Limited due to capacity issues? Intel DG2 GPU will use TSMC 7nm process
According to adoredtv reports, industry sources said that the Intel DG2 GPU uses TSMC's 7nm process technology instead of Intel's own 7nm process, and the new GPU will be launched in 2022.
As early as last October, it was reported that Intel said that its Xe GPUs will use their own 7nm process, so it is unlikely that the upcoming GPUs will be commissioned by TSMC. According to the Intel process roadmap, the 7nm process will be the fourth next year. Ready for the quarter. Adding the two together, Intel suddenly changed its foundry direction and asked for help from TSMC.
The report pointed out that Intel's decision may consider its 7nm capacity issue. In addition, because the new node costs more than the old node, and the GPU used for the game cannot be too expensive, the cost issue is also considered. This may also be related to Intel's crazy hire of former AMD and Radeon employees, who are familiar with TSMC's process.
The source said that Intel's 7nm will be mainly used for the CPU, and the GPU currently only knows that the "Ponte Vecchio" of the Xe_HPC architecture has a play. As it is said, it is not unreasonable for Intel to seek TSMC foundry in order to balance production capacity.
On the other hand, in 2022, it is clear that the capacity of 7nm will be more sufficient. With the continuous advancement of advanced process technology, the price of 7nm will gradually decrease. This may also be the reason why Intel chose to wait. In addition, although Intel has released its tiger lake and DG1, but both are Gen 12 LP, while DG2 is based on Gen 12 HP, which may not have been designed yet.
As early as last October, it was reported that Intel said that its Xe GPUs will use their own 7nm process, so it is unlikely that the upcoming GPUs will be commissioned by TSMC. According to the Intel process roadmap, the 7nm process will be the fourth next year. Ready for the quarter. Adding the two together, Intel suddenly changed its foundry direction and asked for help from TSMC.
The report pointed out that Intel's decision may consider its 7nm capacity issue. In addition, because the new node costs more than the old node, and the GPU used for the game cannot be too expensive, the cost issue is also considered. This may also be related to Intel's crazy hire of former AMD and Radeon employees, who are familiar with TSMC's process.
The source said that Intel's 7nm will be mainly used for the CPU, and the GPU currently only knows that the "Ponte Vecchio" of the Xe_HPC architecture has a play. As it is said, it is not unreasonable for Intel to seek TSMC foundry in order to balance production capacity.
On the other hand, in 2022, it is clear that the capacity of 7nm will be more sufficient. With the continuous advancement of advanced process technology, the price of 7nm will gradually decrease. This may also be the reason why Intel chose to wait. In addition, although Intel has released its tiger lake and DG1, but both are Gen 12 LP, while DG2 is based on Gen 12 HP, which may not have been designed yet.