Semi Accurate's 4XX views

Status
Not open for further replies.
What are the general views on this article :

http://www.semiaccurate.com/2010/03/29/why-nvidia-hacked-gtx480/

What are the technical merits of what this guy has to say?

He was pretty spot on with everything.

He said they couldn't make hte cards and he is right , there are no 512 core cards.

He was right that they were hot as the gtx 480 will idle at 90c with 2 monitors hooked up and its not un usual for it to go into the 100c

Its also slow. Its only a bit faster than the 6 month old 5870.

I dunno. I think the poll we have speaks for itself

http://forum.beyond3d.com/showthread.php?t=57004
 
Regarding the process/effects of binning, he's pretty accurate.

But be careful, you have potentially launched another flame war for/against Charlie outside RPSC.

A flamefest was not my intention. Everything he has said sounds plausible and I have followed his 'rumors' closely on this vcard line.

However one of my professors always gave us answers as incorrect if we got the correct conclusion but our means for getting there was flawed. So as I said I fully recognize that he has been spot on with his conclusions, but do you guys think that the logic/theories to get the answers are accurate as well?
 
However one of my professors always gave us answers as incorrect if we got the correct conclusion but our means for getting there was flawed. So as I said I fully recognize that he has been spot on with his conclusions, but do you guys think that the logic/theories to get the answers are accurate as well?

Apart from the sw tesselation bit (it's a toss up on his performance prediction) he was bang on target with his conclusions.

As far as the unmanufacturable claims are concerned, it remains to be seen how right he was.

The rest seem to be fairly accurate both with his arguments and his conclusions.
 
Apart from the sw tesselation bit (it's a toss up on his performance prediction) he was bang on target with his conclusions.

As far as the unmanufacturable claims are concerned, it remains to be seen how right he was.

The rest seem to be fairly accurate both with his arguments and his conclusions.

Let's see... NVIDIA has just paper-launched two SKUs 6 months late, both of which are not really expected to be available in large quantities when they eventually become available (April 12) and both of which have disabled units, as well as unreasonable thermals.

If that's not a sign of craptastic yields, I don't know what is.
 
Let's see... NVIDIA has just paper-launched two SKUs 6 months late, both of which are not really expected to be available in large quantities when they eventually become available (April 12) and both of which have disabled units, as well as unreasonable thermals.

If that's not a sign of craptastic yields, I don't know what is.

I'll wait to see if the supply is spotty in Q2 before I start believing in the <10% yield rumours.
 
In fact his info and overall picture were the most exact of all news sites. I think the only problem who some people have is the way how it was presented...
 
Thanks for the responses. I agree with some in the thread that the real proof of the pudding as far as the ability to manufacture the card will be a bit later when very few get into the channel. I was really wondering if the yields are as low as semi accurate was suggesting and if his data was correct on the number of wafers Nvidia had in production.

Nvidia staff must be extremely angry with whoever has been giving him his info.
 
Nvidia staff must be extremely angry with whoever has been giving him his info.

Well if you look at what he's right on and what he's wrong it seems his moles are outside of Nvidia. Lots of info on yields and manufacturing related issues, zilch on the products or architecture. There's nothing Nvidia can do about that unless they vertically integrate all steps of the process. And that's not gonna happen so it's just something they have to deal with.
 
Or they could just adopt policies in line with Chairman Jobs and the Cupertino Politburo. Leaks would certainly drop to a minimum.

:)
 
Or they could just adopt policies in line with Chairman Jobs and the Cupertino Politburo. Leaks would certainly drop to a minimum.

:)

Easier to do with Chinese manufacturing than Taiwanese manufacturing. It's a bit easier to be brutal and domineering to your employees in mainland China... A well placed bribe, and authorities will be more than happy to look the other way if you need to beat a few of your employees to prevent leaks.

Regards,
SB
 
Easier to do with Chinese manufacturing than Taiwanese manufacturing. It's a bit easier to be brutal and domineering to your employees in mainland China... A well placed bribe, and authorities will be more than happy to look the other way if you need to beat a few of your employees to prevent leaks.

Its also a LOT easier to switch to a different assembler than a different fab.
 
Last edited by a moderator:
Easier to do with Chinese manufacturing than Taiwanese manufacturing. It's a bit easier to be brutal and domineering to your employees in mainland China... A well placed bribe, and authorities will be more than happy to look the other way if you need to beat a few of your employees to prevent leaks./QUOTE]

Its also a LOT easier to switch to a different assembler than a different fab.

Heh, that's the understatement of the year. There was effectively no competition for TSMC for several years. Only now is there 'kinda' competition.

DK
 
Heh, that's the understatement of the year. There was effectively no competition for TSMC for several years. Only now is there 'kinda' competition.

Competition for fabbing GPUs?

GlobalFoundries hasn't made GPUs (yet), but is quite conceivable, but due to its intense AMD ties, not so likely for NVIDIA to use.

Intel could certainly fab GPUs, but that's clearly not an option for anyone else to use.

SMIC? UMC?... are they up to it technically? Am I missing some other realistic option?
 
Status
Not open for further replies.
Back
Top