|
H******t
发帖数: 687 | 2 1. Ku-band profit margin will be an issue. For ATG gogo does charge quite a
lot, but bandwidth will become an issue. To increase bandwidth cap
investment is needed, no matter which route you go.
2. Agree, but, if it come too soon, bandwidth will become an issue. If it
comes too late, opportunity will be lost.
3. Which airline can they sign? Domestic is no longer in play,
internationally the competition is fierce, and they don't really have
advantage. I agree that if they sign someone then stock ... 阅读全帖 |
|
c******g
发帖数: 123 | 3 ft, your cable tv company doesn't provide cable internet, how can you choose?
if both are availible, i prefer cable modem.
cuz the bandwidth of cable modem is shared by a group.
which means sometimes, you will get a good bandwidth.
maybe you have bad times of getting bad bandwidth too. |
|
E***s
发帖数: 855 | 4 from www.hdmi.org
Overview of HDMI 2.0
What is HDMI 2.0?
HDMI 2.0, which is backwards compatible with earlier versions of the HDMI
specification, significantly increases bandwidth up to 18Gbps and adds key
enhancements to support market requirements for enhancing the consumer video
and audio experience.
What’s new in HDMI 2.0?
HDMI 2.0 significantly increases bandwidth to 18Gbps and includes the
following advanced features:
Resolutions up to 4K@50/60 (2160p), which is 4 times the clarity of
... 阅读全帖 |
|
E***s
发帖数: 855 | 5 from www.hdmi.org
Overview of HDMI 2.0
What is HDMI 2.0?
HDMI 2.0, which is backwards compatible with earlier versions of the HDMI
specification, significantly increases bandwidth up to 18Gbps and adds key
enhancements to support market requirements for enhancing the consumer video
and audio experience.
What’s new in HDMI 2.0?
HDMI 2.0 significantly increases bandwidth to 18Gbps and includes the
following advanced features:
Resolutions up to 4K@50/60 (2160p), which is 4 times the clarity of
... 阅读全帖 |
|
p*****s
发帖数: 1780 | 6 http://www.pcworld.com/article/222039/atandts_uverse_and_dsl_da
AT&T's U-Verse and DSL Data Caps: Good Deal, Bad Precedent
By Jared Newman, PCWorld Mar 14, 2011 8:43 AM
Data caps are coming to AT&T's U-Verse and DSL services, but they're
unlikely to affect the vast majority of customers -- for now.
Starting May 2, AT&T will limit DSL customers to 150 GB per month, and U-
Verse fiber optics subscribers to 250 GB per month, according to DSL Reports
. Overages, which cost $10 for each 50 GB, wil... 阅读全帖 |
|
p*****s
发帖数: 1780 | 7 http://www.pcworld.com/article/222039/atandts_uverse_and_dsl_da
AT&T's U-Verse and DSL Data Caps: Good Deal, Bad Precedent
By Jared Newman, PCWorld Mar 14, 2011 8:43 AM
Data caps are coming to AT&T's U-Verse and DSL services, but they're
unlikely to affect the vast majority of customers -- for now.
