Sign in with
Sign up | Sign in
Your question

What's the Difference between the Althon 4, MP,XP?

Last response: in CPUs
September 27, 2001 9:49:47 PM

What is the Difference?? Is the Althon 4 only mobile?? Is the MP .13 core where the thunderbirds are .18? What's the XP then?? Are they all going to be SLOT A so I don't have to upgrade my MB again? Are any going to Support 333FSB? K.. whatever you know I'd like to!




More about : difference althon

September 27, 2001 9:53:06 PM

Athlon 4 = Palamino for Mobile Systems
Athlon MP = Current Athlon Palamino for Multiprocessor Systems
Athlon XP = Desktop Palamino

All will support 333MHz FSB in the future. All will NOT be Slot A. :)  I really hope you mean Socket A as Slot A has been dead for some time now. All the above stated chips are for SOCKET A alone.
September 27, 2001 10:17:48 PM

Yes that's what I meant.. So is there a map to when these are all going to be released? Will It require different Motherboards for the XP's or just different voltage settings? I had someone tell me that AMD sucks for Muti-CPU systems because something like Microsoft won't certify them? Also he said that a dual AMD CPU system can't have both CPU's run at the same speeds. Isn't it also right that only NT, 2000, and XP support mult-CPU systems? I could have that wrong so fell free to correct me, but let me know what you think!



I Like Cheese<P ID="edit"><FONT SIZE=-1><EM>Edited by o0ospunkieo0o on 09/27/01 05:24 PM.</EM></FONT></P>
Related resources
September 27, 2001 10:37:06 PM

Map Link =

Different motherboards, no. Only a BIOS update, depending on how old the motherboard is. Some newer ones already have support for them.

Your friend is a moron. Microsoft certified? That has nothing AT ALL to do with it. For SMP (symmetric multi-processing) systems, the Athlon/Palamino makes a GREAT choice. On the different speeds issue, your friend is partly correct. Both CPU's will only be utilized on multi-threaded applications. Only a limited number of these are out, and they are most usually Video/Image or Sound editing software, nothing mainstream.

Correct, NT, 2000, and XP only truely support dual CPU's.
September 27, 2001 11:44:11 PM

And Linux. Don't forget Linux support for SMP. In fact, multiple procs are probably what you'd want to use for server apps, so they work well w/ Linux.

"If you teach a child to read, then he or her will be able to pass a literacy test" - George W.
September 28, 2001 12:39:28 AM

Presently ALL the new Athlon 4's MP's or XP's have the 0.18 micron core. they are just improved designs of the old thundabird, plus hardware prefetch and SSE and probably other nice things.
during next year AMD will phase in the die shrink to 0.13 micron. now they are the processors im waiting for :) 

Religious wars are 2 groups of people fighting over who has the best imaginary friend.
September 28, 2001 1:01:09 AM

Ahh yes, I forgot lovely Linux.

.18 micron is the current die size. They will be shrinking it when the new FAB gets all settled and open.
September 28, 2001 8:17:05 AM

They have no new fab opening, they are planning on 2 years away possibly at the campus where I work(fujitsu/amd GMD). The dresden fab is starting a .13 micron line and will provide all .13 micron chips for the next 2 years.

"The Cash Left In My Pocket,The BEST Benchmark"
September 28, 2001 12:32:31 PM

Indeed, at 80% of capacity (figure 80% of chips coming off the lines pass inspection) Dresden can p[roduce something on the order of 50+ million CPU's a year at .13 micron.

That's about 35% of the worldwide production of PC CPUs..out of one fab.


When all else fails, throw your computer out the window!!!
September 28, 2001 1:55:33 PM

80% capacity means the % of TOTAL chips one can produce is ALL tools are used etc, many fabs have idle tools and headcount issues, the capacity of a fab is the average number of chips it could produce if every tool was installed and running.

I have heard from ex techs at AMD(some who came from dresden even) that they had 90% yield middle of this year, and it is up since then. 90% yield on critical logic is AMAZING. That is why I chuckle at the people who insinuate AMD's process/quality is more sloppy than intels.

"The Cash Left In My Pocket,The BEST Benchmark"
September 28, 2001 2:05:46 PM

Thanks for the correction Mat. I meant yield, but had a brain cramp and could only think of capacity.

The determination came from current yields on .13 micron compared to the .18 micron...surprising how many MORE chips AMD will be able to produce.


When all else fails, throw your computer out the window!!!
September 28, 2001 3:50:59 PM

Except that it's harder to keep the same yield when moving to a smaller die. I think I said that right.

Anyway, there's a lot more chance of small particles screwing up the chips, etc.

<font color=green>I post so you don't have to!
9/11 - RIP</font color=green>
September 28, 2001 4:37:07 PM

Yes, but the larger number of die per wafer keeps the yield pretty much near the same, like 90>89 etc, and you wind up generally with MANY more saleable die, unless your fab sucks ass.

"The Cash Left In My Pocket,The BEST Benchmark"
September 28, 2001 4:47:51 PM

Ahh...yield is measured per wafer, not per die?

<font color=green>I post so you don't have to!
9/11 - RIP</font color=green>
September 28, 2001 5:33:56 PM

Yes, yield= number of good die/number of possible die on a wafer

If my 8 inch wafer can fit 100 die, and 90 of them work, my yield is 90%

Some of the products I work with have 1000 die per 8 inch wafer, when these sell for 5 bucks a die, a yield of 90%=4500 per wafer, there are 26 wafers per lot. So a 90% yielding lot would net you 117,000$

now Imagine my yield is 85%, dosent sound like a big difference does it, however you see that.
1000x89%x26x 5$ = 110,500

so a small 5% yield hit cost you 7 thousand per lot, and fabs put out something like 500 lots a month or more(depending on the fab).

Now lets shrink the process.

.13 micron transistion gives you a -3% yield hit, due to smaller feature zise etc etc

so now you yield 82%(we will stick with the 85% as its more realistic than 90) you yield less but you can fit up to doublt the die per wafer(depending on design) lets do a nominal +35% die per wafer.
now the perfect wafer can net you 1350 die.

so we have
1350/82%x26(wafers per lot)x 5$=1107 usable die or 143,910$ per lot, this is a nice fat return. Also the longer a fab runs the higher its yield becomes(unless again your fab sucks ass) because issues in the process are ironed out and streamlined. So yield goes up, die per wafer goes up, profits go up. (bear in mind, besides the tool costs of .13 microns the wafers used are exatly the same as .18 micron, so you always pay a set amount to start a lot(something like 3000 bucks btw). Just figure your average time of completion(usualy 60-90 days per lot, some processes take a lot of time, some devices where I work have 500 stages!)your cost of operation per day, and the number of lots in the fab at any one time, and you can quickly see if you are making or losing money.

Its fun really!

"The Cash Left In My Pocket,The BEST Benchmark"
September 28, 2001 7:39:32 PM

You have an 8 inch wafer? Mine is 9 1/2.

<font color=green>I post so you don't have to!
9/11 - RIP</font color=green>
September 28, 2001 8:29:23 PM

are we still talking computers?;)

<font color=green>
*(k)eep (I)t (S)imple (S)tupid*
</font color=green>
September 29, 2001 4:08:38 AM

Xp = Desktop no SMP
Mp = Server and workstation has SMP
A4 = very hot chip wanting for 0.13 for an AMD mobile chipset to challenge the Mobile P4

Nice Nvidia and ATi users get a Cookie.... :smile: Yummy :smile:
September 29, 2001 4:08:45 AM


Nice Nvidia and ATi users get a Cookie.... :smile: Yummy :smile: