NVIDIA GeForce RTX 4070 Ti Super GPU Review & Benchmarks: Power Efficiency & Gaming

Summary

This video introduces innovative GPU benchmarks with a focus on power efficiency, comparing FPS per watt across different GPUs, including the debut of the 4070 TI Super review. The inclusion of Cyberpunk 2077 RT benchmarks and the anticipation of non-RT Cyberpunk and Alan Wake 2 benchmarks highlight the evolving approach to GPU reviews. The 4070 TI Super is positioned among various Nvidia offerings, with discussions on its performance, price, and comparison to AMD's 7600 XT. The video also teases future content, including the 4080 Super review and continued exploration of GPU efficiency, promising a fresh perspective on GPU performance evaluation.

Takeaways

• 😊 The video reviews the new Nvidia GeForce RTX 4070 Ti Super GPU
• 👍 The 4070 Ti Super shows a ~10% performance uplift over the regular 4070 Ti at 4K resolution
• 🔬 New power efficiency testing methodology introduced - FPS per watt
• 📈 4070 Ti Super is more power efficient than regular 4070 Ti and AMD RX 7800 XT
• 🕹 4070 Ti Super gains increase in games with heavy ray tracing workloads
• 😕 The price remains unchanged from regular 4070 Ti at $799
• 🤔 Performance uplift underwhelming at 1440p and 1080p resolutions
• ⏩ Looking forward to testing GeForce RTX 4080 Super
• 👍 Positive initial results from new GPU testing methodologies
• 🛠 Continuously improving and iterating on GPU benchmarking and reviews

Q & A

What new testing methods were introduced in this GPU review?

This review introduced GPU power efficiency testing, measuring FPS per watt, and latency testing.

How does the 4070 Ti Super compare to the regular 4070 Ti?

The 4070 Ti Super provides around a 10-11% performance uplift at 4K over the regular 4070 Ti due to upgrades to the memory subsystem.

What are the key differences between the 4070 Ti Super and the 7900 XT?

The 4070 Ti Super has better ray tracing performance but the 7900 XT is faster in traditional rasterization. The 4070 Ti Super is more power efficient.

Which GPU was the most power efficient in the testing?

The RTX 4080 was the most power efficient GPU tested.

How did increasing CPU load impact GPU efficiency?

Increasing CPU load actually decreased GPU efficiency, as the GPU power draw remains high but the GPU is unable to fully utilize its capabilities.

How did resolution impact the gains from the 4070 Ti Super's memory upgrades?

The memory upgrades had more impact at higher resolutions like 4K where the memory subsystem is more stressed.

What game showed the biggest performance gains for the 4070 Ti Super?

Cyberpunk 2077 showed very large gains for the 4070 Ti Super over the regular 4070 Ti, up to 70% with ray tracing enabled.

How did the 4070 Ti Super compare to previous generation cards like the 2070?

The 4070 Ti Super provided massive gains over previous cards like the RTX 2070, with performance improvements of 150-200% in many tests.

Did the 4070 Ti Super fully eliminate stuttering compared to the regular 4070 Ti?

No, in some tests like Cyberpunk at 4K Ultra, the 4070 Ti Super had better minimum frame rates but still exhibited some stuttering.

What is the overall assessment of the 4070 Ti Super launch?

It's a modest improvement over the regular 4070 Ti, but pricing remains high. It seems like a middle ground between the more interesting 4080 Super and 4070 launches.

Outlines

00:00

📺 Overview and introduction to GPU testing methodology

Introduces new GPU testing methodologies including power efficiency testing (FPS per watt), adds back Cyberpunk 2077 raytracing benchmarks, and plans to add Alan Wake 2 testing in the future. Also provides an overview of the new RTX 4070 Ti Super launch and how it compares to other RTX 4070 variants.

05:01

💡 Explanation of FPS per watt testing methodology

Explains the FPS per watt testing methodology for measuring GPU power efficiency under gaming loads. Discusses pros/cons compared to Watts per FPS and some observations from test data so far.

10:02

📊 GPU benchmarks and relative performance summary

Summarizes overall performance uplift of 4070 Ti Super over 4070 Ti at different resolutions. At 4K sees ~10-11% faster on average, at 1440p ~7% faster, at 1080p ~5% faster on average.

15:04

🎮 Starfield: 4070 Ti Super 6-12% faster than 4070 Ti

At 4K, 4070 Ti Super is 6.3% faster than 4070 Ti in Starfield. Lead drops to 4.4% at 1440p as resolution decreases. 7900 XT still has advantage in this title.

20:06

🧟‍♂️ Resident Evil 4: Rasterization benchmarks

At 4K without raytracing, 4070 Ti Super is 10.6% faster than 4070 Ti in Resident Evil 4. 7800 XT is 12% faster than the Ti Super. Lead drops to 7.8% at 1440p and 5.1% at 1080p.