Starting May 2, AT&T will limit DSL customers to 150 GB per month, and U-
Verse fiber optics subscribers to 250 GB per month, according to DSL Reports
. Overages, which cost $10 for each 50 GB, wil... 阅读全帖 |
|
w*h
发帖数: 286 | 8 在上面有个论坛，三天两头出问题，从来没有运行一个星期
没有出过事情的。所以建议大家如果要建论坛，千万不要选
择NETFIRMS.COM这种WEBHOSTING，偶现在后悔，要挪地方也
是不方便的。
这个WEBHOSTING的最大一个问题是，他们的BACKBONE PROVIDER
的连接(CONNECTION)常出问题，表现在具体的网站上就是，
有时你无法打开页面，等再一次REFRASH后才行。有时根本进
不了页面，就是进去了也奇慢无比。这样的网站还有谁来访问？
教训！千万不要选择这个WWW.NETFIRMS.COM这个WEBHOSTING。
他们的TRICK也别别具风格。先是说免费，1G的BANDWIDTH，25M
的空间，你如果要真以为给你1G的BANDWIDTH你就大错特错了！
在他们的另行规定中，CGI的点击次数不能超过2000(记不清楚了)
这是是致命的，限制你CGI的点击次数给你再多的BANDWIDTH也
是没有用的，所以说他们的TRICK是别具一格的。等你将所有的
WEB页面和CGI程序送上去的话，你就上钩了，只有花钱买空间
了！花钱给个正规的运行也行啊，可是就是三天两头 |
|
c*r
发帖数: 278 | 9 I/O bandwidth, Memory bandwidth, Control bus bandwidth,
interconnection network, ...... |
|
x*********n
发帖数: 28013 | 10 As long as it is not pure data service, when voice is there, you always need
QoS.No matter how much bandwidth you get.
If it is a pure data service, then it is different story.
Jitter is related to bandwidth, but not depends on bandwidth. |
|
s*****g
发帖数: 1055 | 11 It does not make sense to dedicate one link for outbound and another link
for inbound, the reasons being, ISPs charge you based on your 95th
percentile
bandwidth usage inbound or outbound on top of your committed bandwidth, you
don't want to end up paying both providers for bandwidth outside your
commit, the other reason is unsymmetrical routing will confuse hell of you
when troubleshooting connectivity problems, it is something unavoidable but
you should try to avoid. |
|
e*c
发帖数: 395 | 12 你给出的ubuntu benchmark链接，里面那些测试，都不是专门测memory bandwidth的。
我们已经知道很多程序64bit编译在64bit环境下运行比同样程序32bit编译在32bit环境
下运行要快，因为64bit寄存器更多。不过这个ubuntu benchmark有些结果还是很惊人
的，比如说那个apache的比较，看上去好像不大对头。benchmark后面的讨论里有人说
ubuntu 32bit好像SSE支持得不好？这个得让真懂的专家来说了。比较的结果，光看
32bit，有没有PAE对速度没有什么影响嘛。不过说到底这个ubuntu benchmark不是
memory bandwidth的比较。
要说专门比较memory bandwidth，我这个链接
http://www.geoffchappell.com/viewer.htm?doc=notes/windows/license/memory.htm
还真是。你可以说Windows Experience Index不准确，但是怎么说也是个ballpark
number，不用任何第三方工具。而且你确定作者的32 |
|
a*****a
发帖数: 1429 | 13 早早回家，在换成tomato之前近距离测了一下（2 ft)
[ ID] Interval Transfer Bandwidth
[ 4] 0.0- 1.0 sec 6784 KBytes 55575 Kbits/sec
[ 4] 1.0- 2.0 sec 6912 KBytes 56623 Kbits/sec
[ 4] 2.0- 3.0 sec 6784 KBytes 55575 Kbits/sec
[ 4] 3.0- 4.0 sec 7040 KBytes 57672 Kbits/sec
[ 4] 4.0- 5.0 sec 7040 KBytes 57672 Kbits/sec
[ 4] 5.0- 6.0 sec 6912 KBytes 56623 Kbits/sec
[ 4] 6.0- 7.0 sec 6912 KBytes 56623 Kbits/sec
[ 4] 7.0- 8.0 sec 7040 KBytes 57672 Kbits/sec
[ 4] 8.0- 9.0 sec 6784 ... 阅读全帖 |
|
b*********f
发帖数: 1585 | 14 Nvidia is announcing two products today, two Kepler based HPC / GPGPU cards
- K10 and K20.
NVIDIA Tesla K10
Tesla K10 dedicated to seismic analysis where most valuable resource is
bandwidth. It has 3x single precision of today’s fermi tesla and 1.8x the
memory bandwidth.
