Puff

http://semiaccurate.com/forums/showthread.php?p=168718#post168718
http://www.globalfoundries.com/technology/pdf/GF-14XM-Press-FINAL.pdf

Around 1 year after 20nm-LPM and 20nm-SHP launch.
28nm-SHP (FD-SOI) will launch before the 20nm line.


The roadmap of GlobalFoundries is accelerated. Will AMD follow? I'd say yes, if AMD is still in close collaboration with GF. Meanwhile the high-performance version of 14nm can be delivered quickly after the LPM's launch. Let's say 6-9 months.

Wild Guess to AMD's processor manufacturing roadmap:
Low-power and power-optimized processors: 40nm [11Q1] > 28nm [12Q4] > 20nm [14Q1] > 14nm [15Q2]
High-performance processors: 32nm [12Q2] > 28nm [13Q3] > 20nm [14Q2] > 20nm [15Q2] > 14nm [16Q2]

[ 本帖最後由 Puff 於 2012-9-21 00:10 編輯 ]

qcmadness

I don't think 14nm is feasible before 2016 without EUV AND double-patterning.

Puff

引用:

原帖由 qcmadness 於 2012-9-21 00:09 發表
I don't think 14nm is feasible before 2016 without EUV AND double-patterning.

Well, according to an early AMD manufacturing roadmap published in 2010, 14nm APU is planned for production in late 2015 IIRC.
For double-patterning, GF is implementing it in their 20nm node. They've also started testing EUV equipments.

qcmadness

引用:

原帖由 Puff 於 2012-9-21 00:12 發表

Well, according to AMD's manufacturing roadmap in 2010, 14nm APU is planned for production in late 2015 IIRC.
For double-patterning, GF is implementing it in their 20nm node. They've also started tes ...

20nm其實應該唔駛double patterning
EUV係必須, 但係暫時良率唔ok

Puff

引用:

原帖由 qcmadness 於 2012-9-21 00:12 發表

20nm其實應該唔駛double patterning
EUV係必須, 但係暫時良率唔ok

I am thinking about AMD to push their first phone- and tablet-optimized SoC in 2015. Recently they are developing their own ISP IP core, and hiring architects for wireless infrastructure.

qcmadness

引用:

原帖由 Puff 於 2012-9-21 00:15 發表

I am thinking about AMD to push their first phone- and tablet-optimized SoC in 2015. Recently they are developing their own ISP IP core, and hiring architects for wireless infrastructure.

我諗2015太慢

Puff

引用:

原帖由 qcmadness 於 2012-9-21 00:19 發表

我諗2015太慢

ULP APU for tablets...
2013: Temash down to 3.6-watt TDP
2014: 20nm. Perhaps down to 2.5-watt TDP?
2015: 14nm Low Power? Perhaps down to 1.5-watt TDP?

不過問題是人家 AMD 想不想玩 smartphone，和 x86 入唔入到 smartphone... 又或者 AMD 搞唔搞 ARM ULP APU...


[ 本帖最後由 Puff 於 2012-9-21 18:07 編輯 ]

qcmadness

引用:

原帖由 Puff 於 2012-9-21 18:04 發表

ULP APU for tablets...
2013: Temash down to 3.6-watt TDP
2014: 20nm. Perhaps down to 2.5-watt TDP?
2015: 14nm Low Power? Perhaps down to 1.5-watt TDP?

不過問題是人家 AMD 想不想玩 smartphone，和 x86 入 ...

> 2W is too much for tablet
> 0.5W is too much for smartphone

Puff

引用:

原帖由 qcmadness 於 2012-9-21 19:06 發表

> 2W is too much for tablet
> 0.5W is too much for smartphone

I think maximally ~1-1.5 watt is fine. Tegra 2 is reported to have a power consumption of 1+ watt (below 2) when it's fully loaded.
And the figure refers to the entire SoC. Idle and general use power (e.g. 3G talk, texting, browsing) is more important.



[ 本帖最後由 Puff 於 2012-9-21 21:53 編輯 ]

qcmadness

引用:

原帖由 Puff 於 2012-9-21 21:49 發表

I think maximally ~1-1.5 watt is fine. Tegra 2 is reported to have a power consumption of 1+ watt (below 2) when it's fully loaded.
And the figure refers to the entire SoC. Idle and general use power ...

I don't think AMD can make sub-1.5W SoC in coming 2 years.

siuba

intel 2014年尾都未必出到..唔信其他得..

qcmadness

引用:

原帖由 siuba 於 2012-9-22 02:41 發表
intel 2014年尾都未必出到..唔信其他得..

intel未出既話，equipment會係天價