GeForce RTX 5070 vs RTX 3090 Ti
Aggregate performance score
We've compared GeForce RTX 3090 Ti and GeForce RTX 5070, covering specs and all relevant benchmarks.
RTX 3090 Ti outperforms RTX 5070 by a minimal 2% based on our aggregate benchmark results.
Contents
Primary details
GPU architecture, market segment, value for money and other general parameters compared.
| Place in the ranking | 19 | 22 |
| Place by popularity | not in top-100 | 12 |
| Cost-effectiveness evaluation | 17.60 | 74.34 |
| Power efficiency | 12.07 | 21.33 |
| Architecture | Ampere (2020−2025) | Blackwell 2.0 (2025−2026) |
| GPU code name | GA102 | GB205 |
| Market segment | Desktop | Desktop |
| Release date | 27 January 2022 (4 years ago) | 4 March 2025 (1 year ago) |
| Launch price (MSRP) | $1,999 | $549 |
Cost-effectiveness evaluation
The higher the ratio, the better. We use the manufacturer's recommended prices.
RTX 5070 has 322% better value for money than RTX 3090 Ti.
Performance to price scatter graph
Currently popular graphics cards are shown for comparison.
Detailed specifications
General parameters such as number of shaders, GPU core base clock and boost clock speeds, manufacturing process, texturing and calculation speed. Real power consumption of some graphics cards can well exceed their nominal TDP, especially if overclocked.
| Pipelines / CUDA cores | 10752 | 6144 |
| Core clock speed | 1560 MHz | 2325 MHz |
| Boost clock speed | 1860 MHz | 2512 MHz |
| Number of transistors | 28,300 million | 31,100 million |
| Manufacturing process technology | 8 nm | 5 nm |
| Power consumption (TDP) | 450 Watt | 250 Watt |
| Texture fill rate | 625.0 | 482.3 |
| Floating-point processing power | 40 TFLOPS | 30.87 TFLOPS |
| ROPs | 112 | 80 |
| TMUs | 336 | 192 |
| Tensor Cores | 336 | 192 |
| Ray Tracing Cores | 84 | 48 |
| L1 Cache | 10.5 MB | 6 MB |
| L2 Cache | 6 MB | 48 MB |
Form factor & compatibility
Information on compatibility with other computer components. Useful when choosing a future computer configuration or upgrading an existing one. For desktop graphics cards it's interface and bus (motherboard compatibility), additional power connectors (power supply compatibility).
| Interface | PCIe 4.0 x16 | PCIe 5.0 x16 |
| Length | 336 mm | 245 mm |
| Width | 3-slot | 2-slot |
| Supplementary power connectors | 1x 16-pin | 1x 16-pin |
VRAM capacity and type
Parameters of VRAM installed: its type, size, bus, clock and resulting bandwidth. Integrated GPUs have no dedicated video RAM and use a shared part of system RAM.
| Memory type | GDDR6X | GDDR7 |
| Maximum RAM amount | 24 GB | 12 GB |
| Memory bus width | 384 Bit | 192 Bit |
| Memory clock speed | 1313 MHz | 1750 MHz |
| Memory bandwidth | 1.01 TB/s | 672.0 GB/s |
| Shared memory | - | - |
| Resizable BAR | + | + |
Connectivity and outputs
This section shows the types and number of video connectors on each GPU. The data applies specifically to desktop reference models (for example, NVIDIA’s Founders Edition). OEM partners often modify both the number and types of ports. On notebook GPUs, video‐output options are determined by the laptop’s design rather than the graphics chip itself.
| Display Connectors | 1x HDMI 2.1, 3x DisplayPort 1.4a | 1x HDMI 2.1b, 3x DisplayPort 2.1b |
| HDMI | + | + |
API and SDK support
List of supported 3D and general-purpose computing APIs, including their specific versions.
| DirectX | 12 Ultimate (12_2) | 12 Ultimate (12_2) |
| Shader Model | 6.8 | 6.8 |
| OpenGL | 4.6 | 4.6 |
| OpenCL | 3.0 | 3.0 |
| Vulkan | 1.4 | 1.4 |
| CUDA | 8.6 | 12.0 |
| DLSS | + | + |
Synthetic benchmarks
Non-gaming benchmark results comparison. The combined score is measured on a 0-100 point scale.
Combined synthetic benchmark score
This is our combined benchmark score.
