AllInfo

Nvidia laptop GPUs in the test: GeForce RTX 4050, 4060, 4070, 4080 and 4090 in comparison

admin

Two weeks after the GeForce RTX 4090 and 4080 laptop GPUs, Nvidia and its partners are also launching notebooks with GeForce RTX 4070, 4060 and 4050 laptop GPUs. ComputerBase has tested all five variants and provides detailed comparison benchmarks in games and applications including information on TDP scaling.

Table of contents

  1. GeForce RTX 4070, 4060 & 4050 Laptop GPUs Review
    1. RTX 4090 & 4080 Laptop GPU: A Review
    2. RTX 4070, 4060 & 4050 laptop GPU in detail
  2. The test patterns at a glance
    1. Notebooks from XMG and Gigabyte
    2. Clock rates and consumption under load
  3. Gaming benchmarks
    1. TDP scaling: from 35 to 150 watts
    2. WQHD benchmarks with and without RT
    3. FHD benchmarks with and without RT
    4. Higher FPS with DLSS
    5. Synthetic benchmarks
  4. Application benchmarks
  5. Conclusion

GeForce RTX 4070, 4060 & 4050 laptop GPUs under test

Two weeks ago, Nvidia launched the two top mobile GPU models, the GeForce RTX 4090 & 4080 laptop GPU (test) released on the market, now the smaller variants GeForce RTX 4070, 4060 & 4050 Laptop GPU.

RTX 4090 & 4080 Laptop GPU: A Review

As a reminder: Based on the AD103 GPU known from the GeForce RTX 4080, the GeForce RTX 4090 laptop GPU (16 GB) with a maximum permitted power loss of 150 watts plus up to 25 watts via Dynamic Boost achieved the performance of a GeForce RTX 3090 and was therefore better only 5 percent behind the GeForce RTX 4070 Ti in WQHD at maximum details. The GeForce RTX 4080 laptop GPU (12 GB) based on its AD104 GPU followed by 10 percent – comparatively little if you consider that the 4090 laptop GPU has 30 percent more shaders.

Both graphics chips allowed themselves an average of just over 160 watts in the test in WQHD with high details across all tests, so they made use of the option of tapping into the electrical budget of the CPU via Dynamic Boost and were able to continue the clock in this way increase. Even then, the maximum permissible chip clock was only reached in exceptional cases. Why does this need to be mentioned? More on that in the course of this test.

RTX 4070, 4060 & 4050 laptop GPU in detail

The RTX 4070, 4060 and 4050 laptop GPUs now rely on even smaller GPUs, which are also celebrating their premiere. Because they didn't exist in the desktop either: AD106 (4070) and AD107 (4060 & 4050). The RTX 4070 laptop GPU does without almost 40 percent shaders compared to the RTX 4080 laptop GPU, the 4060 another 33 percent compared to the 4070 and the 4050 compared to the 4060 another 17 percent. The RTX 4070 and RTX 4060 laptop GPU thus offer fewer shaders than the named predecessors, the RTX 4050 is again significantly above the level of the RTX 3050.

Nvidia GeForce RTX 4090, 4080, 4070, 4060 & 4050 laptop GPUs in notebooks from Gigabyte, MSI and XMG tested

When it comes to the memory interface, all three smaller laptop GPUs follow in the footsteps of the RTX 4080 laptop GPU and shift down a gear compared to the last generation: the cut is most noticeable with the 4070, whose Interface has been halved, while the 4060 and 4050 have been cropped by a third and a quarter, respectively. This approach is not uncommon for Ada Lovelace, the significantly larger cache of the GPUs should have enabled Nvidia to take the step.

In terms of storage capacity, the 4070 is flat at 8 GB, while the 4060 (to 8 GB) and 4050 (to 6 GB) have each increased by 2 GB. The subject of storage will come up repeatedly throughout the article.

Another interesting detail of the three new GPUs: Nvidia has intended all three for notebook chassis that must be able to dissipate a maximum of 35 to 115 watts of power loss. Specified, i.e. officially, there are variants from 35 to 140 watts in all three classes. As the test will show, however, theory and practice diverge on this point.

GeForce RTX 4000 laptop GPU vs. desktop vs. RTX 3000 laptop GPU in comparison GPU shader memory model ,
Interface, max. Bandwidth TGP max. Dynamic Boost 2.0 DLSS FG* GeForce RTX 4090 AD102 16,384 24 GB GDDR6X
384 bit, 21.0 Gbps 450 Watt ✓ GeForce RTX 4090 Laptop GPU AD103 9.728 16GB GDDR6
256-bit, 18.0 Gbps 80-150W 25W ✓ GeForce RTX 3080 Ti Laptop GPU GA103M 7424 16GB GDDR6
256-bit, 16.0 Gbps 80-150W 25W GeForce RTX 4080 AD103 9728 16GB GDDR6X< br>256 bit, 22.4 Gbps 320 watts ✓ GeForce RTX 4080 Laptop GPU AD104 7,424 12 GB GDDR6
192 bit, 18.0 Gbps 60 – 150 watts 25 watts ✓ GeForce RTX 3080 laptop GPU GA104 6144 8/16GB GDDR6
256bit, 14.0Gbps 80-150W 15W GeForce RTX 4070 Ti AD104 7680 12GB GDDR6X
192 Bit, 21.0 Gbps 285 Watts ✓ GeForce RTX 3070 Ti Laptop GPU GA104 5,888 8 GB GDDR6
256 Bit, 14.0 Gbps 80 – 125 Watts 25 Watts GeForce RTX 4070 Laptop GPU AD106 4,608 8GB GDDR6
128-bit , ?? Gbps 35 – 115 Watts 25 Watts ✓ GeForce RTX 3070 Laptop GPU GA104 5,120 8 GB GDDR6
256 Bit , 14.0 Gbps 80 – 125 Watts 15 Watts GeForce RTX 4060 Laptop GPU AD107 3,072 8 GB GDDR6
128bit, 16Gbps 35 – 115W 25W ✓ GeForce RTX 3060 Laptop GPU GA106 3,840 6GB GDDR6
192-bit, 14.0Gbps 60-115W 25W GeForce RTX 3050 Ti Laptop GPU GA107 2560 4GB GDDR6
128-bit, 12.0Gbps 35-80W 15W GeForce RTX 4050 Laptop GPU AD107 2,560 6GB GDDR6
96 bit, ??? Gbps 35 – 115 watts 15 watts ✓ GeForce RTX 3050 Laptop GPU GA107 2048 4GB GDDR6
128 bit, 12.0 Gbps 35 – 80 watts 15 watts DLSS frame generation as a new feature of DLSS 3

Overview of test patterns

ComputerBase had two notebooks from XMG and one from Gigabyte available for this test. The already publicly available GeForce 528.49 was used as driver.

