GeForce RTX 3080 Ti vs GeForce RTX 3090: specs and benchmark comparison
ANVIDIA · Ampere
NVIDIA GeForce RTX 3080 Ti
- 12 GB GDDR6X
- 350 W
- PSU 750 W+
June 3, 2021 · $1,199 at launch
BNVIDIA · Ampere
NVIDIA GeForce RTX 3090
- 24 GB GDDR6X
- 350 W
- PSU 750 W+
September 24, 2020 · $1,499 at launch
Short verdict
NVIDIA GeForce RTX 3090 leads in every scenario with comparable benchmarks: rendering and compute. This does not imply an advantage in tasks without benchmarks.
Compare with AIRendering and compute
B +19.1%NVIDIA GeForce RTX 3090 leads in this scenario (up to 19.1% in comparable benchmarks).
No comparable benchmarks yet: gaming: rasterization, gaming: ray tracing, power efficiency.
Advantages
Only measured and rated values: a noticeable benchmark gap and specs where more is better.
Where GeForce RTX 3080 Ti is better1
Launch price $300 lower$1,199 vs $1,499
Where GeForce RTX 3090 is better2
Shader units 256 more10,496 (CUDA cores) vs 10,240 (CUDA cores)
Video memory 12 GB more24 GB vs 12 GB
Benchmarks
Results are compared within one benchmark, one version and the same conditions.
- 0.9% difference, within the threshold
Blender Benchmark 2.93.1
GPU, bmw27 (CUDA) · s, lower is better
GeForce RTX 3080 Ti17.1 sGeForce RTX 309016.9 sPosition among tested models (31, CUDA)
B · 17A · 170100200300Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 0.9% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
17.1 s · community measurement, sample 241 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 241 submissions (32 outliers excluded), interquartile range 16.92–17.32; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
16.9 s · community measurement, sample 279 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 279 submissions (58 outliers excluded), interquartile range 16.76–17.19; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 2.7% difference, within the threshold
Blender Benchmark 2.93.1
GPU, bmw27 (OPTIX) · s, lower is better
GeForce RTX 3080 Ti9.6 sGeForce RTX 30909.3 sPosition among tested models (20, OptiX)
B · 9A · 10050100150Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 2.7% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
9.6 s · community measurement, sample 107 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 107 submissions (20 outliers excluded), interquartile range 9.49–9.69; submissions 2021-07-09 to 2022-03-01; OPTIX backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
9.3 s · community measurement, sample 244 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 244 submissions (56 outliers excluded), interquartile range 9.17–9.43; submissions 2021-07-10 to 2022-02-28; OPTIX backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 0.7% difference, within the threshold
Blender Benchmark 2.93.1
GPU, classroom (CUDA) · s, lower is better
GeForce RTX 3080 Ti80.1 sGeForce RTX 309079.5 sPosition among tested models (26, CUDA)
B · 80A · 800200400600800Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 0.7% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
80.1 s · community measurement, sample 215 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 215 submissions (24 outliers excluded), interquartile range 78.67–81.45; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
79.5 s · community measurement, sample 260 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 260 submissions (43 outliers excluded), interquartile range 78.58–81; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 1.7% difference, within the threshold
Blender Benchmark 2.93.1
GPU, classroom (OPTIX) · s, lower is better
GeForce RTX 3080 Ti34 sGeForce RTX 309033.5 sPosition among tested models (14, OptiX)
B · 33A · 34050100150200Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 1.7% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
34 s · community measurement, sample 121 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 121 submissions (5 outliers excluded), interquartile range 33.4–35; submissions 2021-07-09 to 2022-03-01; OPTIX backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
33.5 s · community measurement, sample 182 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 182 submissions (15 outliers excluded), interquartile range 32.84–34.67; submissions 2021-07-10 to 2022-02-28; OPTIX backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 1.2% difference, within the threshold
Blender Benchmark 2.93.1
GPU, fishy_cat (CUDA) · s, lower is better
GeForce RTX 3080 Ti34.9 sGeForce RTX 309034.4 sPosition among tested models (21, CUDA)
B · 34A · 350100200300Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 1.2% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
34.9 s · community measurement, sample 187 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 187 submissions (23 outliers excluded), interquartile range 34.38–35.32; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
34.4 s · community measurement, sample 213 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 213 submissions (51 outliers excluded), interquartile range 34.13–34.9; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 2.6% difference, within the threshold
Blender Benchmark 2.93.1
GPU, fishy_cat (OPTIX) · s, lower is better
GeForce RTX 3080 Ti18.4 sGeForce RTX 309017.9 sPosition among tested models (13, OptiX)
B · 18A · 18020406080Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 2.6% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
