激情欧美日韩一区,中文av字幕一区,亚洲成色777777女色窝

山寨�U�技 — Fri, 13 Aug 2021 13:17:00 GMT

鉴于��学��p��Reach G(�W?�?上的一��文�?/span>The Girl Who Fell from the Sky�Q?a href="http://m.shnenglu.com/Chipset/archive/2021/08/11/217695.html">在这�?/a>。于是打��了解一下作�?��M�h�?�׃��安娜�U�普�?/span>(�׃��安娜�q�拉)�?/span>人物介绍选自E文维基，以下��文�?/span>
PS: ��译能很好的训练一个�h的英语和汉语句子�l�织能力�Q�时不时的让学英语的娃尝试一下翻译英语课文也��是某种�ȝ��吧�?/span>

�׃��安娜�U�普�?/span>(出生�?/span>1954�q?/span>10�?/span>10�?/span>)�Q�婚后名字朱丽安娜��_拉，是一个�d�?/span>-�U�鲁��Z�^动物学家。她�?/span>1971�q�兰�?/span>508航班�I�难的唯一�q�存者，�I�难发生后独自一人在亚马逊热带雨林生�z�M��11天。他�?/span>3000�c?/span>(9843英尺)坠落�Q�落地时仍然被安全带捆在座椅上�?br />

事前

�׃��安娜�?/span>1954�q�出生于�U�鲁利马�Q�父母都是在利马自然历史博物馆工作的德国人。她是生物学家汉斯威廉科普克和鸟�c�d��家玛丽亚�U�普克的唯一孩子。在�׃��安娜14岁时�Q�其父母带她��M��马逊雨林徏造研�I�基�?/span>--潘古那。在�q�里�Ҏ��了一�?#8220;丛林孩子”�Q�学了野外生存技能。教育当局不同意这样做�Q�要求朱莉安娜回到利马亚历山大冯�z�堡��h��志学校参加考试。她�?/span>1971�q?/span>12�?/span>23日在此校毕业�?br />

�I�难

�׃��安娜�U�普克在�l�历��难上有双重生存故事可供讲述。在1971�q�圣诞节前夕�Q�朱丽安娜搭乘了兰萨508航班。那时她刚从高中毕业�Q�她的母亲玛丽亚本打��在1971�q?/span>12�?/span>19-20日带着奛_��L��古那�Q�但是朱丽安娜希望能�?/span>12�?/span>23日去利马参加毕业典礼。玛丽亚同意�Q�打��圣诞节前夕再飞往潘古那，可是除了里尼思君�Ҏ��航空(兰萨)�q�有�I�Z��其余所有航班的席位均已售罄。尽��丈夫汉斯威廉强烈要求妻子不要乘坐那家航�I�公司的航班�Q�因��航空公司早已臭名昭著�Q�但最�l�她们还是预定了机票。飞��v飞不久遭遇闪电，在空中开始解体�ƈ一头冲向地面。朱丽安娜发现自�׃��直被捆在座椅上，下落��近2英里最�l�落入秘鲁热带雨林�?/span>

在朱丽安娜科普克的此�ơ事件上�Q�专家指出由于在下降中她被安攑ֈ�座椅上才��D��Ҏ��了下来，只是断了锁骨而已。她是该航班的唯一�q�存者，��着溪流在雨林里生活�?/span>11天�?/span>

在她醒过来后丛林里的昆虫不再叮咬她，可是蛆已�l�感染了她的背��Q?/span>9天后�Q�她扑ֈ�了一处露营地。在此，她给自己�q�行了初�U�治疗，例如用汽�Ҏ��到蛆感染的部位。蛆��x��汽油就��d��了伤口。几个小时候后�Q�运木工人发��C��她，�l�与基本�ȝ��q�把她带到更宜居的区域，然后她被�I��到医院�?/span>

伤口恢复后，�׃��安娜帮助搜救队定位失事地点和搜救死难者尸体。她妈妈的尸体于1972�q?/span>1�?/span>12日被发现。朱丽安娜回到�d国后伤口完全恢复了正常。跟父母�c�M��Q�朱丽安娜获得生物学学位后返回秘鲁��l�深入研�I�蝙蝠。她的双重生存故事成了书和电��q��M��Q�其中包括了她的自传和他传，“当我从天而降”�Q�以及导演沃�U��佐格的�U�录�?#8220;��膀上的希望”。他(沃纳赫尔佐格)�?/span>1971�q�也预定了她(�׃��安娜)的那�ơ航班，飞机起飞前最后一分钟改变计划�Q�从而躲�q�一劫�?br />

事后

�׃��安娜的��o�q�存�z�L��了各�U�瞎猜推断的主题�?很明昑֥�被安全带捆在座椅上有了某�U�程度的盄��和缓�Ԍ��而且理论上外围整排的座椅�Q�就是朱丽安娜两边的那些�Q��v��C��降落伞的作用�Q�减�~�了她下落的速度..."�?#8220;风暴漂移以及落地点厚厚的枝叶可以�q�一步降低里冲击�?..”�?/span>

�׃��安娜回到德国��d��康复�q�来�Q�走了父母一��L��路，在基��大学学习生物，1980�q�毕业。在慕尼黑麦克西�c�_��大学获得博士学位后返回到�U�鲁展开��Z�^动物研究�Q�专注于蝙蝠�Q�在1987�q�发表论文，“�U�鲁热带雨林里一个蝙蝠生物王国的研究”�?/span>1989�q�朱莉安娜科普克和一个专注于研究寄生黄蜂的昆虫学家艾力克�q�拉�l�婚�?/span>2000�q�朱丽安娜接��了潘古那当上了��M�Q�Q�父亲去世了�?/span>

现在�׃��安娜�q�拉在慕��黑巴伐利亚动物学蒐藏研�I�中心当��理员。她的自传和他传”当我从天而降”(徯��: Als ich vom Himmel fie)�?/span>2011�q?/span>3�?/span>10日由�z��林格出版�Q��ؓ此她获得�?/span>2011柯琳文学奖�?/span>2019�q�秘鲁政府授予她高��军官的杰出服务奖�?br />

山寨�U�技 2021-08-13 21:17 发表评论

What does “stride�?mean in image processing? [fwd]

山寨�U�技 — Mon, 12 Jul 2021 07:00:00 GMT

https://medium.com/@oleg.shipitko/what-does-stride-mean-in-image-processing-bba158a72bcd

Stride is an important concept in digital image processing. It allows performing several operations with an image in a very fast manner (in constant time) by simple modification of image metadata. If you are interested in finding out what stride is and how to use it stick with us.

Pixel representation in a computer memory

Before we dive into the concept of stride we first need to revise how digital images are stored in a computer memory. We will start from a pixel.

An image pixel is represented in a computer memory by a fixed number of bits. Typical pixel bit depth (amount of bits per pixel) is 32, 16, 8 or, for binary images, 1 bit. In typical RGB images, 8 bits are often used to store the color value of a single channel. Thus, the total bit depth of one pixel is 24. Processing 32 and 16-bit chunks of data is simple and effective on a typical 32 and 64-bit processors. Therefore, the pixels are stored in the format of 32 bits, where the older (or younger, depending on the implementation) 8 bits remain unused. Such an approach to storing pixels requires more memory, but it allows speeding up image processing by using the standard size of the machine word. Thus, a standard RGB image occupies 32 bits in memory and has a depth of 24 bits. We will call another 8 bits necessary to supplement the size of the memory occupied by a pixel to the value of a multiple of degree 2, pixel padding. The total number of bytes occupied by a pixel in memory is called pixel stride (See Image 1).

Image representation in a computer memory

Images are stored in computer memory pixel-by-pixel, line by line. The upper left corner of the image is usually chosen as a coordinate origin (the upper left pixel of the image has the index [0, 0]). The image is stored in memory as a one-dimensional array. Pixels of the first line of the image are first written to the memory, then pixels of the second line and so on up to the last line. Each line in addition to the pixel bytes may also contain additional bytes — line padding. Additional bytes usually do not contain useful information and do not affect the visualization of an image when, for example, displayed on the screen. These additional bytes serve to complement a line, which is necessary for more efficient image processing and is caused by the specificity of the hardware used. For example, Cairo (a popular open source vector graphics software library) requires alignment of rows to multiple 4 bytes, which allows for more efficient image processing algorithms using vectorized processor operations and processing several image pixels simultaneously.

Introducing the term of line padding requires to introduce another closely coupled term — line stride.

Line stride (increment, pitch or step size) is the number of bytes that one needs to add to the address in the first pixel of a row in order to go to the address of the first pixel of the next row. It is important to note that an image width is measured in pixels and describes an image itself (and doesn’t depend on how an image is stored in a computer memory). In contrast, a line stride depends on how an image is represented in memory and is measured in bytes.

In program source code, an image is usually represented by a data structure containing metadata (image width and height, line stride, number of channels, encoding type, etc.), as well as a pointer to the address of the first image pixel in memory (further we will refer to this address as data_begin). This information allows us to unambiguously read and decode an image from memory, as well as to perform a series of fast image operations by changing only a metadata associated with an image.

An image representation in a computer memory. Lines of an image are stored one by one in one-dimensional array.

Image operations:
Let’s summarize all the terms which we introduced to this moment:
pixel_address — a pixel address in memory
pixel depth —the number of bits per pixel (containing valuable information)
pixel_stride — the number of bytes occupied in memory by a pixel of an image
data_begin — the address of the first image pixel in memory
channels — the number of image channels (3 for an RGB-image)
channel_address — the address of a particular pixel channel in memory
height — the image height in pixels
width — the image width in pixels
line_stride — the number of bytes occupied in memory by a line of an image

Operations:

1. Computing pixel address in memory

The equation relating pixel memory address to its coordinates [y, x] in the image coordinate system can be represented as:

pixel_address = data_begin + y * line_stride + x * pixel_stride, (1)
where data_begin — the address of the first image pixel in memory.

Equation (1) is used whenever you access an image in memory. In the rest of operations, presented in this post, we will only change a metadata associated with an image and assume, that the equation (1) is applied after in order to access image.

2. Pixel decoding (for RGB image with an equal amount of bits per channel):

channel_address = pixel_address + n * depth / channels, (2)
where n is a channel index: n = 0, 1, …, channels — 1. Thus, for instance, for the typical RGB-image with an equal amount of bits per channel, a channel address in memory can be computed as follows:
R = pixel_address,
G = pixel_address + depth / channels,
B = pixel_address + 2 * depth / channels.

It is important to note that these equations depend on the type of the image stored. There are formats in which different number of bytes is used to store different channels.

3. Image flip

3.1 Vertical flip
data_begin = data_begin + (height- 1) * line_stride ,
line_stride = -line_stride.

Pointer to the first image pixel for the vertical flip

The negative line stride being inserted into equation (1) allows us to move upwards reading (or visualizing) an image from the last row to the first, thus, realizing vertical flip.

3.2 Horizontal flip
data_begin = data_begin + (width — 1) * pixel_stride ,
pixel_stride = -pixel_stride.

Pointer to the first image pixel for the horizontal flip

In the same manner as with negative line stride in the previous example, the negative pixel stride here allows us to move from right to left and to read (or visualize) an image flipped horizontally.

3.3 Vertical and horizontal flip
The combination of previous two approaches allows to flip an image in both directions at once:
data_begin = data_begin + (height-1) * line_stride + (width-1) * pixel_stride,
line_stride = -line_stride,
pixel_stride = -pixel_stride.