Notebooks from XMG and Gigabyte

GeForce RTX 4060 and RTX 4070 laptop GPUs were each used in the new XMG Neo 17 (E23) with Intel Core i9-13900HX and were configured in this chassis with the maximum power loss specified by Nvidia of 115 watts plus up to 25 watts via Dynamic Boost . This corresponds to the supplier's claim to have created an “uncompromising gaming notebook” with the Neo 17.

The XMG Neo 17 (E23) can also be configured with RTX 4090 and RTX 4080 laptop GPUs up to 150+25 watts. The 17-inch IPS display has a resolution of 2,560 × 1,600 pixels and is 240 Hz fast. Nvidia Advanced Optimus for automatic switching of the display connection from dGPU to iGPU and vice versa is supported. The HDMI connector is stuck on the dGPU.

The XMG Neo 17 or 16 (E23) is available with RTX 4090 to RTX 4060 laptop GPU

The GeForce RTX 4060 laptop GPU came in the Gigabyte Aero 14 OLED in combination with a Core i7-13700H and can consume a maximum of 35 watts plus 10 watts via Dynamic Boost.

The Aero 14 OLED is not aimed at gamers, but at creative people who are looking for maximum performance in such a compact, all-aluminum chassis with a high-quality OLED screen. Anyone who frequently decodes or encodes videos (incl. AV1) and/or uses GPU-accelerated workflows in video, photo or rendering tools should be much better off with the small GeForce RTX 4050 than with comparable notebooks with iGPU. And as the test shows: You can play with it as long as RT is deactivated and the settings don't overwhelm the only 6 GB of memory.

The Gigabyte Aero 14 OLED comes in an ultra-compact chassis with an RTX 4050 laptop GPU with 35+10 watts

< p class="p text-width">The GeForce RTX 4090 laptop GPU and GeForce RTX 4080 laptop GPU, which were also listed again in this test, were again used in notebooks from MSI (Titan GT77HX 13V) and XMG (Neo 16 (E23)). The mobile Radeon RX 6000 was tested in notebooks from Asus (ROG Zephyrus G14 with Radeon RX 6800S) and Corsair (Voyager a1600 with Radeon RX 6800M). The current Adrenalin 23.2.1 was used.

Clock rates and consumption under load

At first, the editors attributed the following results to an error in the BIOS or Control Center of the XMG Neo 17, but it quickly became clear: Unlike the RTX 4090 and 4080 laptop GPU, the RTX 4070 and 4060 laptop GPU irritate the maximum power loss specified by Nvidia, like them which the XMG Neo 17 (E23) also uses, is not used in many games because the maximum clock rate defined by Nvidia is reached beforehand.

With the RTX 4070 laptop GPU, this is 2,400 MHz, which is present in 11 of the 14 games with an average consumption of sometimes well below 115 watts. With the RTX 4060 laptop GPU it is 2,610/2,625 MHz – in 10 out of 14 games this clock rate is already reached at an average of less than 115 watts.

This behavior differs fundamentally from that of the two larger GPUs, which operated at over 150 watts in every game in the test, because they were able to convert the additional electrical budget available via Dynamic Boost into even higher clock rates. The maximum possible clock rate (2,355 MHz for the 4090 and 2,520 MHz for the 4080) was only achieved in Far Cry 6. The clock rate was almost never the limit, but that's the rule with the smaller variants.

RTX 4090 Laptop
150 W + 25 W
(MSI GT77 Titan) RTX 4080 Laptop
150 + 25 W
(XMG Neo 16 (E23)) RTX 4070 Laptop
115 + 25 W
(XMG Neo 17 (E23)) RTX 4060 Laptop
115 + 25 W
(XMG Neo 17 (E23)) Game Consumption
(W, avg/max) clock
(MHz, avg/max) consumption
(W, avg/max) clock
(MHz, avg/max) consumption
(W, avg/max ) Clock
(MHz, avg/max) Consumption
(W, avg/max) Clock
(MHz, AVG/Max) Cyberpunk 2077 162/180 2.302/2.355 2.478/2.520 103/106 2.400 107/110 2.610 Cyberpunk 2077 RT 164/171 2.175 165/175/2.505 108/114 2,400 106/115 2.610/2.610 Eternal 170/172/2.355 174/186 2.475 111/119 2.400 119/124 2.610/2nd610 Eternal RT 169/178/2.340 171/179 2.349/2.490 59/62* 2,400 75/80* 2.625 F1 22 166/173 2.250 168/173 2.502/2.520 111/112/2.400 112/114 2.610 F1 22 RT 169/172 2.019/2.085 171/174 2.368/2.460 125/128 2.400 118/122 2.613/2.625 Far Cry 6 139/187 2.355 140/148 2.520 94/97 97/101 2.610/2.610 Far 6 Rt 143 /182 2.354/2.355 140/148 2.520/2.520 93/96 2.400 59/625/2.625 Ghostwire: Tokyo 165/212 2.355 169/175/2.520 119/122 2.385 121/122 2.610/2.610/2.610 Ghostwire: Tokyo RT 164/184 2130/2355 161/174 2470/2520 102/107 2400/2400 102/105 2610/2610 Guardians of the Galaxy 168/182 2224/2355 166/172 2513 /2520 112/115 2392/2400 111/114 2610/2610 Guardians of the Galaxy RT 170/171 2029/2055 170/174 2436/2505 121/124 2385/2385 116/118 Spider-Man 70/2610/2610 2.265/2.340 171/173 2.331/2.460 92/102 2.400 104/108 2.610/2.610 Spider-Man Remastered RT 166/169/2.355 169/174 2.450/2.400/2.400 93/100 2.625 Average 163/179 2,198/2,291 164/171 2,417/2,503 102/107 2,397/2,398 103/107 2,612/2,614 All benchmarks in WQHD, * The dropped FPS are shown here as a result of the “lack of memory”

It is not known why Nvidia is doing this – not even when asked. The already complicated construction of five GPU classes with very large TDP bandwidths becomes even more confusing with this approach. Because the fact that an RTX 4070 with 115+25 watts, the supposedly fastest model in the series, would not be slower in many games in the test even with 100 watts, can really no longer be explained to anyone. This is also not in the interest of the OEMs, who have so far advertised with the strongest expansion stages.

