The fact is that Intel has never been to IBM in R&D, at least that's what it's shown after years. Of course they only meet in selected markets, but similarly both companies have similar products.
First, that of the CPUs, IBM gives Intel a serious upgrade, though it is also true that it does so with certain nuances.
IBM z15: a company commitment made by GlobalFoundries
Although such a processor was launched in September and there has not been much talk of it, IBM wants to show what position we have in the market within the ISSCC 2020 conference.
We've explained the big details that so far are unknown, but before revealing it, we should highlight the most prominent features of the IBM z15, which is truly amazing.
First, it is designed for the lithographic process of 14 nm from GlobalFoundries and contains 12 cores and 24 threads under buildings Z (CISC). This number of cows reaches amazing speeds 5.2 GHz, which means more than, at least temporarily, its competitor Intel.
It has a brain that the basic area takes for nothing less 696 mm2, which in general and compared to that given by the Swat, means that it is about 3 to 4 times larger than its competitors.
Including 9.2 billion transistors and has an amazing number of 256 MB of L3.
Functionality increases with the cache power
If this were not enough, IBM has enough data that is unknown so far, starting with profitability 10% single cord and 20% cord, a further increase in the height of a single design.
In addition, L2 was increased by 33%, L3 sizes and 43% were L4, numbers from another planet when we compared it to what is offered on a desk or server. In addition, up to 12 cores have been added, which means an additional capacity is 20%.
This leaves us with a +14% CPI if we add an increase in the frequency of a single Thread, which was higher than the product originally expected. Without a doubt, the announcement aims of the brand, which shows the world the size and strength of its engineers and structures.
It is true that death is great which is why they import a cache such as this time, but it is also true that the 14 nm GF process is outdated and the duality is not strictly true, as opposed to the 14 nm process Intel ++, is very well done.
