GT300 just taped out?

[saaya]

http://www.semiaccurate.com/2009/07/...300-tapes-out/

Our sources tell us that the GT300 is beginning it's 7 or 8 week long process through the very expensive machinery at TSMC. We said the target was a mid-July tapeout, and if you are very charitable with 'mid-', Nvidia hit it. A hair under a week ago, the chip had not taped out.

...if everything goes perfectly, not so much as a hiccup, we are looking at mid-December for the first syphilitic trickle of parts.

The chances of this happening are somewhere around zero though. Nvidia can't make GT215s in volume, and that chip is about 1/4 the size of GT300. Toss in that this is going to be their first GDDR5 part, their first DX11 part, and their first part on a new and untested architecture, and you are likely to have at least two spins.

Each spin is the same 8 weeks as the last hot lot. Adding one puts production silicon out of the door, with the previously alluded to trickle, in late February. A second spin puts it into Q2. If there is anything more, well mid-year is very likely. Smart money is on at least two spins.

so nvidia should have early ES cards in 4-8 weeks from now, which means september-october. if those work alright (very unlikely) we might see some performance previews which would be very interesting!

they will most likely need at least 1 respin, and each respin is 6-8 weeks, so best best best case scenario:

a0 silicon works perfect or good enough and comes back in september already, nvidia sends out ES silicon as review cards in september, immediatly kicks off mass production, 10 weeks later in mid november we have mass production gt300 cards.

both nvidia and tsmc have seen better days execution wise, quite some good engineers left nvidia and were bought by intel and its pretty likely that we wont see a best case scenario... having gt300 by christmas is very unlikely, and mid december is pretty much too late to hit the holiday shopping race, from there on its holidays until 2010, and in asia even 2-3 more weeks holidays after that...

id say realistically the earliest we could see gt300 is in mid january... :/

[orangekiwii]

I'm feeling limited production numbers for mid - late December and then yes as you said in your post increased market availability in january

[Cybercat]

BSON talks about yields being low for this chip, now this place says it's just now being taped out.

Who to believe...

[saaya]

BSON talks about yields being low for this chip, now this place says it's just now being taped out.

Who to believe...

lol, if i had to chose between charlie and theo, id def go with charlie
you can say what you want about him being biased, but he has good sources...

i think q2 with 2 respins like he suggested is a little pessimistic... i think mid january on the other hand with one respin is rather optimistic... i think itll probably be something in between, so a cebit launch?

then again, didnt gt200b supposedly tape out in may 2008? and it was launched in january? and that was just a shrink, no new architecture...

i googled for tape out times to see how nvidia did in the past from tape out to launch, but couldnt find much
anybody?

[Oberon]

The whole premise that NVIDIA won't be able to produce GT300 in volume because it can't even get GT215s out the door is a bit flawed. ATI is in the same situation with RV740 and nobody is contradicting the claims that the new chips are getting good yields.

[labs23]

Its good news if this is true. But still, the word is next year.

I'd prefer for nVidia and ATI to market their DX11 cards next year, so I can buy a GTX295 next month with peace in mind.

[perkam]

lol, if i had to chose between charlie and theo, id def go with charlie

If I had to choose beween charlie and theo, I'd pick my cat.

This, along with anything else from Fudzilla, The Inquirer and BSO, should be thrown in the trash.

Perkam

[LordEC911]

The whole premise that NVIDIA won't be able to produce GT300 in volume because it can't even get GT215s out the door is a bit flawed. ATI is in the same situation with RV740 and nobody is contradicting the claims that the new chips are getting good yields.

Because they were actually able to produce RV740 in decent enough yields to at least ship some 40k units for launch, let alone the thousands of other units used in the OEM/mobile market.

The fact is that these G21x were meant for the end of 2008, like RV740, but they still haven't seen. Even if Nvidia was getting the same crappy yields of ~20-30%, why haven't they been able to launch any products?

Also remember, RV740 is ~136mm2 somewhat close to the 181mm2 of RV840.
Supposedly the largest G21x that hasn't been scrapped is ~144mm2, little bit smaller compared to the ~485mm2 of G300...

Just because you have problems with the first test chips doesn't necessarily mean you will have problems with all other chips or larger chips but it does cause a lack of confidence with large chips, especially when the new large chip is also a whole new architecture.

[Smartidiot89]

The whole premise that NVIDIA won't be able to produce GT300 in volume because it can't even get GT215s out the door is a bit flawed. ATI is in the same situation with RV740 and nobody is contradicting the claims that the new chips are getting good yields.

HD4770 was nothing but a learning curve for AMD.

I am not sure about you, but here in sweden I've seen the card in stock for the past few weeks...

[570091D]

If I had to choose beween charlie and theo, I'd pick my cat.

This, along with anything else from Fudzilla, The Inquirer and BSO, should be thrown in the trash.

Perkam

+1

plus, all of these sites have given us totally different release schedules. i remember an article a few months back from BSON predicting that nvidia would beat ati to market with dx11 hardware.

[GAR]

is that Charlie's website? if so why would you quote him on anything he writes about nVidia, we all know how much he loves nVidia, thats sarcasm by the way. I think from all those people, FUD is the closest to being right most of the time.

[saaya]

The whole premise that NVIDIA won't be able to produce GT300 in volume because it can't even get GT215s out the door is a bit flawed. ATI is in the same situation with RV740 and nobody is contradicting the claims that the new chips are getting good yields.

ati mass produced and shipped 40nm parts, nvidia still hasnt... they anounced mobile 40nm parts but are they in mass production? not that i know... and they are very small die sizes, smaller than rv740 afaik...

If I had to choose beween charlie and theo, I'd pick my cat.

