由买买提看人间百态

boards

本页内容为未名空间相应帖子的节选和存档,一周内的贴子最多显示50字,超过一周显示500字 访问原贴
SanFrancisco版 - Intel的东进与ARM的西征(5)
相关主题
Intel的Atom为什么功耗比ARM大那么多? (转载)有人待过这家公司吗 IDT
MSFT又杯具了. (转载)做IC design的是不是只有湾区一个选择?
Good Breaking News[合集] 从Intel EE engineer to a full time Realtor!
Re: 硅谷八道: NVDA, Post PC时代的CPU之王 (转载)Intel Science Talent Search老中第二,踩了阿三.
ARM 来真格的了,CPU 提速 3 倍,Intel 悬了 (转载)Intel Core 2 Quad Q8200S 和 i-7, i-5比如何?
杯具的MIPS 被ARM Imagination拆解收购股市还要涨啊,看看intel 的ER
Google的又一次伟大创举,将革新整个色情产业三藩市要惩罚亚利桑那了
【转载】智能电视系列(1)在说一遍云计算服务 (转载)
相关话题的讨论汇总
话题: intel话题: arm话题: market话题: 电视话题: vpu
进入SanFrancisco版参与讨论
1 (共1页)
w*s
发帖数: 60
1
http://mozilla.com.cn/post/41825/
繁华又算得了什么,不过是星尘的崩碎,那一抹青青的灰。公元79年,意大利维苏威火
山喷发,已然兴盛了600年的庞贝古城被完全湮没[1]。历史只在瞬间,便把那无尽的繁
华刻入了冷峭的火山岩石里。
现如今,发展了近20年的数字电视也正处在繁华的巅峰,然而在这繁华的底部,却有一
个被称作“智慧电视”(SmartTV)的“火山”蓄势待发。我们倒不必过于惊讶,就像
庞贝古城的覆灭一样,或许只在明天,数字电视就会被彻底崩碎,湮灭于智慧电视的青
灰之下。
近日,苹果CEO库克接受NBC专访,大谈电视业务,这无疑笃定了人们关于苹果智慧电视
(iTV)的猜想,甚至于更乐观地估计在2013年10月就能一睹真容。姑且不去谈论智慧
电视如何改进交互技术,以及整个生态系统又怎样构建,我只想先从智慧电视的内核芯
片处说起。
智慧电视的系统芯片里包含有CPU与GPU(Graphic Processing Unit,图形处理单元)
,这一点与智慧手机和平板电脑是相同的。然而不同的是,除了CPU与GPU,智慧电视里
还有另外一个非常重要的处理单元,叫做VPU(Video Processing Unit,视频处理单元
)。在电视芯片里,VPU的复杂程度丝毫不逊于GPU,甚至在整个芯片里占有更大的面积
。VPU的功能主要包括:视频解码,三维处理,画质增强,图像内容识别,基于运动补
偿的帧频转换。
GPU负责在内存里画出菜单或游戏画面,最终则由VPU把它们与视频信号融合,并显示出
来。如果从软件编程的角度看,GPU较为“通用”,VPU则更像是个固化了特定功能的专
用模块,留给程序员的余地并不多。当然,我不知道在未来,会不会有哪个大牛横空出
世,把VPU完全“通用化”,从而给予程序员足够多的编程自由度。
电视芯片里的CPU部分,也至关重要。CPU负责对整个系统资源的调度,既要满足GPU与
VPU的巨大带宽需求,又要提供游戏、网络、文字处理等所需的计算任务。Intel虽然也
曾在电视领域里努力地扑腾了一阵,甚至联合谷歌与索尼推出第一代的GoogleTV,但最
终它还是选择了战略性放弃。而当另一个处理器厂商MIPS也萎靡不振的时候,ARM再次
把握住了机会,将自己在移动领域里的优势肆无忌惮地蔓延到智慧电视里来。
时光虽不能倒流,但历史随时准备重演。一年前,在Intel宣布退出智慧电视市场的同
时,高通却砸出六千万美元收购了IDT公司的电视芯片部门。两大巨头的一出一进会不
会重演六年前移动领域的戏份[2]?这不仅要看高通是否能把视频处理单元与骁龙(
Snapdragon Processor)完美结合起来,还要看智慧电视市场什么时候才能喷薄而出?
而这后一点,则完全取决于苹果何时可以推出那改变世界的iTV了。
在线高清,无限回放,语音控制,体感交互,分享影片,虚拟现实,等等,你尽管把所
能想到的一切都加到智慧电视里来。总之,当智慧真正降临到电视上的时候,我们将会
获得无以复加的完美体验。这一切的体验都是基于天衣无缝的交互方式:你或许会全身
心地沉浸在虚拟世界里,自己担当着男主角,而与社群里另一个并不认识的女主角,共
同欣赏完一部多线索多结局的影片。
智慧电视就像是一个大窗口,我们推窗而望,便立刻置身于一个全新的奇幻世界。这里
动静相宜,风景如画,只有当触感信号从外部传来的时候,我们才发现,原来自己早已
离开了窗口,正身处于亚伦的LC神画里[3]。
l******l
发帖数: 2651
2
>>>>>>会不会有哪个大牛横空出世,把VPU完全“通用化”,从而给予程序员足够多的
编程自由度。
Rumor says that Intel already has prototype.

