RTX 4070 Ti SUPER评测:16G大显存,比4070Ti强10%
Summary
TLDR本视频对NVIDIA 4070Ti SUPER显卡进行了全面评测。相较于4070Ti,SUPER版本在视频内存上进行了升级,从12GB提升至16GB,显著提升了在高分辨率和高负载场景下的性能表现。测试结果显示,4070Ti SUPER在3DMark和多款游戏中的性能均有所提升,尤其是在4K分辨率和光线追踪技术开启时,性能提升更为明显。此外,视频还探讨了4070Ti SUPER在AI绘图和3D渲染软件中的性能,以及超频潜力,最终得出4070Ti SUPER是一款适合4K分辨率玩家和AI应用用户的显卡。
Takeaways
- 🚀 4070Ti SUPER显卡配备了与4080相同的16GB大容量显存,这使得它在高分辨率和高负荷场景下的性能得到显著提升。
- 🔍 与4070Ti相比,4070Ti SUPER在核心规格上有所提升,CUDA核心数量从7680增加到8448,光追单元、纹理单元和张量核心也相应增加,整体性能提升了约10%。
- 💰 4070Ti SUPER与4080相比,虽然性能有所差距,但在价格上提供了更具竞争力的选择,迫使竞争对手7900XT降价应对。
- 🎮 在游戏性能测试中,4070Ti SUPER在2K和4K分辨率下的表现均略优于4070Ti,但仍然落后于4080,尤其在开启光追和DLSS技术后,性能差距更加明显。
- 🛠️ 4070Ti SUPER的散热性能出色,即使在高负载下也能保持较低的核心和显存温度,风扇转速低,噪音小。
- 💡 4070Ti SUPER的显存升级对于使用Stable Diffusion等AI绘图软件的用户来说非常有价值,16GB显存能更好地处理大型项目。
- 🔧 在Blender等3D建模软件中,4070Ti SUPER相比4070Ti有性能提升,尤其在处理大型项目时,更大的显存意味着更好的运行表现。
- 📈 通过超频,4070Ti SUPER的性能可以进一步接近4080,但即便如此,两者之间仍存在性能差距。
- 👀 4070Ti SUPER没有FE版本,但市面上有ASUS的TUF 4070Ti SUPER和ROG Strix 4070Ti SUPER等非公版产品,具有不同的散热设计和性能表现。
- 🔗 4070Ti SUPER的推出,是NVIDIA对市场变化的回应,旨在提供更具性价比的产品,满足不同用户的需求。
- 👍 视频最后提醒观众点赞、评论、关注频道,并推荐访问Taobao和Geek Bay商店查看相关配件。
Q & A
NVIDIA 4070Ti SUPER显卡相较于4080显卡在视频内存上有何不同?
-4070Ti SUPER显卡配备了与4080相同的16GB视频内存,而之前的4070Ti只有12GB 192-bit视频内存。
4070Ti SUPER显卡在核心规格上有哪些提升?
-4070Ti SUPER显卡使用了AD103核心,CUDA核心数量从7680增加到8448,光线追踪单元、纹理单元和张量核心也相应增加,提升了大约10%。
4070Ti SUPER显卡的显存规格与4070Ti有何差异?
-4070Ti SUPER显卡的显存增加到了16GB,256bit规格,与4080保持一致,而4070Ti的显存为12GB。
4070Ti SUPER显卡的默认功耗限制是多少,这与4070Ti有何关系?
-4070Ti SUPER显卡的官方默认功耗限制是285W,与4070Ti相同。
4070Ti SUPER显卡在散热设计上与4070Ti有何区别?
-4070Ti SUPER显卡采用了三风扇设计,散热性能与4070Ti相近,但体积略小,高度也不如4080和4090。
4070Ti SUPER显卡在3Dmark Time Spy测试中的得分如何?
-4070Ti SUPER在3Dmark Time Spy测试中得分为24,723分,比4070Ti高出7.4%,但比4080低约13%。
在实际游戏中,4070Ti SUPER显卡的性能表现如何?
