🛒 Best Price Today on Amazon
Prime members get fast, free shipping on everything above, plus member-only deals. New members get a 30-day free trial — cancel anytime.
- What Changes When You Move From Turing to Blackwell
- Specifications Side by Side
- Gaming Performance: How Big Is the Jump?
- Upscaling and Frame Generation
- VRAM: The Part That Barely Moves
- Power, PSU and Case Fit
- Cost per Performance: A Quick Calculation
- Is the Upgrade Worth It for Your Setup?
- Used-Market Note on the RTX 2080 Ti
- Frequently Asked Questions
If you still game on an RTX 2080 Ti, the RTX 5070 is a worthwhile upgrade for 1440p and high-refresh play: it is roughly 50-65% faster in typical GPU-bound games, adds DLSS 4 Multi Frame Generation, and does it at the same 250 W rating. If you mostly play older or esports titles at 1080p, though, the 2080 Ti still holds up and the money is better saved. The one thing the 5070 does not fix is memory: 12 GB is only 1 GB more than the 11 GB the 2080 Ti shipped with in 2018.
What Changes When You Move From Turing to Blackwell
The RTX 2080 Ti was the flagship of the first RTX generation. It launched on 20 September 2018 at $999 with 4,352 CUDA cores on a 12 nm Turing die and a wide 352-bit memory bus. The RTX 5070 arrived on 4 March 2025 at $549 as a mid-range Blackwell card with 6,144 CUDA cores on a 5 nm process, a narrower 192-bit bus and much faster GDDR7 memory. On paper the two are closer than the generational gap suggests, because the 2080 Ti was built big and wide while the 5070 relies on high clocks (2,512 MHz boost versus 1,545 MHz) and a huge 48 MB L2 cache.
The practical result is that the 5070 is a full performance tier above the old flagship rather than a side-grade, and it brings feature support that Turing cannot get, chiefly frame generation.
Specifications Side by Side
| Spec | RTX 2080 Ti | RTX 5070 |
|---|---|---|
| Architecture | Turing | Blackwell 2.0 |
| Release date | 20 September 2018 | 4 March 2025 |
| Launch MSRP | $999 | $549 |
| CUDA cores | 4,352 | 6,144 |
| Boost clock | 1,545 MHz | 2,512 MHz |
| Process | 12 nm | 5 nm |
| TDP | 250 W | 250 W |
| Memory | 11 GB GDDR6 | 12 GB GDDR7 |
| Memory bus | 352-bit | 192-bit |
| Bandwidth | 616.0 GB/s | 672.0 GB/s |
| L2 cache | 5.5 MB | 48 MB |
| Interface | PCIe 3.0 x16 | PCIe 5.0 x16 |
| Power connectors | 2x 8-pin | 1x 16-pin |
| Length / width | 267 mm / 2-slot | 245 mm / 2-slot |
| Outputs | 1x HDMI 2.0, 3x DP 1.4a, 1x USB-C | 1x HDMI 2.1b, 3x DP 2.1b |
Gaming Performance: How Big Is the Jump?
TechPowerUp’s RTX 5070 Founders Edition review describes the card as rivaling the RTX 4070 Ti in raster performance, which places it a little above the RTX 4070 Super. The RTX 2080 Ti, in the same family of charts, sits roughly at RTX 3070 level. Putting those reference points together, published review data suggests the RTX 5070 is roughly 50-65% faster than the 2080 Ti at 1440p in GPU-limited games. Synthetic-score aggregators such as PassMark show a smaller gap of around 25-35%, largely because those tests are less sensitive to the architectural improvements that show up in modern games, so treat the lower figure as a floor rather than the typical result.
The gap shrinks at 1080p, where the CPU often becomes the bottleneck, and grows at 4K and in ray-traced scenes. Ray tracing is where Turing shows its age most clearly: its first-generation RT cores fall well behind Blackwell’s, so in heavy RT titles such as Cyberpunk 2077 or Alan Wake 2 the 5070 can lead by considerably more than the raster average.
Approximate frame rates
As a rough guide only, and varying with settings, CPU and game patch: in demanding current releases at 1440p high settings, a 2080 Ti typically lands somewhere in the 50-75 fps range where an RTX 5070 lands around 80-120 fps before upscaling. In lighter esports games both cards push well past 144 fps, so the upgrade matters far less there.
Upscaling and Frame Generation
This is one of the less obvious reasons to upgrade. Both cards support DLSS Super Resolution, and NVIDIA’s DLSS 4 transformer-based upscaling model works on the RTX 20 series too, so a 2080 Ti owner already gets cleaner upscaling than in the past. What Turing cannot run is DLSS Frame Generation. The RTX 5070 supports both standard frame generation and DLSS 4 Multi Frame Generation, which can insert up to three generated frames per rendered frame in supported games.
Frame generation is most useful when the base frame rate is already reasonable (around 50-60 fps or more), because it adds smoothness but not responsiveness. It is a genuine advantage for high-refresh 1440p monitors, but it should not be counted as raw performance when comparing the two cards. The 2080 Ti can still use FSR 3 frame generation in games that offer it, which narrows the feature gap somewhat in those titles.
