Nvidia’s launches of graphics cards in the entry-level segment for portable use generally do not bring significant news, as they tend to be positioned as circuits that are sharper than, for example, Intel’s integrated graphics parts. In fact, the Geforce MX350, a “Pascal” -based circuit made of 14 nanometers, has occupied this part of the market on behalf of Nvidia – until now.
The company is now taking almost immediate protection for its countdown to the unveiling of the architecture “Ampere” and is sneaking out its first PCI Express 4.0-compatible graphics card – Geforce MX450. Like the MX350, the new card is not intended to be a performance monster, but still offers 1,024 CUDA cores in a TU117 circuit from the company’s existing “Turing” catalog.
According to PC World, the Geforce MX450 offers support for both GDDR5 and GDDR6, while Techpowerup specifies the power consumption to 50 watts and the performance as comparable to old classics such as the Geforce GTX 580, GTX 1050 and Radeon HD 7950. The 200 mm TU117 circuit is similar to the others “Turing” circuits built at TSMC of 12 nanometers.
The circuit’s compatibility with PCI Express 4.0 is highly likely to meet Intel’s “Tiger Lake”, Geforce MX450’s natural pair horse at the platform’s launch this fall. It is worth mentioning that although the MX450 certainly supports PCI Express 4.0, the card is still limited to a bandwidth of eight lanes, x8, which in practice means the same bandwidth as PCI Express 3.0 x16.
More graphics cards from Nvidia with support for PCI Express 4.0 are expected soon, when the “Ampere” architecture for consumers and the Geforce RTX 3000 series will be unveiled on 1 September.