-在实际游戏中,4070Ti SUPER显卡在2K和4K分辨率下的性能表现接近4070Ti,但在高压力场景下,如4K分辨率加上光线追踪,4070Ti SUPER的性能提升更为明显。
4070Ti SUPER显卡在《赛博朋克2077》游戏中的性能表现如何?
-在《赛博朋克2077》游戏中,4070Ti SUPER在2K分辨率下的性能比4070Ti高出6%,但比4080低17%。在4K分辨率下,性能差距与2K相似。
4070Ti SUPER显卡在视频内存压力较大的游戏中表现如何?
-在视频内存压力较大的游戏中,如《杀手2》(Hitman 2),4070Ti SUPER由于拥有更大的视频内存,在高分辨率和光线追踪开启时表现优于4070Ti。
4070Ti SUPER显卡的散热性能如何?
-4070Ti SUPER显卡的散热性能非常出色,在进行Furmark测试时,核心温度达到61度,显存温度62度,风扇转速低于1400rpm,噪音很小。
4070Ti SUPER显卡在AI绘图性能上有何提升?
-4070Ti SUPER显卡在AI绘图性能上,尤其是在使用Stable Diffusion等模型时,由于配备了更大的16GB视频内存,性能显著优于4070Ti。
4070Ti SUPER显卡的超频潜力如何?
-4070Ti SUPER显卡的超频潜力不错,使用ASUS TUF 4070Ti SUPER可以达到核心频率3045MHz和显存频率24GHz,但即便超频后,性能仍无法与4080相媲美。
4070Ti SUPER显卡的市场定位是什么?
-4070Ti SUPER显卡的市场定位是对标AMD的7900XT,通过提升规格和性能,迫使竞争对手降价应对,同时弥补了4070Ti在高压力场景下的性能不足。
Outlines
🚀 4070Ti SUPER显卡性能评测
本段落主要介绍了4070Ti SUPER显卡的性能特点及其与4070Ti和4080显卡的比较。4070Ti SUPER显卡配备了16GB显存,与4080相同,这在30系列显卡中是一个显著的升级。测试发现,4070Ti SUPER在核心规格上有所提升,包括CUDA核心数量、光线追踪单元、纹理单元和张量核心的增加,显存也从12GB提升到16GB,256位规格。在3Dmark测试中,4070Ti SUPER的性能提升并不显著,但在实际游戏测试中,尤其是在4K分辨率和高压力场景下,4070Ti SUPER表现出了更强的性能,尤其是在显存压力较大的游戏中,如《杀手2》。此外,还提到了4070Ti SUPER的散热性能和噪音控制,以及其在AI绘图和Blender等专业应用中的性能表现。
🔥 4070Ti SUPER在高压游戏场景中的表现
此段落深入探讨了4070Ti SUPER在高压游戏场景下的性能表现。在《荒野大镖客2》和《赛博朋克2077》等游戏中,4070Ti SUPER与4070Ti相比有小幅提升,但与4080相比仍有差距。开启DLSS技术后,性能差距有所缩小。特别是在《杀手2》中,4070Ti SUPER由于拥有更大的显存,在不开启光线追踪时表现优于4070Ti,而在4K分辨率和开启光线追踪的情况下,4070Ti几乎无法正常运行,而4070Ti SUPER则能维持在20fps的帧率。此外,还讨论了DLSS3.0技术对4070Ti SUPER性能的影响,以及其在2K和4K分辨率下的性能对比。总体来看,4070Ti SUPER在高压场景下的性能提升对于4K分辨率玩家来说具有实际意义。
🛠 4070Ti SUPER显卡的超频潜力与市场定位
最后一段讨论了4070Ti SUPER显卡的超频潜力和市场定位。通过超频,4070Ti SUPER的核频和显存频率得到了提升,但即便如此,其性能仍无法与4080相媲美。此外,文中提到了4070Ti SUPER的市场目的,即作为NVIDIA对AMD 7900XT显卡的回应,迫使对手降价。4070Ti SUPER在保持6499元价格的同时,通过搭载AD103核心和16GB 256位显存,弥补了4070Ti的最大短板。尽管在显存充足的情况下,4070Ti SUPER的性能提升大约只有10%,但它在4K游戏和专业应用中的表现使其成为一个有吸引力的选择。最后,作者呼吁观众关注频道并支持他们的淘宝店铺。
Mindmap
Keywords
💡N卡
💡视频内存
💡4070Ti SUPER
💡CUDA核心
💡光追单元
💡张量核心
💡显存位宽
💡功耗墙
💡DLSS
💡Stable Diffusion
💡Blender
Highlights
在30系列时代,3080Ti和3090在显存容量上的差异是其价格差异3000元的主要原因。
4090和RTX 6000 Ada在拥有相同核心的情况下,价格差异显著。
4070Ti SUPER将配备与4080相同大小的16G显存,使其成为三个SUPER中最重要的显卡。
