Chip weather forecast: Price wars are imminent. Squally showers.
|
|
|
Intel never stopped numbering its CPUs
REMEMBER the good old days, when processors were known by their - usually five digit - numbers? So, you could compare Intel 80486 against Motorola 68040, against National 32532, against Alpha 21164 (this last one would win, of course).
Then, when it realised it was unable to trademark numbers, Intel switched to the Pentium medicinal-sounding moniker, and from there we had Itaniums, Celerons, joined by Athlons and Opterons. Then came Centrino, Turion, Core2, Phenom and all other cryptical what-nots. Hard to fathom?
Well, not really! Instead of asking an Intel guy for a Core 2 Extreme QX 9650 3 GHz FSB1333 CPU, simply ask for 80569XJ080. He should know exactly what you want, with far fewer bytes or words.
Basically, Intel continued using the numbers for its CPUs, engraved on the die heat spreaders for years, till now. These are still 5-digit wonders, all smaller than the last 'official' one, the 80586.
For instance, the old favourite, 3.46GHz Presler Extreme dual core, Intel's first CPU to run at 4.27GHz stable without any voltage bump, was just an 80553 CPU. It had an added PH099 statement for desktop 3.46GHz clock version - if the number was 108, it would mean a 3.73GHz version. At the end, there is also a cache size stated as 4M or 8M, but we will ignore it here as it is removed from the 45nm generation engraved inscriptions.
The quad-core 65nm Kentsfields and Clovertowns also have this simplified naming. The X3230 Xeon 2.66GHz FSB1066 will just be an 80562 CPU, with KH067 add-on which translates to "XeonUP 2.66G".
All the desktop Kentsfields also bear the 80562 moniker. The dual-CPU enabled Clovertown flavour is called 80563, though, adding a bit of "value" for the dual socket capability. So, a 3GHz FSB1333 Clovertown from those expensive Apple Mac Pro workstations will be known simply as 80563KJ080.
Finally, the most recent darlings of the press, the lovely Yorkfield and Harpertown 45nm CPUs. All desktop Yorkfields and their uni-CPU server siblings are known as 80569. So, it is 80569XJ080 for the QX9650 Extreme Desktop, or 80569KJ073 for the X3660 XeonUP. The Harpertown dual-socket Xeon is now 80574KL080 - without cache markings this time.
The nostalgic souls among our readers can enjoy it: all the numbers are still
there. And, for simple minds, 80563 will always be clearly better than 80562,
not to mention 80574 outshining them all.
Also, knowing Intel's real numbering system means the unscrupulous chip trader next door won't be able to sell you a Clovertown claiming to be the same-clocked Harpertown before you plug it into the system when it may be too late.
Convincing a layman that "Harpertown" is better than "Clovertown", or " Yorkfield" than "Kentsfield" or "Smithfield" could get the populace of the named places into fistfights.
Who knows, by the time Nehalems come in six months or so, we might see one of them coded 80586 again. Ahem. Just a Back To The Future joke. µ
Tags: Intel
Share this:
Share
"Who knows, by the time Nehalems come in six months or so, we might see one of them coded 80586 again"

You really don't know what in the world you are talking about..

Six months or so?

Yeah,fat chance.Try a year.
When I read computer magazine @ late 1996, I read an article that say:
585 = pentium
686 = pentium pro - pentium III
786 = willmate

In the 1996 pentium is known as P5. All motherboard @ that time use P5 as suffix. That thing is dejavu again with presscott (ie the P5 suffix). AND I Thing when intel lauch it P5x chipset, this will be happend again.
This is quite interesting but with all those letters, at least initially, a chart or crib sheet would be nice to help decider all the numbers and letter, but still much easier than the marketing names. When I read the title I was hoping to see a listing of all the main cpus since the 486 to see the evolution of the numbering system.

The only issue I see is that these numbers do not seem like a continuation of the original 80x86 numbering system, even the x87 match co-processor numbering made sense and fit with the main numbering system.

Not sure what the 80 denoted but the numbers 1-5 denoted the generation and the 86 of course denoted the architecture.

Its too bad these ones aren't something like 80A86 or 801086 with additional letting and numbers to denote smaller differences. & because Netburst was a very different x86 like architecture and should not count, so PII=PentiumPro=686, PIII=786, PM=886, Core=986, and Core2=A86 or 1086 (think hex for the A). That type of number was what I was expecting to be reading about, but I could get used to the newer system.


Enough of my ramblings, great article anyways

"These are still 5-digit wonders, all smaller than the last 'official' one, the 80586."

You do know that the first Pentiums were called 500, 501 and 502?

en.wikipedia.org/wiki/Pentium
...that numbers 1 to 4 in the internal numbering scheme (80569XJ080) give you 9650 for the Core 2 Extreme QX 9650 FSB1333, when spelled backwards? I think not... :)
It's amazing how it takes the Inquirer to write an article talking about Intel's Box Order Codes. They're nothing magical, but hey if it floats your boat!