18.4 s · community measurement, sample 101 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 101 submissions (7 outliers excluded), interquartile range 18.07–18.76; submissions 2021-07-09 to 2022-03-01; OPTIX backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
17.9 s · community measurement, sample 141 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 141 submissions (40 outliers excluded), interquartile range 17.77–18.14; submissions 2021-07-10 to 2022-02-28; OPTIX backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 2.3% difference, within the threshold
Blender Benchmark 2.93.1
GPU, koro (CUDA) · s, lower is better
GeForce RTX 3080 Ti55.8 sGeForce RTX 309054.6 sPosition among tested models (20, CUDA)
B · 55A · 560200400600Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 2.3% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
55.8 s · community measurement, sample 180 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 180 submissions (27 outliers excluded), interquartile range 55.15–56.53; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
54.6 s · community measurement, sample 216 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 216 submissions (47 outliers excluded), interquartile range 54.03–55.35; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 2.7% difference, within the threshold
Blender Benchmark 2.93.1
GPU, koro (OPTIX) · s, lower is better
GeForce RTX 3080 Ti39 sGeForce RTX 309037.9 sPosition among tested models (13, OptiX)
B · 38A · 39050100150200Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 2.7% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
39 s · community measurement, sample 104 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 104 submissions (3 outliers excluded), interquartile range 38.18–40.26; submissions 2021-07-09 to 2022-03-01; OPTIX backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
37.9 s · community measurement, sample 143 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 143 submissions (36 outliers excluded), interquartile range 37.49–38.45; submissions 2021-07-10 to 2022-02-28; OPTIX backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 1.5% difference, within the threshold
Blender Benchmark 2.93.1
GPU, pavillon_barcelona (CUDA) · s, lower is better
GeForce RTX 3080 Ti100.9 sGeForce RTX 309099.3 sPosition among tested models (21, CUDA)
B · 99A · 1010200400600800Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 1.5% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
100.9 s · community measurement, sample 191 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 191 submissions (27 outliers excluded), interquartile range 99.62–102.43; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
99.3 s · community measurement, sample 215 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 215 submissions (50 outliers excluded), interquartile range 98.46–100.87; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 1.8% difference, within the threshold
Blender Benchmark 2.93.1
GPU, pavillon_barcelona (OPTIX) · s, lower is better
GeForce RTX 3080 Ti45.9 sGeForce RTX 309045.1 sPosition among tested models (13, OptiX)
B · 45A · 460100200300Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 1.8% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
45.9 s · community measurement, sample 105 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 105 submissions (9 outliers excluded), interquartile range 45.17–46.74; submissions 2021-07-09 to 2022-03-01; OPTIX backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
45.1 s · community measurement, sample 145 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 145 submissions (26 outliers excluded), interquartile range 44.57–45.81; submissions 2021-07-10 to 2022-02-28; OPTIX backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 1.3% difference, within the threshold
Blender Benchmark 2.93.1
GPU, victor (CUDA) · s, lower is better
GeForce RTX 3080 Ti140.6 sGeForce RTX 3090138.7 sPosition among tested models (19, CUDA)
B · 139A · 14102505007501,000Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 1.3% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
140.6 s · community measurement, sample 180 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 180 submissions (24 outliers excluded), interquartile range 138.92–142.91; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
138.7 s · community measurement, sample 217 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 217 submissions (43 outliers excluded), interquartile range 137.5–140.78; submissions 2021-07-09 to 2022-03-01; CUDA backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 1.7% difference, within the threshold
Blender Benchmark 2.93.1
GPU, victor (OPTIX) · s, lower is better
GeForce RTX 3080 Ti61.5 sGeForce RTX 309060.4 sPosition among tested models (13, OptiX)
B · 60A · 610100200300Each dot is one model’s median in this benchmark. The scale starts at zero, s. The 1.7% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
61.5 s · community measurement, sample 105 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 105 submissions (4 outliers excluded), interquartile range 60.16–63.3; submissions 2021-07-09 to 2022-03-01; OPTIX backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
60.4 s · community measurement, sample 144 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 144 submissions (24 outliers excluded), interquartile range 59.46–61.01; submissions 2021-07-10 to 2022-02-28; OPTIX backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Render time of a single Cycles scene without synchronization. Median of submissions after filtering.
