Verily, after months of whispers and whispers, Nvidia hath at length unveiled the Blackwell GPU unto the masses. This new marvel, the Nvidia GeForce RTX 5090 GPU, doth bringeth forth manifold improvements over its forebear, the RTX 4090. Faster memory, DLSS 4, a plethora of CUDA cores, and much more doth it entail. If thou dost seek to discern the disparity betwixt these two Nvidia GPUs, peruse our meticulous comparison betwixt the Nvidia RTX 5090 and 4090.
Nvidia RTX 5090 vs 4090: Specs Comparison
Nvidia RTX 5090 vs 4090: Architecture
At the outset, the Nvidia RTX 5090 hath been fashioned upon the latest Blackwell architecture, whilst the RTX 4090 doth rest upon the Ada Lovelace architecture, a structure yon two years old. Furthermore, the RTX 5090 hath been wrought upon TSMC’s 4nm (N4P) process node, boasting 92 billion transistors upon its die. Conversely, the RTX 4090 is birthed from TSMC’s 5nm (4N) process node, bearing around 76.3 billion transistors within.
As for CUDA cores, the RTX 5090 doth claim the title of being the foremost Nvidia consumer GPU to house o’er 20,000 CUDA cores (21,760, to be precise). The RTX 4090 shelters 16,386 CUDA cores. Through the grace of the Blackwell architecture and the enhanced 4nm node, the RTX 5090 harbors 32% more transistors.
In its entirety, Nvidia proclaimeth that the RTX 5090 doth deliver nigh unto 2x better performance than its forerunner, the RTX 4090. In the realm of AI workloads, the disparity be almost 3x. And all this hath been achieved at a lower frequency, a marvel indeed. The RTX 5090 card doth attain a peak frequency of 2.41GHz, modestly lower than the RTX 4090’s 2.52GHz peak clock speed.
Nvidia RTX 5090 vs 4090: Performance
Regarding performance, Nvidia hath unveiled figures juxtaposing the RTX 5090 and 4090. The Shader cores upon the RTX 5090 can yield up to 125 TFLOPS, whilst the RTX 4090 can render up to 83 TFLOPS. Herein, doth the RTX 5090 reign supreme by 50%.
The 5th-gen Tensor cores of the RTX 5090 art capable of executing up to a staggering 3,352 trillion operations per second (TOPS) on AI workloads. The 4th-gen Tensor cores belonging to the RTX 4090 can process up to 1,321 AI TOPS, signifying the RTX 5090 be approximately 2.5x swifter than the RTX 4090 in AI tasks.
Nvidia RTX 5090 vs 4090: Memory Bandwidth
Lo, the Nvidia RTX 5090 doth mark the advent of ultra-swift GDDR7 memory unto consumer GPUs. Bearing 32GB of VRAM and operating upon a 512-bit memory bus, the memory reacheth an astonishing bandwidth of 1.8 TBps. Conversely, the RTX 4090 art restricted to a memory bandwidth of 1,008 GBps. Upon the RTX 4090, thou shalt find 24GB of GDDR6X memory, reliant upon a 384-bit memory bus.
Nvidia RTX 5090 vs 4090: Power Consumption and TDP
Concerning power consumption, the Nvidia RTX 5090 GPU doth possess a TDP of 575W, 125W more than the RTX 4090’s 450W TDP. Nvidia doth advise a PSU of up to 1000W for the RTX 5090. Many did anticipate the RTX 5090 would be grander, yet it doth share the same dimensions as the RTX 4090. Unlike the RTX 4090, a 3-slot card, the Founder Edition RTX 5090 is a two-slot GPU, allowing it to fit into a smaller case.
Nvidia RTX 5090 vs 4090: Early Thoughts
The RTX 4090 GPU did make its debut at $1,599, whilst the RTX 5090 hath been priced at $1,999. Many did foresee Nvidia would fix the price of the Blackwell GPU near $2,500, yet the company hath chosen a reasonable cost. With a $400 increment in price, thou dost obtain a much swifter GPU, rendering strong performance in all domains, whether it be gaming, AI tasks, or creative rendering.
Moreover, with the RTX 5090, thou dost receive DLSS 4 with the Multi Frame Generation feature and access to 32GB of ultra-swift GDDR7 memory. Verily, we must await in-depth benchmark testing, yet the Nvidia RTX 5090 GPU doth seem a powerhouse. If thou dost already possess the RTX 4090, perchance thou need not upgrade immediately, but for those with an RTX 30 series card, the RTX 5090 is a recommendation most simple.
