EXTREMES in testing hardware, normal budget combinations never tested

[KedsDead]

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

Why are sensible budget combinations never tested? As a customer you simply cannot find information anywhere about how these combinations perform.

For example graphic cards with budget friendly Intel 14500 CPU. (What is the AMD price equivalent to this CPU?)

Or CPUs with RX 7800 XT or RTX 4070 (super?) ?

So to answer this.

CPU, the 14500 is a little slower... So these numbers are pushing a 4090.

4070 S to 7900 GRE Better price options than a 7800 XT.. But it is also included.

So, since the above CPUs can push a 4090 at 1080p to over 180fps average, Your CPU should have no issue pushing any of these three GPU's to 100% and getting max performance from them.

Edit: Change CPU range to include 12400f.

 

[LabRat 891]

Bottlenecking is overrated anyway.

Indeed.

Even in extreme bottlenecks, there's still a performance uplift.

I scored 3 684 in Fire Strike Ultra

AMD Opteron Processor 880, AMD Radeon Pro WX 9100 x 1, 12288 MB, 64-bit Windows 7}

www.3dmark.com

 

[BoggledBeagle]

2. You look at CPU reviews. You see that the majority of them do absolutely fine with the 4090. Whether you have 280 instead of 300 FPS is inconsequential. With that info, you have your CPU choice.

No, I do not have the CPU choice, speak for yourself. I am not a wizard. When I see how the CPU performs with the most powerful GPU, I have no idea how will they perform with a weaker GPU. I need to see the test with a weaker GPU to see, what CPU still improves the performance and with what CPU the performance does not improve anymore.

See the picture:



From the measured performance with 4090 you learn, that you need A or B CPU, and when you realise, that A is 40% more expensive than B, you will buy B.

From the blue bars you cannot predict orange bars, you need to measure them as well. Only after you measured the performance with 4070, you will learn, that you will be ok with C, unless you need the performance of A or B for something else.

 

[AusWolf]

No, I do not have the CPU choice, speak for yourself. I am not a wizard. When I see how the CPU performs with the most powerful GPU, I have no idea how will they perform with a weaker GPU. I need to see the test with a weaker GPU to see, what CPU still improves the performance and with what CPU the performance does not improve anymore.

No, you don't, because that information is irrelevant. 90% of your gaming experience comes from your GPU, therefore a GPU upgrade will always give you the best results, unless you're on a 2700K or something.

From the blue bars you cannot predict orange bars, you need to measure them as well. Only after you measured the performance with 4070, you will learn, that you will be ok with C, unless you need the performance of A or B for something else.

You don't even need to predict. Once you know the performance of the 4070, and you know that it's way below that of the 4090, you're not gonna fork out the extra cash for CPU A or B, will you?

Edit: On the other hand, you're not gonna look at a 4070 if you can only afford CPU G from your chart.

Edit 2: You don't know where the orange bars are, but you know where this orange line is from the 4070 reviews, which is more than enough info (please excuse my MS Paint skills):

 

[BoggledBeagle]

You don't even need to predict. Once you know the performance of the 4070, and you know that it's way below that of the 4090, you're not gonna fork out the extra cash for CPU A or B, will you? On the other hand, you're not gonna look at a 4070 if you can only afford CPU G from your chart.

If I accept what you wrote, you still have no clue which one of E, D and C to get.

Edit 2: You don't know where the orange bars are, but you know where this orange line is from the 4070 reviews, which is more than enough info (please excuse my MS Paint skills):
View attachment 340838

What you have, if you do not test CPUs with 4070:

• performance of the CPUs with 4090 (from the CPU reviews)
• performance of the 4070 with CPU A (from the review of 4070)
• knowledge that 4070 will not perform better than 4090 with the same CPU
• 4070 with worse CPU will not perform better than 4070 with A.

You have just one performance point of the 4070 clear, other than that you know only where the orange bars cannot end, I marked these areas red. They can end anywhere below these impossible areas.



Unless you really test the CPUs with 4070, you do not know what CPU to get.

 

[joemama]

Are you some perfectionist?
These reviews only need to provide you an estimation of performance. Reviews can never give you an accurate result of what performance you will get.
Even if they have every CPU/GPU combination tested, you won't get the same results as theirs, because the RAM is different, or the mobo is different, or even the PSU, the cooler, the environment temperature, the software, the settings and everything else.

 

[BoggledBeagle]

Are you some perfectionist?
These reviews only need to provide you an estimation of performance. Reviews can never give you an accurate result of what performance you will get.

Why not? Reviewers sometimes perform multiple runs to get a reliable result, they retest with new drivers, they clearly state what are their RAM settings, etc, I think their numbers can be pretty accurate and reproducible.

 

[AusWolf]

If I accept what you wrote, you still have no clue which one of E, D and C to get.

Get the best one within your budget. Easy.

What you have, if you do not test CPUs with 4070:

• performance of the CPUs with 4090 (from the CPU reviews)
• performance of the 4070 with CPU A (from the review of 4070)
• knowledge that 4070 will not perform better than 4090 with the same CPU
• 4070 with worse CPU will not perform better than 4070 with A.

You have just one performance point of the 4070 clear, other than that you know only where the orange bars cannot end, I marked these areas red. They can end anywhere below these impossible areas.

View attachment 340888

Unless you really test the CPUs with 4070, you do not know what CPU to get.

If the performance of CPU X with the 4090 is lower than the performance of the 4070 with CPU A, then get something better than CPU X.