- 1.5% difference, within the threshold
Blender Benchmark 4.5.0
GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) · samples/min, higher is better
GeForce RTX 3080 Ti5,343.7 samples/minGeForce RTX 30905,423.6 samples/minPosition among tested models (67, OptiX)
A · 5,344B · 5,42405,00010,00015,000Each dot is one model’s median in this benchmark. The scale starts at zero, samples/min. The 1.5% difference is below the noticeable threshold (5%).
About the measurement and source
- A · NVIDIA GeForce RTX 3080 Ti
5,343.7 samples/min · community measurement, sample 193 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 193 submissions (16 outliers excluded), interquartile range 5163.7–5447.58; submissions 2025-07-20 to 2026-11-12; OPTIX backend. Snapshot 2026-11-14.
- B · NVIDIA GeForce RTX 3090
5,423.6 samples/min · community measurement, sample 596 · November 14, 2026
Blender Open Data (Blender Foundation)
Median of 596 submissions (34 outliers excluded), interquartile range 5240.15–5558.47; submissions 2025-07-17 to 2026-10-19; OPTIX backend. Snapshot 2026-11-14.
difference = |A − B| ÷ the smaller value × 100%. The 5% noticeable threshold is a UX heuristic, not statistical significance.
Cycles render in Blender Benchmark. Median of the sum of samples per minute over three scenes after removing outliers.
Other comparable benchmarks (18)
| Benchmark | A | B | Difference |
|---|---|---|---|
| Blender Benchmark 2.92GPU, bmw27 (OPTIX) | 9 | 9.1 | A +0.8% |
| Blender Benchmark 2.92GPU, classroom (CUDA) | 56.9 | 55.5 | B +2.4% |
| Blender Benchmark 4.2.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,223.7 | 5,354.8 | B +2.5% |
| Blender Benchmark 5.1.1GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,075.2 | 5,253.3 | B +3.5% |
| Blender Benchmark 5.2.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 4,963.6 | 5,093.8 | B +2.6% |
| Blender Benchmark 3.0.1GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,721.8 | 5,841.8 | B +2.1% |
| Blender Benchmark 3.1.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,557.8 | 6,560.4 | B +18% |
| Blender Benchmark 3.2.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,512.7 | 6,566.2 | B +19.1% |
| Blender Benchmark 3.2.1GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,532.4 | 5,625.7 | B +1.7% |
| Blender Benchmark 3.3.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,968 | 6,018.2 | B +0.8% |
| Blender Benchmark 3.4.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,912.4 | 6,014.6 | B +1.7% |
| Blender Benchmark 3.5.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 6,205.2 | 6,296.9 | B +1.5% |
| Blender Benchmark 3.6.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 6,095.3 | 6,305.5 | B +3.4% |
| Blender Benchmark 4.0.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,440.2 | 5,536.1 | B +1.8% |
| Blender Benchmark 4.1.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,521.3 | 5,536.4 | B +0.3% |
| Blender Benchmark 4.3.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,327.6 | 5,347.6 | B +0.4% |
| Blender Benchmark 4.4.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,276 | 5,333 | B +1.1% |
| Blender Benchmark 5.1.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 5,111.4 | 5,266.1 | B +3% |
Each benchmark version has its own scale: numbers are compared only within a row.