【在 w*s 的大作中提到】
: http://mozilla.com.cn/post/41825/
: 繁华又算得了什么,不过是星尘的崩碎,那一抹青青的灰。公元79年,意大利维苏威火
: 山喷发,已然兴盛了600年的庞贝古城被完全湮没[1]。历史只在瞬间,便把那无尽的繁
: 华刻入了冷峭的火山岩石里。
: 现如今,发展了近20年的数字电视也正处在繁华的巅峰,然而在这繁华的底部,却有一
: 个被称作“智慧电视”(SmartTV)的“火山”蓄势待发。我们倒不必过于惊讶,就像
: 庞贝古城的覆灭一样,或许只在明天,数字电视就会被彻底崩碎,湮灭于智慧电视的青
: 灰之下。
: 近日,苹果CEO库克接受NBC专访,大谈电视业务,这无疑笃定了人们关于苹果智慧电视
: (iTV)的猜想,甚至于更乐观地估计在2013年10月就能一睹真容。姑且不去谈论智慧

f****a
发帖数: 4708
3
VPU不过是搞不定GPU的主硬生生造出来的一个概念。
r**m
发帖数: 1825
4
Very informative post from yahoo finance board:
http://finance.yahoo.com/mbview/threadview/?&bn=23a9aaae-3c87-3
Re: Why is anybody buying here
by mark_templeton_99 . Jan 9, 2013 10:42 PM . Permalink
Ashraf,
I have read your articles and message posts and think that you are a good
writer. However, there is a lot more to business success than technology
leadership.
When Intel beat out Motorola and others for leadership in the 16 bit
microprocessor space (well before your time) they did not have the best
technology at all. In fact the 8088 and 8086 were very awkward architectures
and that awkward legacy continues to this day. Also, at that time (roughly
1982), you could hardly consider Intel to be a process leader by any stretch
of the imagination.
Intel won the PC market by outstanding marketing, customer relationships,
development tool support and most importantly their win of the processor
socket in the original IBM PC. The rest is history and Intel has worked
their position remarkably well. Believe me, their dominance at that time was
not the result of a superior product technology (not even close).
ARM has invented a good processor and a far, far better way to do business.
You need to understand the value of this and the legitimate and lasting
leadership that can come from business model and market position. Your
defense of Intel simply on the basis of a theoretical technology advantage
misses the point and does not reflect the way that technology companies
actually win and hold markets.
During the history of Intel, there have been several processor/manufacturing
combinations that were at the time better than the Intel offering. Sun
SPARC, IBM Power PC and even early MIPS come to mind. In spite of their
technical excellence, none of these companies could compete with Intel's
foothold in PC clones, so they passed into obscurity. Today, ARM has that
market position, ecosystem, etc. equivalent in the mobile space.
Intel may have what it takes to take on ARM, but it will not be based on
technology alone. You need to start looking at their progress and approach
on other, less tangible factors. Also, winning for Intel isn't about taking
on ARM, but about taking on the 1,000 or so companies that are ARM licensees
. The variety, depth, selection, price points and technical excellence from
these companies is truly staggering. I know Intel is great, but not even a
giant can compete with an onslaught this large and varied.
A huge problem that Intel faces is one of business reputation. Companies
know what will happen to their margins if they base their businesses on
Intel technology. Intel caused complete commoditization in the PC space and
potential customers know to expect nothing less in the mobile space. In
other words, if Intel is allowed to build a strong market position, everyone
knows how that story ends and it is not pretty (for anyone but Intel).
I also think that you are pushing your Intel position a little too hard to
be taken seriously. You must be really long Intel or really short ARM.
I like to read unbiased, thoughtful commentary and criticism. Your position
has become so blindly biased that it is no longer helpful and can not be
taken seriously. You are clearly smarter than this and I hope that as you
learn, you will become a more thoughtful and helpful commentator.
Full disclosure: long ARM. Have been long Intel in the past -- this is just
not their time.
All the Best,
Mark

