gpu heirarchy/performance/benchmarks- whos lying?

inquisitor1 · Saturday at 11:46 AM

I do like looking at toms heirarchy page to get a sense of performance but then websites and videos I see show one car run far beter then ranked

for instance saw a used 1080ti in mint condition for a good price. so comparing to a 2070 I also saw. I dont use that trash site many hate but technical city /gpu monkey or yt videos shows the 1080ti on top. so whos not telling the truth? I did notice the rankings not be accurate

this is for a backup pc to replace a 1070 which will replace an annoying rx590

does tpu have A good heirarchy list with performance levels?

Frick · Saturday at 12:01 PM

So someone ranked the 2070 above the 1080ti? Do note that some sites take features into account, like RT amd DLSS. As usual reviews are the best, and if the cards are close enough in time they're usually in the same tables.

inquisitor1 · Saturday at 12:15 PM

Frick said:
So someone ranked the 2070 above the 1080ti?

toms site shows % of performance so thats what its based on in ranking. not "other" features. dont assume either. im talking about toms specifically. gpu monkey and tecnhicalcity dont rank "features". only performance

I pulled back from toms years ago. now the heirarchy list is seperated which makes his site worthless. I only prefer all gpu in one list comparison then for each gpu as that page is for the 2070 and I prefer a full list heirarchy. ive not seen one done well enough for me

DaemonForce · Saturday at 12:23 PM

Frick said:
So someone ranked the 2070 above the 1080ti? Do note that some sites take features into account, like RT amd DLSS. As usual reviews are the best, and if the cards are close enough in time they're usually in the same tables.

This. Feature tables exist for a reason. When making your own comparison, do the same.
Benchmark sites will value a general/total mix of features and tie them in against performance.
You are going to do something completely different, especially types that bought in to the RX590.
Maybe you're a desktop gamer and value raw raster. Maybe you're a dev that needs high core clock, memory, etc.
Maybe you're just picking up a general purpose replacement that can do specialty work like high FPS video encode.
The benchmark sites don't know which camp you're in and neither do we. Gotta decide on what you value up front.

inquisitor1 · Saturday at 12:38 PM

ok, note to self, toms heirarchy chart is shet and not relevant to me. especially now with seperating the gpu list

Bomby569 · Saturday at 12:50 PM

those things are just a average, not all games will follow that hierarchy, it depends on a lot of variables: some cards perform better in 1080p or 1440p, drivers also play a large role in this, within the same gpu there is also variations in the AIB models, etc...
It's just to have an idea, not to take as an absolute truth.

inquisitor1 · Saturday at 1:06 PM

Bomby569 said:
not to take as an absolute truth.

its not even close 1080ti destroys 2070 in every test yet they show the 2070 as better..margibally

The GPU benchmarks hierarchy 2025: Ten years of graphics card hardware tested and ranked

We've run hundreds of GPU benchmarks on Nvidia, AMD, and Intel graphics cards and ranked them in our comprehensive hierarchy, with over 80 GPUs tested.

www.tomshardware.com

now I remember why I dont frequent the site. its just I really hate graphs with lines instead of just stats like tpu show and its missing gpu in the lists. like that linked above so now I goto gpu monkey initially then yt videos.

many innacuracies with their list I have caught. they show x gpu performing higher then y when everywere shown Y destroys X significantly, like the 2 cards I mention. 1080ti destroys the 2070 in all websites I see and all yt videos

reminds me why psu calculators from newegg and mfr are lying ,worthless and innaccurate as they want to upsell u. I found 2-3 accurate ones

Bomby569 · Saturday at 1:47 PM

inquisitor1 said:
its not even close 1080ti destroys 2070 in every test yet they show the 2070 as better..margibally

The GPU benchmarks hierarchy 2025: Ten years of graphics card hardware tested and ranked

We've run hundreds of GPU benchmarks on Nvidia, AMD, and Intel graphics cards and ranked them in our comprehensive hierarchy, with over 80 GPUs tested.

www.tomshardware.com

now I remember why I dont frequent the site. its just I really hate graphs with lines instead of just stats like tpu show and its missing gpu in the lists. like that linked above so now I goto gpu monkey initially then yt videos.

many innacuracies with their list I have caught. they show x gpu performing higher then y when everywere shown Y destroys X significantly, like the 2 cards I mention. 1080ti destroys the 2070 in all websites I see and all yt videos

reminds me why psu calculators from newegg and mfr are lying ,worthless and innaccurate as they want to upsell u. I found 2-3 accurate ones

from here it's about 7% faster at 1080p ultra to be a fair comparison, and it's not a destruction, it's at best a marginal difference.