Reference results (63)
| Benchmark | A | B | Difference |
|---|---|---|---|
| Blender Benchmark 2.90GPU, bmw27 (CUDA) | 20.3 | 21.5 | reference |
| Blender Benchmark 2.92GPU, bmw27 (CUDA) | 19 | 18.5 | reference |
| Blender Benchmark 2.90GPU, bmw27 (OPTIX) | 9 | 8.9 | reference |
| Blender Benchmark 2.90GPU, classroom (CUDA) | 70.8 | 70.6 | reference |
| Blender Benchmark 2.90GPU, classroom (OPTIX) | 33.6 | 35.2 | reference |
| Blender Benchmark 2.92GPU, classroom (OPTIX) | 35.4 | 34.4 | reference |
| Blender Benchmark 2.90GPU, fishy_cat (CUDA) | 37.6 | 39 | reference |
| Blender Benchmark 2.92GPU, fishy_cat (CUDA) | 35.5 | 34.8 | reference |
| Blender Benchmark 2.90GPU, fishy_cat (OPTIX) | 17 | 16.5 | reference |
| Blender Benchmark 2.92GPU, fishy_cat (OPTIX) | 18.1 | 18.4 | reference |
| Blender Benchmark 2.90GPU, koro (CUDA) | 59.4 | 58 | reference |
| Blender Benchmark 2.92GPU, koro (CUDA) | 59.7 | 58 | reference |
| Blender Benchmark 2.90GPU, koro (OPTIX) | 37.6 | 36.5 | reference |
| Blender Benchmark 2.92GPU, koro (OPTIX) | 40 | 57.1 | reference |
| Blender Benchmark 2.90GPU, pavillon_barcelona (CUDA) | 107.6 | 119.6 | reference |
| Blender Benchmark 2.92GPU, pavillon_barcelona (CUDA) | 112.8 | 109.3 | reference |
| Blender Benchmark 2.90GPU, pavillon_barcelona (OPTIX) | 44.5 | 45.4 | reference |
| Blender Benchmark 2.92GPU, pavillon_barcelona (OPTIX) | 44.9 | 45.1 | reference |
| Blender Benchmark 2.90GPU, victor (CUDA) | 142.9 | 159.1 | reference |
| Blender Benchmark 2.92GPU, victor (CUDA) | 183.6 | 178.3 | reference |
| Blender Benchmark 2.90GPU, victor (OPTIX) | 51.3 | 52.8 | reference |
| Blender Benchmark 2.92GPU, victor (OPTIX) | 60.6 | 62.5 | reference |
| Blender Benchmark 4.2.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | 3,559.3 | 3,699.3 | reference |
| Blender Benchmark 4.3.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | 3,676.8 | 3,730.8 | reference |
| Blender Benchmark 5.2.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | 2,942.2 | — | no pair |
| Blender Benchmark 5.0.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 4,612 | 4,875.8 | reference |
| Blender Benchmark 5.3.0GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | 4,626.7 | 5,143.8 | reference |
| Blender Benchmark 2.90.1GPU, bmw27 (CUDA) | — | 20.5 | no pair |
| Blender Benchmark 2.91.2GPU, bmw27 (CUDA) | — | 18.5 | no pair |
| Blender Benchmark 2.90.1GPU, bmw27 (OPTIX) | — | 8.9 | no pair |
| Blender Benchmark 2.91.2GPU, bmw27 (OPTIX) | — | 9.2 | no pair |
| Blender Benchmark 2.90.1GPU, classroom (CUDA) | — | 67.4 | no pair |
| Blender Benchmark 2.91.2GPU, classroom (CUDA) | — | 58.8 | no pair |
| Blender Benchmark 2.90.1GPU, classroom (OPTIX) | — | 38.9 | no pair |
| Blender Benchmark 2.91.2GPU, classroom (OPTIX) | — | 37.1 | no pair |
| Blender Benchmark 2.90.1GPU, fishy_cat (CUDA) | — | 37.3 | no pair |
| Blender Benchmark 2.91.2GPU, fishy_cat (CUDA) | — | 33.9 | no pair |
| Blender Benchmark 2.90.1GPU, fishy_cat (OPTIX) | — | 16.8 | no pair |
| Blender Benchmark 2.91.2GPU, fishy_cat (OPTIX) | — | 16.7 | no pair |
| Blender Benchmark 2.90.1GPU, koro (CUDA) | — | 57.5 | no pair |
| Blender Benchmark 2.91.2GPU, koro (CUDA) | — | 57.5 | no pair |
| Blender Benchmark 2.90.1GPU, koro (OPTIX) | — | 37.1 | no pair |
| Blender Benchmark 2.91.2GPU, koro (OPTIX) | — | 36.3 | no pair |
| Blender Benchmark 2.90.1GPU, pavillon_barcelona (CUDA) | — | 108.8 | no pair |