【在 w*s 的大作中提到】
: http://mozilla.com.cn/post/41825/
: 繁华又算得了什么,不过是星尘的崩碎,那一抹青青的灰。公元79年,意大利维苏威火
: 山喷发,已然兴盛了600年的庞贝古城被完全湮没[1]。历史只在瞬间,便把那无尽的繁
: 华刻入了冷峭的火山岩石里。
: 现如今,发展了近20年的数字电视也正处在繁华的巅峰,然而在这繁华的底部,却有一
: 个被称作“智慧电视”(SmartTV)的“火山”蓄势待发。我们倒不必过于惊讶,就像
: 庞贝古城的覆灭一样,或许只在明天,数字电视就会被彻底崩碎,湮灭于智慧电视的青
: 灰之下。
: 近日,苹果CEO库克接受NBC专访,大谈电视业务,这无疑笃定了人们关于苹果智慧电视
: (iTV)的猜想,甚至于更乐观地估计在2013年10月就能一睹真容。姑且不去谈论智慧

r**m
发帖数: 1825
5
Another one:
How I justify the high P/E
by khitchdee . Dec 6, 2012 5:40 AM . Permalink
If we look at the events of the past 5 years over a 15-20 year span, what's
been happening is ARM's microprocessor IP has proved itself in a hotly
contested mobile device market, warding off challenges from the most
dominant player in the microprocessor technology space, Intel, successfully.
Intel started addressing the "low power" market, which means uses for
microprocessors at around 1W power dissipation, around 2006 after it sold of
its ARM derived XScale business to Marvell. It has tried to displace ARM
for the last 6 years unsuccessfully and from the looks of will be unable to
displace it ever, since ARM in the meantime has firmly entrenched itself in
the industry. The only other player in the microprocessor space, MIPS has
fallen by the wayside. What this means going forward, if we look at it over
a long time span, is that ARM has established a position for itself in
mobile devices as a microprocessor supplier of choice. From this (niche)
vantage point, it can, over time, encroach on other market segments where
microprocessors are used such as servers, which it's started looking into
and eventually PCs. It has the right business model to do this when its
competitor has the deep pockets and staying power that Intel has. ARM's
strength lies in the fact that being a small company that only supplies the
IP, its effective strength against Intel comes from it multiple licensees
who add their domain-specific value to its IP and build chips. So its
entrenched and its got staying power. I think these two factors when
projected forward could lead to a scenario where it dominates microprocessor
technology applications which means its effective market presence could
grow substantially. Of course this is assuming that the demand for
microprocessor based computing remains.
The thing to note is that in this space, changes in the industry play out
over say 5 years. It takes an Intel about 4 years to go from concept to a
new microprocessor. This when they have been developing them for the last 40
years. Therefore, when looking at performance in this industry you have to
take a longer term perspective than what you're used to.
M*******c
发帖数: 4371
6
still rumor?
I thought they already throw them to the trash can.

【在 l******l 的大作中提到】
: >>>>>>会不会有哪个大牛横空出世,把VPU完全“通用化”,从而给予程序员足够多的
: 编程自由度。
: Rumor says that Intel already has prototype.

r*****r
发帖数: 867
7
Thanks for sharing...it is a good one.

【在 r**m 的大作中提到】
: Very informative post from yahoo finance board:
: http://finance.yahoo.com/mbview/threadview/?&bn=23a9aaae-3c87-3
: Re: Why is anybody buying here
: by mark_templeton_99 . Jan 9, 2013 10:42 PM . Permalink
: Ashraf,
: I have read your articles and message posts and think that you are a good
: writer. However, there is a lot more to business success than technology
: leadership.
: When Intel beat out Motorola and others for leadership in the 16 bit
: microprocessor space (well before your time) they did not have the best

i*********e
发帖数: 1010
8
Not an expert. but IMHO, Intel needs high margin, mainly because it has its
own fab. It is a double-edge sword. It gave Intel advantage on chip
optimization and time to market. But it eats so much money to be renovated
to better process.
Intel probably needs to do something about its fab in order to get out of
this dilemma: "winning in mobile market, loosing on margin".
- producing some low margin chips in other non-intel fab
- using fab capacity to take some EXTERNAL high-end high-margin SoC orders
there are many possibilities but it just works against Intel's tradition.
With that said, it is not totally a doomsday for Intel. As in the worst case
, Intel will still sell more servers to handle the ever-growing mobile
market. As long as it can hold its line against ARM64, which is likely so
for at least some time, it will be fine on the bottom line.
p******h
发帖数: 1783
9
其实Intel的情形和Apple有些类似,高margin高端产品只能存在于快速增长变化的市场
中,
一旦市场有所饱和,无论是性能饱和还是占有率饱和,原有的模式都不能持续。
要维持原有模式,就只能存在于高端小众市场,如果要推出低端产品参与竞争,基本是
杀敌
一千自损八百,同时损害自己原有产品的市场和利润率。
而ARM在将要涉足的市场,唯一的优势就是价格。因为伴随着性能的提升,相对于x86低
功耗的优势也荡然无存