NVIDIA GeForce RTX 2070 Founders Edition Review

GeForce RTX 2070 is NVIDIA's most affordable RTX-capable graphics card today. We ran it through 23 games at three resolutions, and the performance results are encouraging. The card sits right between the GTX 1080 and GTX 1080 Ti, with pricing lower than for the GTX 1080 Ti, and you get all the...

www.techpowerup.com

I see that chart has them kind of on par

maybe the issue is doing those nvidia style charts that are very hard to make and to read. But they do have better charts below, so maybe ignore those.

inquisitor1 · Saturday at 1:58 PM

Bomby569 said:
from here it's about 7% faster at 1080p ultra to be a fair comparison, and it's not a destruction, it's at best a marginal difference.

NVIDIA GeForce RTX 2070 Founders Edition Review

GeForce RTX 2070 is NVIDIA's most affordable RTX-capable graphics card today. We ran it through 23 games at three resolutions, and the performance results are encouraging. The card sits right between the GTX 1080 and GTX 1080 Ti, with pricing lower than for the GTX 1080 Ti, and you get all the...

www.techpowerup.com

I see that chart has them kind of on par

maybe the issue is doing those nvidia style charts that are very hard to make and to read. But they do have better charts below, so maybe ignore those.

sorry but yt videos show fps far more the 7% . no disrespect to you but well disagree and not relevant. my point is toms site lying but now I can confirm it

tpus charts is lacking gpus in the graphs so not relevant either. toms percentage list is easier to go back and forth quickly to compare. go from 980ti to2080ti to 4080 as compaison. with all due respect to tpu. too much clicking for fast comparisons but toms site is innacurate and is now out. gpumonkey technicalcity and yt videos now. I wish tpu would have a list like tom did but just % is enough

Carillon · Saturday at 2:14 PM

everybody benches different games at different settings in different parts of the game. But the biggest differentiation is if ray tracing is part of the average, if the 1080ti runs it in software, the 2070 will be vastly superior, if the 1080ti does not run RT at all, the 2070 will be infinite times faster.

inquisitor1 · Saturday at 2:19 PM

Carillon said:
everybody benches different games at different settings in different parts of the game. But the biggest differentiation is if ray tracing is part of the average, if the 1080ti runs it in software, the 2070 will be vastly superior, if the 1080ti does not run RT at all, the 2070 will be infinite times faster.

cool, all yt videos show 1080ti win in 1440 in everthing so not relevant. and 11gb is a plus and I dont worry about rt or fake framing. theyve gaslit people with that scam

Endymio · Saturday at 2:36 PM

inquisitor1 said:
...now I remember why I dont frequent the site. its just I really hate graphs with lines instead of just stats....reminds me why psu calculators from newegg and mfr are lying ,worthless and innaccurate as they want to upsell u. I found 2-3 accurate ones

Sites like Newegg that sell components must provide for a higher safety margin in any calculators and other tools they provide, or potentially face liability issue. The extra $5 in profit they make off a larger PSU isn't a consideration.

As for the video card rankings, if you can't understand that 'performance' is a metric that varies wildly based on the medley of selected games, and the features enabled within those games, then you probably shouldn't read tech sites at all.

inquisitor1 · Saturday at 2:46 PM

Endymio said:
then you probably shouldn't read tech sites at all.

thanks mom

Endymio said:
Sites like Newegg that sell components must provide for a higher safety margin in any calculators and other tools they provide

haha cool story. just an upsell tactic. with gamers nexus video on newegg scamming, havent bought from them

DemonicRyzen666 · Saturday at 10:06 PM

Carillon said:
everybody benches different games at different settings in different parts of the game. But the biggest differentiation is if ray tracing is part of the average, if the 1080ti runs it in software, the 2070 will be vastly superior, if the 1080ti does not run RT at all, the 2070 will be infinite times faster.

The entire point of benchmark is to be able to compare to others using the same hardware. Running a gaming then calling it benchmark isn't a benchmark, it's just average game play.
That is why there is horrible amount of inconsistencies in benchmarking now among reviewers. Canned or built in benchmarks are the only thing that remove those variables other than quality settings.

