RTX 3090
RTX 3090 Review & Specifications
The RTX 3090 is based on the Ampere architecture and manufactured using the 8 nm nm process node. It was launched approximately 5 years, 10 months ago, originally priced at $1499. It features 10496 cores, including 82 compute units (SMs), 328 texture mapping units (TMUs), and 112 render output units (ROPs). For ray tracing and AI-accelerated tasks, it also includes 82 dedicated RT cores and 328 Tensor cores, enabling advanced rendering and upscaling features where supported. It has a base clock of 1395 MHz and a boost clock reaching up to 1695 MHz. The RTX 3090 comes equipped with 24 GB of GDDR6X memory across a 384-bit bus. Memory runs at 1219 MHz and delivers a total memory bandwidth of 936 GB/s.
This GPU has a very high thermal design power (TDP) of 350 W. A PC case with excellent airflow and a high-quality, high-wattage power supply are essential for stable operation.
RTX 3090 Performance
RTX 3090 is a mid-range graphics card, scoring 56.15% in relative gaming performance. Thanks to 24 GB of VRAM, it offers excellent 1440p performance and can handle 4K gaming with optimized settings or the help of DLSS, FSR, and frame generation. Great for users targeting 100+ FPS at 1440p or playable 4K in lighter titles.
In terms of memory capacity, the RTX 3090 features an exceptionally high VRAM size of 24 GB — far exceeding what is typically required for modern games and creative tasks. This allows users to max out VRAM-dependent settings such as textures, caching, and resolution scaling. However, since this is not an ultra high-end GPU in terms of core performance, rendering speed and frame rates in demanding games may be limited by the processing power rather than memory. In the domain of AI and machine learning, the large VRAM capacity enables work with complex neural networks (e.g., large LLMs or diffusion models like Stable Diffusion XL) using high batch sizes without falling back on system memory. It is also ideal for creators running high-resolution video editing, 3D rendering, or simultaneous high-memory applications — even if render times are slower compared to top-tier GPUs. This card offers excellent value for prosumers prioritizing memory capacity over raw graphical horsepower.
RTX 3090 has a gaming performance-per-price ratio of 32.4%, suggesting lower value in this category. This GPU lacks both high performance and value, despite its large VRAM capacity. It is generally not recommended, as better options exist in this price range.
General Information
| GPU Name | RTX 3090 |
|---|---|
| GPU Code Name | GA102 |
| Release Price | $1499 |
| Selected Model | EVGA GeForce RTX 3090 FTW3 Ultra Gaming, 24GB GDDR6X, 10496 CUDA Cores, 1800MHz Boost Clock, 3x Fans, ARGB LED, Metal Backplate, PCIe 4, HDMI, DisplayPort, Desktop Compatible |
| Current Price Amazon | $1500 |
| Amazon Price Updated | 1 week, 4 days ago |
| Launch Date | 5 years, 10 months ago |
| GPU Available | True |
| Architecture | Ampere |
|---|---|