25:07

👾 More rasterization game benchmarks

Additional rasterization game benchmarks at multiple resolutions for Tomb Raider (12% lead over 4070 Ti at 4K), F1 2022 (11% lead at 4K), Rainbow Six Siege (10% lead at 4K). Lead decreases predictably at lower resolutions.

30:08

😈 Cyberpunk 2077: Raytracing benchmarks

Cyberpunk 2077 raytracing benchmarks with Ultra and Medium settings, no upscaling. At 4K Ultra, 4070 Ti Super leads 4070 Ti by 13-14%. Lead decreases to 6-8% at 1080p as memory matters less.

35:08

🧟 More raytracing benchmarks and closing thoughts

Additional raytracing benchmarks for Resident Evil 4 and Dying Light 2 showing 4070 Ti Super lead decreasing as resolution drops. Closing commentary that 4070 Ti Super launch is overall ‘meh’ and squeeze between more interesting launches.

Keywords

💡GPU

GPU stands for Graphics Processing Unit. It is a specialized electronic circuit that is very efficient at manipulating and altering memory to accelerate the creation of images in a frame buffer intended for output to a display device. The script discusses benchmarking different GPU models like the Nvidia RTX 4070 Ti Super and AMD RX 7900 XT to compare gaming performance and efficiency.

💡FPS

FPS stands for frames per second. It refers to the number of consecutive full frames generated by the GPU each second. The script looks at FPS performance of GPUs in games at different resolutions to evaluate gaming experience.

💡Ray Tracing

Ray tracing is an advanced graphics technique that traces the path of light in a 3D scene to simulate complex lighting interactions creating more realistic lighting, shadows and reflections. The script compares ray tracing performance of Nvidia and AMD GPUs in games like Cyberpunk 2077.

💡4K resolution

4K resolution refers to a display resolution of 3840 x 2160 pixels. The script looks at gaming FPS performance of GPUs at 4K resolution to test their capabilities for high-end gaming.

💡Frame rate scaling

Frame rate scaling refers to how the FPS performance of a GPU changes across different resolutions like 4K, 1440p and 1080p. The script examines FPS scaling to evaluate GPU performance as resolution decreases.

💡Upscaling

Upscaling techniques like AMD's FSR and Nvidia's DLSS use advanced reconstruction methods to boost frame rates and resolution quality. The script looks at titles using upscaling to test performance.

💡Memory bandwidth

Memory bandwidth refers to the rate at which data can be read from or stored to a GPU's memory by the rendering engine and cores. The 4070 Ti Super has upgraded memory bandwidth which improves 4K performance.

💡Power efficiency

Power efficiency evaluates performance vs power consumption. The script introduces new GPU testing for FPS per watt to compare efficiency across GPUs under real game loads.

💡Game benchmarks

Game benchmarks measure real game FPS performance across a suite of game titles new and old at different settings. The script looks at benchmarks for games like Cyberpunk, Resident Evil, FFXIV etc.

💡GPU generations

The script compares GPUs across generations, like the new RTX 4070 Ti vs older RTX 3070, to evaluate generational performance gains for potential upgraders.

Highlights

Introduction of GPU power efficiency testing, measuring FPS per watt, as a new benchmarking approach.

Addition of Cyberpunk RT and planning for non-RT Cyberpunk and Alan Wake 2 to the testing suite.

Discussion on the refresh of GPU reviews with new tests to match the evolution in GPU technology.

Overview of the 4070 TI super review, mentioning the lineup of 4070 cards and the introduction of AMD's 7600 XT.

Launch of a limited edition disappointment PC T-shirt to commemorate the most disappointing hardware events of 2023.

Highlighting the memory subsystem overhaul in the 4070 TI super, affecting performance in high memory load scenarios like 4K.

Introduction of FPS per watt testing for GPUs, expanding on the methodology previously used for CPUs.

Explanation of the importance of gaming workloads in power efficiency testing, acknowledging their complexity.

Findings from the power efficiency testing, indicating varying efficiencies among GPUs at different resolutions.

Inclusion of Resident Evil 4 and Cyberpunk in testing, with a focus on ray tracing performance.

Comparison of the 4070 TI super with other GPUs, revealing its placement in terms of power efficiency and performance.

Discussion on the potential impact of memory subsystem changes on performance in specific gaming scenarios.

Announcement of further testing innovations, including a 3D laser scanner for cooler testing.

Review of the 4070 TI super's performance across various games and resolutions, noting its comparative efficiency.

Conclusion on the 4070 TI super review, reflecting on the card's performance, pricing, and positioning within the current GPU market.