Princess Garnet · 2025-04-06T00:44:05+0100

Different sources will use...

1. A different methodology

2. Different applications (read as, games and software)

...So results will vary. It's not (usually) that one is "lying", it's simply that using a different collective of software will naturally give a different average. That's how this stuff works.

Synthetics like Technical City, CPU-Z's "benchmark", and even Passmark can be thrown into the "ballpark at best, terrible at worst" group. I ignore those most of the time. Results that are arrived at using a collection of real world software (TPU, Tom's Hardware, Gamers Nexus, Hardware Unboxed, etc.) will be closer to reality, but you won't be able to compare those results from one source to another source and expect them to match. The reason is the variables/methodology used to arrive at the results will differ. Keep in mind that a larger sample suite will make it accurate, but will also hide the effect of edge cases, which do exist. This stuff is simply too much time to test (and retest for new BIOS, drivers, etc.) so it's not worth it.

As a user, generally, you want to be looking at multiple sources (ignoring the synthetics) and it will paint a picture of where things fall on average. Then be aware that edge cases will exist and any results done with a small sample of games/software may reflect that.

Visible Noise · 2025-04-06T02:42:21+0100

gpu heirarchy/performance/benchmarks- whos lying?

Nobody. Follow the money. Stop watching techtubers to begin with.

ir_cow · 2025-04-06T02:49:30+0100

TPU is telling the truth. Always.

Panther_Seraphin · 2025-04-06T02:50:20+0100

So we are aiming to replace a 1070 with a 1080ti "for a good price" on a backup PC

What sort of money are we talking about? What is available in more modern (think 3070/6700XT/7700XT)?

I dont see much of a point in "upgrading" a card from nearly 10 years ago with something release 6 months later.

Lew Zealand · 2025-04-06T03:18:13+0100

In general, the 1080 Ti's performance is:

as fast as a 2080 in DX11
as fast as a 2070 in DX12

Obviously there will be outliers but this is a decent rule of thumb. Of course once you factor in DLSS with Transformer Model, I might even take the 2060 Super over the 1080 Ti unless I really needed those extra 3GB VRAM. Which the 2070, Super, 2080, Super don't have either.

dgianstefani · 2025-04-06T03:29:31+0100

RTX feature set worth more than even a 20% performance difference, let alone 10%.

inquisitor1 · 2025-04-06T04:26:13+0100

Panther_Seraphin said:
So we are aiming to replace a 1070 with a 1080ti "for a good price" on a backup PC

What sort of money are we talking about? What is available in more modern (think 3070/6700XT/7700XT)?

I dont see much of a point in "upgrading" a card from nearly 10 years ago with something release 6 months later.

post isnt about if its ok to upgrade to that gpu and its rude to say that. you dont know what I need it for, if a person may not need anything more. your opinion isnt relevant on this card and what I should upgrade. keep it on OP

dgianstefani said:
RTX feature set worth more than even a 20% performance difference, let alone 10%.

thanks but have 0 interest in rt . those benches doesnt corrolate to any video against the 2070 and in fact 3060. in every single video 1080ti is better so who are you trying to convince?

LabRat 891 · 2025-04-06T04:31:15+0100

inquisitor1 said:
does tpu have A good heirarchy list with performance levels?

In the GPU Db, there's VERY rough and loose comparisons between all the cards. IIRC, that's mostly scraped from TPU's and TPU's Review Db's reviews, so it's largely old data, from systems built w/ CPUs, etc. of that era. If not, it's probably derived from raw compute specs.

"Take it with a grain of salt" is an understatement.
-Still, probably the best 'at a glance' source for basic performance expectations (relative to other cards listed)

ir_cow said:
TPU is telling the truth. Always.

TPU tries, very hard, to be as accurate as possible. The GPU Db and vBIOS Db are not infallible, however.

inquisitor1 said:
I do like looking at toms heirarchy page to get a sense of performance but then websites and videos I see show one car run far beter then ranked

for instance saw a used 1080ti in mint condition for a good price. so comparing to a 2070 I also saw. I dont use that trash site many hate but technical city /gpu monkey or yt videos shows the 1080ti on top. so whos not telling the truth? I did notice the rankings not be accurate

this is for a backup pc to replace a 1070 which will replace an annoying rx590

[*brushes off flamesuit, dons*]
Ask Grok, use "deepsearch" (for the first or most-detailed prompt), provide it as much detail as you're already aware of, and where you got that info.
From there, you should be able to ask it more-specific-to-your-needs/expectations queries about a given card's performance.