Pointer to the first image pixel for the simultaneous vertical and horizontal flip

4. Extracting image subwindow
data_begin = new_data_begin,
width = new_width,
height = new_height.
With this approach we set a new origin of our image (inside a boundary of the original image) and set a width and height which basically tell us how many time we should apply an equation (1) to read all pixels (width x height) and after which amount of pixels read we should increase the y coordinate (to start reading pixels of the next row). Note that such parameters as line stride remain unchanged.

5. Extracting single image channel
To extract a single image channel we can use a combination of equations (1) and (2):
pixel_address = data_begin + y * line_stride + x * pixel_stride + n * depth / channels,
where n — channel index, n = 0, 1, …, channels — 1.

REFERENCES
1.       A programmer’s view on digital images: the essentials: https://www.collabora.com/news-and-blog/blog/2016/02/16/a-programmers-view-on-digital-images-the-essentials/
2.      Microsoft Media Foundation Programming Guide: Image Stride:Image Stride When a video image is stored in memory, the memory buffer might contain extra padding bytes after each row of pixels…docs.microsoft.com
3.      Wikipedia: Stride of an array: https://en.wikipedia.org/wiki/Stride_of_an_array
4.      Cairo library: https://cairographics.org/

山寨�U�技 2021-07-12 15:00 发表评论

raster and vector image [fwd]

山寨�U�技 — Fri, 09 Jul 2021 12:13:00 GMT

转蝲的，出处不详�?br />
Raster Images

Raster (or bitmap) images are generally what you think about when thinking of images. These are the types of images that are produced when scanning or photographing an object. Raster images are compiled using pixels, or tiny dots, containing unique color and tonal information that come together to create the image.

Since raster images are pixel based, they are resolution dependent. The number of pixels that make up an image as well as how many of those pixels are displayed per inch, both determine the quality of an image. As you may have guessed, the more pixels in the image and the higher the resolution is, the higher quality the image will be.

For example, if we scale a raster image to enlarge it, without changing resolution, it will lose quality and look blurry or pixilated. This is because we are stretching the pixels over a larger area, thus making them look less sharp. This is a common problem but can be remedied by using raster image editing programs such as Photoshop to change resolution and properly scale images.

Common Raster Image Types: JPG, TIFF, GIF, PNG, BMP

Vector Images

Instead of trying to keep track of the millions of tiny pixels in a raster image, vector images, or line art, keep track of points and the equations for the lines that connect them. Generally speaking, vector images are made up of paths or line art that can infinitely scalable because they work based on algorithms rather than pixels.

One of the greatest things about vector images is that you can re-size them infinitely larger or smaller, and they will still print out just as clearly, with no increase (or decrease) in file size. If you remember back to your high school geometry, the equation for a circle of center (h,k) and radius r is (x - h)2 + (y - k)2 = r2. If you want to make the circle bigger, you just increase the value of r - instead of having to keep track of tons more pixels, the computer just has to keep track of a different number. That takes almost no file space at all.

So, what types of graphics would typically be vector? Well, almost all computer font files are based on vector images of the letters - that's why it's possible to scale them WAY up or WAY down and still have the letters be clear. All Microsoft Office clipart uses vector art and most charts and graphs produced by Office or by statistical analysis software are vector-based. Typically vector art is created in illustration applications such as Adobe Illustrator or CorelDRAW. Vector illustrations are great for logos, illustrations/artwork, animations, and text.

Common Vector File Types: EPS, SVG

山寨�U�技 2021-07-09 20:13 发表评论

Base64

山寨�U�技 — Tue, 22 Jun 2021 09:54:00 GMT

摘要: 转自�Q�https://github.com/ReneNyffenegger/cpp-base64 Code highlighting produced by Actipro CodeHighlighter (freeware)http://www.CodeHighlighter.com/-->/* base64.cpp and base64.h... 阅读全文

山寨�U�技 2021-06-22 17:54 发表评论

山寨�U�技 — Mon, 06 Jul 2020 01:49:00 GMT

摘要: 转自知乎�Q�不对真实性负责。基本基于现实生�z�，�l�节有加工。原先写�q�个�{�案�Q�只是想调侃一下当�q�我自己和周围同学们留学以及之后�U�L��的经历，因�ؓ想着以英文字母代替真实�h名，所以只写了不到26个老同学的��故事，但写完后发现太偏颇，都是��留学生转移民的案例�Q�不��以代表�q�大成年工作后再来到加拿大的技术移民，投资�U�L��Q�甚至灰色一点的�Q�难民，��h��客，所以补充了一些故事。但�q�样一来字母就不够用了�Q�我��q��他们的外��h��?.. 阅读全文

山寨�U�技 2020-07-06 09:49 发表评论

PDF里的BOX

山寨�U�技 — Mon, 02 Mar 2020 09:01:00 GMT

The PDF page boxes: MediaBox, CropBox, BleedBox, TrimBox & ArtBox

A PDF describes the content and appearance of one or more pages. It also contains a definition of the physical size of those pages. That page size definition is not as straightforward as you might think. There can in fact be up to 5 different definitions in a PDF that relate to the size of its pages. These are called the boundary boxes or page boxes:

· The MediaBox is used to specify the width and height of the page. For the average user, this probably equals the actual page size. For prepress use, this is not the case as we prefer our pages to be defined slightly oversized so that we can see the bleed (Images or other elements touching an outer edge of a printed page need to extend beyond the edge of the paper to compensate for inaccuracies in trimming the page), the crop marks and useful information such as the file name or the date and time when the file was created. This means that PDF files used in graphic arts usually have a MediaBox which is larger than the trimmed page size.

· The CropBox defines the region that the PDF viewer application is expected to display or print. So if a PDF contains a CropBox definition, Acrobat uses it for screen display and printing. For prepress use, the CropBox is pretty irrelevant. The GWG industry association recommends not to use it at all.

· The TrimBox defines the intended dimensions of the finished page. Contrary to the CropBox, the TrimBox is very important because it defines the actual page size that gets printed. The imposition programs and workflows that I know all use the TrimBox as the basis for positioning pages on a press sheet. By default the TrimBox equals the CropBox.

· The BleedBox determines the region to which the page contents needs to be clipped when output in a production environment. Usually the BleedBox is 3 to 5 millimeters larger than the TrimBox. It is nice to know the size of the BleedBox but it isn’t that important in graphic arts. Most prepress systems allow you to define the amount of bleed yourself and ignore the BleedBox. By default the BleedBox equals the CropBox.

· The ArtBox is a bit of a special case. It was originally added to indicate the area covered by the artwork of the page. It is never used for that but can be handy in a few cases:

· On a PDF page that contains an advertisement, the ArtBox can be used to define the location of that ad. This allows you to place that PDF on another page but only use the area covered by the advert.

· A more common use of the ArtBox is as a means to indicate the safety zone. When creating a poster that will be placed in a lightbox, the designer must make sure text and logo’s aren’t positioned too close to the edge. If the poster is not mounted properly, this could cause that text or logo to disappear behind the frame of the lightbox. In book design, there is also a margin where you cannot put text because the binding might make it difficult to read text that is too close to the spine. The area where it is safe to place graphic elements is called the safety zone or text safe area. The ArtBox can be used to indicate the dimensions of this part of the page.

General rules regarding page boxes

· Each page in a PDF can have different sizes for the various page boxes.

· The page boxes are always rectangular. That may seem logical but artwork is not always rectangular: a PDF can represent an oval label or the foldout of a cardboard box.

· A PDF always has a MediaBox definition. All the other page boxes do not necessarily have to be present in regular PDF files.

· The above rule is not true for the PDF/X file formats:

· PDF/X-1a and PDF/X-3 compliant files need to include the MediaBox, TrimBox, and BleedBox.

· PDF/X-4 files need, next to the MediaBox, a TrimBox or an ArtBox, but not both. The ArtBox or TrimBox cannot be larger that the BleedBox. If a CropBox is present, the ArtBox, TrimBox, and BleedBox need to extend beyond its boundaries.

· The MediaBox is the largest page box in a PDF. The other page boxes can equal the size of the MediaBox but they are not expected to be larger (The latter is explicitly required in the PDF/X-4 requirements). If they are larger, the PDF viewer will use the values of the MediaBox.

BBox

Within PDF files there is another box, the bounding box or BBox, that is used. The bounding box is a rectangular frame that determines the dimensions of an object (such as a graphic, font or pattern) that is placed inside a PDF document. As such, this box has nothing to do with the page boxes.

山寨�U�技 2020-03-02 17:01 发表评论

�~�译器预定义的宏

山寨�U�技 — Wed, 26 Oct 2016 01:01:00 GMT

源自�Q�https://sourceforge.net/p/predef/wiki/Architectures/

Architectures

Please send updates/corrections to predef-contribute.

Alpha

Type	Macro	Description
Identification	`__alpha__`	Defined by GNU C
Version	`__alpha_ev'V'__`	V = Version
Identification	`__alpha`	Defined by DEC C
Identification	`_M_ALPHA`	Defined by Visual Studio

Example

CPU	Macro
Alpha EV4	`__alpha_ev4__`
Alpha EV5	`__alpha_ev5__`
Alpha EV6	`__alpha_ev6__`

AMD64

Type	Macro	Description
Identification	`__amd64__` `__amd64` `__x86_64__` `__x86_64`	Defined by GNU C and Sun Studio
Identification	_M_X64 `_M_AMD64`	Defined by Visual Studio

Notice that x32 can be detected by checking if the CPU uses the ILP32 data model.

ARM

Type	Macro	Description
Identification	`__arm__`	Defined by GNU C and RealView
Identification	`__thumb__`	Defined by GNU C and RealView in Thumb mode
Version	`__ARM_ARCH_'V'__`	V = Version Defined by GNU C 1
Identification	`__TARGET_ARCH_ARM` `__TARGET_ARCH_THUMB`	Defined by RealView
Version	`__TARGET_ARCH_ARM` = V `__TARGET_ARCH_THUMB` = V	V = Version
Version	`__TARGET_ARCH_'VR'`	VR = Version and Revision
Identification	`_ARM`	Defined by ImageCraft C
Identification	`_M_ARM`	Defined by Visual Studio
Identification	`_M_ARMT`	Defined by Visual Studio in Thumb mode
Version	`_M_ARM` = V	V = Version
Identification	`__arm`	Defined by Diab

Example

CPU	Macro	`_M_ARM`
ARM 2	`__ARM_ARCH_2__`
ARM 3	`__ARM_ARCH_3__` `__ARM_ARCH_3M__`
ARM 4T	`__ARM_ARCH_4T__` `__TARGET_ARM_4T`
ARM 5	`__ARM_ARCH_5__` `__ARM_ARCH_5E__`	5
ARM 5T	`__ARM_ARCH_5T__` `__ARM_ARCH_5TE__` `__ARM_ARCH_5TEJ__`
ARM 6	`__ARM_ARCH_6__` `__ARM_ARCH_6J__` `__ARM_ARCH_6K__` `__ARM_ARCH_6Z__` `__ARM_ARCH_6ZK__`	6
ARM 6T2	`__ARM_ARCH_6T2__`
ARM 7	`__ARM_ARCH_7__` `__ARM_ARCH_7A__` `__ARM_ARCH_7R__` `__ARM_ARCH_7M__` `__ARM_ARCH_7S__`	7