Transcripts

[Music]

we have a lot of cool benchmarks today

we're adding GPU power efficiency

testing so FPS per watt like we did for

CPUs except for gpus it's the debut of

this testing it's not anywhere near

final we have a ton more we can do with

it but we're excited to share that with

you in our 4070 TI super review which is

what this is we also added cyberpunk RT

back in so we have some R tracing

benchmarks with cyberp Punk 2077 we're

planning to add non RT cyberp punk in

the future as well and we're looking

into Alan W too for our next iterative

Edition because we're basically doing

kind of one step at a time let's add

some new types of tests to refresh the

GP reviews because that's what the

manufacturers are doing anyway they're

refreshing it with the Super Series and

amd's got the 7600 XT coming out

tomorrow I don't think I get to go home

tonight or tomorrow night but we'll get

them done let's start with the 4070 TI

super review so I don't have to face

reality which is is that I'm staying

here for a few days this video is

brought to you by us and storu Gamers

access.net for our limited edition

disappointment PCT t-shirt recapping all

the most disappointing Hardware events

of 2023 in a commemorative band style

tour date shirt the front features a

memory stick with an exploded diagram

revealing an accurate representation of

memory while the back features our

limited time tour dates with the best

hits such as snake oil the original 4070

TI launch and a Obsession we still have

every size available in our comfortable

charcoal Heather material but we're

running low fast and we won't be making

another run of these your purchases of

shirts like our charcoal Heather

disappointment shirt help directly fund

our new testing Innovations like GPU FPS

per wat power efficiency and our

recently introduced 3D laser scanner for

testing coolers so head over to store.