| Blender Benchmark 2.91.2GPU, pavillon_barcelona (CUDA) | — | 108.5 | no pair |
| Blender Benchmark 2.90.1GPU, pavillon_barcelona (OPTIX) | — | 46.6 | no pair |
| Blender Benchmark 2.91.2GPU, pavillon_barcelona (OPTIX) | — | 46.9 | no pair |
| Blender Benchmark 2.90.1GPU, victor (CUDA) | — | 148.7 | no pair |
| Blender Benchmark 2.91.2GPU, victor (CUDA) | — | 124.3 | no pair |
| Blender Benchmark 2.90.1GPU, victor (OPTIX) | — | 57.7 | no pair |
| Blender Benchmark 2.91.2GPU, victor (OPTIX) | — | 53.9 | no pair |
| Blender Benchmark 3.0.1GPU, sum of scenes monster + junkshop + classroom (CUDA) | — | 3,696.5 | no pair |
| Blender Benchmark 3.1.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | — | 3,418.3 | no pair |
| Blender Benchmark 3.2.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | — | 3,138.2 | no pair |
| Blender Benchmark 3.2.1GPU, sum of scenes monster + junkshop + classroom (CUDA) | — | 3,448.2 | no pair |
| Blender Benchmark 3.3.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | — | 3,729.7 | no pair |
| Blender Benchmark 3.4.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | — | 3,797.9 | no pair |
| Blender Benchmark 3.5.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | — | 3,864 | no pair |
| Blender Benchmark 3.6.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | — | 4,060.8 | no pair |
| Blender Benchmark 4.0.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | — | 3,724.8 | no pair |
| Blender Benchmark 4.1.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | — | 3,735.6 | no pair |
| Blender Benchmark 4.5.0GPU, sum of scenes monster + junkshop + classroom (CUDA) | — | 3,798 | no pair |
| Blender Benchmark 3.1.2GPU, sum of scenes monster + junkshop + classroom (OptiX / HIP / oneAPI / Metal) | — | 5,826.5 | no pair |
Each benchmark version has its own scale: numbers are compared only within a row.
Specifications
| Specification | A · GeForce RTX 3080 Ti | B · GeForce RTX 3090 | Difference |
|---|---|---|---|
| Chip | |||
| Architecture | Ampere | Ampere | equal |
| Shader unitsShader unit counts are comparable only within one architecture. | 10,240 (CUDA cores) | 10,496 (CUDA cores) | B: 256 more |
| Clocks | |||
| Base clock | 1,370 MHz | 1,400 MHz | B: 30 MHz more |
| Boost clockClocks of different architectures do not reflect performance directly. For laptop chips, the clock depends on the laptop’s power limit. | 1,670 MHz | 1,700 MHz | B: 30 MHz more |
| Memory | |||
| Video memory | 12 GB | 24 GB | B: 12 GB more |
| Memory type | GDDR6X | GDDR6X | equal |
| Memory bus | 384 bit | 384 bit | equal |
| Power and dimensions | |||
| Board powerTGP (NVIDIA) and TBP (AMD/Intel) are calculated differently; higher is not better. For laptops, the chip maker sets the range and the laptop maker picks the value, so the same chip behaves differently in different laptops. | 350 W (board power) | 350 W (board power) | equal |
| Recommended PSU | 750 W | 750 W | equal |
| Length (reference / Founders Edition)Partner cards have different dimensions; check the specific SKU. | 285 mm | 313 mm | B: 28 mm higher |
| Thickness (slots) | 2 | 3 | B: 1 higher |
| Power connectors (reference) | 2x 8-pin (adapter to 1x 12-pin included) — FE | 2x 8-pin (adapter to 1x 12-pin included) — FE | equal |
Chip
Architecture
AAmpere
BAmpere
equalShader units
A10,240 (CUDA cores)
B10,496 (CUDA cores)
B: 256 moreShader unit counts are comparable only within one architecture.