Gaming benchmarks

ComputerBase no longer had any notebooks with GeForce RTX 3000 laptop GPUs available for the test like two weeks ago. Because new drivers and in particular game patches ensure that results from the previous year can no longer be compared with current results, a different course had to be used for this test.

Ultimately, ComputerBase decided to test the new laptop GPUs in the course last used for desktop graphics cards – or an excerpt from it. This does not directly show the increase from the 3000 to the 4000 generation in the notebook, but the performance known from the desktop is offered by a current notebook with a GeForce RTX 4000 laptop GPU. It was tested in WQHD and Full HD.

The games used and the graphics details Graphics details in FHD and WQHD Cyberpunk 2077 to Ultra HD Ultra preset, at 5K, medium preset
RT: Ultra Preset Doom Eternal Ultra Nightmare Preset
RT: Reflections “On” F1 22 Ultra High Preset, AO CACAO
RT Shadow + AO + Reflections “High” Far Cry 6 Ultra Preset, HD- Textures
RT: Shadows and reflections “On” Ghostwire: Tokyo Maximum graphic details
RT: Reflections “High” + Culling “High” Guardians of the Galaxy Ultra preset
RT: Reflections “Ultra” + transparent reflections “A” Spider-Man Remastered Maximum graphic details
RT: Resolution + Geometry on “Very High”, view range on “10”

How fast is a separate GPU in the notebook? In many cases, this not only depends on its power loss, but also on whether it outputs the rendered image directly to the (internal) display or via the so-called frame buffer of the iGPU in the processor: The article provides extensive details on this topic Gaming Notebook Benchmarks: Much more FPS in games by one MUX switch click.

To clarify the “problem” again, ComputerBase has the RTX 4090 laptop GPU with 150+25 watts in both “MSHybrid” mode (dGPU can be put to sleep at any time because the display is attached to the iGPU) and in “Discrete GPU” mode “ (dGPU hangs directly on the display). This can be set in the MSI Center. All other mobile GPUs were again measured with a directly connected screen (“dGPU”).

TDP scaling: From 35 to 150 watts

As in the recent review of the GeForce RTX 4080 and RTX 4090 laptop GPU, the editors also have GeForce RTX 4070, 4060 and 4050 laptop GPUs tested over the entire TDP band defined by Nvidia. Shadow Warrior 3 was used again with maximum details in WQHD.

Based on the XMG Neo 17 (E23), the RTX 4070 and 4060 laptop GPUs could be throttled down to 35 watts via the MSI Afterburner, starting from the maximum configuration defined in the notebook of 115 watts (plus 25 watts Dynamic Boost). The 35 watts turned out to be the hard lower limit, below which the performance collapsed. The GeForce RTX 4050 laptop GPU in the Gigabyte Aero 14 OLED, on the other hand, as a 35+10 watt variant, did not allow for higher TDPs, but could be operated with (45 watts) and without (35 watts) Dynamic Boost.

The chart below details the performance history of the five GeForce RTX 4000 laptop GPUs in the game Shadow Warrior 3 in WQHD. The game is one of the titles in which the RTX 4070 and 4060 laptop GPU can reach the maximum TDP of 115 watts. The RTX 4060 even had 120 watts via Dynamic Boost, while the RTX 4070 reached 2,400 MHz with 115 watts.

FPS per watt (Shadow Warrior 3, WQHD, max details) 050100150200250frames per second (FPS) 3035404550556065707580859095100105110115120125130135140145150155160Watt (W)

  • RTX 4090 Laptop GPU (80-150W)
  • RTX 4080 Laptop GPU (60-150W)
  • RTX 4070 Laptop GPU (35-115W)

    • < li class="svgchart__item js-svgchart-item" data-item="385641"> RTX 4060 Laptop GPU (35 – 115W)

  • RTX 4050 Laptop GPU (35 – 115W)
  • not possible (Aero 14 limited to 45W)

    • The following key findings can be derived from the series of measurements:

    1. The mobile GeForce RTX 4090 is even with its Minimum TDP of 80 watts faster than a mobile RTX 4070 – not only in Shadow Warrior 3, but (as the test will show) also in the rest of the course.
    2. The RTX 4080 laptop GPU with only 60 watts also beats the fastest RTX 4070 in this game. However, the following course will also show games in which the RTX 4070 with a TDP of 115 watts is faster.
    3. RTX 4070 & 4060 laptop GPU separate in this title like RTX 4090 & 4080 laptop GPU just over 10 percent. In the overall course it is 17 percent.
    4. The RTX 4050, which could only be operated with 35 and 45 watts in the test, and the RTX 4060 achieve a very similar level of performance in Shadow Warrior 3 – the additional 2 GB of memory in the RTX 4060 may have a negative effect in this case. This disadvantage should have less of an impact with higher power loss.

    WQHD benchmarks with and without RT

    As expected, the gap between the full-fledged flagships RTX 4090 and 4080 laptop GPU at maximum power loss (150+25 watts) and the GeForce RTX 4070 laptop GPU in the largest configuration (115+25 watts) is large: the mobile AD106 GPU ranks 32 percent behind the AD104 variant without RT. Another 15 percent behind is the larger of the two AD107 graphics chips: the RTX 4060 laptop GPU. It is on par with the Radeon RX 6800M (12 GB) with a 120 watt TDP.