NVIDIA is preparing two Tesla GPUs, one is Tesla K10, based on dual-GK104
chips, which is to be devoted to high efficient computing in oil and gas
exploration and defense industry. Card is based on dual-GK104 GPUs and has a
memo... 阅读全帖 |
|
s***g
发帖数: 495 | 15 2.4G:
(%5)(18:03)(air)~> iperf -c punk
------------------------------------------------------------
Client connecting to punk, TCP port 5001
TCP window size: 129 KByte (default)
------------------------------------------------------------
[ 3] local 192.168.0.17 port 50337 connected with 192.168.0.1 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.0 sec 46.4 MBytes 38.9 Mbits/sec
5G:
(%5)(18:03)(air)~> iperf -c punk
---------------------------------------------------------... 阅读全帖 |
|
a*********a
发帖数: 3656 | 16 I see two obvious technological solutions solely on Netflix' part to end
this netflix/Cogent vs Verizon/Comcast situation.
1. Netflix could charge an additional 2-5$ fee for users who demand HD video
quality. Just like you charge $2 more for BluRay delivery service. I am
happy to pay the fee, but after I paid extra, my connection to netflix
better be directed through unhindered hubs, not those crappy Cogent
Communication hubs. Netflix should use the revenue from this extra charge to
pay ISPs for... 阅读全帖 |
|
g*****g
发帖数: 34805 | 17 Net neutrality only covers the last mile from ISP to your home.
It's important for you and bad for ISP. Without it, ISP can throttle content
providers and stir a bidding war for the bandwidth. This is not just
Netflix, it can be BitTorrent, Youtube, or any other app that uses a lot of
bandwidth. Content providers will be forced to pay up and pass down the cost
to you.
Don't forget the bandwidth is shared in your community and it's limited.
What if all Comcast video subscribers get preference? Ca... 阅读全帖 |
|
f********x
发帖数: 99 | 18 The world beyond batch: Streaming 101: A high-level tour of modern data-
processing concept
http://radar.oreilly.com/2015/08/the-world-beyond-batch-streami
by Tyler Akidau August 5, 2015
Editor’s note: This is the first post in a two-part series about the
evolution of data processing, with a focus on streaming systems, unbounded
data sets, and the future of big data.
Streaming data processing is a big deal in big data these days, and for good
reasons. Amongst them:
Businesses crave ever more tim... 阅读全帖 |
|
d*******d
发帖数: 3382 | 19 If high order statistics are taken into consideration,
white noise definition needs to be reconsidered.
We can define white noise if it's not correlated
to other time. In this way, it does not have to be stationary,
and spectrum does not exist.
Also, white noise is relevant to bandwidth. There is no
physical infinite bandwidth signal. Therefore, "whiteness"
is always with respect to the sampling rate of the signal.
If a signal is filtered at B bandwidth continuously,
then sampled at double of th |
|
f*****0
发帖数: 489 | 20 you may remember gain bandwidth product specification for many (voltage
feedback) opamps.
the same thing is here: the higher gain, the more limited bandwidth. 2mv/div
means very high gain, and also very limited bandwidth. |
|
g*z
发帖数: 124 | 21 When you mentioned about glitch, my first reaction is, are you using the
right way to measure your result?
What is the bandwidth of your scope?
Lot's of rookies think, hey, my clock is 25MHz, so a 100MHZ scope is good
enough. WRONG!
because you are using 0.13um process, the fast clock risetime/falltime put
your signal well beyond the GHz region, you need a scope whose bandwidth is
at least in the GHz range.