As-always with LLMs, ask/verify its sources, and work 'alongside' it, never treat it as authoritative or knowing more than you do.
If it starts quoting or using 'bad' sources, tell it to exclude them.

inquisitor1 · 2025-04-06T04:41:11+0100

Princess Garnet said:
Different sources will use...

1. A different methodology

2. Different applications (read as, games and software)

...So results will vary. It's not (usually) that one is "lying", it's simply that using a different collective of software will naturally give a different average. That's how this stuff works.

Synthetics like Technical City, CPU-Z's "benchmark", and even Passmark can be thrown into the "ballpark at best, terrible at worst" group. I ignore those most of the time. Results that are arrived at using a collection of real world software (TPU, Tom's Hardware, Gamers Nexus, Hardware Unboxed, etc.) will be closer to reality, but you won't be able to compare those results from one source to another source and expect them to match. The reason is the variables/methodology used to arrive at the results will differ. Keep in mind that a larger sample suite will make it accurate, but will also hide the effect of edge cases, which do exist. This stuff is simply too much time to test (and retest for new BIOS, drivers, etc.) so it's not worth it.

As a user, generally, you want to be looking at multiple sources (ignoring the synthetics) and it will paint a picture of where things fall on average. Then be aware that edge cases will exist and any results done with a small sample of games/software may reflect that.

Im going to stick by my point and that tom is full of shet as every site and video shows the 1080ti winning and even close to 3060 performance and I saw this with other cards they list is completely off and they are shown in a list where a card scoring % lower is higher on the list. tom is low quality to me and why i dont ever go there. visually the heirarchy list is better then anyones. its wide and not cramped to the left in a graph like tpu And it shows all the cards in one list. tpu u have to click endlessly to see specific scores maybe gn has one

Lew Zealand · 2025-04-06T04:47:01+0100

inquisitor1 said:
those benches doesnt corrolate to any video against the 2070 and in fact 3060. in every single video 1080ti is better so who are you trying to convince?

What games?
What locations?
What settings?
What videos?

Rankings change significantly as these are varied.

The pics from here?

The GPU benchmarks hierarchy 2025: Ten years of graphics card hardware tested and ranked

We've run hundreds of GPU benchmarks on Nvidia, AMD, and Intel graphics cards and ranked them in our comprehensive hierarchy, with over 80 GPUs tested.

www.tomshardware.com

Yep, Pascal is screwy there, especially at 1080p. 1080 is about the same as 2060 and 5600 XT, not the 1080 Ti LOL. 1080 Ti is similar to 5700 XT. 1440p Ultra is closer to reality but Pascal still sits too low.

inquisitor1 · 2025-04-06T04:54:10+0100

LabRat 891 said:
As-always with LLMs, ask/verify its sources, and work 'alongside' it, never treat it as authoritative or knowing more than you do.
If it starts quoting or using 'bad' sources, tell it to exclude them.

you got carried away. its not authrotative and I understand some may get different score but should be in the ballpark. all videos and all sites show the 1080ti on top. and I can see many other cards listed higher but they have a lower % score. but I already knew the answer of whos lying.

theres bias for sure with these things and upselling is the game so putting a newer card higher will help convince some its better. I worked in advertising for 2 years. small things make a larger perceived difference

Lew Zealand said:
What games?
What locations?
What settings?
What videos?

Rankings change significantly as these are varied.

This pic?:

https://cdn.mos.cms.futurecdn.net/qiWnVboCCfkk2JgVern39L-970-80.png.webp

Yep, Pascal is screwy there. 1080 is about the same as 2060 and 5600 XT, not the 1080 Ti LOL. 1080 Ti is similar to 5700 XT.

