pharaoh fortune slot
Welcome, Guest. Please login or register.
December 14, 2018, 10:25:17 AM

Login with username, password and session length
* Home Help Search Login Register
|-+  General discussions
| |-+  General Discussion
| | |-+  SMP using different processors?
0 Members and 1 Guest are viewing this topic. « previous next »
Pages: [1] Print
Author Topic: SMP using different processors?  (Read 255 times)
Wireless Fanatic
Hero Member
Offline Offline

Posts: 2210


View Profile WWW
« on: February 16, 2004, 05:06:05 AM »

Today I pieced together a new server for my lab. It has two P3 processors - one 1GHz, and one 850MHz. Both have a 100MHz FSB and the same stepping pattern. The system boots fine, and I was able to get FreeBSD loaded without any trouble. My question is, can I expect to run into any problems with this setup in the future? I recompiled everything (make world and make kernel), and it seems to have no trouble with the processors, which I'm assuming are both running at 850MHz (what the BIOS and FreeBSD tell me). One thing that concerns me is that about 1/3 of the times that I boot up the system, the BIOS detects 2 1GHz P3s. Does that mean that the 850MHz processor is being overclocked, or is the BIOS just randomly picking a processor to display the speed for? This is a 1U server, so I'm not too confident in the cooling system's ability to properly handle any kind of overclocking. So far I've been immediately switching off the system whever the BIOS shows 2 1GHz processors, so I haven't been able to check what the OS is seeing.

Senior Member
Hero Member
Offline Offline

Posts: 917


View Profile WWW
« Reply #1 on: February 16, 2004, 06:45:55 AM »

well lemme tell ya a story about my time at intel. i was working there when they rolled out the Pentium with MMX, the P-2 and the Celeron.

now generally the time from design to market is minimum 12 months, 18 is average. (yes that means that intel is currently making 5 Ghz processors)

As we all know, intel back when they had no real competition (AMD was around and selling processors, but they were essentially selling the same thing intel had as opposed to now where they have a completely different product) tended to charge as much as they could get away with. As newer and faster processors were introduced, the price of the newest stuff would be the same and the price step down was pretty much consistent from year to year.

so computers were spendy, intel was raking everyone over the coals and building a new $2 billion fab every other year.

but then memory prices started to go down, because companies overseas in an effort to break into the lucrative american market, sold their products dirt cheap long enough to gain market share and recognition. well others in the computer industry saw this and said, "hey we can do this too."

so the "under $1000 pc" was born. well intel at first thought that people wouldnt go for it. up to that time, people were willing to pay premium dollar for even the slightest increase in speed. well intel was wrong...

they saw that the cheapie pc's were making way too much money... so they had to introduce a product that would work in the cheapie market but at the same time maintain the value and integrity of their high end market. so the celeron was born...

but remember, it takes 12-18 months from design to market (assuming no problems...which is a BIG assumption)

well at the time the first celerons came out at 266mhz, the top of the line pentium 2 were at 300 mhz, but what people didnt know was that at the time they were trying to pawn off the 300's, the fabs were putting out 450 mhz chips.

unfortunately, with all new product ramps there is always a learning curve until the process can be ironed out and the bugs resolved. sooo they would make 450 mhz chips that wouldnt run at that, so they would clock em down, or more likely back then, just MARK them down.

also because they are 12 months ahead of the market, sometimes they mark the processors depending on market conditions. suppose that P-3's are about the be released, so everyone is waiting for that. but at the same time, the P-2 450's werent selling. so in order to get rid of the 450's they either had to cut the prices on the 450's (something they never did before because it would be hard to justify selling a 500 for twice what a 450 goes for) so instead they would remark the 450's to 300's and cut the price on the 300's just to get rid of them... (realize that each processor used to cost intel about six bucks to make, so if you thought you were getting over on intel cause you waited and bought cheap...well think again...)

 also P-2's didnt have autosensing to set clock speeds, they were set by jumpers.
so people would look at the marking on the chip and assume that that was the maximum safe operating speed. well that is not true. to insure a good rep and minimal problems, the marking is the MINIMUM safe operating speed. iow, although intel didnt allow it, every chip they make can be overclocked.

another thing you must consider is that the extreme tolerances and adherence to procedure required that a fab be completely retrofitted when the product was changed in any way. the photomasking equipment, slicers, polishers, cylinder cutters, were all custom made to do one job and one job only. these machines dont have adjustments. they cant afford to. that is the reason why it takes so long to bring a product to market. that is also why intel has a 12-18 month supply of processors already made sitting in a warehouse waiting to be sold. and ill bet ya the fastest one they have stored right now will clock over 5 Ghz.
« Last Edit: February 16, 2004, 06:56:08 AM by DaDnDe » Logged

"Nuchi du takara"
"Life is a treasure"
Pages: [1] Print 
« previous next »
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.13 | SMF © 2006-2011, Simple Machines LLC Valid XHTML 1.0! Valid CSS!