OK, you have the GHz scope, are you using the right probe? Don't tell me you
are just usi... 阅读全帖 |
|
a*****1
发帖数: 80 | 22 2. RC low pass filter，bandwidth越小就相当于cut-off
frequency越小 为的是滤掉更多高频
1. close loop的反馈调节（例如负反馈）是通过降低低频处的增益来得到高Bandwidth
的。开环G越大，Bandwidth越大。但是cutoff frequency没变
所以两句不矛盾啊，可以查查Thomas Lee的CMOS RF IC design有一部分讲过这个问题 |
|
b*******g
发帖数: 513 | 23 问题是找一个有两个mode的distribution的highest density region. 有个人写了个文
章和一个R package,链接是：http://robjhyndman.com/papers/hdrgraph.htm。要用的package is "hdrcde", subroutine is hdr.den for one dimension data. 用他的example data which is faithful$eruption，有结果.用自己的data, 总有一个error:Error in bw.SJ(x) : no solution in the specified range of bandwidths。在subroutine hdr.den 设定了bandwidth, "h=0.09", 得到一个还行的结果。而且用不同的bandwidth,得到的highest density region 不同。主要是自己对于一个mode的distribution,写了一个程序，有了一个不错的结果。用自己有一个mode的数据，用hdr.den 分析有同样err |
|
w*******y
发帖数: 60932 | 24 mwave has Asus Motherboard SABERTOOTH 55i Core i7 LGA1156 Intel P55 DDR3 PCI
Express Audio RAID ATX Retail + Asus IO Card PCI-E USB3 2 x USB 3.0 Ports
Support Hot Plug Function Retail:
http://www.mwave.com/mwave/deepsearch_v3.asp?scriteria=SA99004
for $125 Ar + FS for mclub members (free to join)
Description:
Asus SABERTOOTH 55i Motherboard
The TUF series delivers a "tough" image. With the unique design and high
quality components (military-standard), TUF series is born for pursuing the
preemi... 阅读全帖 |
|
l***h
发帖数: 9308 | 25 这个dd-wrt带宽监控里面，有3个实时动态图
Bandwidth Monitoring - LAN
Bandwidth Monitoring - WAN
Bandwidth Monitoring - Wirelss
分别代表什么？LAN是否是路由器后面的有线连接的4个口，唯一连接的HTPC关了，但还
有10-20Kbps持续的连接。
最后的Wireless，好像反映的是跟该从路由器跟主路由器之间的流量，下载(In?)基本
上能保持在我设定的downlink 2M，但上载（Out?）超过了我设置的uplink 0.2M限制，
经常性的达到ISP给的高限0.5M。感觉QoS对uplink没起作用（downlink还是有限制作用） |
|
f****r
发帖数: 5118 | 26 懂信号处理的帮忙分析一下 Wow Signal，俺以前懂，现在虾米了
这个东西说不定正往地球飞过来呢
http://en.wikipedia.org/wiki/Wow!_signal
The Wow! signal was a strong narrowband radio signal detected by Dr. Jerry R
. Ehman on August 15, 1977, while working on a SETI project at The Big Ear
radio telescope of The Ohio State University.[1] The signal bore expected
hallmarks of potential non-terrestrial and non-solar system origin. It
lasted for the full 72 second duration that Big Ear observed it, but has
not been detected again. Much attenti... 阅读全帖 |
|
h******b
发帖数: 6055 | 27 最近我工作最密切的合作伙伴是三哥， 感觉对他们 了解多了很多。 在北美职场，阿
三毕竟是中国人的 最大竞争对手。 在下跟成功两个字拉不上半点关 系，经验只适合
炮灰和底层管理，这里只是抛砖引 玉，希望能得到更多讨论。
先介绍一下自己的情况， 六年前进入公司， 从社会 底层的support, 到后来的