4070Ti在发布时的12GB 192-bit显存是其在4K分辨率下性能不足的主要原因。
4070Ti SUPER的核心CUDA数量从4080的7680增加到8448,性能提升约10%。
4070Ti SUPER的显存提升到16GB,256bit规格,与4080一致。
4070Ti SUPER的默认功耗限制与4070Ti相同,为285W。
4070Ti SUPER没有FE版本,收到的是ASUS的非公版4070Ti SUPER。
ASUS TUF 4070Ti SUPER采用三风扇设计,体积略小于FE版4080和4090。
TUF 4070 Ti SUPER的PCB设计与4070系列的几款TUF几乎相同,表明AD103和AD104可以Pin2Pin共享。
ROG的Strix 4070Ti SUPER在散热和尺寸上与Strix 4070Ti相同,但尺寸大于TUF。
3Dmark Time Spy测试显示,4070Ti SUPER的性能提升并不显著。
在CS2游戏中,4070Ti SUPER在2K和4K分辨率下的性能介于4070Ti和4080之间。
在Red Dead Redemption 2游戏中,4070Ti SUPER的性能提升接近4070Ti,但与4080有差距。
在Cyberpunk 2077游戏中,4070Ti SUPER的性能在2K和4K分辨率下均不如4080。
在Hitman 2游戏中,4070Ti SUPER的16G显存使其在高压力下的性能优于4070Ti。
4070Ti SUPER在Blender中的性能表现优于4070Ti,尤其在大型项目中。
4070Ti SUPER的功耗墙保持在285W,散热性能出色。
4070Ti SUPER的显存从12G升级到16G,对使用Stable Diffusion绘图的创作者具有吸引力。
4070Ti SUPER在超频后性能接近4080,但仍未达到4080的性能水平。
4070Ti SUPER是NVIDIA官方的“体面”降价产品,主要针对竞争对手7900XT。
Transcripts
If the most valuable thing about mid-to-high-end N cards is the video
memory, it can only be the video memory. In
the 30-series era,
the 3080Ti and 3090 have almost only a difference in video memory capacity,
but the price difference is 3,000 yuan.
The price difference between the 4090 and RTX 6000 Ada with the same core is several times
higher. Yellow video memory is as valuable as iPhone memory.
So when I knew that the 4070Ti SUPER would be equipped with a 16G video memory as large as the 4080,
I knew that it would not only be the video card with the longest name
, but also the most important one among the three SUPERs.
The biggest flaw
when the 4070Ti was launched was its 12GB 192-bit video memory
. This gave it more core performance than the 3090Ti
but failed at 4K resolution.
The biggest upgrade of the 4070Ti SUPER is here. If you
simply compare the specifications,
you will find that the 4070Ti SUPER uses The number of AD103 core CUDA from 4080
has increased from 7680 to 8448
light tracing units, texture units and Tensor Cores have also increased in proportion,
which is about 10%. The core specifications have been improved.