Gamers access.net to grab a shirt today

all right the overview so there's a 4070

super we already reviewed that it's on

the channel that's $600 it slots in

alongside the 470 which has dropped

about 50 bucks or so plus or minus a

little bit when this goes up uh then

there's the 470 TI super which is this

one and just as a reminder there's also

a 4070 TI and there's a

470 and then there's the 470 super we

just talked about so we're up to four

4070 cards getting a little confusing

andv video is helping out by

eliminating one of them to think about

the doesn't actually help that much

there's still a lot of 407s and then

there's the 4080 super that is the one

that people in our comments appeared to

be the most excited about or at least

discussing the most when it was all

announced in nvidia's news coverage

that's because it comes down to $11,000

from 1,200 for Baseline uh and that

should be in theory a pretty significant

shift for Value we'll see how it plays

out but it looks a little bit like

Nvidia has maybe couched the less

interesting of these cards between the

two that are potentially more appealing

for value so the 4070 TI super comes in

with a solid meh right in between the

two potentially more appealing cards but

uh there's a lot to talk about today so

a couple of things interesting findings

4K skews higher for the performance

difference between the ti super and the

nons super TI which is because of the

memory subsystem change so the 4070 TI

super we already talked about the specs

in the specs video but it's increased

the memory capacity and in step with

that there's also a memory bandwidth

change memory bus width increase so this

is the entire memory sub system has been

overhauled on the ti super uh and that's

going to reflect more in those higher

memory load scenarios like 4K okay Power

efficiencies first we're not really

going to spend a ton of time setting

this one up so for power efficiency

testing this is something we already

debuted with our CPU bench marks the

video was called Intel problem or

something CP power efficiency and that

looked at primar IL FPS per watt or some

other metric so like mips per watt for

example interestingly FPS per watt could

be simplified into effectively frames

per Jewel cancel the per second cancels

out for both uh and that's what we're

going be looking at now for sake of

using phrases that people are familiar

with and metrics people know watts and

FPS we're going to stick with FPS per

watt but that's kind of the idea the

purpose of FPS per watt testing or

efficiency testing is very simple

typically we'll test a full load power

virus scenario for gpus just like we

test a full load scenario with blender

for CPUs what that misses and what we've

been pretty open about it missing over

the years is gaming workloads which are

not anywhere near that simple it is

never simple with gaming the reasons

it's different are many we explain them

in that CPU piece but as a couple

examples uh where the bottleneck is can

change the efficiency so Shifting the

load more to the CPU to where the GPU

can't even fully stretch its legs that

doesn't necessarily increase the

efficiency and in fact can instead hurt

it we saw this with the CPU test too

where even when we were GPU bound the

CPU was still having to to burn at

fairly High wattage just to keep the GPU

supplied with frames to render now FPS

per watt with gpus produces mostly sub

1.0 numbers and that's just because the

power consumption is higher than

typically the frame rate is in terms of

pure numerical value for what we're

looking at today uh it doesn't

particularly mean or affect anything

it's just that you might ask well why

not watts per FPS as an example then you

get a nice positive number maybe it's

10.5 or something as opposed to

0.47 and the only real answer we have

for that is just consistency we're just

going with FPS per watt because we

already used it for the CPU one people

seem familiar with it they seem to like

it and the response from you all was

positive uh however if everyone would

prefer to see watts per FPS we could do

that too it's kind of all the same it

doesn't really matter at the end of the

day but also FPS per wat allows us to go

for a bigger number better and as we all

know thanks to Intel bigger number

better means better number gets more

bigger better so that's better because

it's bigger let's get started Starfield

at 4K starts us off for efficiency in

this one the most efficient card in the

test was the RTX 480 at 0.27 FPS per

watt or if you want to look at it

another way about 3.7 Watts per FPS or

in another way still 0.27 frames per

jeel there are a lot of options here the

4070 TI super is about equal with the

original 470 TI the latter of which

technically has a 2.1% Improvement in

efficiency here compared to the 7900 XT

the 4070 TI super is about 4% more

efficient in fps per watt at 1440p the

490 slips down the ranks a bit more as

it approaches a CPU bind this reveals

interesting data that shows that

although the CPU may start limiting

performance that doesn't necessarily

make the GPU more efficient it's similar

to the IOD Baseline power requirement

buffer that we saw for AMD CPUs in the

past efficiency test anyway the 4070 TI

super again allows the 4070 TI El lead

in efficiency This Time by about 4.1%

the 4070 TI super lead over the 7800 XT

is relatively large 15% here showing

that AMD is trading power for

performance at 1080p the 480 Remains the

most efficient while the 490 continues

to slip down the ranks its Baseline

power requirement to maintain operation

with its default settings as in the card

settings is reducing its efficiency as

it gets increasingly Tangled Up by other

components this is really interesting

data and it implies that the entire

system will in fact be less efficient as

CPU load increases as we saw in our

previous CPU efficiency test or at least

in the scenarios we've benchmarked so

far the 4070 TI super produced 0.57 FPS

per watt here that has the 4070 TI

original card 4.4% more efficient or the

4070 TI super as 25% more efficient than

the 7800 XT the 7800 XT is also becoming

limited causing a further Gulf so let's

unbind it in Resident Evil 4 at 4K

without RT we found the 4070 TI to be

the least efficient card with the 4070

TI super improving by 6% it could be the

