As expected, Nvidia has announced the RTX 4000 series (Ada Lovelace). Specifically, it was about the RTX 4090 and the 12 and 16 gigabyte variants of the RTX 4080. Nvidia CEO Jensen Huang also presented the new DLSS 3.
Once again we offer you the unique opportunity to ask us your questions about the RTX 4000! So if you have any questions about the new GPUs before our test, please ask them here in the comments below the article. Germany’s supreme graphics deity, our Raff, will then answer your questions completely in an answer video next week.
The new flagship from Nvidia is the RTX 4090. It is expected to be in stores on October 12th. The rumor mill was in good shape recently, because the RTX 4090 relies on the AD102 with 16,384 FP32 shaders and 2,520 MHz boost. The GDDR6X memory with 24 gigabytes should be connected at 21 gigabits per second and with 384 bits, which means around 1 terabyte per second of memory bandwidth – as with the RTX 3090 Ti.
With the RTX 4080, the last assumptions have been confirmed: Nvidia differentiates between a 12 gigabyte and 16 gigabyte version, but the two variants differ not only in the graphics memory, but also in important technical parameters.
Ray tracing, tensor and DLSS3
Third-generation ray tracing cores are said to deliver 200 RT TFLOPs. According to Nvidia, Ada’s third-generation RT Cores have twice the throughput at the intersection of rays and triangles, increasing RT TFLOP performance by more than 2x. The new RT cores also contain a so-called Opacity Micromap (OMM) engine and a new Displaced Micro-Mesh (DMM) engine. The OMM engine enables much faster ray tracing of alpha-tested textures used for foliage, particles, and fences, for example. The main purpose of the DMM engine is to enable ray tracing of geometrically complex scenes in real time.