• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Some AMD news

B

Bluga

Banned
Got it from JC's bbs


<<Quote>>:

Hi,

recently i had the chance to visit AMD's FAB 30 in Dresden
Here's some info that was new to me

-- They're building a design center there. This has existed since 1997 and developed some DSL modem chips and southbridges but they now want to stock up on 200 engineers.
They are going to develop chipsets only there (no processors)

-- They are developing a new southbridge for K8 there with USB 2.0, HyperTransport, 5.1 sound and DSL support there

-- IMHO I don't think they have gotten very far with it, so K8 might still be some time off

-- It looks like the K8 northbridge will be integrated in the processor

-- They are running test circuits (no processors?) in 0.13
i'm not so sure they are happy with 0.13

-- there are 210 processors (TBird I think) on one 200mm wafer

-- i think the yield is slightly higher than 70 percent (on Tbird?)

I don't have any information on palomino, they didn't even mention it

I hope, there's something new for you

regards

CLK
 


<< It looks like the K8 northbridge will be integrated in the processor >>

According to the rumors I've heard it's just the memory controller, not the entire northbridge.
 


<< i think the yield is slightly higher than 70 percent (on Tbird?) >>

I wouldn't believe that for a second. Yields are one of the very most protected statistics. They aren't going to tell some schmuck about their yields. Besides, yields aren't measured in percentages anyway.
 


<<

<< i think the yield is slightly higher than 70 percent (on Tbird?) >>

I wouldn't believe that for a second. Yields are one of the very most
protected statistics. They aren't going to tell some schmuck about their yields. Besides, yields aren't measured in percentages anyway. >>




Did you really have to call him a schmuck? Do people on these forums always have to get an attitude when you correct somebody?
 


<< Did you really have to call him a schmuck? Do people on these forums always have to get an attitude when you correct somebody? >>



It seems to be all the rage these days.

I hope the part about AMD making their own chipsets is correct. We'll finally have high performing stable chipsets probably in line with Intel. No flaming me please. It's just my opinion, I'm entitled to have one.
 

Exactly - I'm sure WingZ just meant to bring the point home that "mere mortals" (i.e.: non AMD employees) wouldn't be told what the yields are.

Out of curiousity Wingz, what ARE yields measured in? 🙂

Percentage seemed like a good enough way for me.

i.e.; 10 wafers (say) with 210 (say) processors.

50% @ 1400 MHz
10% @ 1333 MHz
5% @ 1000 MHz
30% Broken
5% Eaten by stray dog/abducted by aliens 🙂

Or something like that?
 
Thanks, guys. And yes, I definitely didn't mean any offense by it. I tend to call everyone a "schmuck" or "putz." Just like others use terms like "Joe Blow" or "Joe Average". If I've ever sent you a PM or ICQ, I've probably called you a schmuck. But I see your point, and I'll keep that in mind for the future schmucks that I talk about. 😉

Shathal, yields are measure in RISO. An arbitrary, universal algorithm. The easiest way to explain it, is that it's pretty much akin to defects per square centimeter. The reason RISO is used, is so that the numbers can be compared to different die sizes or even wafer sizes. The way you are talking about catagorizing die speeds, would be called "bin splits". Different than yield.

A4A, I agree... Having the memory controller onboard is pretty cool. Although I'm not exactly sure what the drawbacks of that design would be.
 


<< Still though, even if just the mem controller is on the K8 CPU that ought to be great. >>

True. I didn't mean to imply "well, it's only a memory controller.", but was making a correction that only the memory controller and not the entire Northbridge will be included. Including the memory controller should/will result in lower latency for main memory accesses.

Shalthal: frequency - beyond a bare minimum operating frequency - is not included in the yield calculation. And the frequency distribution is a normal/Gaussian distribution - so you'd expect something more like:
10% @ 1400 MHz
50% @ 1333 MHz
30% @ 1000 MHz

...for example.
 