Calls into question everything else on the chart.

link looks dodgy, I didnt click

System Name	Black MC in Tokyo
Processor	Ryzen 5 7600
Motherboard	MSI X670E Gaming Plus Wifi
Cooling	Be Quiet! Pure Rock 2
Memory	2 x 16GB Corsair Vengeance @ 6000Mhz
Video Card(s)	XFX 6950XT Speedster MERC 319
Storage	Kingston KC3000 1TB \| WD Black SN750 2TB \|WD Blue 1TB x 2 \| Toshiba P300 2TB \| Seagate Expansion 8TB
Display(s)	Samsung U32J590U 4K + BenQ GL2450HT 1080p
Case	Fractal Design Define R4
Audio Device(s)	Plantronics 5220, Nektar SE61 keyboard
Power Supply	Corsair RM850x v3
Mouse	Logitech G602
Keyboard	Dell SK3205
Software	Windows 10 Pro
Benchmark Scores	Rimworld 4K ready!

System Name	DevKit
Processor	AMD Ryzen 5 3600 ↗4.0GHz
Motherboard	Asus TUF Gaming X570-Plus WiFi
Cooling	Koolance CPU-300-H06, Koolance GPU-180-L06, SC800 Pump
Memory	4x16GB Ballistix 3200MT/s ↗3800
Video Card(s)	PowerColor RX 580 Red Devil 8GB ↗1380MHz ↘1105mV, PowerColor RX 7900 XT Hellhound 20GB
Storage	240GB Corsair MP510, 120GB KingDian S280
Display(s)	Nixeus VUE-24 (1080p144)
Case	Koolance PC2-601BLW + Koolance EHX1020CUV Radiator Kit
Audio Device(s)	Oculus CV-1
Power Supply	Antec Earthwatts EA-750 Semi-Modular
Mouse	Easterntimes Tech X-08, Zelotes C-12
Keyboard	Logitech 106-key, Romoral 15-Key Macro, Royal Kludge RK84
VR HMD	Oculus CV-1
Software	Windows 10 Pro Workstation, VMware Workstation 16 Pro, MS SQL Server 2016, Fan Control v120, Blender
Benchmark Scores	Cinebench R15: 1590cb Cinebench R20: 3530cb (7.83x451cb) CPU-Z 17.01.64: 481.2/3896.8 VRMark: 8009

Processor	Ryzen 5 5700x
Motherboard	B550 Elite
Cooling	Thermalright Perless Assassin 120 SE
Memory	32GB Fury Beast DDR4 3200Mhz
Video Card(s)	Gigabyte 3060 ti gaming oc pro
Storage	Samsung 970 Evo 1TB, WD SN850x 1TB, plus some random HDDs
Display(s)	LG 27gp850 1440p 165Hz 27''
Case	Lian Li Lancool II performance
Power Supply	MSI 750w
Mouse	G502

Processor	Ryzen 5 5700x
Motherboard	B550 Elite
Cooling	Thermalright Perless Assassin 120 SE
Memory	32GB Fury Beast DDR4 3200Mhz
Video Card(s)	Gigabyte 3060 ti gaming oc pro
Storage	Samsung 970 Evo 1TB, WD SN850x 1TB, plus some random HDDs
Display(s)	LG 27gp850 1440p 165Hz 27''
Case	Lian Li Lancool II performance
Power Supply	MSI 750w
Mouse	G502

System Name	S.L.I + RTX research rig
Processor	Ryzen 7 5800X 3D.
Motherboard	MSI MEG ACE X570
Cooling	Corsair H150i Cappellx
Memory	Corsair Vengeance pro RGB 3200mhz 32Gbs
Video Card(s)	2x Dell RTX 2080 Ti in S.L.I
Storage	Western digital Sata 6.0 SDD 500gb + fanxiang S660 4TB PCIe 4.0 NVMe M.2
Display(s)	HP X24i
Case	Corsair 7000D Airflow
Power Supply	EVGA G+1600watts
Mouse	Corsair Scimitar
Keyboard	Cosair K55 Pro RGB

gpu heirarchy/performance/benchmarks- whos lying?

inquisitor1

Frick

Fishfaced Nincompoop

inquisitor1

DaemonForce

inquisitor1

Bomby569

inquisitor1

The GPU benchmarks hierarchy 2025: Ten years of graphics card hardware tested and ranked

Bomby569

The GPU benchmarks hierarchy 2025: Ten years of graphics card hardware tested and ranked

NVIDIA GeForce RTX 2070 Founders Edition Review

inquisitor1

NVIDIA GeForce RTX 2070 Founders Edition Review

Carillon

inquisitor1

Endymio

inquisitor1

DemonicRyzen666

Princess Garnet

Visible Noise