Remote desktop protocol compression acceleration using single instruction, multiple dispatch instructions

A method for compressing a source image on a processor that supports vector processing, comprising:

• receiving a source image comprising pixels, each pixel comprising red-green-blue-alpha (RGBA) image data;
• planarizing the image into its R, G and B components by sending a series of processor instructions to the processor, each processor instruction operating on data plurality of pixels simultaneously;
• converting the planarized image into luminance-orange-green (YCoCg) image data by sending a series of processor instructions to the processor, each processor instruction operating on data plurality of pixels simultaneously;
• compressing the Co and Co components of the YCoCg image by sending a series of processor instructions to the processor, each processor instruction operating on data plurality of pixels simultaneously;
• and
• storing the compressed image data as an output image.

A computer-readable storage medium, comprising computer-readable instructions that when executed upon a single-instruction multiple dispatch (SIMD) processor, the processor being capable of operating upon 128 bits simultaneously, perform operations comprising:

• 16-bit aligning each channel of the source image, the source image comprising at least one 32-bit pixel, each pixel comprising an 8-bit red (R) value, an 8-bit green (G) value, an 8-bit blue (B) value, and an 8-bit alpha (A) value;
• planarizing the source image to produce a R vector, a G vector and a B vector, the R vector comprising the R value of each pixel, the G vector comprising the G value of each pixel and the B vector comprising the B value of each pixel;
• unpacking each of the R vector, the G vector, and the B vector, such that eight zeros precede each 8-bit value in each vector;
• setting a Co vector to equal the R vector minus the B vector through a series of processor instructions that operate on 128-bits of each vector simultaneously;
• setting a Y vector to equal the sum of the B vector Co vector arithmetically bitwise shifted once to the right through a series of processor instructions that operate on 128-bits of each vector simultaneously;
• setting a Cg vector to equal the Y vector minus the G vector through a series of processor instructions that operate on 128-bits of each vector simultaneously;
• setting the Y vector to equal the Y vector added to the Cg vector arithmetically bitwise shifted once to the right through a series of processor instructions that operate on 128-bits of each vector simultaneously;
• setting the Co vector to equal the Co vector arithmetically bitwise shifted to the right by a color loss level and then added to a chroma mask through a series of processor instructions that operate on 128-bits of each vector simultaneously;
• setting the Cg vector to equal the Cg vector bitwise shifted to the right by a color loss level and then added to a chroma mask;
• through a series of processor instructions that operate on 128-bits of each vector simultaneously
• run-length encoding the Y vector through a series of processor instructions that operate on 128-bits of each vector simultaneously;
• run-length encoding the Co vector through a series of processor instructions that operate on 128-bits of each vector simultaneously;
• run-length encoding the Cg vector through a series of processor instructions that operate on 128-bits of each vector simultaneously;
• and
• storing an output image, wherein the output image comprises the Y vector, the Co vector and the Cg vector, the output image is in luminance-orange-green (YCoCg) colorspace, and the output image corresponds to the source image.