[ad_1]
In short: Nvidia divides its Turing silicon into A-matrices and not A for RTX 2080 Ti, 2080 and 2070 graphics cards. This has allowed Nvidia to classify high-quality Turing Matrix A matrices into the overclocked factory cards of choice. board partners. It also allowed the use of non-A silicon for non-overclocked discounted cards. Now, Nvidia will only produce the TU104-410 and TU106-410 matrices, stopping production on lower quality non-A matrices. This implies that the manufacturing process has been optimized and that we can expect the silicon Turing to be of better quality in the future.
Igor Wallossek and Tom & # 39; s Hardware Germany announced that Nvidia would stop classifying Turing A chips and would also stop producing non-A chips on RTX 2080 and 2070 cards. This means that for the future, Nvidia will only offer one variant of Turing Silicon for cards.
Theoretically, this is good news for players, because the non-A silicon is the cheapest and least powerful variant that limits the margin of overclocking. This should mean that TSMC's 12 nm "FFN" process is mature enough to maximize yields, which means Nvidia no longer needs to separate silicon from Turing. The "new" silicon, which is already in production, will be the TU104-410 (RTX 2080) and the TU106-410 (RTX 2070). Nvidia would apparently sell the new matrices at the same price as the previous non-A counterparties.
TechPowerUp spotted the separate Turing matrices some time ago, as it was discovered that there were two different device IDs per Turing graphics card. For example, the RTX 2080 uses both the TU104-400-A1 and TU104-400A-A1 variants. This allowed Nvidia to provide card partners with higher quality chips for flagship models, as well as cheaper non-A chips at lower prices. We do not know if Nvidia plans to do the same with the RTX 2080 Ti and TU102.
[ad_2]
Source link