    When ray tracing is used, the RTX 4070 falls another 10 percentage points behind the RTX 4080. That's largely due to Doom, Spider-Man Remastered, and Cyberpunk 2077, where the 8GB of storage is tight. In Doom, both the RTX 4070 and RTX 4060 lag behind the Radeon RX 6800M, which can only boast 12GB of memory. This is also noticeable in the overall RT rating, where the 4070 is only 41 percent ahead of the RDNA-2 specimen and the 4060 is only 24 percent ahead.

    It also fits that the RTX 4070 in the fastest configuration without RT is 6 percent faster than the RTX 4080 with 60 watts, including RT it is then 7 percent behind.

    In Doom Eternal and Spider-Man Remastered, the documented results with and without RT were each taken after a game restart, and were even worse when changing settings on the fly. Nvidia's laptop GPUs underline once again that 8 GB is no longer an all-round carefree equipment, even in WQHD.

    Performance Rating WQHD

    Edit Performance Rating WQHD – FPS Input Charts All None

    Unit: Percent Edit Performance Rating WQHD – FPS (RT) Incoming Graphs All None

    Unit: Percent Edit Performance Rating WQHD – Frametimes Input Charts All None

    Unit: percent Edit Performance Rating WQHD – Frametimes (RT) Input Charts All None

    Unit: percent < p class="p text-width">The RTX 4050 laptop GPU is missing in this course because it would be simply too slow with a maximum of 45 watts and even fewer shaders and only 6 GB of memory.

    2,560 × 1,440

    Cyberpunk 2077 – 2,560 × 1,440

  • FPS, 1% percentile:
    • GeForce RTX 4090 FE
      450 W88.0
    • GeForce RTX 4080 FE
      320 W86 ,4
    • GeForce RTX 4070 Ti
      285W, Gigabyte Eagle OC75,5
    • GeForce RTX 3090 Ti
      450 W72,2
    • GeForce RTX 3090
      Asus Strix OC67,6
    • Radeon RX 7900 XTX
      355 W63.5
    • RTX 4080 Laptop (150+25W)
      dGPU, XMG Neo 16 , 528,3759.0
    • GeForce RTX 3080 FE
      320 W56.6
    • Radeon RX 7900 XT
      305 W56.3
    • RTX 4090 Laptop (150+25W)
      MS Hybrid, MSI GT77 Titan, 528,3754.0
    • Radeon RX 6900 XTXH
      XFX Black49.0
    • Radeon RX 6900 XT
      300 W48.2
    • RTX 4090 Laptop (150+25W) II
      dGPU, MSI GT77 Titan, 528.3746.6
    • RTX 4070 Laptop (115+25W)
      dGPU, XMG Neo 17, 528.4945.7
    • Radeon RX 6800 XT
      300 W45.2
    • RTX 4090 Laptop (80W)
      dGPU, MSI GT77 Titanium, 528.3744.5
    • RX 6800M (120 watts)
      dGPU, Corsair Voyager a1600, 23.2.143.0
    • GeForce RTX 3070 FE
      220 W41.7
    • Radeon RX 6800
      250 W39.0
    • GeForce RTX 3060 Ti FE
      230 W35.5
    • Radeon RX 6700 XT
      230 W33.1
    • RTX 4060 Laptop (115+25W)
      dGPU, XMG Neo 17, 528.4932.7
    • RTX 4080 Laptop (105W)
      dGPU, XMG Neo 16, 528.3731.1
    • GeForce RTX 3080 Ti FE
      350 W29.5
    • RTX 4080 Laptop (60W)
      dGPU, XMG Neo 16, 528,3727.9

    • unit : Frames per second (FPS) Cyberpunk 2077 (RT) – 2,560 × 1,440

  • FPS, 1% percentile:
    • GeForce RTX 4090 FE
      450 W54.0
    • GeForce RTX 4080 FE
      320 W43.2
    • GeForce RTX 4070 Ti
      285 W, Gigabyte Eagle OC35.3
    • GeForce RTX 3090 Ti
      450 W34.4
    • GeForce RTX 3090< br>Asus Strix OC32.5
    • RTX 4090 Laptop (150+25W)
      MSHybrid, MSI GT77 Titan, 528.3730.8
    • RTX 4090 Laptop (150+25W) II
      dGPU, MSI GT77 Titanium, 528,3730.2
    • GeForce RTX 3080 Ti FE
      350 W29.5
    • GeForce RTX 3080 FE< br>320 W26.9
    • RTX 4080 Laptop (150+25W)
      dGPU, XMG Neo 16, 528.3726.4

      • < li class="chart__row">Radeon RX 7900 XTX
      355 W23.0

  • Radeon RX 7900 XT
    305 W21.8
  • RTX 4090 Laptop (80W)
    dGPU, MSI GT77 Titanium, 528.3720.7
  • Radeon RX 6900 XTXH
    XFX Black19.5
  • RTX 4080 Laptop (105W)
    dGPU, XMG Neo 16, 528,3718.8

    • li>

  • Radeon RX 6900 XT
    300 W18.7
  • GeForce RTX 3070 FE
    220 W17.3
  • Radeon RX 6800 XT
    300 W17.2
  • GeForce RTX 3060 Ti FE
    230 W17.1
  • Radeon RX 6800
    250 W15.2
  • RTX 4080 Laptop (60W)
    dGPU, XMG Neo 16, 528,3714.9
  • RTX 4070 Laptop (115+25W)
    dGPU, XMG Neo 17, 528,4912.8

    • li>

  • RTX 4060 Laptop (115+25W)
    dGPU, XMG Neo 17, 528,4910.9
  • Radeon RX 6700 XT
    230 W9.4
  • RX 6800M (120 Watt)
    dGPU, Corsair Voyager a1600, 23.2.18.6< /li>