| Foundry | Samsung |
| Process | 8 nm |
| Transistors | 28300 million |
| Die Size | 628 mm² |
| Bus Interface | PCIe 4.0 x16 |
| TDP | 350 W |
| Cores | 10496 |
|---|---|
| CUs (SMs) | 82 |
| TMUs | 328 |
| ROPs | 112 |
| RT Cores | 82 |
| Tensor Cores | 328 |
| L1 Cache | 128 KB (per SM) |
| L2 Cache | 6 MB |
| Base Clock | 1395 MHz |
|---|---|
| Boost Clock | 1695 MHz |
| Memory Size - Type | 24 GB GDDR6X |
|---|---|
| Memory Bus Width | 384 bit |
| Memory Clock | 1219 MHz |
| Memory Bandwidth | 936 GB/s |
| DirectX | 12 Ultimate (12_2) |
|---|---|
| OpenGL | 4.6 |
| OpenCL | 3.0 |
| Vulkan | 1.4 |
| Shader Model | 6.8 |
Theoretical Performance
| Pixel Rate | 189.80 GPixel/s |
|---|---|
| Texture Rate | 556.00 GTexel/s |
| FP16 | 35.58 TFLOPS |
| FP32 | 35.58 TFLOPS |
| FP64 | 556.02 GFLOPS |
Gaming Performance Chart
RTX 3090 Compared to Similar Performance GPUs
GPU Price / Performance Table
| GPU (Desc) | Price (Desc) | Gaming_1080p (Desc) | Gaming_1440p (Desc) | Gaming_2160p (Desc) | Game_1440p/Price (Desc) |
|---|---|---|---|---|---|
|
$1650
|
↓
-2.9%
|
60.8%
|
55.5%
|
50.5%
|
33.3%
|
|
|
$670
|
0.0%
|
61.5%
|
54.6%
|
44.7%
|
80.7%
|
|
|
$700
|
0.0%
|
60.0%
|
52.2%
|
45.4%
|
73.8%
|
|
|
$839
|
0.0%
|
57.5%
|
49.4%
|
41.2%
|
58.3%
|
|
|
$1500
|
↓
-3.2%
|
56.2%
|
49.0%
|
44.8%
|
32.4%
|
|
|
$799❗
|
↓
-5.9%
|
56.0%
|
48.8%
|
41.3%
|
60.4%
|
|
|
$0
|
0.0%
|
56.0%
|
48.5%
|
40.9%
|
N/A | |
|
$590
|
↓
-4.8%
|
55.1%
|
47.9%
|
43.7%
|
80.4%
|
|
|
$1098
|
0.0%
|
55.7%
|
47.8%
|
41.3%
|
43.1%
|
GPU Specification Table
| GPU (Desc) | Architecture | Shader/TMU/ROP/RT | Base/Boost Clock | Memory | Mem Clock/Bandwidth | TDP |
|---|---|---|---|---|---|---|
| RTX 3090 Ti | GA102 / 8 nm nm / Ampere | 10752 / 336 / 112 / 84 | 1560 MHz / 1860 MHz | 24 GB GDDR6X / 384 bit | 1313 MHz / 1 GB/s | 450 W |
| RTX 4070 Ti | AD104 / 5 nm nm / Ada Lovelace | 7680 / 240 / 80 / 60 | 2310 MHz / 2610 MHz | 12 GB GDDR6X / 192 bit | 1313 MHz / 504 GB/s | 285 W |
| RTX 5070 | GB205 / 5 nm nm / Blackwell 2.0 | 6144 / 192 / 80 / 48 | 2325 MHz / 2512 MHz | 12 GB GDDR7 / 192 bit | 1750 MHz / 672 GB/s | 250 W |
| RTX 4070 SUPER | AD104 / 5 nm nm / Ada Lovelace | 7168 / 224 / 80 / 56 | 1980 MHz / 2475 MHz | 12 GB GDDR6X / 192 bit | 1313 MHz / 504 GB/s | 220 W |
| RTX 3090 | GA102 / 8 nm nm / Ampere | 10496 / 328 / 112 / 82 | 1395 MHz / 1695 MHz | 24 GB GDDR6X / 384 bit | 1219 MHz / 936 GB/s | 350 W |
| RX 7900 GRE | Navi 31 / 5 nm / RDNA 3.0 | 5120 / 320 / 160 / 80 | 1287 MHz / 2245 MHz | 16 GB GDDR6 / 256 bit | 2250 MHz / 576 GB/s | 260 W |
| RX 9070 GRE | Navi 48 / 5 nm / RDNA 4.0 | 3072 / 192 / 96 / 48 | 1420 MHz / 2790 MHz | 12 GB GDDR6 / 192 bit | 2250 MHz / 432 GB/s | 220 W |
| RTX 3080 Ti | GA102 / 8 nm nm / Ampere | 10240 / 320 / 112 / 80 | 1365 MHz / 1665 MHz | 12 GB GDDR6X / 384 bit | 1188 MHz / 912 GB/s | 350 W |
| RX 6950 XT | Navi 21 / 7 nm / RDNA 2.0 | 5120 / 320 / 128 / 80 | 1860 MHz / 2310 MHz | 16 GB GDDR6 / 256 bit | 2250 MHz / 576 GB/s | 335 W |
Add Comment on RTX 3090