Performance of an RTX 4000 versus the increasingly aging Quadro P4000
| Features of NVIDIA Quadro Professional Graphics Cards | |||||||
| Video card | Number of cores | Base frequency, MHz | Maximum computing power FP32, TFLOPS | Memory capacity, GB | The throughput capacity of the memory, GB / s | TDP, W | Price, $ |
| GV100 | 5120 | 1200 | 14.9 | 32 8 | 870 | 185 | 8999 |
| RTX 8000 | 4608 | 1440 | 16.3 | 48 5 | 624 | ? | 10000 |
| RTX 6000 | 4608 | 1440 | 16.3 | 24 5 | 624 | 295 | 6300 |
| RTX 5000 | 3072 | 1350 | 11.2 | 16 5 | 448 | 265 | 2300 |
| RTX 4000 | 2304 | ? | 7.1 | 8 1 | 416 | 160 | 900 |
| TITAN V | 5120 | 1200 | 14.9 | 12 4 | 653 | 250 | 2999 |
| P6000 | 3840 | 1417 | 11.8 | 24 6 | 432 | 250 | 4999 |
| P5000 | 2560 | 1607 | 8.9 | 16 6 | 288 | 180 | 1999 |
| P4000 | 1792 | 1227 | 5.3 | 8 3 | 243 | 105 | 799 |
| P2000 | 1024 | 1370 | 3.0 | 5 3 | 140 | 75 | 399 |
| P1000 | 640 | 1354 | 1.9 | 4 3 | 80 | 47 | 299 |
| P620 | 512 | 1354 | 1.4 | 2 3 | 80 | 40 | 199 |
| P600 | 384 | 1354 | 1.2 | 2 3 | 64 | 40 | 179 |
| P400 | 256 | 1070 | 0.6 | 2 3 | 32 | thirty | 139 |
| Notes (edit) | Memory: 1 GDDR6; 2 GDDR5X; 3 GDDR5; 4 HBM2; 5 GDDR6 (ECC); 6 GDDR5X (ECC); 7 GDDR5 (ECC); 8 HBM2 (ECC) Architecture: P = Pascal; V = Volta; RTX = Turing |
We do not have accurate base clock data for the RTX 4000, but we do know its maximum processing power when performing 32-bit floating point operations – 7.1 TFLOPS, which means the level of performance corresponding to the level of the GeForce RTX 2070 in the gaming segment. From this it follows that the RTX 4000 GPU can provide excellent gameplay at 1080p and 1440p resolutions, and in some games even at 4K resolutions.
Quadro RTX cards with larger numbers offer even better performance but are also more expensive; the apotheosis of this is the $ 10K RTX 8000 card with a mind-boggling 48GB GDDR6 ECC memory. If you have higher demands on speed and memory, or need above-average performance, then you should look at the RTX 5000 – if your budget allows.
The Quadro P4000 card has 5.3 TFLOPS processing power, and in this indicator alone, the new RTX 4000 card is 34% faster and at the same time not much more expensive. True, the increase in performance comes with a slightly higher power consumption, but the TDP of 160 W allows the new 4000-series card to remain in the single-slot category. The power connector is located at the rear end (not the top), which makes this card suitable for relatively small cases.
Unlike previous generations of graphics cards in the Quadro RTX series, the transition from Pascal architecture to Turing architecture does not just mean a noticeable increase in speed (and efficiency), but, thanks to the inclusion of Tensor and RT cores in the GPU, essentially translates RTX models into a separate league for market of video cards. As we mentioned, Tensor kernels are of great practical importance for the development of AI and deep learning technologies, and RT kernels provide significant advantages in implementing real-time ray tracing in applications that use this technology.
The table below summarizes the performance characteristics of the Quadro RTX models available today. For Turing processors, NVIDIA introduced a new parameter – RTX-OPS: the higher the value, the higher the potential capabilities of the video card.
