Briefly Study Case - Intel CPU trend (2004 – Q2, 2010)

Technoart

Distinguished
Apr 8, 2009
61
0
18,640
Briefly Study Case - Intel CPU trend (2004 – Q2, 2010)

This Study Case is dedicated to Intel Pentium 4 1st generation Willamette chip until Intel i7-980X Gulftown for mainstream and high end processors (not including Celeron and Atom).

Still remember the good old days when everything is sufficient for SimCity 2000 and Windows ME (2000) with all parts are still in MB capacity and not like on GBs? The improvements of CPU along with these years are fast and furious. Why am I saying like that? Within roughly 6 years of time, Intel had changed a single or solo core CPU into a same TDP or lower but higher output power with 6 cores inside a CPU.

Since everybody familiar with Core 2 CPUs, I will not mention about it.

Let’s begin with Intel’s Pentium 4 1st generation Willamette chip which released on Nov 2000. The Willamette chip uses NetBrust Architecture so as Northwood (October 2001), Prescott (February, 1 2004), and Cedar Mill (January, 1 2006). All these CPU are single core with the fabrication process from 180nm to 65nm (Willamette to Cedar Mill) and with micro Pin Grid Array of 478 pins or mPGA socket 478. The last Pentium 4 production is Cedar Mill including the extreme CPU with HT under Pentium 4 which last approximately Dec 2006 after first Core based CPU are out on Aug 2006.

Some of Pentium 4’s Prescott and Cedar Mill CPU and also Pentium D’s Smithfield and Presler used Land Grid Array 775. The mentioned processors are still single core. The trend of the Intel is to end all single cores for during that period to allow PC user to migrate into dual cores or more cores. Back into now, after a series of dual core and quad core processors Core 2 based are out, the i3, i5 and until i7-975 which also had HT represents the ending of dual and quad core just after socket LGA 775 transform to socket LGA 1156 and socket LGA 1366 in the next coming years .The i7-980X means that the starting of six cores CPU will be soon into mainstream until users are not ready (premature market demand) to make way for six cores mainstream in their PCs since it is too near in time.

The point is when a HT processors are released, the product with old n cores (n = number) and new form of socket with be terminated along with the new n + n cores product and new form socket are produce massively. This represent that the ending of n cores CPU which had HT ready are the highest available in n cores CPU which applies for both Pentium 4 and I series CPU in the study case.

Finally, after 6 years, we can summaries the trend for now into below:

Formula:
A HT CPU out, n cores CPUs ends, n+1 socket forms and n+1 CPUs in.

Stay on for the next topic – Hard disk, a revolutionary change. Magnetic Disk and SSD outdated, use atomic instead for storage!
 

SidVicious

Distinguished
Jan 15, 2002
1,271
0
19,280
Decent read, if a bit convoluted even for someone who has been following the hardware industry daily since before the first P4 were released.

You should have spent more time on structuring your thoughts and making the timeline stand out on its own, it look a bit jumbled and could use important events such as process shrinks and product introductions as landmarks.

BTW, the first Willamettes were Socket 423, all Netburst based "dual cores" were in fact two dies in a single package, LGA 1366 came before 1156 and LGA775 was in fact two different generations of motherboard, the first, initially named Socket T, being exclusively meant for P4s due to their chipset, the second, know as Socket 775, with full C2D/C2Q support and backward compatible with P4s.

Also, a CPU fitting a socket was no guarantee of motherboard/chipset compatibility, for example, a number of early enthusiast class "Bad Axe" Socket 775 motherboards were unable to support Core 2 based CPUs without some modification to their power regulation due to electrical limitations, even if the chipset and BIOS were adequate, this was even more of an headache with Socket 478.

That formula doesn't have any merit for being made after the fact nor predictive value as the introduction of sockets and new products happen in the real world for factors that are outside the scope of your formula.