If I remember correctly, Ollila said:
"Millions of GPRS phones, and then we do not
mean 0.9 millions, but more than 2 millions"
Linguistically there is this finnish way of
expressing the number between 1.0+epsilon up
to at least 2.0-epsilon millions, very precisely.
Not just one or many, clearly excluding this
roundoff problem of 0.5 million and 1.5 million.
Ilmarinen
Roundoff and truncation, a common problem in DSP, as
well as broken antennas for RF guys and girls.
Avoiding getting into the cumulative statistics of
QCOM, cumulativ truncation errors can be a real
problem, but one can always hope they keep them
within the marketing department, out of the DSPs
and project time schedules. |
