by Anton Shilov
11/03/2004 | 03:58 AM
ATI Technologies’ next-generation high-end visual processing unit code-named R520 has been taped out, reports Beyond3D web-site citing analysts from Goldman Sachs research firm who had presumably been at ATI’s Analyst Day last week. The news confirms ATI is on-track to launch its new high-end product in the first half of 2005.
A high-ranking source close to ATI Technologies confirmed X-bit labs that the chip had been taped out, which means that the completed design of the processor has been sent to an ATI’s foundry partner. The source said the chip would be made using 90nm process technology, presumably at TSMC, and will be launched commercially in the first half of next year.
Not much is known about the architecture and capabilities of the code-named R520 product that was initially referred as the R500. What is clear at the moment is that the new graphics chip will sport Shader Model 3.0 – pixel shaders 3.0 and vertex shaders 3.0 – bringing additional programming capabilities to ATI’s graphics processors as well as some other innovations. ATI’s architecture will not resemble that of the previous generation products and NVIDIA’s GeForce 6 architecture known as NV4x, particularly ATI will implement efficient flow-control, a crucial feature for pixel shaders 3.0, that will not bring speed penalty it does on existing SM3.0 hardware, according to sources.
Some sources claim that the R500 is a code-name of ATI’s graphics processor that will be submitted for Microsoft’s next Xbox console. The shader core of the R500 was reported to have 48 Arithmetic Logic Units (ALUs) that can execute 64 simultaneous threads on groups of 64 vertices or pixels. ALUs are automatically and dynamically assigned to either pixel or vertex processing depending on load. The ALUs can each perform one vector and one scalar operation per clock cycle, for a total of 96 shader operations per clock cycle. Texture loads can be done in parallel to ALU operations. At peak performance, the GPU can issue 48 billion shader operations per second, it was said.
Earlier this year sources close to ATI said the company had been developing GDDR4, a new standard for graphics. The GDDR4 memory builds upon the GDDR3 standard, just like the latter evolved from the GDDR2 specification, therefore, it is possible to expect the technology to utilize the point-to-point nature. There are no revolutions, it was said, but special tweaks to bolster clock-speeds of DRAMs used on graphics cards brought by GDDR4. The initial goals for the GDDR4 were to complete the process of standardization by the end of 2004 and push up the frequencies towards the 1.40GHz (2.80GHz effective) level. Lower clock-speeds, e.g. 1.00GHz (2.00GHz effective) are achievable by the GDDR3 technology, according to Samsung Electronics, who plans to debut such memory by the end of the year.
Usually it takes graphics companies from 90 to 120 days to start commercial production of graphics processors from the initial tape-out. In case the first silicon of a 0.13 micron chips did not work correctly, it might take up to 14 weeks to tape out another one with the issue addressed. Typically with thinner manufacturing technologies it takes lower time to address certain design flaws, a process that is known as re-spin.
ATI’s previous-generation RADEON X800-series of graphics processors was taped out in late December, 2003, and was launched formally in May, 2004.
Officials from ATI Technologies did not comment for the news-story.