The silicon supports up to 24 GB of GDDR6X or up to 48 GB of GDDR6+ECC memory (the latter will be seen in the RTX Ada professional-visualization card), across a 384-bit wide memory bus. The streaming multiprocessors (SM) come with special components that enable the Shader Execution Reordering optimization, which has a significant performance impact on both raster- and ray traced graphics rendering performance. The AD102 physically features 18,176 CUDA cores, 568 fourth-generation Tensor cores, and 142 third-generation RT cores. This is thanks to TSMC 4N offering nearly thrice the transistor-density of the Samsung 8LPP node on which the GA102 is based. Built on the 4 nm EUV (TSMC 4N) silicon fabrication process, the chip has a gargantuan transistor-count of 76.3 billion, a nearly 170% increase over the previous GA102, and a die-size of 608 mm², which is in fact smaller than the 628 mm² die-area of the GA102. The AD102 silicon on which NVIDIA's new flagship graphics card, the GeForce RTX 4090, is based, is a marvel of semiconductor engineering. The Tensor cores deployed on Ada are functionally identical to the ones on the Hopper H100 Tensor Core HPC processor, featuring the new FP8 Transformer Engine, which delivers up to 5X the AI inference performance over the previous generation Ampere Tensor Core (which itself delivered a similar leap by leveraging sparsity).
With Ada, NVIDIA is introducing its 4th generation of Tensor core (after Volta, Turing, and Ampere). With traditional raster graphics, SER contributes a meaty 25% performance uplift. Besides generational IPC and clock speed improvements, the latest CUDA core benefits from SER (shader execution reordering), an SM or GPC-level feature that reorders execution waves/threads to optimally load each CUDA core and improve parallelism.ĭespite using specialized hardware such as the RT cores, the ray tracing pipeline still relies on CUDA cores and the CPU for a handful tasks, and here NVIDIA claims that SER contributes to a 3X ray tracing performance uplift (the performance contribution of CUDA cores). We're yet to receive a technical briefing about the architecture itself, and the various hardware components that make up the silicon but NVIDIA on its website gave us a first look at what's in store with the key number-crunching components of "Ada," namely the Ada CUDA core, 4th generation Tensor core, and 3rd generation RT core. Yesterday, NVIDIA launched its GeForce RTX 40-series, based on the "Ada" graphics architecture. There are no plans for EVGA to partner with AMD or Intel at this time when it comes to graphics cards and the company stressed they will continue to sell and support current-gen GPUs having retained enough units for RMA purposes too. The company's CEO confirmed as much to a few media channels citing poor margins and a challenging, stressful relationship that was no longer fruitful. Given EVGA's revenue sheets point to nearly 80% contribution from being an NVIDIA add-in card partner, this effectively also means an end to a long partnership with NVIDIA.
Today EVGA decided to throw a massive curve ball by formally announcing the company is canceling its plans to carry the next generation of graphics cards. TechPowerUp was doing due diligence in collecting the facts while keeping emotions aside from contacts who were understandably not in the best of moods, and one thing common across the board was there was something major coming up dealing with the EVGA GPU product line.
Some even went far enough to say they would share more in a few weeks time about how they felt exactly about their time there, the various issues that kept them from doing their best, and also that at least a couple of ex-employees were let go. When pushed further, several hinted towards some decisions being made by EVGA's management, including CEO Andrew Han, that would jeopardize their future. Towards the latter half of August, multiple EVGA employees involved in technical marketing and engineering had let us know privately that they were leaving the company for other ventures.
0 kommentar(er)