    • Unit: frames per second (FPS) Doom Eternal – 2560 × 1440

    Unit: frames per second (FPS) Doom Eternal (RT) – 2560 × 1440

    Unit: frames per second (FPS) F1 22 – 2,560 × 1,440

    Unit: frames per second (FPS) F1 22 (RT) – 2,560 × 1,440

    Unit: frames per second (FPS) Far Cry 6 – 2560 × 1440

  • FPS, 1% percentile:
    • Radeon RX 7900 XTX
      355 W117.5
    • Radeon RX 7900 XT
      305 W117.4
    • GeForce RTX 4090 FE
      450 W111.0
    • GeForce RTX 4080 FE
      320 W103.2
    • GeForce RTX 4070 Ti
      285W, Gigabyte Eagle OC101,0
    • Radeon RX 6900 XTXH
      XFX Black100,1
    • GeForce RTX 3090 Ti
      450 W99,0
    • GeForce RTX 3090
      Asus Strix OC98,5
    • GeForce RTX 3080 Ti FE
      350 W97, 0
    • Radeon RX 6900 XT
      300 W95.2
    • Radeon RX 6800 XT
      300 W92.7
    • RTX 4080 Laptop (150+25W)
      dGPU, XMG Neo 16, 528,3792.5
    • GeForce RTX 3080 FE
      320 W90.5
    • RTX 4090 Laptop (150+25W) II
      dGPU, MSI GT77 Titan, 528,3789.8
    • RTX 4090 Laptop (150+25W)
      MSHybrid, MSI GT77 Titan, 528,3784.1

      • li>

    • RTX 4080 Laptop (105W)
      dGPU, XMG Neo 16, 528,3782.4
    • Radeon RX 6800
      250 W78.6
    • RTX 4090 Laptop (80W)
      dGPU, MSI GT77 Titanium, 528,3770.2
    • RTX 4080 Laptop (60W)
      dGPU, XMG Neo 16, 528,3767.7
    • Radeon RX 6700 XT
      230 W66.0
    • RTX 4070 Laptop (115+25W)
      dGPU , XMG Neo 17, 528,4962.2
    • RTX 4060 Laptop (115+25W)
      dGPU, XMG Neo 17, 528,4955.6
    • RX 6800M (120 watts)
      dGPU, Corsair Voyager a1600, 23.2.154.3
    • GeForce RTX 3070 FE
      220 W53.9
    • GeForce RTX 3060 Ti FE
      230 W52.7

    • Unit: frames per second (FPS) Far Cry 6 (RT) – 2,560 × 1,440

    Unit: frames per second (FPS) Ghostwire Tokyo – 2,560 × 1.440

  • FPS, 1% percentile:
    • Radeon RX 7900 XTX
      355 W81.7
    • Radeon RX 7900 XT
      305 W81.1
    • Radeon RX 6900 XTXH
      XFX Black81.0
    • GeForce RTX 4090 FE
      450 W80.0
    • GeForce RTX 4080 FE
      320 W80.0
    • GeForce RTX 3090 Ti
      450 W78.8
    • Radeon RX 6900 XT
      300 W78.6
    • GeForce RTX 3090
      Asus Strix OC78.0
    • GeForce RTX 3080 Ti FE
      350 W77.9
    • GeForce RTX 4070 Ti
      285 W, Gigabyte Eagle OC76.4
    • Radeon RX 6800 XT
      300 W71.7
    • GeForce RTX 3080 FE
      320 W71.3
    • RTX 4090 Laptop (150+25W) II
      dGPU, MSI GT77 Titanium, 528.3768.5
    • RTX 4080 Laptop (105W)
      dGPU, XMG Neo 16, 528.3764.6
    • RTX 4080 Laptop (150+25W)
      dGPU, XMG Neo 16, 528.3763.3
    • RTX 4090 Laptop (150+25W)MSHybrid, MSI GT77 Titanium, 528.3762.7
    • Radeon RX 6800
      250 W60.5
    • GeForce RTX 3070 FE
      220 W55.1
    • Radeon RX 6700 XT
      230 W47.9
    • RTX 4080 Laptop ( 60W)
      dGPU, XMG Neo 16, 528.3746.5
    • RTX 4070 Laptop (115+25W)
      dGPU, XMG Neo 17, 528,4946.2
    • GeForce RTX 3060 Ti FE
      230 W44.5
    • RX 6800M (120 watts)
      dGPU, Corsair Voyager a1600, 23.2.142.0
    • < strong>RTX 4060 Laptop (115+25W)
      dGPU, XMG Neo 17, 528,4940.8

    • Unit: frames per second (FPS) Ghostwire Tokyo (RT) – 2,560 × 1,440

  • FPS, 1% percentile:
    • GeForce RTX 4090 FE
      450 W59.7
    • GeForce RTX 4080 FE
      320 W54.2
    • GeForce RTX 4070 Ti
      285W, Gigabyte Eagle OC48.1
    • GeForce RTX 3090 Ti
      450 W48.1
    • Radeon RX 7900 XTX
      355 W47,9
    • GeForce RTX 3090
      Asus Strix OC46,4
    • GeForce RTX 3080 Ti FE
      350 W43.6
    • Radeon RX 7900 XT
      305 W43.6
    • RTX 4090 Laptop (150+25W)
      MS Hybrid, MSI GT77 Titan, 528.3742.1
    • RTX 4090 Laptop (150+25W) II
      dGPU, MSI GT77 Titan, 528.3740.8
    • RTX 4080 Laptop (150+25W)dGPU, XMG Neo 16, 528,3739.9
    • GeForce RTX 3080 FE
      320 W36.8
    • < strong>RTX 4080 Laptop (105W)
      dGPU, XMG Neo 16, 528,3734.1
    • Radeon RX 6900 XTXH
      XFX Black32,4
    • Radeon RX 6900 XT
      300 W30,6
    • Radeon RX 6800 XT
      300 W28.3
    • GeForce RTX 3070 FE
      220 W26.6
    • RTX 4080 Laptop (60W )
      dGPU, XMG Neo 16, 528.3724.5
    • RTX 4070 Laptop (115+25W)
      dGPU, XMG Neo 17, 528.4924.3
    • GeForce RTX 3060 Ti FE
      230 W22.9
    • RTX 4060 Laptop (115+25W)
      dGPU, XMG Neo 17, 528,4921.7
    • Radeon RX 6800
      250 W19.9< /li>
    • Radeon RX 6700 XT
      230 W15.5
    • RX 6800M (120 watts)
      dGPU, Corsair Voyager a1600, 23.2.114.7