case that the uplift in this title comes

from the memory subsystem change as

opposed to frequency the 7800 XT is

significantly more efficient here due to

overall higher performance than

previously allowing it to roughly tie

the R TX 480 the 7800 XT sits between

the ti super and the original TI at

1440p the 470 TI super still leads the

470 TI Now by about 2.7% so it's reduced

with the 7800 XT still relatively high

up on the chart the 470 super ends up

striking the strongest balance overall

on this one at 1080p the 480 and the

7900 XT still have some scaling Headroom

for frame rate the90s overall

comparative efficiency has fallen but it

still pulls 357 watts and it remains

It's relatively heavily loaded despite

being limited on total CPU scale the

4070 TI super is roughly tied with the

4070 TI at this resolution so it's lead

shrank with each resolution decrease the

7800 XT allows the 47 TI super a lead of

1.4% so they're about tied and the 470

super strikes the best balance so far

once again in Rainbow Six Siege at 4K

the RTX 480 Remains the most efficient

of the group followed by the 470 super

which is now establishing a trend and

then the 490 the 490 is slightly lower

on the efficiency curve for volt

frequency tuning to maximize performance

the 4070 TI super leads the 4070 TI by

8.3% in this title a result of the

average FPS increase being higher than

we've seen in some other games where

that 8.3% lead is for efficiency the 700

XT ends up between the two TI class

cards now we're going to quickly look at

some 10p and 30p point non-c consecutive

highs from the entire test run including

loading screens in this chart we're

looking at just Watts the 490 pulled 52

Watts for its average on the highest 10

points of data with 498 Watts for the

highest 30 points averaged this won't

capture the smallest transin which will

Spike higher still but that's something

we looked at in the past the 700 XT

pulled 385 Watts as it's 10o high with

the 470 TI super at 330 watts to the

TI's 321 watts and Starfield at 1440p

that drops to 382 Watts on the 700 XT

372 on the 490 which is becoming CPU

constrained and 26 6 6 on the 4070 TI

super the 480 shows why it was so power

efficient here so that's the foundation

of efficiency testing I am really

excited to see where this goes we put

this together just in the last week so

uh again there's a lot that can be

changed there's a lot we can improve

we're aware of it this is nowhere near

sort of like the perfect end game of how

we want to represent this but it's an

initial representation gives you some

ideas to play around with Post in the

comments and we'll keep tweaking it as

well and post a separate Standalone

piece as we continue to figure it out

but um we did this just in the last week

alongside adding cyberpunk rage tracing

and starting to explore Allen wake 2

testing so there's a lot going on for

our GPU reviews but of all of that we

think the GPU efficiency stuff is going

to be the most interesting and promising

anyway we'll start with some game

Benchmark summary charts and then we'll

get into the individual game benchmarks

here's a summary chart of the 4070 TI

super versus the 4070 TI nons super non

non-ti the bars indicate percent

Improvement by moving two the 4070 TI

super from the 4070 TI in general we're

seeing about a 10 to 11% uplift at 4K

our testing tends to be at highly

intensive settings to minimize CPU load

so especially when at 4K this makes

sense with the memory subsystem and

frame BFF for improvement two of the

older games here Horizon zero Dawn and

Total War Warhammer 3 posted gains

higher than we saw in other titles on

this chart at 1440p we saw about a 7%

average or so uplift in the 470 TI super

versus the 470 TI that's not

particularly exciting it reduces from

the 4K gains because the memory just

isn't as impactful at 1080p we typically

saw from 2% to 6% gains with a few

breakouts higher generally speaking it's

call it about 5% or so uplift at 1080p

now we have two identical 4070 TI Super

Cards the tough models and so we also

decided to run these through Port Royal

for frequency validation the brief test

established an average frequency 2746

MHz for the unit we tested for our

review with the more recently tested

unit that wasn't used in the review

today at 2725 MHz this is about a 20 MHz

or so Delta it's within manufacturing

variants but the Silicon quality on the

unit we chose at random appears to be a

little higher than the secondary model

and obviously the one that ran 20 MHz

higher smelled freshly of leather

jackets handpicked definitely a golden

sample that was SAR

ASM but someone was typing it it'll SN

it a little bit though but let's move on

to the 7900 XT versus the 4070 TI we

broke this into RT and non-rt charts

because the scale slides opposing

directions and we'd prefer to do as much

of that math in a single direction or

percent Improvement as possible that

just keeps it cleaner now that said when

you're talking about a percentage Point

change of

0.5 between going positive or negative

Direction the impact isn't that big

regardless but we are trying to keep it

all showing as much as possible in

percent Improvement regardless the RT

performance strongly benefits the 4070

TI super we noticed that cyberpunk

scales insanely in favor of Nvidia as

all graphic settings increase with

customized ultra settings without

upscaling pushing the ti super in the

range of 70 plus% for its lead over the

7900 XT with cyberpunk using medium

settings instead but still RT we saw the

gains have down to something like 35%

for the 4070 TI super over the 700 XD as

a workload for RT amplifies the favor

for NVIDIA likewise seems to

disproportionately amplify in less

intensive RT scenarios the lead is often

below 20% there's one 4K upscaled result

favoring the 7900 XT in Resident Evil 4

which progressively moves toward Nvidia

favor as the resolution decreases in

non-rt workload switching the chart to

show to the 7900 XT from the 4070 TI

super we see that 7900 XT starts pulling

ahead on average depending on resolution

it's anywhere from equal to 14% ahead

with losses toward the bottom of the

chart time to move into specific game

Benchmark charts we have a lot of cards

on here so depending on which chart you

look at goes back to the 2070 uh there's

1080ti in there 3060 variants are in

