ATI... yield at TSMC’s Fab 6... has surpassed 90% - DT

[T2k]

http://www.digitimes.com/NewsShow/MailHome.asp?datePublish=2003/6/24&pages=06&seq=27

This part caught my eyes: The average manufacturing yield at TSMC’s Fab 6 – one of its six 8-inch wafer fabs – has surpassed 90%, according to company sources.


I never thought it's possible.
Is it? Or it's just "company sources"?

PS: And the article has more interesting points...

 

[RussSchultz]

Note, however, that it says the average yield at FAB 6 exceeds 90%.

It does not say that ATIs yield in fab 6 exceeds 90%.

However, 90%+ yields are possible, if your die is small enough, and there's some sort of mechanism in place for dealing with random defects (like repairable ram, for example)

 

[T2k]

Note, however, that it says the average yield at FAB 6 exceeds 90%.

It does not say that ATIs yield in fab 6 exceeds 90%.

Yes, of course. Thx for the fix and sorry for my phrasing.

However, 90%+ yields are possible, if your die is small enough, and there's some sort of mechanism in place for dealing with random defects (like repairable ram, for example)

I see. BTW, it's about the 130-nano process, according to DT.

 

[Slides]

Head for the hills!!

What does it mean?

 

[RussSchultz]

Fab 6, I believe, is a 200mm wafer fab, with a single 'pilot' line for 300mm/.13u wafers. (I believe all of TSMCs .13u production is on 300mm wafers)

So I'd be suprised if these 90% yields that they're talking about are in .13u.

 

[Sabastian]

Fab 6, I believe, is a 200mm wafer fab, with a single 'pilot' line for 300mm/.13u wafers. (I believe all of TSMCs .13u production is on 300mm wafers)

So I'd be suprised if these 90% yields that they're talking about are in .13u.

Yeah, I seriously doubt that ATi is able to get such great yeilds out of the .13um process where nvidia continually insists that they are having so many tribulations with that process. Interestingly enough though, ATi management has stated that they are not having any problems with the .13um process. I am waiting to hear something about this out of the ATi CC tomorrow morning, they ought to clear up this if their yields are that good they will say so without even being pressured on it.

 

[KimB]

Well, Sebastian, just remember that ATI's .13 micron chips are quite a bit smaller.

For example, if nVidia's yield rate at .13 micron with a 100 million transistor chip is 50%, then a similar chip on the same process at 50 million transistors should have a yield of around 75%.

 

[Sabastian]

Well, Sebastian, just remember that ATI's .13 micron chips are quite a bit smaller.

For example, if nVidia's yield rate at .13 micron with a 100 million transistor chip is 50%, then a similar chip on the same process at 50 million transistors should have a yield of around 75%.

Yeah I do realize the RV350 is a smaller core. I am not so sure about your math though as I don't really think that the RV350 is done on the exact same conditions as the NV35, in other words there are other variables then the core size that effect the outcome even thought they are both done on the .13um process. Anyhow like I said I will be listening for this info tomorrow morning in ATIs CC.

 

[Pete]

The NV31 should be close in size to the RV350, no? And yet nV is also reporting poor yields with that.

 

[Sabastian]

The NV31 should be close in size to the RV350, no? And yet nV is also reporting poor yields with that.

heh, yes it is nearly as small AFAIK. I didn't know they were reporting poor yeilds with that particular chip though. Not that it matters I don't think there will be a strong demand for it.