LONDON – The Semiconductor Industry Association has confirmed that it mis-stated the quarterly sequential comparisons for Japan and the global semiconductor market in its statement of global semiconductor sales for September issued Monday (Oct. 31).
The effect was that the SIA under-reported the size of the sequential growth in third quarter sales compared with the second quarter both in Japan and globally. The global chip market grew 3.5 percent in the third quarter compared with the second quarter, considerably more than the 2.1 percent the SIA announced and reported by EE Times on Monday (Oct. 31). The SIA said it is preparing an updated statement.
The error does not affect the September three-month average data posted by the SIA based on World Semiconductor Trade Statistics (WSTS) data, but was due to comparison error with earlier data.
It appears that the SIA accidently compared September data – the average of September, August and July, which effectively represents the actual chip sales in the third quarter – not with June's average, as it said it had and representing the second quarter, but with the May averaged data.
As a result the Japanese region's $3.80 billion averaged data in September should be compared with $3.30 billion averaged data recorded for June 2011, which produces a 14.9 percent sequential growth rather than 13.7 percent growth as stated by the SIA on Monday (Oct. 31).
Similarly the total chip market averaged in September remains $25.76 billion, as stated by the SIA, but is some 3.5 percent ahead of the equivalent figure for June, not 2.1 percent as previously stated.
The SIA and the European Semiconductor Industry Association publish monthly data as three-month-average figures. This makes the sales of any given month the average of the sales for that month and the previous two months. The SIA and ESIA prefer to present data in this way as it smoothes out the actual data that usually show troughs at the beginnings of the quarters and peaks at the ends of the quarters.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.