implementation, sales， account manager，project manager都做过。 我 自己编程
技术不过硬， 没法吃纯R&D的饭，学历如 果同级别的人是少林武当那我就是斧头帮出
来的， 情商用我妻子的话来说就是没法在国内混那种人。
优点是jack of all trades，相当于RPG里面的混合型 职业，能输出伤害能加血能控制
但什么都不精通那 种。 而且直觉不错，复杂的事情能简单处理，众说 纷纭的会议里
或一潭死水的项目里产生一剑封喉的 作用。Presentation skill不错。 部门一年前搞
了个 产品presentation大赛，有solo也有team performance，结果我二十分钟的solo
部分拿了那 次比赛唯一的满分，让我当时... 阅读全帖 |
|
r****n
发帖数: 179 | 28 再贴个牛人的对印策略总结：
标 题: 职场感想：如何在北美职场下游对抗阿三
发信站: BBS 未名空间站 (Sat Apr 20 10:33:32 2013, 美东)
最近我工作最密切的合作伙伴是三哥， 感觉对他们 了解多了很多。 在北美职场，阿
三毕竟是中国人的 最大竞争对手。 在下跟成功两个字拉不上半点关 系，经验只适合
炮灰和底层管理，这里只是抛砖引 玉，希望能得到更多讨论。
先介绍一下自己的情况， 六年前进入公司， 从社会 底层的support, 到后来的
implementation, sales， account manager，project manager都做过。 我 自己编程
技术不过硬， 没法吃纯R&D的饭，学历如 果同级别的人是少林武当那我就是斧头帮出
来的， 情商用我妻子的话来说就是没法在国内混那种人。
优点是jack of all trades，相当于RPG里面的混合型 职业，能输出伤害能加血能控制
但什么都不精通那 种。 而且直觉不错，复杂的事情能简单处理，众说 纷纭的会议里
或一潭死水的项目里产生一剑封喉的 作用。Presentation skill不错... 阅读全帖 |
|
a***l
发帖数: 264 | 29 你所谓性能不就是指的bandwidth吗,
top10里面有9种不同的interconnect方法, 比IB带宽高的多了去了.
IB的优势是可扩展性, 不是raw bandwidth. |
|
p****s
发帖数: 3184 | 30 你哪只眼看到有人说过OFDM最能抗干扰？除了你没人说过这话。
大傻把提高bandwidth和frequency利用效率这些额外的话题搅和进来。OFDM是提高
bandwidth和frequency利用效率的，和抗干扰没关系。 |
|
p****s
发帖数: 3184 | 31 在美军现役通信设备里，FHSS就是用来抗干扰的(anti-jamming)。
FHSS当然还有一些其它特点，比如LPD/I/E，比如同时使用多个正交的频段提高
bandwidth和frequency利用率，但这都是书呆子掉书袋，实际情况是虽然书本上如此说
，但FHSS用来干这些事是不趁手的。搞LPD/I/E的会首选DSSS而不是FHSS，提高
bandwidth和frequency利用率要靠OFDM而不是FHSS。 |
|
D*****i
发帖数: 8922 | 32 看了一下老胡今年在ISSCC上发的文章，3B主要指标如下：
Die area: 299.8mm^2
L2-cache: 4MB
Cores: 8
Max frequency: 1.05GHz
Peak performance: 128GFLOPS/s (double precision)
Compatibility: MIPS64, 200+ X86 emulation instructions
Each core includes 2 fixed-point units, 2 floating-point/vector units,
1 memory functional unit.
Vector unit： 256-bit/each. 2 vector units on each core can perform 8
multiply-and-add (MADD) double precision operations per clock cycle, that is
, 16GFLOPS/core (or 16*8=128GFLOPS/chip) peak perfo... 阅读全帖 |
|
m*r
发帖数: 37612 | 33 Bandwidth Limit Exceeded
The server is temporarily unable to service your request due to the site own
er reaching his/her bandwidth limit. Please try again later.
Apache mod_fcgid/2.3.6 mod_auth_passthrough/2.1 mod_bwlimited/1.4 FrontPage/
5.0.2.2635 Server at www.mulinosny.com Port 80 |
|
y*****u
发帖数: 4807 | 34 http://www.mulinosny.com/
Bandwidth Limit Exceeded
The server is temporarily unable to service your request due to the site
owner reaching his/her bandwidth limit. Please try again later.