The video memory has increased to 16GB, 256bit specifications,
which are consistent with the 4080
cache and 4070Ti. 48M
will be smaller than the 64M of 4080. The
official default power consumption wall is 285W,
which is consistent with 4070Ti.
From the overall specifications,
compared to the big brother of 4070Ti,
this card is more like the younger brother of 4080.
It is a pity that
4070Ti SUPER does not have an FE version .
What I received is the non-public version 4070Ti SUPER from ASUS.
This picture is the ASUS TUF 4070Ti SUPER.
The three-fan design is close to 3 slots thick.
The volume is slightly
smaller than the FE version 4080 and 4090. The over-the-shoulder height is not that high.
Compared with the TUF version of 4080 and 4090. The power supply interface is
obviously smaller. It is no surprise that a 16pin interface is used
to disassemble the radiator.
The PCB layout of TUF 4070 Ti SUPER
is almost the same as that
of several TUFs in the 4070 series , indicating that AD103 and AD104 can share Pin2Pin.
The core code of the board design is AD103-275
. A total of 10 phases of core power supply and 2 phases of memory power supply are arranged on the PCB. The
inductor is also pressed with a thermal pad to reduce howling.
We also received this picture of ROG's Strix 4070Ti SUPER.
The appearance design of the card is exactly the same as that of Strix 4070Ti. It
has the same heat dissipation and
the size is much larger than TUF.
Of course, as a 70-level product,
its size is still much smaller than that of 80 and 90 Raptors.
In terms of thickness, it has lost a lot of weight. But
the design language is basically the same,
so let’s start performance testing directly!
The first is 3Dmark. In Time Spy,
4070Ti SUPER ran a score of 24,723 points.
This score is only 7.4% higher than 4070Ti
and almost 13% lower than 4080.
It seems that the improvement of 4070Ti SUPER is not that big.
Time has higher pressure. In Spy Extreme,
4070Ti SUPER ran a score of 12145,
which is 10.2% higher than 4070Ti.
This is in line with the specification improvement,
but the gap between it and 4080 is still 14%.
In the light tracing performance test,
4070Ti SUPER is also about 10% better than 4070Ti. 10%
but still 12% weaker than the 4080.
The theoretical test results seem to indicate that
the 4070Ti SUPER is a card better at high resolution and high pressure.
But it is obviously still far behind the 4080.
So in actual games,
the 4070Ti SUPER What kind of performance will it look like?
Let’s run a few games to see.
The first is CS2.
The test method we use here is still the demo produced by CSGO-Purple.
This scene includes two parts, Inferno point B and point A,
as well as common throwing objects, which
is more in line with the actual game situation.
The game CS2 proved to us last time how stressful it is. In
the end,
the 4070Ti SUPER ran an average frame of 252fps at 2K,
11% stronger than the 4070Ti, 11%
weaker than the 4080,
and perfectly nailed between the two cards at
4K. A similar situation continues below.
The average frame rate of 127fps is 12% stronger than 4070Ti and 12%
weaker than 4080.
Lao Huang’s knife skills are really amazing.
But how do I remember that 4070Ti used to be called 4080 12G,
so what should this card be called?
4080 16G SE?
Next is the veteran 3A Red Dead Redemption 2.
In Cousin 2 with 2K resolution,
the 4070Ti SUPER is 9% stronger than the 4070Ti
but 12% weaker than the 4080.
It seems that it is still closer to the 4070Ti.
After turning on the DLSS balance file
The gap between the three has been reduced.
Will turning the resolution higher make it closer to 4080?
In the end,
the 4070Ti SUPER still maintained the same performance gap as that under 2K resolution at 4K resolution
. On the contrary, after turning on DLSS,
it was even more powerful than the 4070Ti
. Overall, it was an improvement that was neither surprising nor pleasant. So
in TGA's best performance?
Among the best continuously operating games, Cyberpunk 2077, the 4070Ti SUPER
is only 6% better than the 4070Ti and 17% weaker than the 4080 at 2K with neither ray tracing nor DLSS turned on.
It seems that the number of CUDA is the most important here.