    • Unit: frames per second (FPS) Guardians of the Galaxy – 2560 × 1440

    Unit: frames per second (FPS) Guardians of the Galaxy (RT ) – 2,560 × 1,440

  • FPS, 1% percentile:
    • GeForce RTX 4090 FE
      450 W129.6
    • GeForce RTX 4080 FE
      320 W101.0
    • GeForce RTX 3090 Ti
      450 W81.1
    • RTX 4090 Laptop (150+25W)
      MSHybrid, MSI GT77 Titan , 528.3780.7
    • GeForce RTX 4070 Ti
      285 W, Gigabyte Eagle OC80.2
    • RTX 4090 Laptop ( 150+25W) II
      dGPU, MSI GT77 Titanium, 528.3777.9
    • GeForce RTX 3090
      Asus Strix OC77,3
    • GeForce RTX 3080 Ti FE
      350 W72,8
    • GeForce RTX 3080 FE
      320 W66.2
    • RTX 4080 Laptop (150+25W)
      dGPU, XMG Neo 16, 528.3760.2
    • RTX 4080 Laptop (105W)
      dGPU, XMG Neo 16, 528.3753.9
    • GeForce RTX 3070 FE
      220 W48.8
    • RTX 4070 Laptop (115+25W)
      dGPU, XMG Neo 17, 528,4944.1
    • Radeon RX 6900 XTXH
      XFX Black43.3
    • GeForce RTX 3060 Ti FE< br>230 W43.2
    • RTX 4080 Laptop (60W)
      dGPU, XMG Neo 16, 528.3741.6
    • Radeon RX 6900 XT
      300 W40.4
    • RTX 4060 Laptop (115+25W)
      dGPU, XMG Neo 17, 528.4938.0
    • Radeon RX 6800 XT
      300 W37.8
    • Radeon RX 7900 XTX
      355 W36.0
    • Radeon RX 7900 XT
      305 W35.7
    • Radeon RX 6800< br>250 W31.6
    • Radeon RX 6700 XT
      230 W24.5
    • RX 6800M (120 Watt)
      dGPU, Corsair Voyager a1600, 23.2.122.7

    • Unit: frames per second (FPS) Spider-Man Remastered – 2560×1440

    Unit: frames per second (FPS) Spider Man Remastered (RT) – 2,560 × 1,440

  • FPS, 1% percentile:
    • GeForce RTX 4090 FE
      450 W97.8
    • GeForce RTX 4080 FE
      320 W91.1
    • Radeon RX 7900 XTX
      355 W83.3
    • Radeon RX 7900 XT
      305 W78.9
    • GeForce RTX 3090 Ti
      450 W76,8
    • GeForce RTX 4070 Ti
      285 W, Gigabyte Eagle OC76,7
    • GeForce RTX 3090
      Asus Strix OC72,7
    • RTX 4090 Laptop (150+25W)
      MS Hybrid, MSI GT77 Titanium, 528.3771.5
    • RTX 4090 Laptop (150+25W) II
      dGPU, MSI GT77 Titanium, 528.3770.9< /li>
    • GeForce RTX 3080 Ti FE
      350 W68.9
    • RTX 4080 Laptop (150+25W)
      dGPU, XMG Neo 16, 528,3763.1
    • GeForce RTX 3080 FE
      320 W61.4
    • Radeon RX 6900 XTXH
      XFX Black56,5
    • Radeon RX 6900 XT
      300 W53.2
    • Radeon RX 6800 XT
      300 W49.9
    • Radeon RX 6800
      250 W44.3< /li>
    • RTX 4080 Laptop (60W)
      dGPU, XMG Neo 16, 528.3742.3
    • GeForce RTX 3070 FE
      220 W38.1
    • Radeon RX 6700 XT
      230 W34.5
    • GeForce RTX 3060 Ti FE< br>230 W34.1
    • RTX 4070 Laptop (115+25W)
      dGPU, XMG Neo 17, 528,4933.4
    • RX 6800M (120 Watt)
      dGPU, Corsair Voyager a1600, 23.2.131.0
    • RTX 4060 Laptop (115+25W)
      dGPU, XMG Neo 17, 528,4930.2
    • < strong>RTX 4080 Laptop (105W)
      dGPU, XMG Neo 16, 528,3728.9

    • Unit: frames per second (FPS)

    FHD benchmarks with and without RT

    In Full HD without RT, the distances shrink a bit and 4070 and 4060 can hold their own even better against the Radeon RX 6800M. This course also includes the RTX 4050 laptop GPU with 35+10 watts and the Radeon RX 6800S 8 GB with 80 watts, the Radeon leaves the field with 77 percent higher consumption with a 44 percent lead as the winner – in the chassis of a Gigabyte Aero 14 but there would be no place for this variant.

    Performance Rating FHD

    Edit Performance Rating FHD – FPS Input Charts All None

    Unit: Percent Edit Performance Rating FHD – FPS (RT) Inflow Charts All None

    Unit: Percent Edit Performance Rating FHD – Frametimes Input Charts All None

    Unit: Percent Edit Performance Rating FHD – Frametimes (RT) Incoming Graphs All None

    Unit: percent

    With ray tracing, the RTX 4070 and 4060 laptop GPUs stand out better against the Radeon RX 6800M in FHD, but in games like Doom Eternal and Spider-Man the 8 GB remains a shortcoming. However, this hits the Radeon RX 6800S even harder, which sometimes mercilessly thirsts for memory when using raytracing. The GeForce RTX 4050 laptop GPU with 6 GB is also out.