Passmark
This is the most ubiquitous GPU benchmark. It gives the graphics card a thorough evaluation under various types of load, providing four separate benchmarks for Direct3D versions 9, 10, 11 and 12 (the last being done in 4K resolution if possible), and few more tests engaging DirectCompute capabilities.
3DMark 11 Performance GPU
3DMark 11 is an obsolete DirectX 11 benchmark by Futuremark. It used four tests based on two scenes, one being few submarines exploring the submerged wreck of a sunken ship, the other is an abandoned temple deep in the jungle. All the tests are heavy with volumetric lighting and tessellation, and despite being done in 1280x720 resolution, are relatively taxing. Discontinued in January 2020, 3DMark 11 is now superseded by Time Spy.
3DMark Fire Strike Graphics
Fire Strike is a DirectX 11 benchmark for gaming PCs. It features two separate tests displaying a fight between a humanoid and a fiery creature made of lava. Using 1920x1080 resolution, Fire Strike shows off some realistic graphics and is quite taxing on hardware.
3DMark Cloud Gate GPU
Cloud Gate is an outdated DirectX 11 feature level 10 benchmark that was used for home PCs and basic notebooks. It displays a few scenes of some weird space teleportation device launching spaceships into unknown, using fixed resolution of 1280x720. Just like Ice Storm benchmark, it has been discontinued in January 2020 and replaced by 3DMark Night Raid.
3DMark Ice Storm GPU
Ice Storm Graphics is an obsolete benchmark, part of 3DMark suite. Ice Storm was used to measure entry level laptops and Windows-based tablets performance. It utilizes DirectX 11 feature level 9 to display a battle between two space fleets near a frozen planet in 1280x720 resolution. Discontinued in January 2020, it is now superseded by 3DMark Night Raid.
Gaming performance
Let's see how good the compared graphics cards are for gaming. Particular gaming benchmark results are measured in FPS.
Average FPS across all PC games
Here are the average frames per second in a large set of popular games across different resolutions:
| Full HD | 210 | 217
+3.3%
|
| 1440p | 142
+14.5%
| 124 |
| 4K | 101
+31.2%
| 77 |
Cost per frame, $
| 1080p | 9.52 | 2.53
+73.4%
|
| 1440p | 14.08 | 4.43
+68.5%
|
| 4K | 19.79 | 7.13
+64%
|
- RTX 5070 has 73% lower cost per frame in 1080p
- RTX 5070 has 69% lower cost per frame in 1440p
- RTX 5070 has 64% lower cost per frame in 4K
FPS performance in popular games
Full HD
Low
| Atomic Heart | 210−220
+1.9%
|
210−220 |
| Counter-Strike 2 | 300−350
+0.6%
|
300−350 |
| Cyberpunk 2077 | 219
+25.1%
|
170−180 |
| Resident Evil 4 Remake | 210−220
+1.9%
|
210−220 |
Full HD
Medium
| Atomic Heart | 210−220
+1.9%
|
210−220 |
| Battlefield 5 | 180−190
+1.1%
|
180−190 |
| Counter-Strike 2 | 300−350
+0.6%
|
300−350 |
| Cyberpunk 2077 | 201
+14.9%
|
170−180 |
| Far Cry 5 | 200−210 | 322
+57.1%
|
| Fortnite | 300−350 | 300−350 |
| Forza Horizon 4 | 280−290
+2.2%
|
270−280 |
| PLAYERUNKNOWN'S BATTLEGROUNDS | 170−180 | 170−180 |
| Valorant | 400−450
+2.9%
|
400−450 |
Full HD
High
| Atomic Heart | 210−220
+1.9%
|
210−220 |
| Battlefield 5 | 180−190
+1.1%
|
180−190 |
| Counter-Strike 2 | 300−350
+0.6%
|
300−350 |
| Counter-Strike: Global Offensive | 270−280 | 270−280 |