After the resolution is turned to 4K,
the 4070Ti SUPER clearly has a gap with the 4070Ti. It
is 13% stronger than the 4070Ti,
but still 18% weaker than the 4080.
The big brother is still the big brother.
Then we continue to increase the intensity and turn on
the ray tracing. After 2K,
the 4070Ti SUPER is stronger than the 4070Ti. It is 8%
weaker than the 4080 and 13% weaker
than the 4080. The gap between it and the 4080 is obviously smaller than without light pursuit.
So we further increase the resolution to 4K.
At this time, the gap between the 4070Ti SUPER and the 4070Ti directly widens to 18%,
the gap between the 4080 and the 4080. It has also been reduced to 10%
, just as we expected.
The greater the pressure, the farther away it is from the 4070Ti.
After turning on the DLSS balance file,
there is still a similar gap between the three cards
, and further turning on the light reconstruction and frame generation of DLSS3.5. After technology,
the 4070Ti SUPER can also exceed the average frame rate of 60fps in 4K
, but the 4080 will leave the 4070Ti SUPER behind at this time.
Finally
, let’s run again
the heavyweight work that made the three brothers of the 4070 family 12G video memory directly explode the video memory - Soul The video memory usage of the game Hitman 2
is simply devilish.
First of all, without turning on ray tracing,
the 4070Ti SUPER is 10% stronger than the 4070Ti
at 2K resolution
and 8% weaker than the 4080.
At 4K, the gap is 7% and 11% respectively, and
the overall gap is between After turning on ray tracing between the two cards
, the situation is different. At 2K,
the 4070Ti SUPER is 20% better than the 4070Ti.
At 4K,
the 4070Ti can no longer run normally. The 4070Ti SUPER
still maintains a frame rate of 20fps with its
larger video memory. Although this frame rate
is still far from being playable,
compared with 3fps e-sports, it is still too much.
Future games will inevitably consume more video memory.
This is the most meaningful improvement of 4070Ti SUPER.
In order to be able to To play this game normally with
this level of graphics card, you still have to turn on DLSS.
After turning on the balanced DLSS,
the frame rate of each graphics card is doubled. More than
2K,
the 4070Ti SUPER is 15% stronger than the 4070Ti and
10% weaker than the 4080
at 4K. is to further widen the gap to 24% and 9%
Therefore, it is not difficult to see that the greater the pressure of the game,
the greater the lead of 4070Ti SUPER.
We continue to enable the frame generation technology of DLSS3.
The lead of 4070Ti SUPER under 2K has narrowed.
However, under 4K, it is due to the relationship between the video memory capacity and
the already exploded video memory. The 4070Ti 4070Ti SUPER that is "complementing frames"
can obviously work normally.
At this time, the memory usage has exceeded 13G and
the frame rate has reached an average of more than 60fps.
In this case,
the gaming experience of 4070Ti SUPER will be much better than that of 4070Ti.
Summary Looking at the results of the game test,
it is obvious that the 4070Ti SUPER is a card that has more advantages the higher the pressure.
Compared to the 4070Ti, which is weak in high-pressure scenarios,
the 4070Ti SUPER has a higher graphics memory configuration and
is obviously better than the 4K+ ray tracing scenario. 4070Ti is more calm.
In the end,
4070Ti SUPER has a comprehensive improvement of 8.44% over 4070Ti in 2K,
10.3% in
4K, and 11.3%
in 2K and 14.93%
in ray tracing.
It is the only 70-level graphics card of this generation that can be said to be good at it. 4K game model
, so if you are a 4K resolution player,
the improvement of 4070Ti SUPER is still very meaningful.
The power consumption wall of 4070Ti SUPER maintains the same 285W as 4070Ti
, so will it not have the same power consumption wall
as 4070 SUPER? Woolen cloth?
In fact, this 285W power consumption wall is still very suitable for it.
In the 3DMark Time Spy test, it basically ran close to this power consumption wall
, but at the same time, it rarely downclocked due to encountering the power consumption wall.