there uh 6600 variants are in there it

depends on the resolution but there's a

lot to look at so feel free to pause for

the specific card you might have and if

your specific card is not there remember

you can go and grab a a reference to a

card that is on our charts with a known

percent difference from that card and

you can start kind of calculating it

roughly on your own it's not perfect but

I'll get you pretty close to it okay

let's get into those charts Starfield is

up next another of our 2023 titles with

2024 patches to test at 4K the RTX 470

TI super landed at 62 FPS average with

lows proportional and well timed that

has it ahead of the 4070 TI by

relatively uneventful 6.3% the lead over

the 4070 super is 18% and compared to

some cards you might upgrade from it's

ordered as follows 79% over the 3070 and

161% over the 2070 the RX 7900 XT leads

the 470 TI super by 10% here so AMD is

still leveraging its advantage in this

particular game and has a wider lead

than it does in some others and just for

reference the 490 leads the 4070 TI

super by 47% here at 92 FPS average if

you were curious what that jump would do

at 1440p the 4070 TI super still has

full scaling in this Benchmark its lead

over the nons super variant 95 FPS

average is just 4.4% now down from the

already low 6% at 4K and the ti super

now only leads the 470 super by 14.7% so

it's even less exciting than before

let's move on Resident Evil 4 is another

2023 game that we've added recently and

this one tested at 4K and rasterized

first the 4070 TI super ran at 87 FPS

average and led the 4070 TI by 10. 6%

with the 7800 XT leading the ti super by

12% 480 is just ahead of that looking

down the stack the 6800 XT remains one

of the most powerful and viable options

at the $500 Mark that's assuming it's

still there when this goes up although

it's not the most viable in every game

on the Nvidia side the ti super leads

the new RTX 470 Sans TI super at 71 FPS

for that 4070 super by 23% over possible

upgrade Pathways the lead over the 3070

the 20 70 non super and the 2080 is

significant respectively at 86% 190% on

the 2070 and 139% on the 2080 scaling is

clean at 1440p lows are proportional so

since the 0.1% lows don't indicate

anything noteworthy here we've removed

them to instead accommodate more gpus on

this chart that's the trade-off the 470

TI super is 172 FPS average as at 7.8%

ahead of the 4070 TI reduced from the

10.6% result at 4K scaling Titans as

resolution fall for these two the 4070

TI super is still bed by the 7800 XT

here this time by 6.8% this chart added

a bunch of mid-range cards from ages

past so for some potential Pathways

there's the RX 6600 shown at 58 FPS

average the 6600 XT at 70 and the 360 at

57 FPS average with the 360 TI at 78

respectively these would SE Improvement

of 200% on the 6600 147% on the 66 XT

and then 2011% and 121% for the 2360

class cards at 1080p the 4070 TI super

is still scaling with the CPU it's at

253 FPS average leading the 470 TI by

5.1% predictably less than at 1440p and

4K the 7800 xt's lead also reduces here

down to 3.4% d light 2 is a 2022 game

that's up now tested at 4K first the

4070 TI super leads the 470 TI by 12.6%

better than we saw in the last game the

700 XT is about equal with the 4070 TI

super and can't be told apart at a

functional level owners of the 2070 nons

super would see massive gains under

these test conditions at about

178% a lot of that is from the Intensive

4K load that the 2070 is struggling with

the 2080 would likewise get multiples of

uplift alternatives to the 4070 TI super

might include the 7800 XT at equivalent

performance or the 470 super at reduced

performance but it still achieves 83% of

the performance of the ti super the 6800

XT would be another option here and is

one that has aged relatively well with

its price drops up next 1440p 0.1% lows

scaled perfectly fine with this there

were no abnormalities so we removed them

to permit space for more cards on the

chart this includes fresh retest of the

6600 the 6600 XT and the 6700 XT in this

test the 470 TI suers lead over the 470

TI is reduced to 8.4% with gains against

the 6600 XT and 6600 at 152% and 197%

respectively mentioned just in case

that's something you're considering

upgrading from Alternatives still

include the 700 XT ahead by 3.1% and the

480 more meaningfully ahead but we would

strongly advise against a 480 given the

480 sup is inbound status in basically a

week and lower price and the 470 super

is another alternative at 85% of the

performance of the 4070 TI super at

1080p the 4070 TI supers lead over the

the 4070 TI is now even lower at 5.8%

the 700 XT is tied so its lead has also

Fallen scaling is clean from a testing

standpoint just that this isn't

particularly interesting so let's move

forward now we're moving on to our set

of benchmarks that we keep around for

reliability and long-term validation

because the numbers rarely change for

these so it's easier to keep a huge

amount of cards on the charts in Tomb

Raider at 4K the 4070 TI super tough

leads the 4070 TI by 12% in this regard

it's roughly Tied by the 7 00 XT for

upgrade references the lead over the

2070 XC is 167% it's the same over the

6600 XT and the RTX 3060 XC at 1440p the

4070 TI super led the 470 TI by 85%

posting full scaling with the lead over

the new 4070 super at about

16.5% compared to older generation cards

you might upgrade from options include a

doubling over the RTX 360 TI and RTX

2080 a 171 1% boost over the 6600 non XT

at 82 FPS and similar over the original

RTX 360 in f-122 at 4K the RTX 470 TI

super ran at 137 FPS average leading the

470 TI by 11% that has the 7800 XT ahead

of the 4070 TI super by

8.7% and the 480 is just past that

compared to upgrade options the lead of

the 4070 TI super over the 2070 XC is

163% against the 52 FPS average result

it's about the same Boost from the 6600

XD and 360 XC Alternatives might include

the 4070 super which achieves 82% of the

performance of the ti super for 75% of

the price or the 6800 XT which achieves

84% of the performance for 63% of the

cost in this game although that changes

with rt at 1440p the 4070 TI super post

full scaling at 228 FPS average leading

the 470 TI by 6. . 3% reduction against

4K despite Unbound scaling the 7800 XT

is closer to the 480 than the ti super