ARM64

Type	Macro	Description
Identification	`__aarch64__`	Defined by GNU C 1

Blackfin

Type	Macro	Description
Identification	`__bfin` `__BFIN__`	Defined by GNU C

Convex

Type	Macro	Description
Identification	`__convex__`	Defined by GNU C
Version	`__convex_'V'__`	V = Version

Example

CPU	Macro
Convex C1	`__convex_c1__`
Convex C2	`__convex_c2__`
Convex C32xx series	`__convex_c32__`
Convex C34xx series	`__convex_c34__`
Convex C38xx series	`__convex_c38__`

Epiphany

Type	Macro
Identification	`__epiphany__`

HP/PA RISC

Type	Macro	Description
Identification	`__hppa__`	Defined by GNU C
Identification	`__HPPA__`	Defined by Stratus VOS C
Identification	`__hppa`
Version	`_PA_RISC'V'_'R'`	V = Version R = Revision

Example

CPU	Macro
PA RISC 1.0	`_PA_RISC1_0`
PA RISC 1.1	`_PA_RISC1_1` `__HPPA11__` `__PA7100__`
PA RISC 2.0	`_PA_RISC2_0` `__RISC2_0__` `__HPPA20__` `__PA8000__`

Intel x86

Type	Macro	Format	Description
Identification	`i386` `__i386` `__i386__`		Defined by GNU C
Version	`__i386__` `__i486__` `__i586__` `__i686__`		Defined by GNU C
Identification	`__i386`		Defined by Sun Studio
Identification	`__i386` `__IA32__`		Defined by Stratus VOS C
Identification	`_M_I86`		Only defined for 16-bits architectures Defined by Visual Studio, Digital Mars, and Watcom C/C++ (see note below)
Identification	`_M_IX86`		Only defined for 32-bits architectures Defined by Visual Studio, Intel C/C++, Digital Mars, and Watcom C/C++
Version	`_M_IX86`	V00	V = Version
Identification	`__X86__`		Defined by Watcom C/C++
Identification	`_X86_`		Defined by MinGW32
Identification	`__THW_INTEL__`		Defined by XL C/C++
Identification	`__I86__`		Defined by Digital Mars
Version	`__I86__`	V	V = Version
Identification	`__INTEL__`		Defined by CodeWarrior
Identification	`__386`		Defined by Diab

Notice that Watcom C/C++ defines _M_IX86 for both 16-bits and 32-bits architectures. Use __386__ or _M_I386 to detect 32-bits architectures in this case.

Notice that the Stratus VOS is big-endian on IA32, so these macros cannot be used to detect endianness if __VOS__ is set.

Example

CPU	`_M_IX86`	`__I86__`
80386	300	3
80486	400	4
Pentium	500	5
Pentium Pro/II	600	6

Intel Itanium (IA-64)

Type	Macro	Format	Description
Identification	`__ia64__` `_IA64` `__IA64__`		Defined by GNU C
Identification	`__ia64`		Defined by HP aCC
Identification	`_M_IA64`		Defined by Visual Studio
Identification	`_M_IA64`		Defined by Intel C/C++
Version	`_M_IA64`	?
Identification	`__itanium__`		Defined by Intel C/C++

Example

CPU	`_M_IA64` (Intel C/C++)
	64100

Motorola 68k

Type	Macro	Description
Identification	`__m68k__`	Defined by GNU C
Version	`__mc'V'__` `__mc'V'` `mc'V'`	V = Version
Identification	`M68000`	Defined by SAS/C
Identification	`__MC68K__`	Defined by Stratus VOS C
Version	`__MC'V'__`	V = Version

Example

CPU	Macro
68000	`__mc68000__` `__MC68000__`
68010	`__mc68010__`
68020	`__mc68020__` `__MC68020__`
68030	`__mc68030__` `__MC68030__`
68040	`__mc68040__`
68060	`__mc68060__`

MIPS

Type	Macro	Description
Identification	`__mips__` `mips`	Defined by GNU C
Version	`_MIPS_ISA` = `_MIPS_ISA_MIPS'V'`	V = MIPS ISA level
Version	`_R3000` `_R4000` `_R5900`
Identification	`__mips`	Defined by MIPSpro and GNU C
Version	`__mips`	The value indicates the MIPS ISA (Instruction Set Architecture) level
Version	`__MIPS_ISA'V'__`	V = MIPS ISA level
Identification	`__MIPS__`	Defined by Metrowerks

Example

CPU	`_MIPS_ISA`	GNU C Macro	`__mips`	MIPSpro Macro
R2000	`_MIPS_ISA_MIPS1`		1
R3000	`_MIPS_ISA_MIPS1`	`_R3000`	1
R6000	`_MIPS_ISA_MIPS2`		2	`__MIPS_ISA2__`
R4000		`_R4000`
R4400	`_MIPS_ISA_MIPS3`		3	`__MIPS_ISA3__`
R8000	`_MIPS_ISA_MIPS4`		4	`__MIPS_ISA4__`
R10000	`_MIPS_ISA_MIPS4`		4	`__MIPS_ISA4__`

PowerPC

Type	Macro	Description
Identification	`__powerpc` `__powerpc__` `__powerpc64__` `__POWERPC__` `__ppc__` `__ppc64__` `__PPC__` `__PPC64__` `_ARCH_PPC` `_ARCH_PPC64`	Defined by GNU C
Version	`__ppc'V'__`	V = Version
Identification	`_M_PPC`	Defined by Visual Studio
Version	`_M_PPC`	?
Identification	`_ARCH_PPC` `_ARCH_PPC64`	Defined by XL C/C++
Version	`_ARCH_'V'`	V = Version
Version	`__PPCGECKO__`	Gekko Defined by CodeWarrior
Version	`__PPCBROADWAY__`	Broadway Defined by CodeWarrior
Version	`_XENON`	Xenon
Identification	`__ppc`	Defined by Diab

Example

CPU	`_M_PPC`	Macro	XL Macro
PowerPC 440			`_ARCH_440`
PowerPC 450			`_ARCH_450`
PowerPC 601	601	`__ppc601__`	`_ARCH_601`
PowerPC 603	603	`__ppc603__`	`_ARCH_603`
PowerPC 604	604	`__ppc604__`	`_ARCH_604`
PowerPC 620	620

Pyramid 9810

Type	Macro
Identification	`pyr`

RS/6000

Type	Macro	Description
Identification	`__THW_RS6000`	Defined by XL C/C++
Identification	`_IBMR2`
Identification	`_POWER`
Identification	`_ARCH_PWR` `_ARCH_PWR2` `_ARCH_PWR3` `_ARCH_PWR4`

SPARC

Type	Macro	Description
Identification	`__sparc__`	Defined by GNU C
Identification	`__sparc`	Defined by Sun Studio
Version	`__sparc_v8__` `__sparc_v9__`	Defined by GNU C
Version	`__sparcv8` `__sparcv9`	Defined by Sun Studio

Example

CPU	Sun Studio Macro	GNU C Macro
SPARC v8 (SuperSPARC)	`__sparcv8`	`__sparc_v8__`
SPARC v9 (UltraSPARC)	`__sparcv9`	`__sparc_v9__`

SuperH

Type	Macro	Description
Identification	`__sh__`	Defined by GNU C
Version	`__sh1__` `__sh2__` `__sh3__` `__SH3__` `__SH4__` `__SH5__`

SystemZ

Type	Macro	Description
Identification	`__370__` `__THW_370__`	Identifies System/370 Defined by XL C/C++
Identification	`__s390__`	Identifies System/390 Defined by GNU C
Identification	`__s390x__`	Identifies z/Architecture Defined by GNU C
Identification	`__zarch__`	Identifies z/Architecture Defined by clang
Identification	`__SYSC_ZARCH__`	Identifies z/Architecture Defined by Systems/C

TMS320

Type	Macro	Description
Identification	`_TMS320C2XX` `__TMS320C2000__`	C2000 series
Identification	`_TMS320C5X` `__TMS320C55X__`	C5000 series
Identification	`_TMS320C6X` `__TMS320C6X__`	C6000 series

Example

DSP	Macro
C28xx	`_TMS320C28X`
C54x	`_TMS320C5XX`
C55x	`__TMS320C55X__`
C6200	`_TMS320C6200`
C6400	`_TMS320C6400`
C6400+	`_TMS320C6400_PLUS`
C6600	`_TMS320C6600`
C6700	`_TMS320C6700`
C6700+	`_TMS320C6700_PLUS`
C6740	`_TMS320C6740`

TMS470

Type	Macro
Identification	`__TMS470__`

山寨�U�技 2016-10-26 09:01 发表评论

When someone gives you a bug [FWD]

山寨�U�技 — Wed, 01 Apr 2015 06:39:00 GMT

(I'm a front-end developer, but I thought this might resonate)

Someone gives you a bug. "The light in the conference room on the 26th floor is on. It needs to be off."

A note on the bug says, “This will take you like 5 minutes. It's just flipping a switch."

You go to the conference room on 26. The light is on, but there's no light switch in the room.

So you prepare to install one. But the designer says it would ruin the room’s aesthetic. Plus, the walls are concrete. With the proper tools, you could install the switch. But no one will approve the purchase of the proper tools. Without the proper tools it will take two days. And they want it done now, because they're afraid that any minute the CEO might decide to go to the 26th floor and happen to walk by the conference room and ask why the hell that light is on.

And now you're getting emails asking why the light isn't off yet.

So now you have to stop and send a group email to explain the situation, and several people start up a panicked email chain.

You know if you wait for the problem to be resolved by anyone discussing it in the email chain, it won’t get fixed. The bug has your name on it, and it's dated today, so you're the one in trouble if it isn't resolved. So you go up into the hallway ceiling on 26, find the wires leading to the light, cut and cap them. Finally. Problem solved.

In order to quell the panic in the email thread, you report back how you solved the issue.

You don’t hear anything for a while. When you do, everyone is concerned that now the light can’t be turned on and off. What if the CEO wants to have a meeting in there? So here’s what they ask you to do: They want you to run wires from the light down into the basement. When someone needs the light to be on or off, they’ll contact you, and have you run to the basement and either connect or disconnect the wires.

You protest the ridiculousness of this solution. Your boss says, “Yeah, I know it’s not ideal. But it’s the only solution we have right now."

At this point you realize you have a choice. You could do this. Or you could quit in protest, and find another job. But you realize that once you start that new job, they’re likely to ask you to do something just as idiotic, if not more so.

So you go run the wires from floor 26 down to the basement. When you get to the basement, you see dozens of wires hanging out of the walls, from all the people who have had to do this exact same thing before. (So that’s where the idea came from.) You set up the wires and label them as best you can, with a short apology to whoever has to deal with this next.

When you get back to your desk, you have a message. QA has reopened the bug. It says, “I see light."

You head back up to the conference room on 26. The light is off. You go back to your desk and close the bug, reporting that you checked on it in person.

QA reopens the bug again. “Room still lit” it says. After looking at the unlit bulb one more time, you tell your boss, who suggests you go back down to the basement and check the wires. You protest that you are looking at the light right now and it’s off. “I know, but this way you can tell QA you checked out absolutely everything."

So you sigh and head to the basement. Sure enough, the wires are not connected. The ends are capped. They are not resting on anything that could conduct electricity.