Ummmm. Mac, are suggesting that for some reason Wingz won't be workin' @ Intel for much longer? :| Back the Hell off...
 
Northbridge intergrated into the Cpu? If that is true the need to buy a new motherboard will decrease since you could simply buy a processor with a better intergrated northbridge. this is truly interesting! I was under the impression that the K8 will be out 1st half 2002 and that AMD was quoted as saying the Hammer line will be out earlier than expected.
 
Once again another post completely ignorant of the facts containing something completely irrelevant to the thread with the sole purpose of irritating someone else. Macro2, do you ever post anything useful? Like something helpful? something that contributes in some fashion? A search of your prior posts didn't turn up much.
 
I've heard some scuttlebut that the Hammer may also have more advanced aspects like SMT (Simultaneous muti-threading). Apparently Dirk Meyer, who heads up some of the AMD architects, was intimately involved in the design of the Alpha. Guess we'll just have to wait and see around mid-October. As far as the K8, it looks like it will be sampling broadly in November but will probably not be product until the end of March, 2002 or later.
 
Well, it is logical that you don't need a northbridge, cause a northbridge interfaces with 3 things right? The processor, memory and the southbridge, since the memory controller is built in, and doesn't need a external interface to talk to the proocessor obviously, what AMD will use to connnect to a southbridge is what they made it for in the first place: Hypertransport.

I also read a part at Ace's Hardware on an article about processors in general, not about the hammer, and one person mentioned that just having a built in memory controller adds very little die to a chip, and vastly improves performance, since you don't have to go through a northbridge. He said it is better to have that than out of order execution units, which takes a lot of space on a die. He also said why hasn't anyone done this? Well, here is a first for AMD.

oops, forgot about agp, I think that is going into the southbridge anyhow, that's what VIA's hammer chipset will do from what I heard.
 
Macro2,

I think you need to reread Wingznut's orginal post. He wasn't calling Bluga a schmuck, merely anyone who might be considered beyond a company's circle of "people in the know".

I'm not even going to comment on fkloster's posts, tempted though I might be...
 
Thanks guys.

Macro, why the shot at me? I don't understand what provoked that. I said nothing but good things about AMD in this thread. And if you understood what I was talking about, you would've known that.

I suppose that you don't like me, because I work for Intel. I'm sorry that you feel that way, although it's hard to feel sorry for such an irrational feeling. But you aren't alone. There's three others just like you, on this msg board.
 
Did you really have to call him a schmuck? Do people on these forums always have to get an attitude when you correct somebody?

Lol. Wingz probably bashs the least out of anyone on these forums.

I must agree with Athlon4all and pm that the idea of an integrated north bridge on the processor sounds pretty cool. 🙂 I wonder what the exact advantages of this type of integration is. We'll see on October 15th when AMD reveals more specifics on the Hammer line.
 
Guys, I really don't see how Wingz was calling Bluga a shmuck, considering Bluga said he was quoting someone else from JC's BBS. But Wingz is right (he would know), yields are tightly kept secrets that wouldn't be revealed on a factory tour.
 
I am used to people disliking me, simply because of who I work for. Like I said, I can think of three or four people who take shots at me all the time. I am used to it.

But what I'm not used to is people passively-agressively threatening my job. This is the second time that you've done it. Possibly third, depending on how another one of your posts was interpreted.

I'm not ashamed of the job that I have. I'm actually quite proud that I was an auto mechanic last year, and I have greatly improved my (and my family's) life. I don't hide who I work for, yet many times that works against me. I suppose you'd be happier if I didn't say who I was employed by, or if I quit posting altogether.


"I get in knock down drag out arguements with my Intel employee friends all the time..."

What's the argument? I was basically defending AMD here. This is what I don't understand about you taking shots at me.

Enable your PM's or ICQ me. I'd love to carry on this conversation, but the GH forum is not the appropriate place.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top