meanwhile the 470 super gives the ti

super a 15% lead bringing upgrades back

into the picture the 1080 TI would gain

more than double the frame rate although

the 108ti is still plenty good for a lot

of rasterized games RX 6600 users would

jump 167% from 86 FPS average we're

still not CPU bound at 1080p for the 407

TI super the ti super leads the ti by

5.8% now down again and the 470 super by

12% the 700xt has climbed the ranks a

little and established a 14% lead over

the 4070 TI super revisiting those

upgrade options the 2070 was down at

around 120 FPS average so there'd be a

huge boost here with the 3060 at 126 FPS

average the 1080t at 142 and the 6600 XT

at 152 FPS average in any of those scen

scenarios the ti super would be a big

jump with rt jumping even more whether

or not that's worth it really depends on

the games you play and if you notice

that your card is feeling slow if you're

not playing heavier newer games it's

obviously less likely to be critical

Rainbow Six Siege is up now we ran

efficiency tests for this earlier but

remember the FPS numbers aren't taken

from this chart they were tested

separately with the background tasks the

4070 TI sup 184 FPS average had it 10%

ahead of the 4070 TI and 21% ahead of

the 407 super the AMD 7900 XT puts up a

strong fight here technically anyway but

it's not particularly noticeable it does

however outpace the 470 TI super price

is what'll matter there compared to the

2070 super at 84 FPS average we're seing

a 120% Improvement in the ti super the

3060 isn't distant from that math down

at 71 FPS average 1440p reduces the ti

Super's lead to just 3 and a half% over

the 4070 TI an unexciting move the 7900

XTX and 490 indicate that we still have

plenty of CPU Headroom so this is an

unfortunately real and boring change

from the 4070 TI there's no point

spending more time here it doesn't

meaningfully move the needle from the

4070 TI it's boring and it certainly

arrests the price of the card at the

$800 Mark Final Fantasy 14 is up now

first at 4K the 470 TI super ran at 129

FPS average here establishing a 15 FPS

lead over the 4070 TI or about 12.5% the

7800 XT sits close to the ti super but

behind it and AMD has worst .1% lows in

this game then would be proportional to

the average against the 470 super the ti

super is 23% better and against the

Baseline 2070 it's 178% improved we

haven't retested it in a while but the

RTX 2060 is on this chart as a point of

reference its frame rate may have moved

slightly in the past year generally

speaking though we don't see meaningful

gaps to developed with this game at this

point which is why it's here finally for

the Old Reliable benchmarks GTA 5 at 4K

has the 4070 TI super at 140 FPS average

leading the 4070 TI by 11.6% and the 470

super by 19% if you bought an RTX 26d

when it launched an upgrade here would

yield a 185% uplift the 7900 XT is just

behind the ti super here as well we'll

open our rate tracing testing with

cyberpunk Phantom Liberty which was only

just added to the test suite and

therefore has fewer cars than the RT

test we'll look at after it we're

iterating and adding to this each review

and we'll keep adding new types of tests

too as we continue to do more reviews

all of these tests were conducted in the

last two days and were done across two

groups of settings the first batch uses

RT Ultra with upscaling disabled Ray

reconstruction disabled and overdrive

off the second batch uses a much lighter

RT medium which we found to have lower

scaling impact overall and more

favorable but still weak AMD performance

and these performance deficit widens as

RT load increases and again upscaling is

off for both of these here's the chart

at 4K first the 4070 TI super ran at

about 23 FPS average this average is low

but the lows are spaced proportionally

and well as compared to the average

despite being unplayable this is a good

academic study of bandwidth and why we

included this test the 4070 TI ran at

20.6 FPS average allowing the ti super a

lead of 133% that breaks rank from some

of the others we've seen critically the

original TI has significantly worse lows

here you wouldn't use these settings for

either card but it does show that the

memory bandwidth and capacity change

influenced the performance in some key

areas the closest AMD card here is the

7800 XTX which underperformed against

the 4070 super but held stronger lows at

1440p Ultra the 4070 TI Super's lead

over the 4070 TI drops to 8% the cards

diverge as load increases as a result of

the memory subsystem changes so it does

in fact do something but the impact

becomes largely academic since we were

in unplayable territory at 4K Ultra the

7800 XTX runs below the 4070 super here

the 490 posts a massive lead of 58% over

the 470 TI super just to set the ceiling

at 1080p Ultra the 470 TI Super's lead

was reduced to 6% the memory subsystem

overhaul doesn't matter much here it's

not as heavily loaded as at 4K the 480

leads by 15% here with the 7800 XTX

still tied with the 470 look at RT

medium briefly here the 4070 TI super

lead the ti by 12% at 57.7 to 51.6 the

7800 XTX is much more competitive here

than it was at Ultra now surpassing the

470 super and roughly tying with the 470

TI it's still not competitive on the

whole in this test but this illustrates

that heavier RT loads disproportionately

separate AMD and Nvidia favoring Nvidia

as RT workload goes higher finally at

1080p medium the 4070 TI super leads the

4070 TI by 7% the lowest of this game in

settings so far the 7800 XTX is about

equal with the 470 super here with the

XT about equal to the 470 and's relative

rank overall has fallen at this

resolution it's benefited by the higher

resolution combination with lower rate

tracing settings in Resident Evil 4 with

rate tracing and at 4K we're using FSR

as a test of both upscaling and RT

impact we measured the 4070 TI Supra at

101 FPS average here leading the 4070 TI

by 11% the 700xt manages to outperform

the 470 super in this one although that

doesn't persist to heavier RT games like

cyberpunk as we saw earlier compared to

the 4070 super the ti super is about 22%

better and compared to the older 2070

back when RTX was new it's about 150%

better at 1440p the 4070 TI super runs

at 165 FPS average with upscaling and RT

that has it just ahead of the 7800 XT