You report back to QA that you checked the wires, which are not connected, and that you looked at the bulb, which was unlit.

“I didn’t mean the bulb,” says QA. “The bug is about light in the room. There’s still too much light. Shouldn’t you close the blinds?"

You respond that the blinds don’t fall under your control, and that the bug specifies the light being turned off.

Not believing you, QA sends out a group email asking if the blinds are covered by the bug.

Some time passes before you hear from anyone. Finally someone from the email chain calls you.

“Theoretically,” they ask, “could someone participating in a meeting in the conference room on 26 open or close the blinds by themselves if it was too bright or too dark?"

Yes they could, you reply.

“Like, an ordinary person? They wouldn’t need you to do it?"

Yes, an ordinary person. No, they wouldn’t need you. Anyone can do it.

“Great. Excellent. Then we’ll leave that for now. I’ll schedule a stand-up meeting about the blinds issue."

So the bug is closed. Now, the CEO, possibly having caught wind of all the discussion and furtive activity surrounding the conference room on 26, wants to have a meeting there. You get several panicked emails that they need the light on.

You go to the basement, connect the wires, and return to your desk, to find 32 new messages in your inbox. “Something’s wrong—the light’s not on!” “There’s a problem — no light!” “Are you getting these emails?” and so on.

The 32nd email says, “Nevermind—the light’s on."

This process is repeated more or less exactly when it’s time to turn the light off again.

But if there’s any good news, it’s this: after the meeting, everyone forgets that there even is a conference room on 26, so you never have to do anything about it again.

Wow, gold! Thanks!

山寨�U�技 2015-04-01 14:39 发表评论

using gcc_x86 assembly to scan the most significant one bit and the least significant one bit

山寨�U�技 — Sun, 15 Jun 2014 09:38:00 GMT

原来发过搜烦�W�一位的帖子�Q�这�ơ算再补充一下吧。以下取自网�l�，作者不详�?br />

1 #ifdef __GNUC__
2
3 #if (defined _WIN64 || defined __x86_64__)
4
5 /**
6  * ../bit_scan_forward
7  * @return index (0..63) of least significant one bit
8  */
9 unsigned bit_scan_forward(unsigned long long x)
10 {
11    asm("bsfq %0, %0" : "=r" (x) : "0" (x));
12    return unsigned(x);
13 }
14
15 /**
16  * bit_scan_reverse
17  * @return index (0..63) of most significant one bit
18  */
19 unsigned bit_scan_reverse(unsigned long long x)
20 {
21    asm("bsrq %0, %0" : "=r" (x) : "0" (x));
22    return unsigned(x);
23 }
24
25 #elif (defined _WIN32) || (defined __linux__)
26
27 /**
28  * ../bit_scan_forward
29  * @return index (0..63) of least significant one bit
30  */
31 unsigned bit_scan_forward(unsigned long long x)
32 {
33    asm("bsf %0, %0" : "=r" (x) : "0" (x));
34    return unsigned(x);
35 }
36
37 /**
38  * bit_scan_reverse
39  * @return index (0..63) of most significant one bit
40  */
41 unsigned bit_scan_reverse(unsigned long long x)
42 {
43    asm("bsr %0, %0" : "=r" (x) : "0" (x));
44    return unsigned(x);
45 }
46
47 #endif
48
49 #endif

山寨�U�技 2014-06-15 17:38 发表评论

Is there an ISO standard for C sharp language?

山寨�U�技 — Tue, 27 Sep 2011 06:43:00 GMT

今天看老外写的文章提到�?/span>C#国际标准…

什么，C#�q�有国际标准�Q�不会吧�Q�谷歌一下，果然如此�Q?/span>06�q�就有了ISO标准…居然�q�能�?/span>ISO官网下蝲到电子版�?/span>

看来金钱��是好东西啊�Q�有��p��佉K��推磨道理在地球上哪里都好�?/span>…

�l�箋��h��Q�在MSDN上发��C��下面的文字�?/span>

�q�接�Q?/span>http://msdn.microsoft.com/en-us/netframework/aa569283

In June 2005, the General Assembly of the international standardization organization Ecma approved edition 3 of the C# Language and the Common Language Infrastructure (CLI) specifications, as updated Ecma-334 and Ecma-335, respectively (see press release). The updated technical report on the CLI, Ecma TR-84, and a new technical report on the CLI, Ecma TR-89, were also ratified.

In July 2005, Ecma submitted the Standards and TRs to ISO/IEC JTC 1 via the ISO Fast-Track process. The Standards were adopted in April 2006 as ISO/IEC 23270:2006 (C#), ISO/IEC 23270:2006 (CLI), ISO/IEC TR 23272:2006 (CLI, XML Libraries) and ISO ISO/IEC TR 25438:2006 (CLI, Common Generics).

In July 2006 the General Assembly of Ecma approved edition 4 of the Standards which correspond to the ISO 2006 versions.

Latest Standards

The following official Ecma documents are available for C# and the CLI (TR-84, TR-89). These links are direct from Ecma:

File name	Size (Bytes)	Content
ECMA-334.pdf	2 614 003	C# Language Specification
ECMA-335.pdf	3 219 107	Common Language Infrastructure
ECMA-335.zip	754 982	XML-based Library Specification
TR-084.pdf	187 450	Information Derived from Partition IV XML File
TR-084.zip	19 329 610	XML Tool, Libraries in Microsoft© Word and PDF
TR-089.pdf	589 400	Common Generics Library
TR-089.zip	461 074	Common Generics Library Reference Implementation

Reference implementation for TR-89
Reference implementation for the Parallel API

The official ISO/IEC documents are available from the ISO/IEC Freely Available Standards page. These links are direct from that page:

	File name		Content
	ISO/IEC 23270:2006		Information technology -- Programming languages -- C#
	ISO/IEC 23271:2006 - Software		Information technology -- Common Language Infrastructure (CLI) Partitions I to VI
	ISO/IEC TR 23272:2006 - Software		Information technology -- Common Language Infrastructure (CLI) -- Technical Report on Information Derived from Partition IV XML File
	ISO/IEC TR 25438:2006 - Software		Information technology -- Common Language Infrastructure (CLI) -- Technical Report: Common Generics

Current Working Draft

Work on the 5th edition of Ecma-335 CLI standard began in mid-2009. The TC49-TG3 task group is working on extending both the virtual machine and class libraries of the CLI specification. In addition, improvements are being made to clarify existing elements of the specification. Many of these improvements are the result of feedback received from outside the task group, for which the task group is grateful.

Posted below is a snapshot of the committee's work as of 27 March 2010.

The participants in TC49/TG3 are providing these working documents to the public for informational purposes only. The contents are subject to change as often as once a month. To participate in the standardization process, contact your organization's Ecma representative. If your company does not currently participate in Ecma and wishes to do so, please contact ECMA directly.

The following organizations and contributors are actively participating in the work of TC49/TG3:
Eiffel Software, Microsoft Corporation, Novell Corporation, Kahu Research, and Twin Roots.

Many of the organizations that are currently participating in the TC49/TG3 work have volunteered to mirror this site. The URLs for the mirror sites are:
- Eiffel Software
- Microsoft Corporation
- Novell Corporation
- Kahu Research
- Twin Roots

Available Documents (Documents current as of 27 March 2010)
The following working draft documents are available:

- CLI Partition I - Architecture (word/pdf zip)
- CLI Partition II - Metadata and File Format (word/pdf zip)
- CLI Partition III - CIL (word/pdf zip)
- CLI Partition IV - Library (word/pdf zip)
- CLI Partition V - Binary Formats (word/pdf zip)
- CLI Partition VI - Annexes (word/pdf zip)
- Class Library XML (xml zip)
- Class Library Detailed Specifications (word/pdf zip)

Annotated Standards

Members of the Standard committees and others have combined to produce annotated versions of the Standards. These are:

The Common Language Infrastructure Annotated Standard, James S. Miller & Susann Ragsdale, Addison-Wesley, 2004, ISBN 0-321-15493-2 (based on Edition 2 of Ecma-335)
C# Annotated Standard, Jon Jagger, Nigel Perry & Peter Sestoft, Morgan Kaufmann, 2007, ISBN 978-0-12-372511-0 (based on Edition 4 of Ecma-334)

Microsoft Implementation Specific Versions

The following documents are versions of the Standards with Microsoft implementation-specific notes added. These notes provide extra information about Microsoft's Common Language Runtime (CLR) implementation of the CLI.

File name	Size (Bytes)	Content
MS Partition I.pdf	815 983	Common Language Infrastructure, Partition I: Concepts and Architecture
MS Partition II.pdf	1 758 195	Common Language Infrastructure, Partition II: Metadata Definition and Semantics
MS Partition III.pdf	661 414	Common Language Infrastructure, Partition III: CIL Instruction Set
MS Partition IV.pdf		Common Language Infrastructure, Partition IV: Profiles and Libraries
MS Partition V.pdf		Common Language Infrastructure, Partition V: Binary Formats
MS Partition VI.pdf		Common Language Infrastructure, Partition VI: Annexes

The Ecma 4^th and ISO 2^nd Editions

Aside from bug fixes, major enhancements from previous editions include:

CLI

First-class support for generics at the runtime and class library level
An API to help developers begin multithreaded and parallel programming
Enhancements to the Common Intermediate Language (CIL) and Common Language Specification (CLS)
An interchangeable debug format

C#

First-class language support for generics
Anonymous methods
Iterators
Nullable Types

Previous Editions Background

In August, 2000, Microsoft Corporation, Hewlett-Packard and Intel Corporation co-sponsored the submission of specifications for the Common Language Infrastructure (CLI) and C# programming language to the international standardization organization Ecma. As a result, Ecma formed two task groups (TG3 and TG2, respectively) within TC39, its technical committee responsible for programming languages and application development.

During the next year, the co-sponsor companies, in conjunction with other Ecma members and guests (including IBM, Fujitsu Software, Plum Hall, Monash University and ISE), refined these specifications into standards. In December, 2001, the Ecma General Assembly ratified the 1^st edition of the C# and CLI standards as Ecma-334 and Ecma-335, respectively. A technical report on the CLI, Ecma TR-84, was also ratified.

In late December, 2001, Ecma submitted the standards and TR to ISO/IEC JTC 1 via the latter's Fast-Track process. The subsequent 6-month evaluation and comment period resulted in two NO votes (Japan and UK) on the draft standards, and one NO vote (Japan) on the draft TR. All comments resulting from this review were considered at a ballot resolution meeting held in October, 2002. The two NO votes on the standards were resolved, making acceptance unanimous. However, Japan did not change its NO vote on the draft TR (Japan would like to see a formatted/readable rendering of the CLI class library as part of the standard, not as a TR; this will be considered for a future edition).

The ISO/IEC standards and TR were published in April, 2003, and are known formally as ISO/IEC 23270 (C#), ISO/IEC 23271 (CLI) and ISO/IEC 23272 (CLI TR). Equivalent specifications were adopted as 2^nd edition standards and TR by Ecma at its December, 2002, General Assembly.

Joining Ecma

To participate in the standardization process, contact your organization’s Ecma representative. If your company does not currently participate in Ecma and wishes to do so, please contact Ecma directly.

