Some pretty interesting rumors about AMD’s upcoming Zen 2 architecture are surfacing from none other than Chiphell – the enigmatic engineering forum responsible for the vast majority of leaks. This latest post by someone testing AMD Zen 2 claims that IPC improvements will be in the 10-15% range. IPC is an important metric for clock-to-clock performance and a 10-15% increase is very impressive from one generation to the next.
AMD’s 7nm based Zen 2 architecture will have IPC improvement between 10% to 15%
AMD has previously talked about the Zen 2 uArach but most of the details were not finalized at that point or they simply did not wish to discuss them. Considering there is an event going on in Maranello right now regarding the Threadripper 2 platform, its not surprising that we are hearing more details of future platforms leak out.
Related AMD Next-Gen 7nm Zen 2 CPUs Rumored To Feature 10-15% IPC Increase, Up To 16 Cores on AM4, 32 Cores on TR4 and 64 Cores on SP3
A 10-15% performance gain in IPC would make Zen 2 a much much higher performing architectural path then the 7th gen to 8th gen shift for example. It looks like AMD features revolutionary leaps over evolutionary – which is perfectly in-line with their philosophy. The post further states that at Zen 2 AMD will move on to 16 cores on the mainstream platform (it is 8 currently) and they will achieve this by increasing the core count of the CCX from 4 to 8. Its worth noting that this is in direct contradiction to some leaks which had suggested a 6-core CCX.
Keep in mind that any IPC gains that we are talking about here will only add on top of the gains that we can expect by simply moving to a smaller process (and the economies of power and scale that come along with it).
AMD has preparing the 7nm version of its Zen architecture for quite some time now and word on the rumor mill is that it will TSMC will get a major share of fabrication – which will be a big win for AMD considering their processes are usually fairly high performance. This also has interesting implicaitons for the Wafer Supply Agreement which usually prohibits AMD from reaching out to other foundries.
Related TSMC Begins Mass Production of 7nm Process, AMD Vega 7nm Confirmed, Zen 2 CPUs Expected Too – Production Capacity To Increase By 3 Times Next Year
The 7nm TSMC process has 35% more power efficiency and yields compared to the standard 16FF+ process. According to the US technology forum, 50% of the wafers will be completed before the year ends with production capacity expected to multiply by 3 by next year. TSMC will also be beginning mass production on the 7nm+ using EUV technology next year, which will further reduce power consumption by 10%, compared to 7nm.
Compared to its 10nm FinFET process, TSMC’s 7nm FinFET features 1.6X logic density, ~20% speed improvement, and ~40% power reduction. TSMC set another industry record by launching two separate 7nm FinFET tracks: one optimized for mobile applications, the other for high-performance computing applications. via TMSC
This means that TSMC will be able to deliver a mass total of 1.1 million units by the next year, a three-fold increase over the current year. The AMD 7nm OEM orders are expected to being production in Q4 2018 and TSMC is expected 7nm to account for up to 20% of their Q4 2018 revenue.
AMD can select and use Fabs of both TSMC and Global Foundries to create their next-gen processors. This is made possible by Global Foundries using similar 7nm pitches and SRAM cells that are very close in design to TSMC, allowing Zen 2 7nm processors to be developed on either Fab without major differences.
Later this year, GF will use immersion steppers to tape out its first 7-nm chip, an AMD processor. An IBM processor will follow with ASICs coming in 2019, said Patton.
GF made the size of its 7-nm pitches and SRAM cells similar to those of TSMC to let designers like AMD use both foundries. AMD “will have more demand than we have capacity, so I have no issues with that,” he said of AMD using the Taiwan foundry. via EETimes