and 7.7% ahead of the 4070 TI finally

for Resident Evil RT 1080p has the 47 TI

super at 2011 FPS average with FSR

reducing the lead over the ti original

to 5.7% the 7900 XT sits closer at this

point to the original 4070 TI than the

4070 TI super but overall it's between

them in dlight 2 with r tracing and

equivalent upscaling between all the

vendors we saw the 4070 TI super at 51.8

FPS average leading the 4070 TI by just

9% the 700 XT Trails the 4070 super here

down at 38 FPS a average and just ahead

of the 4070 Baseline card so the XT is

not in a great spot competitively Nvidia

is advantaged it's not as Extreme as in

cyberp Punk or Allen wake 2 but it's

definitely more than in Resident Evil 4

generationally the 470 TI super leads

the 370 by 86% and the RTX 2080 by 182%

at 1440p we've added some mid-range past

gen cards to the list the RTX 2070 now

appears down at 30 FPS PS average ahead

of the 6600 xt's 27 FPS average result

and behind the 3060 TI's 46 result the

4070 TI super is 96 FPS average leads

the 4070 TI by about 7% so it's dropping

once again the 700xt is behind the 380

and the 4070 super in this one and ahead

of the 4070 it's not competing with the

ti class cards directly at 1080p the

4070 TI super lead over the 4070 TI is

reduced to just 4.4% the 7800 XT is

about 5 to 6% ahead of the 470 and it

continues to yield an advantage to

Nvidia not much has changed for these

ranks finally we're on power although we

looked at Power efficiency earlier now

we'll briefly look at our standardized

total power consumption against more

cards the 4070 TI super tough card

pulled 291 Watts with the Quiet bios or

296 Watts with the default performance

feed bios that are shipped in

overclocking allows a maximum power

consumption of 321 Watts with the OC on

the performance bios the 296 watt figure

has it about the same as a 6800 XT and

less than an RX 7900 XT by 10 to 25

watts depending on which card you look

at the 4070 TI was about 5 Watts less at

290 watts in our original testing and

that contributes to some of the

performance Gap as well well it's a lot

of numbers first on our side of things

we're excited not because of this this

is like we'll talk about it we're

excited because the efficiency testing

the latency stuff's really cool um it's

coming back in for the 480 super like

stated but we wanted to kind of get the

efficiency stuff in this one so we're

experimenting with a couple different

approaches to reviewing gpus that aren't

just frame rate which is what we've

historically done there's obviously uh

plenty of different approaches out there

but our primary focus has been frame

times frame pacing frame rate and then

power in total uh power consumption

workloads like power virus so we're

overhauling that and uh the efficiency

stuff I'm personally really excited

about now uh as for the rest of this so

for the cards you have all the numbers

we'll let you make your own decision but

if you want some stances on it as

briefly as possible uh this launch can

largely be summed up with a very

Advanced

word me it's really not particularly

strong but it's also not offensively bad

the price is locked so probably the

worst or biggest downside there is is

that we might traditionally have

received the price cut here with the ti

but instead the ASP or the average

selling price remains up as a result of

this launch but the performance also

goes up a little bit with it and so

we're back to me it's not the the worst

a card could receive on this channel

it's just kind of in the middle so it

seems like Nvidia as we said in the

beginning has chosen to couch this car

in the middle of maybe two more

interesting launches but especially the

480 super uh looks interesting now that

said we haven't tested it yet so we'll

see we'll know how that pans out

somewhere around January 31st or so as

when it's supposed to be launching so

reviews plus or minus day on that but

the 4070 TI super puts pressure on AMD

7900 xte if you care a lot about Super

heavy rage tracing like Ultra or high

settings in games like cyberp Punk or

Allen wake then that's where the

difference really starts to appear and

you've also got things like Ray

reconstruction to consider uh that is

basically entirely subjective we have a

separate video on that where we just

show you the uh A and B comparison for

it if you don't care about Ray tracing

or you play games that are Ray traced

but typically with something like a

single RT feature Shadows Reflections

things like that but not typically

combined or generally lower uh RT

intensity then the 7900 XT remains a

good consideration especially because of

its new found price

drops little hesitant to commit to

calling a particular price for it right

now cuz it's kind of in flux but the

last we checked for Hardware news uh

there's evidence of it on New Egg it was

like 710 bucks uh sometimes 750 but 710

seem to be starting to become at least

available in the US North American

market so anyway that's been changing a

lot it's largely going to come down the

price on those and if you care about RT

or particularly heavy RT like uh as high

settings as you might be able to

realistically set with a particular card

but that'll be it for the review You've

got a lot of numbers to work with and

make a choice on it um we've been

experimenting with totally over hauling

the test one Benchmark at a time so next

one we're hoping to bring the latency

back in we debuted again the 70 super uh

we've got some changes we're do into

that to further improve it hopefully

make it even more useful and explain it

some more further and then we also have

the 7600 XT review next it is about 30

hours be well just give everyone some

perspective uh it's less than 12 hours

before this review has to go live so I

need to finish editing the rest of it uh

and then I need need to start writing

the 7600 XT which will go up 24 hours

after that

so see you soon and otherwise the 480

super should be the one where we have

about a week to work on it so check back

for that as well okay we'll see you all

tomorrow about 24 hours I think I don't

I don't know thanks for watching as

always go to store. Gamers access.net

grab one of these shirts to support our

in-depth testing efforts and our

overhauls and and you can go to

patreon.com Gamers ACC helps out

directly we'll see you all next time