Acknowledgements

The following organizations have participated in the work of Ecma TC39/TG2 and TC39/TG3 and their contributions are gratefully acknowledged: Borland, Fujitsu, Hewlett-Packard, Intel Corporation, International Business Machines, ISE, IT University Copenhagen, JSL (UK), Kahu Research (New Zealand), Microsoft Corporation, Monash University, Netscape, Novell Corporation, OpenWave, Plum Hall, Sun Microsystems.

Many of the organizations that have participated in the TC39/TG2 and TC39/TG3 work have volunteered to mirror this site. The links for the mirror sites are:

Intel Corporation
ISE (Eiffel)
IT University, Copenhagen
Kahu Research, New Zealand
Microsoft Corporation
Novell (Mono)

山寨�U�技 2011-09-27 14:43 发表评论

一个母亲一生的8个谎�a�(�?

山寨�U�技 — Wed, 07 Sep 2011 03:12:00 GMT

转脓�Q�已�l�记不清出处�Q�抱歉�?/p>

1、儿�Ӟ��男孩家很穷�Q�吃饭时�Q�饭常常不够吃，母亲��把自己��里的饭分给孩子吃。母亲说�Q�孩子们�Q�快吃吧�Q�我不饿�Q?#8212;—母亲撒的�W�一个谎

2、男孩长�w�体的时候，勤劳的母亲常用周日休息时间去厉K��农村��x��里捞些鱼来给孩子们补钙。鱼很好吃，鱼汤也很鲜。孩子们吃鱼的时候，母亲��在一旁啃鱼骨��_��用舌头舔鱼骨头上的肉渍。男孩心��|��把自己��里的鱼夹到母亲��里�Q�请母亲吃鱼。母亲不吃，母亲又用�{�子把鱼夹回男孩的碗里。母亲说�Q�孩子，快吃吧，我不爱吃��|��——母亲撒的�W�二个谎

3、上初中了，��Z��~�够男孩和哥姐的学费�Q�当�~��n工的母亲��去居委会领些火柴盒拿回家来�Q�晚�?�p�了挣点分分��p��点家用。有个冬天，男孩半夜醒来�Q�看到母亲还�w�着�w�子在��a灯下�p�火柴盒。男孩说�Q�母�Ԍ��睡了吧，明早您还要上班呢。母亲笑�W�，��_��孩子�Q?快睡吧，我不囎ͼ�——母亲撒的�W�三个谎

4、高考那�q�_��母亲请了假天天站在考点门口为参加高考的男孩助阵。时逢盛夏，烈日当头�Q�固执的母亲在烈日下一站就是几个小时。考试�l�束的铃声响了，母亲�q�上去递过一杯用�|�头瓶��好的��茶叮嘱孩子喝了�Q�茶亦浓�Q�情更浓。望着母亲�q�裂的嘴唇和满头的汗珠，男孩��手中的�|�头瓶反递过去请母亲喝。母亲说�Q�孩子，快喝吧，我不��_��——母亲撒的四个�?

5、父亲病逝之后，母亲又当爹又当娘�Q�靠着自己在缝�U�社里那点微薄收入含辛茹苦拉扯着几个�?子，供他们念书，日子�q�得苦不堪言。胡同�\口电�U�杆下修表的李叔叔知道后�Q�大事小事就扑ֲ��q�来打个帮手�Q�搬搬煤�Q�挑挑水�Q�送些��q��来帮补男孩的安��。�h�?草木�Q�孰能无情。左��d��舍对此看在眼里，记在心里�Q�都劝母亲再嫁，何必苦了自己。然而母亲多�q�来却守�w�如玉，始终不嫁�Q�别人再劝，母亲也断然不听，母亲 ��_��我不爱！——撒的五个�?

6、男孩和她的哥姐大学毕业参加工作后，下了岗的母亲��在附近农��N市场摆了个小摊维持生�z�R��n在外地工作的孩子们知道后��常常寄钱回来补贴母�Ԍ��母亲坚决不要�Q��ƈ��钱退了回厅R��母亲说�Q�我有钱�Q?#8212;—撒的六个�?

7、男孩留校�Q教两�q�_��后又考取了美国一所名牌大学的博士生�Q�毕业后留在��国一家科研机构工作，待遇相当丰厚�Q�条件好了，�w�在异国的男孩想把母亲接来��n享清��却被老�h回绝了。母亲说�Q�我不习惯！——撒的七个�?

8、晚�q�_��母亲患了胃癌�Q�住�q�了医院�Q�远在大西洋彼岸的男孩乘飞机赶回来时�Q�术后的母亲已是奄奄一息了。母亲老了�Q�望着被病��折��得��d��z�L��的母�Ԍ��男孩悲痛�Ʋ绝�Q�潸然泪下。母亲却��_��孩子�Q�别哭，我不疹{�?#8212;—撒的最后一个谎

山寨�U�技 2011-09-07 11:12 发表评论

查找一�?4位整��C��q�制�W�一�?

山寨�U�技 — Tue, 23 Aug 2011 04:41:00 GMT

有时候用��C��q�算。需要快速找��C��个整数的二进制第一�?/span>1�?/span>0在哪个位(下标)�Q�例如：十进制数100的二�q�制�?/span>1100100�Q�那么它的第一�?/span>1在下�?/span> �?/span>2的位�|?/span>(bsf, bit scan forward)�?/span>6的位�|?/span>(bsr, bit scan in reverse order)�Q�由于只用于存储一个状态，至于�?/span>bsf�q�是bsr则无所谓�?/span>

解决�q�个问题的第一个想法就是用内联汇编的做法，使用特别�?/span>CPU指��o��L��Q�但汇编的可�U�L��性比较差�Q�不同的CPU型号使用的指令可能不一��P��执行速度也不一栗��?/span>
假设找一�?/span>64位无�W�号整数二进制的�W�一�?/span>1�Q�用bsf, AT& T汇编(gcc汇编)可以�q�样做：

1 // bit scan forward for 64 bit integral number
2 /* ============================================ */
3 inline int bsf_asm (uint64_t w)
4 {
5   int x1, x2;
6   asm ("bsf %0,%0\n" "jnz 1f\n" "bsf %1,%0\n" "jz 1f\n" "addl $32,%0\n"
7      "1:": "=&q" (x1), "=&q" (x2):"1" ((int) (w >> 32)),
8      "0" ((int) w));
9   return x1;
10 }

如果�?/span>C来实现的话，那就有点�ȝ��了，在此不讲复杂的数学原理，仅仅�l�出代码�?/span>

1 // bit scan forward for 64 bit integral number
2 /* ============================================ */
3 inline int bsf_folded (uint64_t bb)
4 {
5      static const int lsb_64_table[64] =
6    {
7       63, 30,  3, 32, 59, 14, 11, 33,
8       60, 24, 50,  9, 55, 19, 21, 34,
9       61, 29,  2, 53, 51, 23, 41, 18,
10       56, 28,  1, 43, 46, 27,  0, 35,
11       62, 31, 58,  4,  5, 49, 54,  6,
12       15, 52, 12, 40,  7, 42, 45, 16,
13       25, 57, 48, 13, 10, 39,  8, 44,
14       20, 47, 38, 22, 17, 37, 36, 26
15    };
16    unsigned int folded;
17    bb ^= bb - 1;
18    folded = (int) bb ^ (bb >> 32);
19    return lsb_64_table[folded * 0x78291ACF >> 26];
20 }

如果想从后往前搜索一个整数的二进制第一�?/span>1的下标，用汇�~�可以这样做�?/span>

1 // bit scan in reverse order for 64 bit integral number
2 /* ============================================ */
3 inline int bsr_asm (uint64_t w)
4 {
5   int x1, x2;
6   asm ("bsr %1,%0\n" "jnz 1f\n" "bsr %0,%0\n" "subl $32,%0\n"
7      "1: addl $32,%0\n": "=&q" (x1), "=&q" (x2):"1" ((int) (w >> 32)),
8      "0" ((int) w));
9   return x1;
10 }

如果�?/span>C来实现的话，也比较简单，�?/span>divide and conquer 的原理就不会太慢�?/span>

1 // a logn (n == 32)algorithm for bit scan in reverse order
2 /* ============================================ */
3 inline int bsr32(uint32_t bb)
4 {
5      static const char msb_256_table[256] =
6    {
7       0, 0, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3,
8       4, 4, 4, 4, 4, 4, 4, 4,4, 4, 4, 4,4, 4, 4, 4,
9       5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5,
10       6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,
11       6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,
12       7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
13       7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
14       7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
15       7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
16    };
17    int result = 0;
18
19    if (bb > 0xFFFF)
20      {
21       bb >>= 16;
22       result += 16;
23    }
24    if (bb > 0xFF)
25      {
26       bb >>= 8;
27       result += 8;
28    }
29
30    return (result + msb_256_table[bb]);
31 }
32
33 /* ============================================ */
34 inline int bsr_in_C(uint64_t bb)
35 {
36   const uint32_t hb = bb >> 32;
37   return hb ? 32 + bsr32((uint32_t)hb) : bsr32((uint32_t)bb);
38
39 }
40

下面�q�个��g��也可以，��p�|时返�?/span>-1023�Q�至于速度快慢��p��看编译器的嗜好了�?/span>

1 // bit scan in reverse order for 64 bit integral number
2 /* ============================================ */
3 inline int bsr_double (uint64_t bb)
4 {
5    union
6    {
7       double d;
8       struct
9       {
10          unsigned int mantissal : 32;
11          unsigned int mantissah : 20;
12          unsigned int exponent : 11;
13          unsigned int sign : 1;
14       };
15    } ud;
16    ud.d = (double)(bb & ~(bb >> 32));
17    return ud.exponent - 1023;
18 }
19

以上uint64_t�?/span>uint32_t是新C++标准可以支持的整型，分别相当�?/span>旧的unsigned long long�?/span>unsigned long�c�d��?span style="color:blue">以上代码不是我的原创�Q�而是来自国外某位朋友�Q�我�E�微加工了一下脓到这里，版权属于原作者，如果我没有记错的话应该是GNU�?/span>

山寨�U�技 2011-08-23 12:41 发表评论

慈善的是心而不是钱

山寨�U�技 — Fri, 19 Aug 2011 06:47:00 GMT

2007�q�初�Q�刚刚卸�ȝ��联合国秘书长安南�Q�在德克萨斯州的一个庄园里举行了一场慈善晚��_��旨在为非�z��困儿童募捐。应邀参加晚宴的都是富商和�C�会名流。在晚宴��要开始的时候，一位老妇人领着一个小奛_��来到了庄园的入口处，��女孩手里捧着一个看上去很精致的��L��?/span>

守在庄园入口处的保安安东��拦住了�q�一老一��?#8220;�Ƣ迎你们�Q�请出示��h��Q�谢谢�?#8221;�?/span>

“��h��Q�对不�v�Q�我们没有接到邀��P��是她要来�Q�我陪她来的�?#8221;老妇人抚摸着��女孩的头�?/span>

“很抱歉，除了工作人员�Q�没有请柬的��Z��能进厅R�?#8221;安东��D��?/span>

“��Z��么？�q�里不是举行慈善晚宴吗？我们是来表示我们的心意的�Q�难道不可以吗？”老妇人的表情很严肃，“可爱的小露西�Q�从电视上知道了�q�里要�ؓ非洲的孩子们举行慈善�z�d��Q�她很想为那些可怜的孩子做点事，军_��把自己储��q��里所有的钱都拿出来，我可以不�q�去�Q�真的不能让她进��d��Q?#8221;