its
case

【在 i*********e 的大作中提到】
: Not an expert. but IMHO, Intel needs high margin, mainly because it has its
: own fab. It is a double-edge sword. It gave Intel advantage on chip
: optimization and time to market. But it eats so much money to be renovated
: to better process.
: Intel probably needs to do something about its fab in order to get out of
: this dilemma: "winning in mobile market, loosing on margin".
: - producing some low margin chips in other non-intel fab
: - using fab capacity to take some EXTERNAL high-end high-margin SoC orders
: there are many possibilities but it just works against Intel's tradition.
: With that said, it is not totally a doomsday for Intel. As in the worst case

l******l
发帖数: 2651
10
今年CES展厅,Central Hall一进门右手边就一Intel powered 机顶盒 for Comcast。
可同时支持6个屏1080P。
问前两周,那个报道说 "Intel will get in set-top box business, and will show
the industry how to do it right",是咋回事。
The project manager said, it is rumor, Intel had not issued such press
release. Because so many ppl had asked to verify that report, he had to deny
that very firmly.
机顶盒was designed for Comcast, and is a prototype.
You make the call to believe or not.

【在 M*******c 的大作中提到】
: still rumor?
: I thought they already throw them to the trash can.

l******l
发帖数: 2651
11
Intel will sell TV set-top box
http://www.sfgate.com/business/article/Intel-will-sell-TV-set-t

【在 M*******c 的大作中提到】
: still rumor?
: I thought they already throw them to the trash can.

a***e
发帖数: 27968
12
顶这句
数字电视要什么VPU,通通GPU搞定

【在 f****a 的大作中提到】
: VPU不过是搞不定GPU的主硬生生造出来的一个概念。
a***e
发帖数: 27968
13
for fab owners, both intel and TSMC etc, the key is to keep the fab running.
therefore the volume is what really matters. in term of cost, look at what
tsmc, 3* spending in last couple years, it is not any less than intel. in
this sense costwise for two camps are actually close.
for intel, even it could take 100% of ARM market, it will only feed maybe 1/
4 of its capacities. the bigger problem is it is not as neutual as TSMC in
eyes of apple, TI, QCOMM etc so for the mobile products, it is all by itself.
The only chance now for intel is performance and time to market. if they
could beat out 3*, apple etc own brews, it might recover. it doesn't look
very likely.
worse case, it could become 2xTSMC, a $30B company.

its
case

【在 i*********e 的大作中提到】
: Not an expert. but IMHO, Intel needs high margin, mainly because it has its
: own fab. It is a double-edge sword. It gave Intel advantage on chip
: optimization and time to market. But it eats so much money to be renovated
: to better process.
: Intel probably needs to do something about its fab in order to get out of
: this dilemma: "winning in mobile market, loosing on margin".
: - producing some low margin chips in other non-intel fab
: - using fab capacity to take some EXTERNAL high-end high-margin SoC orders
: there are many possibilities but it just works against Intel's tradition.
: With that said, it is not totally a doomsday for Intel. As in the worst case

1 (共1页)
进入SanFrancisco版参与讨论
相关主题
在说一遍云计算服务 (转载)ARM 来真格的了,CPU 提速 3 倍,Intel 悬了 (转载)
NVDA咋了?杯具的MIPS 被ARM Imagination拆解收购
INTC/NVDA盘后大涨。。。还在涨Google的又一次伟大创举,将革新整个色情产业
要给家里老人买笔记本,请推荐deal【转载】智能电视系列(1)
Intel的Atom为什么功耗比ARM大那么多? (转载)有人待过这家公司吗 IDT
MSFT又杯具了. (转载)做IC design的是不是只有湾区一个选择?
Good Breaking News[合集] 从Intel EE engineer to a full time Realtor!
Re: 硅谷八道: NVDA, Post PC时代的CPU之王 (转载)Intel Science Talent Search老中第二,踩了阿三.
相关话题的讨论汇总
话题: intel话题: arm话题: market话题: 电视话题: vpu