits.

www.techpowerup.com

If you can't look and compare numbers and come up with an idea of how things will work together you should ask questions at this forum, its free, everyone here is willing to help and a lot of the members will spend time to spec out a system given a budget, and what goals the user has.

 

[Gica]

1. 14700KF, all cores enable/28t, "7900X"mode preset (~104W in Cinebench R23)
2. 14700KF, 4 cores/8t, max 4.5GHz, PL 88W
RTX 3070 Ti

Graphics Score:
1. 14672
2. 14609
The conclusion is simple, the war for +/- 1 fps is fought by those who pay extravagantly for an X3D to turn an entry-middle-mainstream video card into an entry-middle-mainstream Batman.

 

[cvaldes]

Why would I buy a 7900XTX or 4090 and pair it with a 9700K or 10 series, or a low power APU, or 3800X or old hardware that $100 or 150 more gets me a 7800X3D? I wouldn't

Why would I buy a system and not load it with RAM when its cheap? I wouldn't

Why would I buy a system and not understand that without everything being equal it will be as fast as the slowest component? I wouldn't

Yes, let's start by testing RTX 4090 and RX 7900 XTX graphics cards with 15-year-old 500W power supplies! In poorly ventilated cases with no fans. With 1GB of RAM! Put the game a 5400 rpm spinner!

I can't wait to see the test results!

 

[LabRat 891]

Yes, let's start by testing RTX 4090 and RX 7900 XTX graphics cards with 15-year-old 500W power supplies! In poorly ventilated cases with no fans. With 1GB of RAM! Put the game a 5400 rpm spinner!

I can't wait to see the test results!

Are you reading my content notes (that I haven't written down)?
Heck, skip the PSU and run the +12V straight off a car battery!

Regardless, I believe that would qualify as: "Destructive Testing"


Also, yes. That kind of content does exist on YouTube, to some degree.

 

[Veseleil]

To give my answer to your initial question:
probably because between two extremes there are so many possible combinations it would demand dozens of staff members, huge stockpile of various hardware at hand, and probably few weeks or months to test anything properly. Not to mention need for precise and proven methodology to be followed step by step each time for any meaningful and valuable results. Just look at Gamer's Nexus and their never ending struggle with improving their methodology. It's a pain.
But I can understand your questions, the sentiment behind them, and why the majority of people are asking those all the time. This hobby is getting ridiculously expensive.

 

[nomdeplume]

Okay, You are right. Perhaps that was a bad example/analogy on my part.

HOWEVER - is that the ONLY way they test the Bugatti? And is the results of that "real-world" scenario the ONLY criterial they use to (1) rate the Bugatti and (2) compare with competing cars?

No. Not just no, but hell no! - as you well know.

You are missing the point. Testing is what happens before it goes to the track to make record attempts or gets pulled up in front of the cameras for publicity. Or for our purposes consider well prepared examples being overclocked and dispensed for review. Once delivered a product gets rated for the badge it wears or from the actual use of it.

The basis OP was formed on was actually testing how far expectations diverge in real life from these highly orchestrated test routines. I don't personally expect much given how suppressed and tortured current hardware presented to the market is. W1zz probably added all he felt necessary on that topic, opinions showed more meaningful variety and change across conditions.


While suggesting the reason he tested these combinations was because they had relevance to a larger crossection of the public than i9 K or top tier GPU paired with all else. Interactions can be more important to that audience than umpteen data charts of best case examples. It quite simply is not covered because it has no marketable value when the most expensive and often less suitable items are given a seamless media blitz. Yet he did cover it when the site was filled with enthusisasts and quit when it drew complaints.


I was jerking your chain just a bit so thanks for being a good sport.

 

[Steevo]

Yes, let's start by testing RTX 4090 and RX 7900 XTX graphics cards with 15-year-old 500W power supplies! In poorly ventilated cases with no fans. With 1GB of RAM! Put the game a 5400 rpm spinner!

I can't wait to see the test results!

God the number of computers still existing in closed desks and users wondering why it's slow as everything is thermally throttling....

Or laptops with dead fans, packed with lint, dust and pet hair....

I had a HP server a few years ago in a cabinet, 1 disk was dead, another dying on a RAID 5 array, the system/CPU fan was half dead and packed. WD blues don't like heat, it was the third set installed, a SSD, a couple cans of air, and closing the heat vent under the cabinet it was in and a few holes to let airflow and it lasted until I replaced the whole thing.

1. 14700KF, all cores enable/28t, "7900X"mode preset (~104W in Cinebench R23)
2. 14700KF, 4 cores/8t, max 4.5GHz, PL 88W
RTX 3070 Ti

Graphics Score:
1. 14672
2. 14609
The conclusion is simple, the war for +/- 1 fps is fought by those who pay extravagantly for an X3D to turn an entry-middle-mainstream video card into an entry-middle-mainstream Batman.
View attachment 340636View attachment 340637

The X3D CPUs use 1/3 the power as 14 series Intel and cost about the same, maybe I'm missing your meaning...

Intel Core i7-14700K Review - Catching the 13900K

Intel Core i7-14700K is the only processor released today with an upgraded core configuration. Thanks to an additional four E-Cores, more cache and a 200 MHz frequency bump, the 14700K reaches performance quite close to the 13900K, at much better pricing.

www.techpowerup.com

 

[Lew Zealand]

The X3D CPUs use 1/3 the power as 14 series Intel and cost about the same, maybe I'm missing your meaning...