“是的�Q�这里将要�D行一场慈善晚��_��应邀参加的都是很重要的�h士，他们��ؓ非洲的孩子慷慨解囊。很高兴你们带着爱心来到�q�里�Q�但是，我想�q�场合不适合你们�q�去�?#8221;安东��D��释说�?/span>

“叔叔�Q�慈善的不是钱，是心�Q�对吗？”一直没有说话的��女孩问安东��|��她的话让安东��愣住了�?#8220;我知道收到邀��L��人有很多钱，他们会拿出很多钱�Q�我没有那么多，但这是我所有的钱啊�Q�如果我真的不能�q�去�Q�请帮我把这个带�q�去吧！”��女孩露西说完，��手中的储钱�|�递给安东��{�?/span>

安东��g��知道是接�q�是不接�Q�正在他不知所措的时候，�H�然有�h��_��“不用了，孩子�Q�你说得对，慈善的不是钱�Q�是心，你可以进去，所有有爱心的�h都可以进厅R�?#8221;说话的是一位老头�Q�他面带微笑�Q�站在小露西�w�旁。他�w��n和小露西交谈了几句，然后直��n��h��Q�拿��Z��份请柬递给安东��|��“我可以带她进��d��Q?#8221;

安东��接�q�请柬，打开一看，忙向老头敬了个礼�Q?#8220;当然可以了，沃��u·巴菲特先生�?#8221;

当天慈善晚宴的主角不是倡议者的安南�Q�不是捐�?00万美元的巴菲特，也不是捐�?00万美元的比尔·盖茨�Q�而是仅仅捐出30��元�?5��分的小露西�Q�她赢得了最多最热烈的掌声。而晚宴的主题标语也变成了�q�样一句话“慈善的不是钱�Q�是心�?#8221;�W�二天，��国各大媒体�U�L��以这句话作�ؓ标题�Q�报道了�q�次慈善晚宴。看到报道后�Q�许多普普通通的��国人纷�U��C��为非�z�那些��I�L��孩子捐赠�?/span>

山寨�U�技 2011-08-19 14:47 发表评论

B+树的C++实现

山寨�U�技 — Tue, 16 Aug 2011 07:05:00 GMT

关于B+树的实现和测试，我见�q�的最详细的C++代码在：
http://idlebox.net/2007/stx-btree/
一位�d国朋友的��C��?/div>

山寨�U�技 2011-08-16 15:05 发表评论

架构师已�?转脓)

山寨�U�技 — Tue, 01 Feb 2011 00:58:00 GMT

2006�q�的职场出奇的冷清，相比前几�q�_��历的数量和质量都大�ؓ不如�Q�很隑־�扑ֈ�三年工作�l�验以上的�h�Q�有一�?/span> 不是特别�W�，��是特别怪。就是么�Q�干得好谁没事换工作啊！Simon是一家外企��Y件公司的�ȝ��理，最�q�给�q�个问题愁坏了。项目一个接一个的接下来，人手��?/span> 来越紧张。虽�?/span>Simon是个极限�~�程的粉丝，但也不得不批准了一份又一份的加班甌��?/span>HR�l�理把这个问题归�l�到房�h上，他的妙论�?/span>“怕失业了�q�不上房 �ƾ，不敢��x��”�?/span>

�q�天�Q?/span>K��目�l�长Allen�l�于忍不住了�Q�带了一个只有一�q�工作经验的��伙子要Simon面试�Q?/span>“很聪明！�l�验��了炏V�?/span>”

Simon�׃��q��毛，��_��“你不知道�q�个职位最低要求是三年工作�l�验吗？”

Allen��_��“�q�已�l�是三个月里通过技术考试中最好的一个了�Q�老大�Q�试试吧�?/span>”Allen�?/span>Simon多年的哥们，比较随便�?/span>

抵到面子上来�Q?/span>Simon只好�?/span>Allen把小伙子带进来�?/span>

Simon的面试通常是三步曲�Q?/span>

问题一�Q�你能说说毕业后的主要工作经历吗�Q?/span>

问题二：再说说你在公司的��C��Q?/span>

问题三：你的发展目标是什么？�{�回�{�后�Q�比如说架构师，他就跟着问：惌��一下你当架构师的一天，说给我听听？

��伙子回�{�第一问题很快很清楚，一�q�工作当然没什么东�ѝ�?/span>Simon觉得��伙子挺聪明。所以在��伙子回�{�了�W�二个问题后�Q�问了一个发散性的问题�Q?/span>“你刚才说你在公司里处于中�{�水�q�I��那比你差的�h��Z��么会比你差呢�Q?/span>”

�q�个问题是个陷阱�?/span>

��伙子冒冒失失回�{�说�Q?/span>“我觉得他们每天工作是为工作而工作，工作没有责�Q感�?/span>”

Simon点点头说�Q?/span>“是吗�Q�那真是�p�糕的员工。那你刚好比�p�糕的员工好一点了�Q?/span>”

��伙子的�怸�下子�U�了�Q?/span>“我不是这个意�?/span>……”

“好了�Q�那你说说比你好的�h��Z��么比你强�Q?/span>”

“我觉得他非常努力�Q�工作很多年了还在学习各�U�构�Ӟ��水��^很高�?/span>”于是Simon��问那最后一个问题。果�Ӟ��伙子回�{�的是要成�ؓ架构师。大�?/span>70�Q�的人想成�ؓ架构师。但是架构师是什么呢�Q?/span>

Simon问道�Q?/span>“那你��Z��么要成�ؓ架构师呢�Q?/span>”

��伙子一愣，大概�q�没有�h�q�么�|�疑�q�他�?/span>“�q�纪大了�Q�不能老写�E�序吧�?/span>”�q�个回答�Q�让Simon惌��v关于他对什么是老的定义�Q�当你希望做�q�轻人做的事情时�Q�你��p��q�轻�Q�如果你希望做老年人做的事情，你就老了。这和你出生了多长时间是没有关系的�?/span>

Simon接着问：“好吧�Q�那你说说你成�ؓ架构师以后，每天都会做什么？”

��伙子说�Q?/span>“我还没想�q�，不过�Q�我惛_��该主要是需求分析，设计构架�?/span>……”�q�大概是现在�q�轻人的通病�Q�年��M�h很容易追逐一些自�׃��不清楚的目标�?/span>

Simon问：“那设计构架具体都做些什么呢�Q?/span>”

��伙子这�ơ的回答是：“比如�Q�选择�E�序框架�Q�决定用Spring�?/span>Struts�{�等�?/span>”

“哦，那我问你�Q�你怎么说服别�h是用Spring�q�是Struts呢？”

“如果我有�l�验�Q�我会知道哪个更�?/span>……”

“是吗�Q�但关于Spring�?/span>Struts的知识�Q谁都可以很容易得到。如果别��Z��同意你的��Q�你怎么说服他？如果同意你的��Q�那你不�q�是作出了和别�h一��L��认识�Q�别人又凭什么认可你呢？”

��伙子没惌��架构师日子里�q�有一个说服�h的工作，��_��“我是架构师，我应该有权力做决定吧�Q?/span>”

Simon惌��v权力的三�U�层�ơ，�W�一层，��d��Q�第二层�Q�专业；�W�三层，品�d�?/span>

Simon问：“如果在一个成熟的软�g企业里没有你所惌��的架构师呢？或者说�Q�架构师�q�种职业已经��M��或消�׃��呢？你会怎么定位你的职业�Q?/span>”

��伙子显得很震惊�?/span>

Simon��M��一个系�l�构�Ӟ��然后又给��伙子看了一�D�代码�?/span>

“那一个更难懂�Q?/span>”Simon问�?/span>

��伙子指着代码��_��“代码难懂�?/span>”

Simon的解释是�Q?/span>“�q�就是�ؓ什么实际上所谓的架构师不存在的原因。一个更��单的东西怎么会更有�h值呢�Q�每个�h都能够画��U�构架图�Q�但不是每个人都能写出好的代码�?/span>”

送走了小伙子�Q?/span>Simon有点隑֏�。他有点喜欢�q�个��伙子，但是�Q�这又是一个被愚蠢的教育和误�h子弟的技术杂志污染的家伙�?/span>Simon在自��q��W�记本中加了一句话�Q�中国程序员最愚蠢的认识之三：我想当架构师。前面两个赫然是�Q?/span>

35岁后写不动程序了�Q?/span>

我只要做Java�Q?/span>C�Q�＋�Q�；

山寨�U�技 2011-02-01 08:58 发表评论

C++/C#/F#/Java/JS/Lua/Python/Ruby渲染速度比试

山寨�U�技 — Thu, 02 Dec 2010 05:23:00 GMT

之前常听说C#和Java与C++的速度接近�Q�更有甚者说很多情况下他们都比C++快，而且丑և�一大堆的范�?多是些IO操作�Q�测量误差超�U�大�Q�因此很难��o��Z��?�Q�于是听到很多�h出来圆场�Q�说对于语言内徏�c�d��(整�Ş、��Q点型�{?�Q�编译成二进制应该相差不大，�q�似乎有些道理，但我仍然有些怀疑�?br>
�q�曾�l�听不少人鼓吹过脚本�Q�说脚本�E�序比C++�E�序慢不了多��，有�h甚至�l?0%�Q�对此我不加评论了，看看�q�里的测试结果就一目了然�?br>
下面有个��点密集型的计算�E�序�Q�没有��用blitz++和MTL�Q�很�W�合一般性应用，如果用上他们那就不好说怎么��P��因�ؓ主要是和Fortran比科学计��速度时才用。已�l�有人编码测试了�?br>只讲速度�Q�如果再比内存，其他几种语言��没有必要比下去了�?br>

不同语言版本的代码到原作者提供的地址��M��载：http://files.cnblogs.com/miloyip/smallpt20100623.zip
下面是测试用的系�l�配�|�：

��试配置

��g: Intel Core i7 920@2.67Ghz(4 core, HyperThread), 12GB RAM
操作�pȝ��: Microsoft Windows 7 64-bit

��试名称	�~�译�?/span>/解译�?/span>	�~�译/�q�行选项
VC++	Visual C++ 2008 (32-bit)	/Ox /Ob2 /Oi /Ot /GL /FD /MD /GS- /Gy /arch:SSE /fp:fast
VC++_OpenMP	Visual C++ 2008 (32-bit)	/Ox /Ob2 /Oi /Ot /GL /FD /MD /GS- /Gy /arch:SSE /fp:fast /openmp
IC++	Intel C++ Compiler (32-bit)	/Ox /Og /Ob2 /Oi /Ot /Qipo /GA /MD /GS- /Gy /arch:SSE2 /fp:fast /Zi /QxHost
IC++_OpenMP	Intel C++ Compiler (32-bit)	/Ox /Og /Ob2 /Oi /Ot /Qipo /GA /MD /GS- /Gy /arch:SSE2 /fp:fast /Zi /QxHost /Qopenmp
GCC	GCC 4.3.4 in Cygwin (32-bit)	-O3 -march=native -ffast-math
GCC_OpenMP	GCC 4.3.4 in Cygwin (32-bit)	-O3 -march=native -ffast-math -fopenmp
C++/CLI	Visual C++ 2008 (32-bit), .Net Framework 3.5	/Ox /Ob2 /Oi /Ot /GL /FD /MD /GS- /fp:fast /Zi /clr /TP
C++/CLI_OpenMP	Visual C++ 2008 (32-bit), .Net Framework 3.5	/Ox /Ob2 /Oi /Ot /GL /FD /MD /GS- /fp:fast /Zi /clr /TP /openmp
C#	Visual C# 2008 (32-bit), .Net Framework 3.5
*C#_outref	Visual C# 2008 (32-bit), .Net Framework 3.5
F#	F# 2.0 (32-bit), .Net Framework 3.5
Java	Java SE 1.6.0_17	-server
JsChrome	Chrome 5.0.375.86
JsFirefox	Firefox 3.6
LuaJIT	LuaJIT 2.0.0-beta4 (32-bit)
Lua	LuaJIT (32-bit)	-joff
Python	Python 3.1.2 (32-bit)
*IronPython	IronPython 2.6 for .Net 4
*Jython	Jython 2.5.1
Ruby	Ruby 1.9.1p378