    1,920 × 1,080

    Cyberpunk 2077 – 1,920 × 1,080

    Unit: frames per second (FPS) Cyberpunk 2077 (RT) – 1920 × 1080

    Unit: frames per second (FPS) Doom Eternal – 1920 × 1080

    Unit: frames per second ( FPS) Doom Eternal (RT) – 1920 × 1080

    Unit: frames per second (FPS) F1 22 – 1,920 × 1,080

    Unit: frames per second (FPS) F1 22 (RT) – 1,920 × 1,080

    Unit: frames per second (FPS) Far Cry 6 – 1,920 × 1,080

    Unit: frames per second (FPS) Far Cry 6 (RT) – 1,920 × 1,080

    Unit: frames per second (FPS) Ghostwire Tokyo – 1,920 × 1,080

    Unit: frames per second (FPS) Ghostwire Tokyo ( RT) – 1920 × 1080

    Unit: frames per second (FPS) Guardians of the Galaxy – 1,920 × 1,080

    Unit: frames per second (FPS) Guardians of the Galaxy (RT) – 1920 × 1080

    Unit: frames per second (FPS) Spider-Man Remastered – 1920 × 1080

    Unit: frames per second (FPS) Spider-Man Remastered (RT) – 1,920 × 1,080

    Unit: frames per second (FPS)

    Higher FPS through DLSS

    With frame generation, Nvidia offers a second, AI-supported technology exclusively for the GeForce RTX 4000 (laptop GPUs) in addition to classic upsampling.

    Because in contrast to DLSS Super Resolution (formerly DLSS 2) cannot positively influence the input lag, it should not be used if the initial FPS level does not offer a noticeably good gaming experience. However, it can be used, especially in the CPU limit, to take advantage of the high refresh rates of current gaming notebooks.

    The following example shows how the GeForce RTX 4000 laptop GPUs use DLSS 3 (DLSS 2 + Frame -Generation) are able to partially double the FPS.

    While changing the settings on the RTX 4090 with 16 GB and RTX 4080 with 12 GB could be done on the fly, Spider-Man on the RTX 4070 and 4060 had to be restarted between runs in order to achieve wildly fluctuating results at a lower level to prevent – the memory, which is too small for WQHD with maximum details at 8 GB, was shown here again.

    Spider-Man Remastered with DLSS2/3

    Spider-Man Remastered with DLSS2/3 – FPS

    Unit: Frames Per Second (FPS) Spider-Man Remastered with DLSS2/3 – Frametimes

    Unit: frames per second (FPS)

    Synthetic benchmarks

    Finally, a handful of synthetic ones for information Benchmarks, in this case also with other high-end mobile GPUs from Nvidia and AMD.

    3DMark

    3DMark – TimeSpy

  • Graphics result:
    • RTX 4090 Laptop GPU
      dGPU, 150+25W, 13950HX22,363.0
    • RTX 4080 Laptop GPU
      dGPU, 150+25W, 13700HX19,648.0
    • RTX 3080 Ti Laptop GPU Boost
      dGPU, 150+25W, 12900HK13,533.0
    • RTX 3080 Ti Laptop GPU
      dGPU, 150W, 12900HK12,632.0
    • RTX 4070 Laptop GPU
      dGPU, 115+25W, 13900HX12,606.0
    • Radeon RX 6800M
      dGPU, ~150W, 5900HX12,235.0
    • RTX 3080 Laptop GPU Boost
      dGPU, 150+15W, 11800H11822.0
    • RTX 3080 Laptop GPU
      dGPU, 150W, 11800H11302.0
    • RTX 4060 Laptop GPU
      dGPU, 115+25W, 13900HX11008.0
    • Radeon RX 6800M III
      dGPU, ~120W, 6900HS, 23.2.110.504.0
    • RTX 4050 Laptop GPU
      dGPU, 45W, 13700H6 .371.0
  • CPU result:
    • RTX 4060 Laptop GPU
      dGPU, 115+25W, 13900HX15,631.0
    • RTX 4080 Laptop GPU
      dGPU, 150 +25W, 13700HX15105.0
    • RTX 4070 Laptop GPU
      dGPU, 115+25W, 13900HX15,010.0
    • RTX 4090 Laptop GPU
      dGPU, 150+25W, 13950HX14,919.0

      • li>

    • RTX 3080 Ti Laptop GPU Boost
      dGPU, 150+25W, 12900HK12,135.0
    • RTX 3080 Ti Laptop GPU
      dGPU, 150W, 12900HK12,039.0
    • Radeon RX 6800M III
      dGPU, ~120W, 6900HS, 23.2,110,393.0
    • RTX 3080 Laptop GPU
      dGPU, 150W, 11800H10,099.0
    • RTX 3080 Laptop GPU Boost
      dGPU, 150+15W, 11800H10,084.0
    • Radeon RX 6800M
      dGPU, ~150W, 5900HX9,368.0
    • RTX 4050 Laptop GPU
      dGPU, 45W, 13700H7,535.0

    • Unit: Frames Per Second (FPS) 3DMark – Port Royal

    Unit: frames per second (FPS) Superposition Benchmark – FHD Extreme

    Unit: frames per second (FPS)

    Application -Benchmarks

    Ada Lovelace also proves its strength in applications in the notebook, although the performance gap between the five classes and – if the CPU also plays a role – in comparison to the desktop PC fails or can fail.

    Charts

    Blender Benchmark 3.4.0

    Unit: Points 8 entries Adobe Premiere Pro 22.5

    Unit: minutes, seconds 8 entries Agisoft Metashape 1.8.4

    Unit: Minutes, Seconds 8 entries HandBrake Nightly – 2160p60 H.264 to 2160p30 H.265

    Unit: minutes, seconds

    Conclusion

    Nvidia is carrying out the model update in the notebook faster than ever before: From the RTX 4090 to the RTX 4050, the group has updated the entire range of mobile GeForce RTX graphics chips in just two weeks.

    At the head of the series, the RTX 4090 Laptop GPU and RTX 4080 Laptop GPU are two extremely fast, extremely efficient variants that are only available in systems from 2,600 euros (4080) and 4,300 euros (4090) respectively. With a maximum TDP of 150 watts, the RTX 4090 laptop GPU has almost the performance of a GeForce RTX 4070 Ti. The mobile 4080 is only a little over 10 percent behind, although the difference in the shaders is over 30 percent. However, one should not deduce from this that the new GeForce RTX 4070 laptop GPU is also closer to the RTX 4080 laptop GPU in practice than the data sheet suggests.

    Nvidia GeForce RTX 4090, 4080, 4070, 4060 & 4050 laptop GPUs in notebooks from Gigabyte, MSI and XMG tested

    The test confirms the assumption that there is a large performance gap between the RTX 4080 and 4070 in the notebook: In WQHD without ray tracing, the 4070 is 32 percent slower, in FHD it is still 21 percent. With ray tracing, there are even 42 and 36 percent between the two series. In some titles, the 8 GB memory (RTX 4080 laptop: 12 GB) still plays a role in the 70s class.