Your red bar does not exist, as the performance of the 4070 is already maxed out with CPU A. How much performance you're leaving on the table with the 4070 is irrelevant if you know that even the 4090 is slower with your chosen CPU than the 4070 with a much better one.

 

[BoggledBeagle]

Your red bar does not exist

??? Red bar means the only certain information you have: that the orange bar will not overlap it.

 

[Chrispy_]

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" ..

The whiners who don't understand bottlenecks should be ignored.

If you read a GPU comparison review, you want all GPUs to be tested fairly, ie at their best. If *your* CPU is weak, then you have to go and look at a CPU comparison review to work out how much (if any) your CPU is hindering the GPU.

Asking people to check both types of review provides them with meaningful information for any combination of CPU and GPU.
Trying to appease people with a specific combination of CPU and GPU is a fool's errand and for every one person you please you will be wrong for ten more.

Like any scientific method, attempt to isolate all variables except the one you're testing, that means CPUs need to be tested with the best GPU available and vice-versa. Attempting to appease specific people for anything else is a huge waste of time for almost everyone else who doesn't have that exact combination of CPU+GPU, and there are plenty of unscientific YouTubers demonstrating various pairings if you're too lazy to read a CPU and GPU review of the particular pairing you're interested in.

 

[BoggledBeagle]

If you read a GPU comparison review, you want all GPUs to be tested fairly, ie at their best. If *your* CPU is weak, then you have to go and look at a CPU comparison review to work out how much (if any) your CPU is hindering the GPU.

You will find out, how is the CPU hindering ONE GPU (most likely 4090). No information about other cards is available, except their performance with the test CPU (7800X3D or 14900K). The only saving grace in this messy situation is that 7800X3D is pretty affordable.

Asking people to check both types of review provides them with meaningful information for any combination of CPU and GPU.

This is simply not true.

 

[Chrispy_]

This is simply not true.

It's not true if that person is too dumb to understand what they're looking at, correct. Otherwise, no, you are wrong.

Anyone with an ounce of understanding will be able to draw meaningful data for any hardware combination. I didn't say exact data, I said meaningful data.

The "exact" data is even narrower than it first seems. Say a reviewer tests a bunch of games on a 12700K and a 4070S.

Which 4070? Factory OC or FE at stock values?
Which RAM, what timings, what gear? Tuned or XMP defaults?
Which motherboard BIOS, What PL1/PL2 limits? What cooling is available? What cooling profiles? Open test-bench or case with warm air in it?

All of these things can impact the results a huge amount or none at all, depending on the games, application, ambient room temperature, and what you had for lunch yesterday....

 

[BoggledBeagle]

Anyone with an ounce of understanding will be able to draw meaningful data for any hardware combination. I didn't say exact data, I said meaningful data.

How? You cannot make things up.

How I illustrated above, you have only one performance point for a midrange GPU with one CPU and then you know where the performance definitely cannot be. And where it really is? Reviews do not tell.

 

[chrcoluk]

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

Why are sensible budget combinations never tested? As a customer you simply cannot find information anywhere about how these combinations perform.

For example graphic cards with budget friendly Intel 14500 CPU. (What is the AMD price equivalent to this CPU?)

Or CPUs with RX 7800 XT or RTX 4070 (super?) ?

I think you going to get a "there isnt enough time", the review industry will say the publish date is a prime factor as eyeballs drop off sharply after that date, and once that date is passed the motivation for further testing drops off.

Might also say "we use the best hardware to try and avoid bottle necking the product we reviewing".

I would love what you ask for, as seeing how the hardware behaves when bottle necked, when used in different combinations is what I expect many will want to see. It needs one major reviewer to decide to be "different" and to test things that the others are not testing, but I think thats seen as a risky strategy. Hopefully TPU maybe experiments with this idea, and if proves popular then they can do it repeatedly.

It would also need more than one test bed, as they not going to keep rebuilding PCs for it, so maybe keep older gen test bed's intact, and have something 2-3 test bed's per generation.

I expect review guides are also likely a factor. Most vendors want their products reviewed in a specific way.

My suggestion would be to stop running a full test suite on all partner cards (they all going to be very close to each other, its the same chip with a different brand on the shroud, and different cooling), just run some basic tests to determine cooling performance and performance differential of bios, and call it a day on those after market cards, and only run full suite on the main card. The time saved can then be used to test with different CPU/RAM combinations.

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" ..

Thats sad, I would appreciate it if you resumed it, and I dont think I would be the only one, I would even be promoting the reviews on every community I post on.

 

[KedsDead]

There are plenty of Youtube Reviewers doing what you are asking. I did all kinds of research, watching vids, of different GPU combinations with the CPU I wanted to get. Then also checked the others with the CPU a step up and down. Sure it took me a little longer than reading one of @W1zzard posts. All the information is there, you just have to look for it.

Wiz, I don't think you need to cater to the 1% of people who refuse to do their own research.

 

[AusWolf]

??? Red bar means the only certain information you have: that the orange bar will not overlap it.

Look at the orange line in my post above. It's more than enough information.

How? You cannot make things up.

How I illustrated above, you have only one performance point for a midrange GPU with one CPU and then you know where the performance definitely cannot be. And where it really is? Reviews do not tell.

If that one performance point is higher than that of a CPU with a much more powerful GPU, then that CPU isn't good enough for your needs. I don't know why you need more performance points to deduct this simple piece of information.

 

[Toothless]

I really don't think it's that hard to do research. Just look at reviews of "my CPU can handle up to this FPS with this GPU" and vice versa. Is Google search really that hard now or are we talking to a brick?