渲染的解像度�?/span>256x256�Q�每象素�?/span>100�ơ采栗��?/span>

�l�果及分�?/span>

下表中预讄��相对旉��以最快的单线�E�测�?/span>(IC++)作基准，用鼠标按列可改变基准。由�?/span>Ruby�q�行旉��太长�Q�只每象素作4�ơ采��P��把时间乘�?/span>25。另外，因�ؓ各测试的渲染旉��相差很远�Q�所以用了两个棒形图��L��C�数据，分别昄��旉��于4000�U�和��于60�U�的��试(Ruby�?/span>4000�U�以外，不予�?/span> �C?/span>)�?/span>

Test	Time(sec)	Relative time
IC++_OpenMP	2.861	0.19x
VC++_OpenMP	3.140	0.21x
GCC_OpenMP	3.359	0.23x
C++/CLI_OpenMP	5.147	0.35x
IC++	14.761	1.00x
VC++	17.632	1.19x
GCC	19.500	1.32x
C++/CLI	27.634	1.87x
Java	30.527	2.07x
C#_outref	44.220	3.00x
F#	47.172	3.20x
C#	48.194	3.26x
JsChrome	237.880	16.12x
LuaJIT	829.777	56.21x
Lua	1,227.656	83.17x
IronPython	2,921.573	197.93x
JsFirefox	3,588.778	243.13x
Python	3,920.556	265.60x
Jython	6,211.550	420.81x
Ruby	77,859.653	5,274.69x

C++/.Net/Java�l�别

静态语�a�和动态语�a�在此��试下的性能不在同一数量�U�。先比较静态语�a��?/span>

C++�?/span>.Net的测试结果和上一��博文相若，�?/span>C#�?/span>F#无显著区别。但是，C++/CLI虽然同样产生IL�Q�于括管�?/span>.Net�q�_��上执行，其渲染时�?/span> 却只�?/span>C#/F#�?/span>55%左右。�ؓ什么呢�Q��?/span>ildasm��d��汇编C++/CLI�?/span>C#的可执行文�g后，可以发现�Q�程序的热点函数 Sphere.Intersect()在两个版本中�Q?/span>C++/CLI版本的代码大��?/span>(code size)�?/span>201字节�Q?/span> C#则�ؓ125字节�Q?/span> C++/CLI版本在编译时�Q�已把函数内所�?/span>Vec�cȝ��Ҏ��调用全部内联�Q��?/span>C#版本则��?/span>callvirt调用Vec的方法。估�?/span>JIT没有把这函数�q?/span> 行内联，做成�q�个性能差异。另外，C++/CLI版本使用了值类型，�q��用指�?/span>(代码中�ؓ引用)托管代码(C++/CLI)的渲染时��_��仅�ؓ原生非括��代�?/span>(IC++)�?/span>1.91倍，个�h觉得.Net�?/span>JIT已经非常不错�?/span>

另一斚w��Q?/span>Java的性能表现非常�H�出�Q�只�?/span>C++/CLI�E�慢一点，Java版本的渲染时间�ؓC#/F#�?/span>65%左右。以前一直认为，C#不少设计会��其性能高于Java�Q�例�?/span>C#的方法预设�ؓ非虚�Q?/span>Java则预设�ؓ虚；又例�?/span>C#支持struct作值类�?/span>(value type)�Q?/span>Java则只�?/span>class引用�c�d��(reference type)�Q�后者必��M��?/span>GC。但是，�q�个��试昄��Q?/span>Java VM应该�?/span>JIT中做了大量优化，估计也应用了内联�Q�才能��其性能��D��C++/CLI�?/span>

�U?/span>C++斚w��Q?/span>Intel C++�~�译器最快，Visual C++慢一点点(1.19x)�Q?/span>GCC再慢一点点(1.32x)。这�l�果�W�合本�h预期�?/span> Intel C++�?/span>OpenMP版本和单�U�程比较�Q�达5.16加速比(speedup)�Q�对�?/span>4�?/span>Hyper Threading来说��是不错的结果。读者若有兴��，也可以自行测�?/span>C# 4.0的�ƈ行新�Ҏ��?/span>

动态语�a��l�别

首先�Q�要说一句，Google太强了，难以惛_��JsChome的渲染时间仅�?/span>IC++�?/span>16.12倍，C#�?/span>4.94倍�?/span>

以下比较各动态语�a�的相�Ҏ��_��?/span>JsChrome为基准�?/span> Chrome�?/span>V8 JavaScript引擎(1.00x)大幅抛离Firefox�?/span>SpiderMonkey引擎(15.09x)。�?/span>LuaJIT(3.49x)�?/span>Lua(5.16x)则排�W�二和第三名�?/span> Lua�?/span>JIT版本是没�?/span>JIT�?/span>68%�Q��ƈ没有惛_��中的快，但是也比Python(16.48x)快得多。曾听说�q?/span>Ruby有效能问题，没想到问题竟然如此严�?/span>(327.31x)�Q�其渲染旉��差不多是Python�?/span>20�?

山寨�U�技 2010-12-02 13:23 发表评论

C#, C++, Java性能�Ҏ��

山寨�U�技 — Sun, 01 Mar 2009 14:22:00 GMT

摘选自Onur Gumus的博客�?br>http://reverseblade.blogspot.com/2009/02/c-versus-c-versus-java-performance.html

C# vs C++ vs Java性能�Ҏ��

C#, Java 和C++�Ҏ��性能常常是引发争议的一个话题。那些不熟悉JIT的�h常常声称JIT上运行的��M��E�序跟C++比速度都不行。而那些��用Java和C#的�h�? 常常声称性能差别甚小�Q�几乎可以忽略不计。这里有一个测试对比结果图�Q�来自freenode IRC服务器，不知道谁��试的，但是�l�果��g��很真实�?nbsp;

对于我来��_��因�ؓ性能原因选择C++��g��理由不��。记住：工程师比服务器要昂贵的多�Q?br>

部分其他人的评论�Q?br>
单线�E�测试的�Q�多�U�程如何�Q�内存耗费多少�Q�测试程序数据��用的语言内徏�c�d��(比如int)�q�是自定义类型，如果语言本��n内徏的类型那�~�译成二�q�制��试旉��消耗差别不会多大是正常的�?br>PHP没有针对Windows�pȝ��优化�Q�因此速度肯定上不厅R�?br>在Windows上测试Java是否�q�行在虚拟机上？默认情况可�ƈ非如�?跟Linux可不�?�Q�再者，��试�l�果耗费旉��太短误差相对可能很大�Q�很难说明实际情形�?br>
更多评论...

山寨�U�技 2009-03-01 22:22 发表评论

山寨�U�技 — Fri, 20 Feb 2009 06:58:00 GMT

�q�里有一�D�非常简单的代码�Q�取自网�l�，我稍加修改，贴在�q�里。用来检查CPU的生产商和品牌，以及当前工作频率�Q�如果是台式机CPU�Q�频率应该恒定，但是�U�d��版本的CPU�Q�频率不停地在变。以下代码用Visual C++�~�译�Q�因为内嵌一�Ҏ��~�，造成�U�L��性变�?例如�Q�GCC汇编跟AT&T汇编语法�c�M��Q�但是MS汇编跟Intel汇编语法�c�M��)�Q�以下代码如果希望在MinGW(GCC)下编译，需要修攚w��点内嵌的汇编�?br>
#include "stdafx.h"
#ifndef CPUID_HPP_
#define CPUID_HPP_

#include <string>
#include <cstring>
#include <iostream>
#include <windows.h>

typedef unsigned long DWORD;
typedef __int64 LONGLONG;

class CPUID
{
public:
  CPUID() : m_eax(0), m_ebx(0), m_ecx(0), m_edx(0){}
  std::string vendor(); //哪个厂商出的�Q�Intel, AMD, VIA ?
  std::string name();   //�q�是CPU的整体描�q?/span>
  LONGLONG frequency(DWORD stime = 1000);//CPU的当前工作频�?/span>

private:
  void  Executecpuid(DWORD veax);
  LONGLONG cycles() const
  {
    DWORD h, l;         //不能直接��d��Q�用变量�q�渡一�?/span>
    __asm
    {
      rdtsc             //把当前CPU周期��d��寄存�?/span>
      mov l, eax
      mov h, edx
    }
    LONGLONG high = h, low = l;
    return high << 32 | low;
  }

  DWORD  m_eax;
  DWORD  m_ebx;
  DWORD  m_ecx;
  DWORD  m_edx;
};

void CPUID::Executecpuid(DWORD veax)
{
  DWORD deax;
  DWORD debx;
  DWORD decx;
  DWORD dedx;

  __asm
  {
    mov eax, veax
    cpuid
    mov deax, eax
    mov debx, ebx
    mov decx, ecx
    mov dedx, edx
  }

  m_eax = deax;
  m_ebx = debx;
  m_ecx = decx;
  m_edx = dedx;
}

std::string CPUID::vendor()
{
  const DWORD S = sizeof(DWORD);
  char cVID[S*3+1];
  std::memset(cVID, 0, sizeof(cVID));
  Executecpuid(0);
  std::memcpy(cVID+S*0, &m_ebx, S);
  std::memcpy(cVID+S*1, &m_edx, S);
  std::memcpy(cVID+S*2, &m_ecx, S);

  return std::string(cVID);
}

std::string CPUID::name()
{
  const DWORD vendorID = 0x80000002;
  const DWORD S = sizeof(DWORD);
  char cvendor[S*4*3+1];
  std::memset(cvendor, 0, sizeof(cvendor));

  for(DWORD i = 0; i < 3; i++)
  {
    Executecpuid(vendorID + i);
    // 每次执行�l�束后，保存四个寄存器里的ascii码到数组
    std::memcpy(cvendor + i*S*4 + S*0, &m_eax, S);
    std::memcpy(cvendor + i*S*4 + S*1, &m_ebx, S);
    std::memcpy(cvendor + i*S*4 + S*2, &m_ecx, S);
    std::memcpy(cvendor + i*S*4 + S*3, &m_edx, S);
  }
  return std::string(cvendor);
}

LONGLONG CPUID::frequency(DWORD stime)
{
  HANDLE hp = GetCurrentProcess();
  HANDLE ht = GetCurrentThread();

  DWORD pc = GetPriorityClass(hp);
  DWORD tp = GetThreadPriority(ht);