Clocks
Base clock
A1,370 MHz
B1,400 MHz
B: 30 MHz moreBoost clock
A1,670 MHz
B1,700 MHz
B: 30 MHz moreClocks of different architectures do not reflect performance directly. For laptop chips, the clock depends on the laptop’s power limit.
Memory
Video memory
A12 GB
B24 GB
B: 12 GB moreMemory type
AGDDR6X
BGDDR6X
equalMemory bus
A384 bit
B384 bit
equal
Power and dimensions
Board power
A350 W (board power)
B350 W (board power)
equalTGP (NVIDIA) and TBP (AMD/Intel) are calculated differently; higher is not better. For laptops, the chip maker sets the range and the laptop maker picks the value, so the same chip behaves differently in different laptops.
Recommended PSU
A750 W
B750 W
equalLength (reference / Founders Edition)
A285 mm
B313 mm
B: 28 mm higherPartner cards have different dimensions; check the specific SKU.
Thickness (slots)
A2
B3
B: 1 higherPower connectors (reference)
A2x 8-pin (adapter to 1x 12-pin included) — FE
B2x 8-pin (adapter to 1x 12-pin included) — FE
equal
“More” does not always mean “better”: higher power draw or size is a system requirement, not an advantage (such bars are dimmed).
Compare with AI
An AI model reviews the specs and benchmarks of this pair and gives a plain-language verdict on who each model suits.
Questions and answers
How much video memory do NVIDIA GeForce RTX 3080 Ti and NVIDIA GeForce RTX 3090 have?
NVIDIA GeForce RTX 3080 Ti: 12 GB GDDR6X, 384-bit bus. NVIDIA GeForce RTX 3090: 24 GB GDDR6X, 384-bit bus.
What power supply does the manufacturer recommend?
For NVIDIA GeForce RTX 3080 Ti, 750 W or more; for NVIDIA GeForce RTX 3090, 750 W or more. This is a recommendation for a typical system; also check the power connectors of the specific card model.
Which is faster: NVIDIA GeForce RTX 3080 Ti or NVIDIA GeForce RTX 3090?
NVIDIA GeForce RTX 3090 leads in every scenario with comparable benchmarks: rendering and compute. This does not imply an advantage in tasks without benchmarks.
See also
Data sources (3)
- NVIDIA — specifications of NVIDIA GeForce RTX 3080 Ti, specifications of NVIDIA GeForce RTX 3090checked November 7, 2026
- Blender Open Data (Blender Foundation) — benchmark Blender Benchmark 2.90, benchmark Blender Benchmark 2.90.1, benchmark Blender Benchmark 2.91.2, benchmark Blender Benchmark 2.92, benchmark Blender Benchmark 2.93.1, benchmark Blender Benchmark 3.0.1, benchmark Blender Benchmark 3.1.0, benchmark Blender Benchmark 3.2.0, benchmark Blender Benchmark 3.2.1, benchmark Blender Benchmark 3.3.0, benchmark Blender Benchmark 3.4.0, benchmark Blender Benchmark 3.5.0, benchmark Blender Benchmark 3.6.0, benchmark Blender Benchmark 4.0.0, benchmark Blender Benchmark 4.1.0, benchmark Blender Benchmark 4.2.0, benchmark Blender Benchmark 4.3.0, benchmark Blender Benchmark 4.5.0, benchmark Blender Benchmark 5.2.0, benchmark Blender Benchmark 5.1.1, benchmark Blender Benchmark 3.1.2, benchmark Blender Benchmark 4.4.0, benchmark Blender Benchmark 5.0.0, benchmark Blender Benchmark 5.1.0, benchmark Blender Benchmark 5.3.0checked November 14, 2026
- Wikipedia (English) — release date and launch pricechecked November 14, 2026
How we compare: methodology. Where the data comes from: sources.