    The gap is so large that the RTX 4090 and RTX 4080 laptop GPU, even with 80 and 60 watts respectively, are still faster in many games than the fastest variant of the RTX 4070 – this is more often the case with active ray tracing than without. So they exist, the titles in which an RTX 4070 beats an RTX 4080 in the notebook. The RTX 4060 laptop GPU no longer succeeds in this in the WQHD test course. With the same power loss, it is on average 17 percent slower than the 4070 with a third fewer shaders but 10 percent higher clock rates on the identical memory interface in both the WQHD and FHD benchmark courses. Depending on the game, the distance can also be significantly smaller or significantly larger. High TDP RTX 4060 laptop GPUs beating low TDP RTX 4070 laptop GPUs are possible – at the top end of the scale the 4060 requires 30 watts, at the bottom end of the scale only 15 watts more (Shadow Warrior 3).

    RTX 4070 and 4060 laptop GPUs have in common that they didn't go through the WQHD test course with maximum details as carefree as the RTX 4090 and 4080 laptop GPUs did recently. The reason is the 8 GB VRAM, which is a showstopper in Doom Eternal including RT, but also gnaws at both graphics chips in Spider-Man Remastered and Cyberpunk 2077. Raw performance is okay, but you won't experience the smooth gameplay of a 12GB RTX 4080 laptop GPU or a 16GB RTX 4090 laptop GPU at the same maximum settings. In Doom Eternal, both models still have feathers even in Full HD. To get consistent benchmark results, the game had to be restarted after changing settings.

    The competition can at least benefit from this in the form of the 12 GB Radeon RX 6800M, while the Radeon RX 6800S with 8 GB does even worse on the course, once again underlining that Nvidia knows how to do more with the same storage capacity. Nevertheless, it stays the same: 8 GB of graphics memory is often not enough in current games in WQHD with maximum details or when ray tracing is used, even if the raw GPU performance is sufficient for the selected settings.

    Nvidia has given the RTX 4050 laptop GPU the same 2 GB increase in storage capacity as the RTX 4060, in this case from 4 to 6 GB. That's enough to go through the Full HD test course without ray tracing (Doom Eternal refuses “Ultra Nightmare”). The 45-watt GeForce RTX 4050 used by Gigabyte in the Aero 14 OLED then achieves 50 percent of the performance of the RTX 4070 laptop GPU with maximum power loss. That's enough to be able to play current titles (with reduced details) in Full HD with 60 FPS. However, ray tracing and WQHD are not an option.

    With the RTX 4050 laptop GPU, however, Nvidia and the OEMs are primarily targeting a different market segment: that of compact yet powerful notebooks for creative professionals. With the smallest mobile GeForce RTX 4000, you get all the features of the big variants, including AV1 decoding and encoding as well as ray tracing and tensor cores, which can be used in more and more applications. There, the ultra-compact Aero 14 OLED is undoubtedly faster than a pure iGPU notebook.

    The TDP classification of the three smaller GeForce RTX 4000 laptop GPUs caused additional confusion in the test (and in case of doubt also with the customer in the future). As it turned out, RTX 4070 and 4060 laptop GPUs do not use the maximum configurable power loss in many games because they already reach the maximum usable clock. While an RTX 4090 or 4080 laptop GPU with “advertised” 150+25 watts is always faster than a variant with 125 watts, this is not necessarily the case with the 4060 and 4070: On average in the WQHD course, 100 watts were enough for both models, um to call up the maximum possible performance – 40 watts less than are actually allowed in the XMG Neo 17 (E23) with 115+25 watts. More confusion is always possible.

    Not only the performance, but also the price difference between 4090 and 4080 on the one hand and 4070, 4060 and 4050 on the other hand is large. In contrast to the RTX 4090 laptop GPU (currently only from 4,300 euros in the price comparison) and RTX 4080 laptop GPU (from 2,600 euros), there are notebooks with the three smaller versions from 1,200 euros, 1,350 euros and 1,670 euros respectively. The offer in the golden mean is not available in this generation for the time being.

    Variant Price Notebooks with RTX 4090 laptop GPU, 16 GB from 4,300 euros Notebooks with RTX 4080 laptop GPU, 12 GB from 2,600 euros Notebooks with RTX 4070 laptop GPU, 8 GB from 1,670 euros Notebooks with RTX 4060 laptop GPU, 8 GB from 1,350 euros Notebooks with RTX 4050 laptop GPU, 6 GB from 1,200 euros

    Before buying (after the mandatory look at the power loss in the respective offer!), you should also take a look at the predecessors and the competition. Notebooks with the Radeon RX 6800M are available from 1,600 euros and are therefore in the price range of the new GeForce RTX 4070 laptop GPU models. Creators should generally prefer the GeForce, but the RDNA 2 example is still an alternative for gamers, even if it is a bit slower on average in the rasterizer course and significantly slower in the ray tracing course. On the other hand, this class with 12 GB of memory did not record a memory-related total failure in the test. In the end, however, the rest of the equipment should also speak in favor of the Radeon variant.

    Overall, the GeForce RTX 4070, 4060 and 4050 also deliver more performance than their respective predecessors. However, the three smaller representatives do not come close to the presentation of the two flagships and this is not only, but to a large extent, due to the built-in memory. Ultimately, this shows how dominant Nvidia still is in the notebook sector – and RDNA 3 should change little or nothing this year.

    ComputerBase became the XMG Neo 17 (E23) with RTX 4070 laptop GPU, the XMG Neo 17 (23) with RTX 4060 laptop GPU and the Gigabyte Aero 14 OLED with RTX 4050 laptop GPU on loan from the respective manufacturer under NDA. There was no press driver in this case, the publicly available GeForce 528.49 already supports the new mobile graphics chip. The only requirement was the earliest possible publication date. The manufacturer did not influence the test report, there was no obligation to publish it.

    This article was interesting, helpful or both? The Editors are happy about any support from ComputerBase Pro and disabled ad blockers. More about ads on ComputerBase.

    Exit mobile version