| Cyberpunk 2077 | 173 | 170−180
+1.2%
|
| Dota 2 | 217
+3.3%
|
210−220 |
| Far Cry 5 | 200−210 | 306
+49.3%
|
| Fortnite | 300−350 | 300−350 |
| Forza Horizon 4 | 280−290
+2.2%
|
270−280 |
| Grand Theft Auto V | 170 | 170−180
+1.2%
|
| Metro Exodus | 178 | 170−180 |
| PLAYERUNKNOWN'S BATTLEGROUNDS | 170−180 | 170−180 |
| The Witcher 3: Wild Hunt | 394 | 436
+10.7%
|
| Valorant | 400−450
+2.9%
|
400−450 |
Full HD
Ultra
| Battlefield 5 | 180−190
+1.1%
|
180−190 |
| Cyberpunk 2077 | 152 | 170−180
+15.1%
|
| Dota 2 | 195
+2.6%
|
190−200 |
| Far Cry 5 | 200−210 | 290
+41.5%
|
| Forza Horizon 4 | 280−290
+2.2%
|
270−280 |
| PLAYERUNKNOWN'S BATTLEGROUNDS | 170−180 | 170−180 |
| The Witcher 3: Wild Hunt | 193 | 210
+8.8%
|
| Valorant | 400−450
+2.9%
|
400−450 |
Full HD
Epic
| Fortnite | 300−350 | 300−350 |
1440p
High
| Counter-Strike 2 | 220−230
+2.3%
|
210−220 |
| Counter-Strike: Global Offensive | 500−550 | 500−550 |
| Grand Theft Auto V | 151
+2.7%
|
140−150 |
| Metro Exodus | 125
+3.3%
|
120−130 |
| PLAYERUNKNOWN'S BATTLEGROUNDS | 170−180 | 170−180 |
| Valorant | 450−500 | 450−500 |
1440p
Ultra
| Battlefield 5 | 180−190
+2.8%
|
180−190 |
| Cyberpunk 2077 | 104
+2%
|
100−110 |
| Far Cry 5 | 180−190 | 222
+21.3%
|
| Forza Horizon 4 | 240−250
+2.9%
|
240−250 |
| The Witcher 3: Wild Hunt | 180−190
+8.4%
|
166 |
1440p
Epic
| Fortnite | 150−160 | 150−160 |
4K
High
| Atomic Heart | 65−70
+3%
|
65−70 |
| Counter-Strike 2 | 100−105
+3.1%
|
95−100 |
| Grand Theft Auto V | 181
+7.1%
|
160−170 |
| Metro Exodus | 84
+5%
|
80−85 |
| The Witcher 3: Wild Hunt | 173
+15.3%
|
150 |
| Valorant | 300−350 | 300−350 |
4K
Ultra
| Battlefield 5 | 130−140
+2.3%
|
130−140 |
| Counter-Strike 2 | 100−105
+3.1%
|
95−100 |
| Cyberpunk 2077 | 53
+8.2%
|
45−50 |
| Dota 2 | 184
+2.2%
|
180−190 |
| Far Cry 5 | 120−130
+4.3%
|
116 |
| Forza Horizon 4 | 200−210
+4.1%
|
190−200 |
| PLAYERUNKNOWN'S BATTLEGROUNDS | 95−100 | 95−100 |
4K
Epic
| Fortnite | 75−80 | 75−80 |
This is how RTX 3090 Ti and RTX 5070 compete in popular games:
- RTX 5070 is 3% faster in 1080p
- RTX 3090 Ti is 15% faster in 1440p
- RTX 3090 Ti is 31% faster in 4K
Here's the range of performance differences observed across popular games:
- in Cyberpunk 2077, with 1080p resolution and the Low Preset, the RTX 3090 Ti is 25% faster.
- in Far Cry 5, with 1080p resolution and the Medium Preset, the RTX 5070 is 57% faster.
All in all, in popular games:
- RTX 3090 Ti performs better in 35 tests (59%)
- RTX 5070 performs better in 9 tests (15%)
- there's a draw in 15 tests (25%)
Pros & cons summary
| Performance score | 69.95 | 68.65 |
| Recency | 27 January 2022 | 4 March 2025 |
| Maximum RAM amount | 24 GB | 12 GB |
| Chip lithography | 8 nm | 5 nm |
| Power consumption (TDP) | 450 Watt | 250 Watt |
RTX 3090 Ti has a 2% higher aggregate performance score, and a 100% higher maximum VRAM amount.
RTX 5070, on the other hand, has an age advantage of 3 years, a 38% more advanced lithography process, and 44% lower power consumption.
Given the minimal performance differences, no clear winner can be declared between GeForce RTX 3090 Ti and GeForce RTX 5070.
Other comparisons
We selected several comparisons of graphics cards with performance close to those reviewed, providing you with more options to consider.