Compared with the 4070Ti The power consumption is slightly high
, but it is very limited.
I also tested the heat dissipation performance of ASUS's TUF 4070Ti SUPER.
I used Furmark to maintain a power consumption of 285W
in an open environment at 25 degrees Celsius. After 5 minutes
, the BIOS was in the default performance mode. It finally
reached the core temperature of 61 degrees and the memory temperature of 62 degrees.
At this time, the fan speed was less than 1400 rpm.
The sound was perceptible but very small.
The temperature performance was quite cool.
After switching to the quiet mode BIOS,
the GPU temperature correspondingly reached 65 degrees
of the video memory . The temperature is 66 degrees.
At this time, the fan speed is only 1100rpm and there is
almost no noise.
I am quite impressed by the cooling performance of TUF 4070Ti SUPER. The performance
is outstanding
. The upgrade of the video memory of 4070Ti SUPER from 12G to 16G
is not only attractive to gamers. Besides,
the most in demand is probably the cyber painters who use Stable Diffusion to draw.
I tested the performance under the SD1.5 model and the performance under the SDXL 1.0 model respectively.
The Sampling Steps were set to 50 times.
Sampling method we ran Euler a
SD1. 5 We generate 512x512 graphs
and SDXL generates 512x512 and 1024x1024 graphs respectively
and use the TensorRT plug-in
to maximize their performance .
In the end
, the 4070Ti SUPER is significantly better than the other three members of the 4070 family .
The overall performance is significantly closer to the 4080
, and the 16G video memory and 12G video memory is sometimes not only a difference in speed, but also
a difference in whether it can run.
Therefore, for users with AI needs,
the purchase value of 4070Ti SUPER is obviously much higher than that of 4070Ti.
In Blender
, 4070Ti SUPER is also better than
There is still a certain gap between 4070Ti and 4080
, but in large projects,
16G video memory and 12G video memory will still face the difference between running and not running,
so if your project is relatively
large , 4G video memory will be stronger than 4070Ti. Performance is more valuable.
This test is almost completed,
but I feel like there is still something unfinished
!
I haven’t overclocked this 4070Ti SUPER yet!
Since it is also the AD103 core
4070Ti SUPER overclocking, can it catch up with the big brother 4080?
I used this ASUS TUF 4070Ti SUPER and finally reached
the core frequency of 3045MHz and the memory frequency of 24GHz.
At this time, the 3Dmark Time Spy score has exceeded 26600 points.
Although it is a lot higher,
it is obviously still different from the 4080 default score of more than 28000 points. The gap was not small
, so I took out ROG's 4070Ti SUPER Raptor.
The power consumption of this card is as high as over 360W
. What surprised me the most was that
the performance of this ROG Raptor was so good that
it easily exceeded the core frequency of 3165MHz.
In the end, it successfully reached the Time Spy graphics score of 27,500.
This score is very close to the default 4080
, but unfortunately it still cannot reach the performance of 4080.
So if you want to buy a 4070Ti SUPER to use with 4080,
I would advise you to give up. This idea,
then the above is the entire content of this issue's 4070Ti SUPER first test.
As I said before in the 4070 SUPER first test video,
SUPER is largely NVIDIA's official "decent" price reduction product.
The main market purpose of 4070Ti SUPER is The disgusting rival 7900XT
forced the 7900XT to cut its price to cope with the situation
. Compared with the 4070Ti,
the 4070Ti SUPER maintains the price of 6,499 yuan and
directly makes up for the biggest shortcomings of the 4070Ti
by carrying the AD103 core and 256bit 16G video memory.
However, in scenarios where the video memory is sufficient,
The absolute performance improvement of 4070Ti Super is only about 10%, and
then only the performance time of 4080 SUPER is left
. So we will see you next week.
If you like our show, don’t forget to press and hold the like button to comment. Do
n’t forget to follow our channel.
Finally, don’t forget to go to our Taobao store and Geek Bay store
to look at our peripherals
. See you in the next episode.
5.0 / 5 (0 votes)