  BOOL flag1 = FALSE, flag2 = FALSE;

  flag1 = SetPriorityClass(hp, REALTIME_PRIORITY_CLASS); //优先�U�设�|�最�?/span>
  flag2 = SetThreadPriority(ht, THREAD_PRIORITY_HIGHEST);//优先�U�设�|�最�?br>
  //�׃��CPU本��n旉��波动较大�Q�因此时间从�pȝ��d��Q�这��h��较��^�?/span>
//周期除以旉��是工作频率�Q�即单位旉��内的周期
  Sleep(stime);
  LARGE_INTEGER fq, st, ed;
  QueryPerformanceFrequency(&fq);
  QueryPerformanceCounter(&st);
  LONGLONG start = cycles();
  Sleep(stime);
  QueryPerformanceCounter(&ed);
  LONGLONG end = cycles();
  if(flag1)
    SetPriorityClass(hp, pc);
  if(flag2)
    SetThreadPriority(ht, tp);

  CloseHandle(hp);
  CloseHandle(ht);

  return (end - start) * fq.QuadPart / (ed.QuadPart - st.QuadPart);
}

#endif

#include <cstdio>
#include <cstdlib>

int main()
{
  CPUID cpu;
  std::printf("Vendor: %s\n", cpu.vendor().c_str());
  std::printf("Full name: %s\n", cpu.name().c_str());
  const unsigned TIMES = 3;
  std::printf("Frequency(testing %d times): ", TIMES);
  const double UNIT = 1000.0;
  for(unsigned i = 0; i < TIMES; ++i)
  {
    double fre = cpu.frequency() / UNIT; // 假设CPU主频高于1000Hz
    if(fre > UNIT)
      fre /= UNIT;
    std::printf("%f%s ", fre, (fre > 10.0 ? "MHz" : "GHz"));
  }
  printf("\n");
  std::system("PAUSE");
  return 0;
}

山寨�U�技 2009-02-20 14:58 发表评论

山寨�U�技 — Mon, 19 Jan 2009 10:58:00 GMT

摘要: v\:* {behavior:url(#default#VML);} o\:* {behavior:url(#default#VML);} w\:* {behavior:url(#default#VML);} .shape {behavior:url(#default#VML);} Normal 0 7.8 pt 0 2 false false false... 阅读全文

山寨�U�技 2009-01-19 18:58 发表评论

B树、B-树、B+树、B*树都是什�?转蝲)

山寨�U�技 — Mon, 05 Jan 2009 15:45:00 GMT

B�?/span>

即二叉搜索树�Q?/span>

1.所有非叶子�l�点臛_��拥有两个儿子�Q?/span>Left�?/span>Right�Q�；

2.所有结点存储一个关键字�Q?/span>

3.非叶子结点的左指针指向小于其关键字的子树�Q�右指针指向大于其关键字的子树；

如：

B树的搜烦�Q�从根结点开始，如果查询的关键字与结点的关键字相�{�，那么��命中；否则�Q�如果查询关键字比结点关键字��，��p��入左儿子�Q�如果比�l�点关键字大�Q�就�q�入叛_��子；如果左儿子或叛_��子的指针为空�Q�则报告找不到相应的关键字；

如果B树的所有非叶子�l�点的左叛_��树的�l�点数目均保持差不多�Q��^衡）�Q�那�?/span>B树的搜烦性能��D��二分查找�Q�但它比�q�箋内存�I�间的二分查扄��优点是，改变B树结构（插入与删除结点）不需要移动大�D늚�内存数据�Q�甚至通常是常数开销�Q?/span>

�?/span>B树在�l�过多次插入与删除后�Q�有可能��D��不同的结构：

双��也是一�?/span>B树，但它的搜索性能已经是线性的了；同样的关键字集合有可能导致不同的树结构烦引；所以，使用B树还要考虑��可能让B树保持左囄��l�构�Q�和避免叛_��的结构，也就是所谓的“�q��”问题�Q?/span>

实际使用�?/span>B树都是在�?/span>B树的基础上加上��^衡算法，�?#8220;�q��二叉�?#8221;�Q�如何保�?/span>B树结点分布均匀的��^衡算法是�q��二叉树的关键�Q��^衡算法是一�U�在B树中插入和删除结点的�{�略�Q?/span>

B-�?/span>

是一�U�多路搜索树�Q��ƈ不是二叉的）�Q?/span>

1.定义��L��非叶子结�Ҏ��多只�?/span>M个儿子；�?/span>M>2�Q?/span>

2.根结点的儿子��Cؓ[2, M]�Q?/span>

3.除根�l�点以外的非叶子�l�点的儿子数�?/span>[M/2, M]�Q?/span>

4.每个�l�点存放臛_��M/2-1�Q�取上整�Q�和臛_��M-1个关键字�Q�（臛_��2个关键字�Q?/span>

5.非叶子结点的关键字个�?/span>=指向儿子的指针个�?/span>-1�Q?/span>

6.非叶子结点的关键字：K[1], K[2], …, K[M-1]�Q�且K[i] < K[i+1]�Q?/span>

7.非叶子结点的指针�Q?/span>P[1], P[2], …, P[M]�Q�其�?/span>P[1]指向关键字小�?/span>K[1]的子树，P[M]指向关键字大�?/span>K[M-1]的子树，其它P[i]指向关键字属�?/span>(K[i-1], K[i])的子树；

8.所有叶子结点位于同一层；

如：�Q?/span>M=3�Q?/span>

B-树的搜烦�Q�从根结点开始，对结点内的关键字�Q�有序）序列�q�行二分查找�Q�如果命中则�l�束�Q�否则进入查询关键字所属范围的儿子�l�点�Q�重复，直到所对应的儿子指针�ؓ�I�，或已�l�是叶子�l�点�Q?/span>

B-树的�Ҏ��：

1.关键字集合分布在整颗树中�Q?/span>

2.��M��一个关键字出现且只出现在一个结点中�Q?/span>

3.搜烦有可能在非叶子结点结束；

4.其搜索性能�{��h于在关键字全集内做一�ơ二分查找；

5.自动层次控制�Q?/span>

�׃��限制了除根结点以外的非叶子结点，臛_��含有M/2个儿子，��保了结点的臛_��利用率，其最底搜索性能为：

其中�Q?/span>M��定的非叶子结�Ҏ��多子树个敎ͼ�N为关键字��L��Q?/span>

所�?/span>B-树的性能��L��{��h于二分查找（�?/span>M值无养I��Q�也��没�?/span>B树��^衡的问题�Q?/span>

�׃��M/2的限�Ӟ��在插入结�Ҏ��Q�如果结点已满，需要将�l�点分裂��Z��个各�?/span>M/2的结点；删除�l�点�Ӟ��需��两个不��?/span>M/2的兄弟结点合�qӞ��

B+�?/span>

B+树是B-树的变体�Q�也是一�U�多路搜索树�Q?/span>

1.其定义基本与B-树同�Q�除了：

2.非叶子结点的子树指针与关键字个数相同�Q?/span>

3.非叶子结点的子树指针P[i]�Q�指向关键字值属�?/span>[K[i], K[i+1])的子树（B-树是开区间�Q�；

5.为所有叶子结点增加一个链指针�Q?/span>

6.所有关键字都在叶子�l�点出现�Q?/span>

如：�Q?/span>M=3�Q?/span>

B+的搜索与B-树也基本相同�Q�区别是B+树只有达到叶子结�Ҏ��命中�Q?/span>B-树可以在非叶子结点命中）�Q�其性能也等价于在关键字全集做一�ơ二分查找；

B+的特性：

1.所有关键字都出现在叶子�l�点的链表中�Q�稠密烦引）�Q�且链表中的关键字恰好是有序的；

2.不可能在非叶子结点命中；

3.非叶子结点相当于是叶子结点的索引�Q�稀疏烦引）�Q�叶子结点相当于是存储（关键字）数据的数据层�Q?/span>

4.更适合文�g索引�pȝ��Q?/span>

B*�?/span>

�?/span>B+树的变体�Q�在B+树的非根和非叶子�l�点再增加指向兄弟的指针�Q?/span>

B*树定义了非叶子结点关键字个数臛_��?/span>(2/3)*M�Q�即块的最低��用率�?/span>2/3�Q�代�?/span>B+树的1/2�Q�；

B+树的分裂�Q�当一个结�Ҏ��Ӟ��分配一个新的结点，�q�将原结点中1/2的数据复制到新结点，最后在父结点中增加新结点的指针�Q?/span>B+树的分裂只媄响原�l�点和父�l�点�Q�而不会媄响兄弟结点，所以它不需要指向兄弟的指针�Q?/span>

B*树的分裂�Q�当一个结�Ҏ��Ӟ��如果它的下一个兄弟结�Ҏ��满，那么��一部分数据�U�d��兄弟�l�点中，再在原结�Ҏ��入关键字�Q�最后修改父�l�点中兄弟结点的关键字（因�ؓ兄弟�l�点的关键字范围改变了）�Q�如果兄弟也满了�Q�则在原�l�点与兄弟结点之间增加新�l�点�Q��ƈ各复�?/span>1/3的数据到新结点，最后在父结点增加新�l�点的指针；

所以，B*树分配新�l�点的概率比B+树要低，�I�间使用率更高；

��结

B树：二叉树，每个�l�点只存储一个关键字�Q�等于则命中�Q�小于走左结点，大于走右�l�点�Q?/span>

B-树：多�\搜烦树，每个�l�点存储M/2�?/span>M个关键字�Q�非叶子�l�点存储指向关键字范围的子结点；

所有关键字在整颗树中出玎ͼ�且只出现一�ơ，非叶子结点可以命中；

B+树：�?/span>B-树基��上，为叶子结点增加链表指针，所有关键字都在叶子�l�点中出玎ͼ�非叶子结点作为叶子结点的索引�Q?/span>B+树��L��到叶子结�Ҏ��命中�Q?/span>

B*树：�?/span>B+树基��上，为非叶子�l�点也增加链表指针，��结点的最低利用率�?/span>1/2提高�?/span>2/3�?/span>

转自: http://blog.csdn.net/manesking/archive/2007/02/09/1505979.aspx

山寨�U�技 2009-01-05 23:45 发表评论

激情欧美日韩一区,中文av字幕一区,亚洲成色777777女色窝

What does “stride�?mean in image processing? [fwd]

raster and vector image [fwd]

Base64

PDF里的BOX

�~�译器预定义的宏

源自�Q�https://sourceforge.net/p/predef/wiki/Architectures/

Architectures

Example

Example

Example

Example

Example

Example

Example

Example

Example

Pyramid 9810

Example

Example

TMS470

When someone gives you a bug [FWD]

using gcc_x86 assembly to scan the most significant one bit and the least significant one bit

Is there an ISO standard for C sharp language?

Latest Standards

Current Working Draft

Annotated Standards

Microsoft Implementation Specific Versions

The Ecma 4th and ISO 2nd Editions

CLI

C#

Previous Editions Background

Joining Ecma

Acknowledgements

一个母亲一生的8个谎�a�(�?

查找一�?4位整��C���q�制�W�一�?

慈善的是心而不是钱

B+树的C++实现

架构师已�?转脓)

C++/C#/F#/Java/JS/Lua/Python/Ruby渲染速度比试

���试配置

�l�果及分�?/span>

C++/.Net/Java�l�别

动态语�a��l�别

C#, C++, Java性能�Ҏ��

B树、B-树、B+树、B*树都是什�?转蝲)

The Ecma 4^th and ISO 2^nd Editions

查找一�?4位整��C��q�制�W�一�?

��试配置