Nvidia Korea CEO Expands On RTX 4080 12GB Cancellation

    13
    Nvidia Korea CEO Expands On RTX 4080 12GB Cancellation

    Nvidia’s Korea CEO Kim Seung-gyu provided further details on the cancellation of the GeForce RTX 4080 12GB, according to a report by Bodnara. In essence, Kim Seung-gyu said that the RTX 4080 16GB had much higher performance than the RTX 4080 12GB and that the latter’s moniker was “confusing.” According to Seung-gyu, Nvidia usually gives GPUs model names so that performance can be predicted, but the stark performance differences between the two 4080 models would perplex buyers.

    The official USA release from Nvidia wasn’t quite as explicit; it was reduced to a single phrase that stated that the categorization was unclear without outlining the reasoning behind it.

    The 4080 cancellation appears to increase the entry-level price of the 4000-series to $1,200, and there is no information on when a less expensive option would hit the market, making it significant that the cancellation could be self-serving. That guarantees a high entry price and will also drive plenty of customers to Ampere cards, helping the business get rid of its excess of those types. Recently, Nvidia deducted $1.32 billion from reserves for excess Ampere inventory (opens in new tab).

    In case you missed it, Nvidia terminated the RTX 4080 12GB SKU permanently seven days ago after acknowledging that the card’s nomenclature was unclear. Between the 12GB and 16GB variants, there were also significant changes in performance and specs. This was corroborated by Nvidia’s own testing, which revealed a startling 30% performance difference between the two SKUs of the 4080 12GB and 16GB.

    Clearly, there was more going on than merely a change in memory size. In contrast to the 4080 16GB, which utilised AD103 with 9728 CUDA cores and 76 SMs, the “unlaunched” 4080 12GB used a different chip, AD104, with 7680 CUDA cores and 60 Streaming Multiprocessors (SMs). The 16GB card’s memory was also clocked faster, which suggests that 24Gbps GDDR6X chips from Micron are almost probably being used.

    However, we do know that AIB partner versions of the RTX 4080 12GB were already developed, and Nvidia will be “refunding” those partners — probably for the packaging, which will need to change. Nvidia hasn’t yet disclosed a replacement plan for the RTX 4080 12GB. These “ex-RTX 4080 12GB” cards will ultimately receive a new moniker from Nvidia, most likely sooner rather than later. We anticipate that the fundamental design will eventually manifest as an RTX 4070 or RTX 4070 Ti, however Nvidia may make a little modification to the specifications to avoid allegations that it is only renaming a 4070 GPU.

    You may learn more about the “unlaunched” RTX 4080 12GB’s full specifications in the Nvidia Ada Architecture section. In the meantime, the GeForce RTX 4090 has risen to the top of our GPU benchmarks hierarchy and is currently considered the quickest of the top graphics cards. When the 4080 12GB successor materialises, let’s hope the price will be reduced to reflect the performance and feature reductions.