尼玛，多少人在点这个link阿！！ |
|
o**********e
发帖数: 18403 | 35 【 以下文字转载自 ITRelief 俱乐部 】
发信人: onetiemyshoe (onetiemyshoe), 信区: ITRelief
标 题: 斧头帮大牛：谈职场经验123
发信站: BBS 未名空间站 (Thu Apr 10 14:26:18 2014, 美东)
很看好！
【 以下文字转载自 Seattle 讨论区 】
发信人: rayman (rayman), 信区: Seattle
标 题: 敢冒西版版规“大不惟”，就是要讨论阿三 ------职场经验谈123
发信站: BBS 未名空间站 (Thu Apr 10 14:14:30 2014, 美东)
zz
标 题: 职场感想：如何在北美职场下游对抗阿三
发信站: BBS 未名空间站 (Sat Apr 20 10:33:32 2013, 美东)
最近我工作最密切的合作伙伴是三哥， 感觉对他们 了解多了很多。 在北美职场，阿
三毕竟是中国人的 最大竞争对手。 在下跟成功两个字拉不上半点关 系，经验只适合
炮灰和底层管理，这里只是抛砖引 玉，希望能得到更多讨论。
先介绍一下自己的情况， 六年前进入公司， 从社会 底层... 阅读全帖 |
|
m****r
发帖数: 283 | 36 Denon AVR-684 AV Surround Receiver. All the usual bells and whistles. 75
Watt. Six Channels.
Velodyne CHT-8 Subwoofer. Frequency response: 35-149 Hz. A versatile 8"
sub, with a clear, deep sound.
Two B & W(Bowers & Wilkins) DM600 S3 bookshelf loudspeakers. Bandwidth 75
Hz - 22 kHz. 120 Watts at 8 Ohms.
Tweeter: 25 mm. Boomer: 180 mm.
Two B & W DM602 S3 bookshelf loudspeakers. Bandwidth 49 Hz - 22 kHz. 100
Watts at 8 Ohms. Tweeter: 25 mm.
Boomer: 135 mm. All the speakers have a clean sound with g |
|
g**d
发帖数: 77 | 37 【 以下文字转载自 Stock 讨论区 】
发信人: good (哈哈), 信区: Stock
标 题: Telecom的复苏(2)
发信站: BBS 未名空间站 (Wed Sep 6 20:01:20 2006)
2. Demand,
说了这么多的supply overcapacity,我们来讲一讲Demand.
2.1 Last mile
2000年Telecom Bubble的一个主要原因就是Last mile access bandwidth太低.多数用
户还在使用Dial up,只有56K bps.就好象大家都在骑自行车,你修再多高速公路也毫无
意义.
Last mile的情况在2006年有所好转,其中最关键的因素就是北美宽带用户的普及率在05
-06年达到50%以上,随着越来越多的Dial up用户在半强迫下转向Broadband,普及率会加
速提高.以前Dial up用户的带宽一般是56K,而一般的DSL/Cable Modem在1-2M左右,增长
幅度在20-40倍,相当吻合平均3%的Bandwidth利用率.