VRAM: The Part That Barely Moves
The 5070’s 12 GB is its weakest point for anyone upgrading from an 11 GB card. In practice the extra gigabyte matters less than the bandwidth and cache improvements, but it means the upgrade does not buy texture headroom. A rough VRAM guide for current games:
- 1080p: 8 GB is the minimum, 10-12 GB comfortable at high settings.
- 1440p: 12 GB covers most games at high settings; a few path-traced or ultra-texture titles can push past it.
- 4K: 16 GB is the safer target, especially with ray tracing and frame generation, which itself uses extra memory.
If 4K is the goal or you keep cards for five years or more, the RTX 5070 Ti with 16 GB is the more future-proof jump from a 2080 Ti.
Power, PSU and Case Fit
Both cards carry a 250 W TDP, so an existing 2080 Ti power supply is usually adequate. A simple sizing calculation: 250 W for the GPU, about 125 W for a typical gaming CPU, plus roughly 150 W of headroom for the rest of the system and transient spikes gives about 525 W. A quality 650 W unit is the comfortable choice; a good 550-600 W unit is the practical minimum.
The connector changes. The 2080 Ti uses two 8-pin plugs, while the 5070 uses a single 16-pin connector. Most cards include an adapter that takes two 8-pin PCIe cables, so an older PSU works, but a native 12V-2×6 cable from a newer PSU is tidier. Seat the 16-pin plug fully and avoid sharp bends near the connector.
Case fit is not a concern for the reference designs: the 5070 Founders Edition is 245 mm long and two slots thick, shorter than the 267 mm 2080 Ti. Partner models vary, so measure if your case is compact.
The 5070 runs on PCIe 5.0 x16 but works fine in PCIe 3.0 and 4.0 slots. Because it uses all 16 lanes, the bandwidth penalty on an older PCIe 3.0 board is small.
Cost per Performance: A Quick Calculation
Using launch prices and the approximate relative performance above, set the 2080 Ti at 100 points and the 5070 at about 155-165 points:
- RTX 2080 Ti: $999 / 100 = about $10.00 per performance point at launch.
- RTX 5070: $549 / 160 = about $3.40 per performance point at launch.
The more relevant number for an owner is net upgrade cost. A used 2080 Ti usually sells for around $250-$350 depending on condition and cooler, which can cut the effective cost of a 5070 nearly in half.
Is the Upgrade Worth It for Your Setup?
| Your situation | Better move | Why |
|---|---|---|
| 1080p, esports or older games | Keep the 2080 Ti | Already CPU-limited in many titles; gains are small |
| 1080p, new AAA with ray tracing | RTX 5070 | Much stronger RT and frame generation |
| 1440p high refresh | RTX 5070 | Roughly 50-65% more raster performance plus frame generation |
| 4K or long ownership | Look at RTX 5070 Ti | 16 GB VRAM avoids the 12 GB ceiling |
| Tight budget, PCIe 3.0 platform | RTX 5070, possibly with a CPU check | Older CPUs may limit the gain at 1080p |
One caveat: if your CPU is from the same era as the 2080 Ti, such as an older quad- or six-core chip, it may hold the 5070 back at 1080p and lower 1440p settings. Pairing the new card with at least a modern six-core processor lets it reach its potential.
Used-Market Note on the RTX 2080 Ti
Anyone weighing a used 2080 Ti against a new 5070 should be cautious. These cards are now seven or more years old, and many ran warm for long periods. Typical used pricing falls in the $250-$350 range, which is good value for 1080p and light 1440p gaming, but there is no warranty, thermal paste and pads may need replacing, and the card will never support DLSS Frame Generation. For a new build, the 5070 is the safer purchase.
Frequently Asked Questions
Is the RTX 5070 faster than the RTX 2080 Ti?
Yes. Published review data places the RTX 5070 roughly 50-65% ahead of the RTX 2080 Ti at 1440p in GPU-limited games, with a larger lead in ray tracing. The gap is smaller at 1080p, where the processor often limits both cards.
Can the RTX 2080 Ti use DLSS 4?
The RTX 2080 Ti supports DLSS 4 Super Resolution, including the newer transformer upscaling model. It does not support DLSS Frame Generation or Multi Frame Generation, which require RTX 40 and RTX 50 hardware respectively.
Do I need a new power supply to go from a 2080 Ti to an RTX 5070?
Usually not, since both cards are rated at 250 W. A quality 650 W unit is comfortable, and an adapter for the 16-pin connector that uses two 8-pin PCIe cables is typically included in the box.
Is 12 GB of VRAM enough for an upgrade from 11 GB?
For 1440p it is enough in most current games at high settings. It does not add meaningful headroom over the 2080 Ti, so buyers targeting 4K or heavy texture mods should consider a 16 GB card instead.
Will the RTX 5070 work on a PCIe 3.0 motherboard?
Yes. The card is backward compatible and uses a full x16 link, so the performance loss on PCIe 3.0 is small. An older CPU on that platform is more likely to limit performance than the slot itself.