Intel Core i7-14700K Review - Catching the 13900K

Intel Core i7-14700K is the only processor released today with an upgraded core configuration. Thanks to an additional four E-Cores, more cache and a 200 MHz frequency bump, the 14700K reaches performance quite close to the 13900K, at much better pricing.

www.techpowerup.com

He seems to be suggesting you should spend $400 on a K-series CPU, but then power limit it to pair with a $350 used GPU? I don't get it either.

 

[Solaris17]

maybe I'm missing your meaning...

You arent dont let them bait you.

 

[Steevo]

You arent dont let them bait you.

I remember being that troll, but way better at it. Now we just get goobers.

 

[BoggledBeagle]

To give my answer to your initial question:
probably because between two extremes there are so many possible combinations it would demand dozens of staff members, huge stockpile of various hardware at hand, and probably few weeks or months to test anything properly. Not to mention need for precise and proven methodology to be followed step by step each time for any meaningful and valuable results.

I do not think that the demand would be that high. I would start with analysis of the games - you really do not need to test dozens of games each time. You can analyze how the games perform, and then test just one of the group of the games that perform very close to each other.

You do not need to test every CPU available from each manufacturer, you may limit the number to 3 or 4 CPUS from both manufacturers to cover the price spectrum.

The same with GPUs. 3 or 4 from each manufacturer.

That is max 64 tests. Less than that, because few people will want to use the most expensive GPU with the cheapest CPU and vice versa.

Let us say 40 tests per game, with 5 games it makes 200 tests, which can be automated somehow I believe.

And the consumers would get a lot of useful information and would buy just what they really need.

 

[DaemonForce]

I prefer certain extremes because they're perfectly reasonable situations for those of us that spend literal YEARS on the same loadout between upgrades.
When I was still on a Phenom II X4 and switched AM3 boards, I suddenly needed a GPU because the DX10 IGP that was pushed so heavily was no longer relevant. Picked up the most awful 2GB HD6570 and that held me over well into 2018. No joke. So what was my next jump? RX 580 8GB. Finding benchmarks of my old configuration and then comparing similar ones to the new config was enough to prevent the hardware compass from perpetually spinning. Moved to an FX-8370, which was fine. Then made the move to a R5 3600 and that was good too. My next jump is the 7900 XT, which hopefully bottlenecks the snot out of this Ryzen. I can find similar video demo benchmarks all over YT and they "might" influence the purchasing decisions but play less of a part compared to the full list of numbers. This is fine.

 

[masterdeejay]

There is some driver overhead on nvidia cards. So it is possible the nvidia card more limited by weaker cpu than amd cards on the same level.

I think that overhead is happening to the intel cards with the very old games, where dx9 is somewhat emulated/translated in driver level. (but i didnt tested it, dont have any proof just my theory, still didnt matter if a dx9 title runs 200 fps instead of 400)

 

[agent_x007]

You do not need to test every CPU available from each manufacturer, you may limit the number to 3 or 4 CPUS from both manufacturers to cover the price spectrum.

That is max 64 tests. Less than that, because few people will want to use the most expensive GPU with the cheapest CPU and vice versa.

And the consumers would get a lot of useful information and would buy just what they really need.

First you want someone to test 14500, and then you say this... OK, I'll bite.

A 4 CPUs from one gen test :
1) 12100F, 12400F, 12600K, last for either 12700k or 12900k (can't have both).
I guess 12500 doesn't qualify in this gen, better luck next time.
2) 13400, 13600k, 13700k, 13900k (again, sorry about no 13500 this time either, there is always next time)
Current gen then ?
3) 14100, 14600k, 14700k, 14900k, nope - no place for 14500 in this lineup either.
You can force to test 14500, instead of 14600k (but that means no OC results for your hex P-core though) or 14100 (however that leaves low budget price point hangin', and that was done already on 13-th gen).

Not to mention, price difference between xx500 and xx600k/xx700F may, or may not be worth it for actual buyers (depending on use case).

If you truly cared about consumers, you would do this tests yourself instead of just highlighting your "care" (unless you want to keep trolling ?).

 

[BoggledBeagle]

I would test these Intel CPUs of the current gen (no reason to test older, some are just renamed older gen CPUs anyway): 14100 (4 cores, the cheapest CPU), 14500 (14 cores, better value than 14 core 14600K), 14700K (20C - price in the middle between 14500 and 14900K) and 14900K (24C - the best standard CPU Intel makes, 14900KS is overpriced limited edition).

AMD has now two generations running concurrently with AM4 CPUs being the budget option, I am slightly confused by that and I will not spend an hour learning their lineup, comparing prices and thinking about the best 4 CPUs to choose. Actualy AMD could be represented by 5 CPUs, because they have the special enlarged cache gaming model.

 

[maxfly]

@OP, this forum is a rich resource of information, literally at your fingertips. Use it. Asking specific questions here is far more likely to quickly garner you the answers your looking for vs expecting reviewers to upend their tried and true testing methodology.

If there is a CPU/GPU combo that your considering, I can almost guarantee you someone on the forum has some experience with them and can help you find the answers you seek.