The world’s largest foundry or chip factory, Taiwan Semiconductors, sees its dominance threatened by Intel’s announcement that it wants to compete with the Asian foundry and Samsung’s new nodes. TSMC has therefore developed N4P, a new manufacturing node in order to attract more and more new processor designs.
There is no doubt that there is a particular race between the big three foundries where new manufacturing nodes are created for future chips, the aim is to get the big designers from different segments of the hardware market to design their future products in these new processes. This market is so important that it has also become a strategic market for manufacturing countries and at this time Taiwan with its TSMC foundry became the market leader after the Intel misadventures with its 10nm node, now renamed Intel 7, but in a competitive and short-lived market, there is no pause.
TSMC presents its N4P node
The Taiwanese foundry has just introduced a new chipmaking node to its portfolio. Under the name of N4P TSMC has just presented a manufacturing node which will share space with nodes N5, N4 and N3 of the same foundry. it is a process conforms to the design rules of node N5, and therefore, it could be used to manufacture the same processor designs.
N4P node is therefore a intermediate knot, which is itself an improved version of the N4. CPUs and GPUs manufactured under this node will have a 11% improvement in clock speed compared to the same part manufactured under node N5 and 6% compared to N4. At the same clock speed and relative to N5, the new N4P node has the same c chipwill consume up to 22% less and all this occupying a 6% less space. They also talked about improvements in the manufacturing process, and these translate into a reduction in the number of masks, which will speed up production time and therefore
The launch date of the new node? TSMC expects N4P to begin manufacturing in the second half of 2022, so it will be used to launch more energy efficient versions of TSMC’s 5nm node designs. Given the high costs of deploying new chip-making technology, it’s clear that the big deals are already in place. Are we on the trail of better performing versions of Zen 4, RDNA 3 and RTX 40 by 2023? Who knows.
TSMC Intermediate Nodes
Intermediate nodes allow designs from a previous node to be used as part of a more advanced manufacturing process and therefore release versions with better technical specifications and CPU, GPU or APU performance that already exist on the market. TSMC’s N4P manufacturing process is one of them.
TSMC’s general strategy for some time has been to create intermediate nodes between its large nodes to offer customers whose products do not require a full node hopping a more technologically advanced version than the current version. This was the case with the 12FF which was an advanced version of its 16FF node, also with the current N6 which is an optimized version of the N7 which takes part of the technology of the N5 and of course the future N4 and its N4P variant.
The reason these fabrication nodes exist is that if you want to port a design from a current node to a more advanced node, the entire chip needs to be completely redesigned as design standards change. In many cases, to cover certain market segments, it is better to go for an improved version of current technology. The problem is, the manufacturing node used by the original models can be a limitation on the density or number of transistors, clock speed, consumption or all of the above in combination.