Battle still in process
I agree with Richard Barry in the article "PC Goliath, the last stand?" that the battle of processors seems to be over. However, I disagree with his view of the outcome. While memory speed is holding the PC back, the processor market seems to have matured - gone are the dramatic leaps in performance of the early 90's. Instead, we have a number of manufacturers providing roughly equal levels of performance with varying prices. Each year the manufacturers bring out a slightly improved model.
The current manoeuvering seems to be for position within the market place.
AMD and Cyrix are competing for the small business workstation and home PC. Intel is still showing interest in this area with its Pentium range.
However, with the PPII and Slot One, Intel seems to be aiming for what may be a more lucrative server market - especially given physical problems of fitting Slot One into a PC box.
Given Barry's definitions, Slot One cannot be aimed at being a replacement for Socket 7. As he says: "Events in PC history involve setting standards which everyone can follow." This being true, how can Intel possibly hope to lead the field with a proprietary technology which "Cyrix and AMD are not allowed to use"?
According to Intel, the technological advances planned for inclusion in future Slot One motherboards are numerous (AGP, USB, IEEE 1394). Each of the technologies Intel mentions will also be available on the Socket 7 motherboards. There does not seem to be any single feature planned for Slot One which will not become available on Socket 7.
This leads me to two possible conclusions.
The first possibility is that Intel is aiming for the server market.
By tying businesses to proprietary Slot One, all future processor upgrades must come from Intel. In this case Intel and Slot One - rather than trying to beat the desktop/home PC competition - are withdrawing from it.
The second possibility is that Intel is hoping to sell Slot One to AMD and Cyrix, at a cost which will be passed on to the consumer. If this happens then Barry can wave goodbye to his dream of "high-end workstations at a fraction of the price".
If PPII and Slot One survive, in some form, after Digital's patent law suit, then I would imagine that the first possibility is most likely - otherwise Intel may be in further trouble from the monopolies people.
Leaping to a solution
I read David Morrison's letter (PC Week 22 July) about the year 2000 problem and whether it's a leap year or not.
I had exactly the same problem some six years ago when researching this point for an application rebuild. I always remembered the fairly simple rule that if the value of the year was exactly divisible by four then the year would be a leap year - so therefore 1900 and 2000 would be leap years via that rule.
However, it is also true that to make one final correction to the earth's annual rotation about the sun, then every 400 years the leap year on a new century value is omitted. This designation was made sometime around 1670 as far as I can remember, so the years 1700, 1800 and 1900 were leap years, but the year 2000 is not. Not that it will worry any of us, but the year 2100, 2200 and 2300 will be leap years.
The above determination is a remarkable indication of the accuracy available to the astronomers of the 17th century and also dictates that your PC BIOS is wrong to change from 28/2/2000 to 29/2/2000 and not 1/3/2000.
Leap seconds occur every so often to account for the minute but measurable slowing of the earth's daily rotation. When and if the leap year rule is revised, I regret I have no knowledge, nor can I think I need worry.
My source was some most respected Almanac whose name I now don't remember, but I suspect that Encyclopaedia Brittannica will have a reference somewhere.
I hope that this is enlightening.
The two major problems of year 2000 compliance - the first of which I have had fixed in our applications for some years now - are ensuring that historical items are properly aged. Hence the two-digit 00 for year is correctly assigned to either 1900 or 2000 as should be the case, and the second is the apparent backward jump in time which will occur if the change from 12:59:59 on 31/12/1999 goes to 00:00:00 on 1/1/1900.
For applications and dedicated systems that compute motion and position the result of this huge backward time jump will be to produce answers that no programmer ever anticipated and the likely result is a programmed abort or reset.
So while I don't believe that aeroplanes will fall out of the sky at midnight on 31/12/1999, I do believe that there will be many unanticipated crises occurring in time-critical applications unless, and until, the potential problem is eliminated and/or humans can react and control those very situations.
In reply to David Morrison's letter (PC Week 22 July) AD 2000 is a leap year. In the Gregorian calendar all years divisible by four are leap years except century years that are not divisible by 400.
Thus AD 1600 and AD 2000 are leap years but AD 1700, AD 1800, AD 1900 and AD 2100 are not. Doing this drops three leap days every four centuries and is needed because the year is 365.2422 days long (approx) rather than 365.25 days exactly.
This calendar replaced the Julian calendar (which assumed a year of 365.25 days) over 400 years ago, and to backdate it to the start of the AD calendar 10 days were dropped altogether at the time it was adopted. In the modern age the calendar is kept on track by adding the occasional leap second.
In the same vein, I would like to take the opportunity to point out that since there was no 'year zero' between the BC and AD calendars the AD calendar started on 1 January, AD 1. Thus all millennia in the AD calendar start in a year ending in '1' and end in a year ending '0'. This means that AD 2000 is the last year of a millennium and AD 2001 starts a new one, a fact that many people writing to the technical press seem unaware of.
The worst effects of the so-called millennium problem will therefore have come (and hopefully cured) before the next millennium starts!
IBM hopes that its new tool will avoid bias in artificial intelligence
Found by calculating the strength of the material deep inside the crust of neutron stars
Can highlight in real-time the relevant regions of an image being described
Double legal trouble for Musk as he also faces civil lawsuit over renewed British pot-holer 'paedo' claims