目前AT&T(SBC/BLS)和Verizon都在推广 |
|
n******6
发帖数: 1829 | 38 【 以下文字转载自 Military 讨论区 】
发信人: hitmantb (hitmantb), 信区: Military
标 题: 职场感想：如何在北美职场下游对抗阿三
发信站: BBS 未名空间站 (Sat Apr 20 10:33:32 2013, 美东)
最近我工作最密切的合作伙伴是三哥， 感觉对他们 了解多了很多。 在北美职场，阿
三毕竟是中国人的 最大竞争对手。 在下跟成功两个字拉不上半点关 系，经验只适合
炮灰和底层管理，这里只是抛砖引 玉，希望能得到更多讨论。
先介绍一下自己的情况， 六年前进入公司， 从社会 底层的support, 到后来的
implementation, sales， account manager，project manager都做过。 我 自己编程
技术不过硬， 没法吃纯R&D的饭，学历如 果同级别的人是少林武当那我就是斧头帮出
来的， 情商用我妻子的话来说就是没法在国内混那种人。
优点是jack of all trades，相当于RPG里面的混合型 职业，能输出伤害能加血能控制
但什么都不精通那 种。 而且直觉不错，复杂的事情能简单处理，... 阅读全帖 |
|
r****n
发帖数: 179 | 39 再贴个牛人的对印策略总结：
标 题: 职场感想：如何在北美职场下游对抗阿三
发信站: BBS 未名空间站 (Sat Apr 20 10:33:32 2013, 美东)
最近我工作最密切的合作伙伴是三哥， 感觉对他们 了解多了很多。 在北美职场，阿
三毕竟是中国人的 最大竞争对手。 在下跟成功两个字拉不上半点关 系，经验只适合
炮灰和底层管理，这里只是抛砖引 玉，希望能得到更多讨论。
先介绍一下自己的情况， 六年前进入公司， 从社会 底层的support, 到后来的
implementation, sales， account manager，project manager都做过。 我 自己编程
技术不过硬， 没法吃纯R&D的饭，学历如 果同级别的人是少林武当那我就是斧头帮出
来的， 情商用我妻子的话来说就是没法在国内混那种人。
优点是jack of all trades，相当于RPG里面的混合型 职业，能输出伤害能加血能控制
但什么都不精通那 种。 而且直觉不错，复杂的事情能简单处理，众说 纷纭的会议里
或一潭死水的项目里产生一剑封喉的 作用。Presentation skill不错... 阅读全帖 |
|
j******1
发帖数: 25 | 40 回报本版，新鲜G的电面面经，顺便求bless。
(1) Suppose each request has some data, calculate the average bandwidth in
the last 1 second, 1minute, 1 hour.
这题跟careercup那道recent average request很像。
首先，不能有错误，只想出来了use a queue as moving window, and then take
average over last 1 sec, 1 minute, 1 hour. Maintain the queue by deleting
all requests information outside the 1 hour window.
其次，允许一定的错误量。这样就可以用circular array. For example, take a
circular array to store the bandwidth in each second, sample_per_sec[3600].
Then circul... 阅读全帖 |
|
j******1
发帖数: 25 | 41 第二种方法是有误差的（这个误差取决于implementation。）假设用一个sample_per_
sec[3600] to store bandwidth in each second in the last hour.
考虑一下情况：
1. msg at 0.02 sec with 1kb data
2. msg at 3600.01 sec with 1kb data.
3. query at 3600.015 for the bandwidth in last hour.
算法一会给你2kb，算法二会给你1kb。 |
|
o**********e
发帖数: 18403 | 42 【 以下文字转载自 ITRelief 俱乐部 】
发信人: onetiemyshoe (onetiemyshoe), 信区: ITRelief
标 题: 斧头帮大牛：谈职场经验123
发信站: BBS 未名空间站 (Thu Apr 10 14:26:18 2014, 美东)
很看好！
【 以下文字转载自 Seattle 讨论区 】
发信人: rayman (rayman), 信区: Seattle
标 题: 敢冒西版版规“大不惟”，就是要讨论阿三 ------职场经验谈123
发信站: BBS 未名空间站 (Thu Apr 10 14:14:30 2014, 美东)
zz
标 题: 职场感想：如何在北美职场下游对抗阿三
发信站: BBS 未名空间站 (Sat Apr 20 10:33:32 2013, 美东)
最近我工作最密切的合作伙伴是三哥， 感觉对他们 了解多了很多。 在北美职场，阿
三毕竟是中国人的 最大竞争对手。 在下跟成功两个字拉不上半点关 系，经验只适合
炮灰和底层管理，这里只是抛砖引 玉，希望能得到更多讨论。
先介绍一下自己的情况， 六年前进入公司， 从社会 底层... 阅读全帖 |
|
o**********e
发帖数: 18403 | 43 【 以下文字转载自 ITRelief 俱乐部 】
发信人: onetiemyshoe (onetiemyshoe), 信区: ITRelief
标 题: 斧头帮大牛：谈职场经验123 (转载)
发信站: BBS 未名空间站 (Fri Jun 13 13:45:00 2014, 美东)
【 以下文字转载自 SanFrancisco 讨论区 】
