GloFlo "yields are world class"?

[Elixer]

mods, I really have no clue where this post should be stuck in... feel free to move it.

Gary started the interview by pointing out that it has now been a year since the GlobalFoundries purchase of many of IBM's semiconductor assets and they have hit every commitment they made. They had a black eye from the ramp up of 28nm in Dresden, they canceled 20nm and had to license 14nm from Samsung. Last year they said they would qualify 14nm at the beginning of this year and they did. They now have a ton of tape-outs in-line, they are in production on multiple parts and yields are world class. The 14nm process they are running now will also provide a baseline for 7nm development.

Gary confirmed GlobalFoundries will not be offering a 10nm process. They believe it will be short lived node and don't see the value proposition in it (authors note, at 20nm TSMC is really the only foundry that offered it and they quickly transitioned to 16nm, many believe the 10nm to 7nm transition will be similar).
...

https://www.semiwiki.com/forum/cont...pdate.html?s=3444a7362f674688bb6aeab46e0bb531

I suppose since they are a private company, they can say whatever they want, nobody is going to bust them on it (they got NDA's with everyone).
However, the facts don't support that claim, all AMD's chips so far produced don't have very good ASIC values, so, something isn't adding up.

 

[wahdangun]

nah i don't think good yield and good ASIC value have any correlation at all.

 

[3DVagabond]

nah i don't think good yield and good ASIC value have any correlation at all.

I don't think a good asic correlates with anything. lol

 

[Phynaz]

Ask them how they define world class.

Typical wishywashy pr speak.

 

[Magee_MC]

Ask them how they define world class.

Typical wishywashy pr speak.

They're a foundry in the world and in one of the classes of foundries in that world.

 

[Arachnotronic]

Ask them how they define world class.

Typical wishywashy pr speak.

This. SEC can't bust them for misleading shareholders because they are private. If they ever go public they'll have to tone down the chest puffing, especially when their major customers outright say in their own conference calls to shareholders that yields aren't great.

 

[krumme]

Are they only producing on sundays?

 

[Qwertilot]

They're a foundry in the world and in one of the classes of foundries in that world.

Its really quite a small class of foundries too It is easy to get a bit dismissive of people who aren't absolutely at the leading edge, but producing any kind of even vaguely modern silicon is still an objectively awe inspiring achievement.

 

[plopke]

I am kinda curious what indeed their yields and production capacities are , could be yields are fine but their production capacities just sucks. You could maybe call them world class also , the RX 480 kinda stretches it to its limits , wonder how this will look for mobile parts. But like with everything from global foundries , be very sceptical.

 

[raghu78]

Glofo is the worst of the leading edge foundries . TSMC is the best followed by Samsung. Even after licensing 14LPP they have the worst possible implementation. Nobody in their right mind will take what they say seriously. We will see what they deliver at 7nm using IBM R&D. Right now TSMC is the world's best foundry by a good margin and they will be probably have the best foundry 7nm process . TSMC 7nm HP version should be right up there with Intel's 10nm process in transistor density and performance. TSMC 7nm is scheduled for early 2018 ramp. So even if we take Q2 2018 that would put it 6-9 months behind Intel 10nm which I think is commendable.

 

[Headfoot]

"World Class" -> extreme lol. That means nothing. World class compared to what? Certainly not compared to Intel or TSMC yields I would expect.

Of course making silicon machines is probably the most advanced piece of technology we have as humans - it is irksome to see them boasting how great they are when they're unambiguously in last place among their competitors.

 

[96Firebird]

Just curious, what data is there to show one fab is better than another?

 

[raghu78]

Just curious, what data is there to show one fab is better than another?

as always the list of customers and products being fabbed at the respective foundries is one measure. The second is the same product being fabbed at multiple foundries (like Apple A9) gives a good measure of which process is better.The other is revenue numbers. TSMC is the only one which gives detailed financial information. Using all these measures TSMC is by far the best with roughly 65-70% of overall foundry FINFET revenue.

 

[Azix]

as always the list of customers and products being fabbed at the respective foundries is one measure. The second is the same product being fabbed at multiple foundries (like Apple A9) gives a good measure of which process is better.The other is revenue numbers. TSMC is the only one which gives detailed financial information. Using all these measures TSMC is by far the best with roughly 65-70% of overall foundry FINFET revenue.

nvidia going to samsung then suggests otherwise. not to mention 1070/1080 stock

 

[Elixer]

The other is revenue numbers. TSMC is the only one which gives detailed financial information.

They only do that since they are a public company, and not a private company like GloFlo.
That is why everyone is speculating what is going on, there is no way to get hard facts from GloFlo until someone leaks the information.

 

[Elixer]

Just curious, what data is there to show one fab is better than another?

Well...there is this.

http://anysilicon.com/top-20-semiconductor-companies-2016/

Of course, that isn't only foundries... but, it gives you a rough estimate of where they are going to build their chips.
I know there are other reports out there that talk about fabs in detail, but they want $10K+ for that information.

 

[sandorski]

Well...there is this.

http://anysilicon.com/top-20-semiconductor-companies-2016/

Of course, that isn't only foundries... but, it gives you a rough estimate of where they are going to build their chips.
I know there are other reports out there that talk about fabs in detail, but they want $10K+ for that information.

I wonder how that reflects total capacity? If there's a correlation between units produced and capacity, that would largely explain why so many more choose TSMC over GF.

 

[Arachnotronic]

Well...there is this.

http://anysilicon.com/top-20-semiconductor-companies-2016/

Of course, that isn't only foundries... but, it gives you a rough estimate of where they are going to build their chips.
I know there are other reports out there that talk about fabs in detail, but they want $10K+ for that information.

Most of Samsung's semi sales are DRAM/NAND.