New Products and Future Products for 2002

[bloom25]

There's some more news out today from nVidia:

They are releasing another nForce chipset for motherboards called nForce 415-D.  It's basically the same as the 420-D I talked about in my chipset guide for Athlon, but doesn't include the integrated GF 2 MX core.  Boards using this chipset are due to be released early next month.  The only real benefit of this new chipset is that it should bring the cost of nForce based boards down a little.  If you have your own GF 2 Mx video card (or higher), the 420 costs you a little bit more for something you aren't using.

AKDejaVu, I've learned that when it comes to production, Intel is hard to beat.    I'm sure AMD is going to see significant issues getting their own .13 micron line up and running.  AMD does have one huge advantage over Intel though:  The Athlon @ .13 micron will have a much smaller die size than Northwood P4s.  Even though they are still using 200 mm wafers, it will cost them quite a bit less to make an Athlon than it costs Intel to make a P4.  I'm sure you are right that Intel has higher yields though...

I have no clue how many processors need to be produced to release a chip.  It seems to vary wildly, with AMD and Intel both being guilty of "paper releases" of higher clockspeed chips that take months to actually show up in stores.  The 1 Ghz P3 is a recent example I can think of.  AMD hasn't really had this problem with the Athlon, but in the past they were well known for delays and production problems.

[AKDejaVu]

. I think it has to be over 50,000 for them to consider it as a release of a new processor. Although it was suggested that Intel released the first versions of the P4 at numbers less than 20,000 and scaled up production from there. HOWEVER I could be wrong on both accounts since it was nearly three years ago that I read that about the Athlons.

For the P-4's... double it and add a zero... that's a minimum.  I'd tend to say that Intel seldomely releases a new product unless there are close to 1 mill available.  And they still have shortages.

AMD my be less.. towards the 200,000 if % marketshare is any kind of guage.  But that's a tough number to hit.. and to sustain.  Especially if you can't get more than 20 off of 1 wafer.

AKDejaVu

[Thorns]

Thanks for all the info!  This is getting better than going to the geek sites.

Thorns

[AKDejaVu]

Wow.. I missed this post Bloom25.

AKDejaVu, I've learned that when it comes to production, Intel is hard to beat.

We do our best:D

I'm sure AMD is going to see significant issues getting their own .13 micron line up and running.

This is a given.  The main problem is, they have significant issues with thier .18 um process.  The .13 is going to need a miracle.  Oh yea... and they are going to do it on SOI.  LOL!

AMD does have one huge advantage over Intel though: The Athlon @ .13 micron will have a much smaller die size than Northwood P4s.

The Athlon on .18 already has a smaller die size than the Northwood.  And it still cost more for AMD to make:D  Die size is important, but yield rules supreme.  It doesn't matter how many chips you put on a wafer if only 10% of them work.

Even though they are still using 200 mm wafers, it will cost them quite a bit less to make an Athlon than it costs Intel to make a P4.

There is so much more to it that this is a pretty flawed statement.  Its possible, given the smaller die size, that if AMD had 90% yield, they might be able to make a processor for less money than Intel does... but I doubt it.

Really... this is an area we are so far ahead in that is is simply impossible to describe.

Don't let the cost of the processors fool you... there is a difference between a company willing to opperate at a loss and one that seeks actual profit.

I'm sure you are right that Intel has higher yields though...

Sources say that from an Intel wafer with some 250 chips (die) on it, 238 will be functional.  An AMD with a similar number of die would be lucky to have 120.

We used to call it "making jewelry".  Back in the days when Intel was counting how many wafers it took to get a good die.

AKDejaVu

[bloom25]

There's some more news out today:

VIA is due to announce it's Socket A (AMD Athlon) KT333 chipset, which supports DDR PC2700 ram.  In addition it seems to have support for AGP 8x.  SiS also seems to be working on a chipset called 745 with similar specs.

In a related story, it seems nVidia's upcoming GeForce 4 may have an AGP 8x mode.

2002 is definately shaping up to be a year of big performance boosts.

It also seems that both Intel and AMD did better than analysts predicted for Q4 2001.