发信人: onetiemyshoe (onetiemyshoe), 信区: SanFrancisco
标 题: 斧头帮大牛：谈职场经验123 (转载)
发信站: BBS 未名空间站 (Thu Apr 10 17:19:34 2014, 美东)
发信人: onetiemyshoe (onetiemyshoe), 信区: ITRelief
标 题: 斧头帮大牛：谈职场经验123
发信站: BBS 未名空间站 (Thu Apr 10 14:26:18 2014, 美东)
很看好！
发信人: rayman (rayman), 信区: Seattle
标 题: 敢冒西版版规“大不惟”，就是要讨论阿三 ------职场经验谈123
发信站: BBS ... 阅读全帖 |
|
m*****b
发帖数: 13 | 44 QUANTIL is a high growth, pre-IPO startup in Silicon Valley, backed by the
world’s second largest CDN, ChinaNetCenter. QUANTIL helps customers reach
new markets by accelerating their content on our global network. We control
over 500 POPs and 30,000 servers, which generate 7 Tbps of throughput.
Mobile, Website, Download, Video, Audio, Enterprise Application and Carrier
Acceleration, we do it all.
Our team strives to simplify IT. Our long term focus on creating true value
safeguards both our cust... 阅读全帖 |
|
l*****u
发帖数: 12114 | 45 dd-wrt _CAN_ set upper limit of bandwidth by running an IPTable rule set,
very easy to do. I limit my PPStream lappy's bandwidth to 784/64, works
great.
个和 |
|
g******n
发帖数: 53185 | 46 是很牛气。
Netflix CEO Reed Hastings Responds to Whitney Tilson: Cover Your Short
Position. Now.
A great investor and a wonderful human being, Whitney Tilson recently posted
an article about why he is short Netflix (NFLX). Whitney, who is a major co
-donor with me to charter public schools like KIPP, writes that he has lost
money betting against Netflix, and that he is still short Netflix in a big
way.
At Netflix we mostly focus on building our business and letting the numbers
do the talking. But Whi... 阅读全帖 |
|
s******v
发帖数: 4495 | 47 2. 成本
=======
I. DVD Rental
平均一个DVD Mail Delivery的成本是$1，NFLX在全美有25个Process Center处理邮件
的分发和回收，目前USPS可能提价，这方面NFLX有成本上升的压力。所以如果一个客户
，一个月看了10个DVD，NFLX是亏钱的。我看到的数字，每个NFLX用户平均一个月是5个
DVD Delivery。
Streaming一个DVD Video，成本是$0.05，没错是5美分！这个是AKAM给的Exclusive
Deal，相信LVLT的价格应当还有所下调。而且北美市场的Bandwidth价格还在以每年40-
50%的速度下降，这个Streaming的价格，即使换成HD，相比Mail Delivery，价格会保
持在相当低的水平。NFLX的Video Server是在AMZN Cloud上面，成本不是很清楚，相信
和Bandwidth差不多。
NFLX之所以狂涨，除了这个Subscriber的增长，另外就是随着客户慢慢转向Streaming
，传输成本而大幅下降，说的通俗一些，就是用白粉价格交钱的客户，NFLX慢慢... 阅读全帖 |
|
i**********0
发帖数: 73 | 48 FYI
Despite significant revision, we still see downside to 1Q consensus EPS
estimates
We expect non-GAAP revenue to rise 19% yoy to US$95m, which is in line with
company
guidance (US$93m-96m). This is mainly driven by 34% yoy growth in ad revenue
. We expect
the GAAP gross margin to further decline 2ppt qoq to 56% on higher video
content and
bandwidth costs, and the non-GAAP op margin to narrow 4ppt qoq to 23% on
higher S&M
expenditure on miniblog. We note that Bloomberg consensus estimates for ... 阅读全帖 |
|