This, along with anything else from Fudzilla, The Inquirer and BSO, should be thrown in the trash.

Perkam

ok, so what news do you trust?

[metro.cl]

Funny how their first DX11 should be a huge issue to make because it is their first, no one said that on the first DX10 card.

I believe NVIDIA is not in a hurry with all these parts market is in a downturn and they spend quite a bit on marketing cards which might not sell good enough.

[TheCarLessDriven]

WTF is your title supposed to mean?

[trinibwoy]

ati mass produced and shipped 40nm parts, nvidia still hasnt... they announced mobile 40nm parts but are they in mass production? not that i know... and they are very small die sizes, smaller than rv740 afaik...

I'm not sure why people are using the HD 4770 launch as a comparison point. That wasn't (and still isn't) a success by any measure. I'm sure if AMD knew availability was going to be this bad they would've never launched it in the first place. So whether or not Nvidia had a part ready is just one part of the equation. Even if they did they could simply have read the 40nm situation better and decided not to release a $100 card that they couldn't keep in stock (which would be harder for them since they sell more GPUs than AMD).

I guess the point is that just because AMD released a trickle of cards into the market doesn't mean that Nvidia couldn't. Because obviously doing so intentionally wouldn't make sense which leads me to believe AMD was just as suprised as everyone else when yields didn't improve (hence the unplanned 48xx price drop to fill the gap).

[Jamesrt2004]

The whole premise that NVIDIA won't be able to produce GT300 in volume because it can't even get GT215s out the door is a bit flawed. ATI is in the same situation with RV740 and nobody is contradicting the claims that the new chips are getting good yields.

i believe the yields are going to be down to the actual die's size if its like last time ati's will be much MUCH smaller, which means they will have a higher success rate

[trinibwoy]

ati mass produced and shipped 40nm parts, nvidia still hasnt.

Oh btw a quick search would've answered your question. There are already laptops available for sale with 40nm GT21x parts. So like I said above, AMD's rushed launch of RV740 doesn't give you any insight into Nvidia's 40nm situation or strategy (though it's pretty obvious that all of 40nm was/is pretty f'ed up).

http://www.jr.com/asus/pe/ASU_N51VNA1/
http://www.shoppingfever.de/product_...paign=geizhals

[Talonman]

A truly entertaining thread.

[largon]

^And the "" means what?
You _know_ something?

[Talonman]

No... But rather like the responses.

[SubZero.it]

If I had to choose beween charlie and theo, I'd pick my cat.

This, along with anything else from Fudzilla, The Inquirer and BSO, should be thrown in the trash.

Perkam

+1

[Oberon]

I'm not sure why people are using the HD 4770 launch as a comparison point. That wasn't (and still isn't) a success by any measure. I'm sure if AMD knew availability was going to be this bad they would've never launched it in the first place. So whether or not Nvidia had a part ready is just one part of the equation. Even if they did they could simply have read the 40nm situation better and decided not to release a $100 card that they couldn't keep in stock (which would be harder for them since they sell more GPUs than AMD).

I guess the point is that just because AMD released a trickle of cards into the market doesn't mean that Nvidia couldn't. Because obviously doing so intentionally wouldn't make sense which leads me to believe AMD was just as suprised as everyone else when yields didn't improve (hence the unplanned 48xx price drop to fill the gap).

This.

[Farinorco]

RV740 is nothing else than a step to master the new 40nm process, exactly like those new mobile parts are to NVIDIA. The temporal difference between them (both manufacturing and commercializing) could be a hint of AMD having a slight advantage here, that could translate to a temporal advantage for the next GPU generation, or could be due to other things. It's a little hint anyway.

Anyway, I hope both AMD and NVIDIA release their DX11 40nm parts ASAP, and I hope they both do it as near in time between them as possible. Given the previous rumors, I'd say a possible Q4 09 launch is good news, and if they could get it sooner, I'd be even happier.

ok, so what news do you trust?

Trust no1

Seriously though, those not-verified-by-facts "news" are entertaining to read and comment and somewhat mitigate the absence of constrasted news on the new technologies, but you shouldn't put any "trust" on them, specially on those coming from sites with a reputation of being biased and/or specially "inventive".

[saaya]

Oh btw a quick search would've answered your question. There are already laptops available for sale with 40nm GT21x parts. So like I said above, AMD's rushed launch of RV740 doesn't give you any insight into Nvidia's 40nm situation or strategy (though it's pretty obvious that all of 40nm was/is pretty f'ed up).

http://www.jr.com/asus/pe/ASU_N51VNA1/
http://www.shoppingfever.de/product_...paign=geizhals

good point...

there is one problem though, gt300 will be big, and on 40nm thatll be a problem, even now that its... but maybe nvidia will have a medium sized gt300 style chip... who knows...

and metro, your right, the market slowed down... but according to what i just read, i think on xbitlabs, the gpu market is supposed to surpass 100m in q3 09 again, and some speculate q4 09 will set a new all time record...
and ati seems to be close to a launch and intel seems to be polishing off larrabee as well and isnt that far from launching it, its a pure strategy decision for intel, they could have launched already if they really wanted i think...

[saaya]

Seriously though, those not-verified-by-facts "news" are entertaining to read and comment and somewhat mitigate the absence of constrasted news on the new technologies, but you shouldn't put any "trust" on them, specially on those coming from sites with a reputation of being biased and/or specially "inventive".

everybody is biased and every site is biased...
i prefer an obvious bias over a hidden pretending to be unbiased bias...
its easier to tell apart the actual information and the bias that way...

and not verified by facts... then the only news would be what people either already know or forwarding press releases and anouncements... thats not really news in MY opinion