At the last Intel Developer Forum (IDF) during Intel's big product and promotion gala, Intel did a brief demo of Larrabee, but did little in terms of discussing the product or answering questions other than indicating that it would be introduced in mid-2010.
Intel has now indicated that it has scrapped plans to introduce Larrabee, the first versions of the product family, but that it will be continuing development efforts with subsequent products. And, Intel will still make the current silicon available as a graphics software development vehicle (SDV), and will also promote and support Larrabee as a SDV for a future co-processor/accelerator for high-performance computing (HPC) applications. No further details were provided about future development plans or product introductions.
The big question is: what happened? First let's consider what Intel was trying to do. They were trying to create a graphics architecture that was programmable just like a standard x86 processor. This not only required a new hardware architecture, but also a new programming model. Developing a new hardware architecture is difficult enough, but developing a new programming model is exponentially more challenging because it requires support from the rest of the industry. History in the electronics industry indicates that few new technologies meet their initial schedules and adoption of new technologies and methodologies takes two to three times longer than anticipated. In addition, Intel set a very high bar for the product. As one Intel executive indicated in 2008 "If it does not set performance records, it will not be introduced." After the Teraflop performance demonstrated at the SC09 conference just a few weeks ago, it seemed like Intel was on track to meet its commitment for the product, but it appears that the ecosystem was not ready for the solution at the given time. As a result, Intel had to make a choice to continue moving forward because they face very stiff competition in graphics from AMD and NVIDIA. Intel will provide an update on their product plans some time in 2010, possibly in the third quarter at IDF.
Intel has indicated that the setback will not have any financial or personnel impact on its investment in developing graphics technology. With Intel's growing focus on consumer electronics, the company can not consider ending graphics development, because graphics is a critical technology to all the major consumer and computing platforms. Intel currently has three graphics solutions and development efforts: The "gen" architecture for graphics integrated into core-logic chipsets and the up coming Westmere processor generation, PowerVR (licensed from Imagination) which is integrated with x86 cores for handsets and consumer electronics, and Larrabee for discrete graphics processing units (GPUs). The postponement, delay, or cancelling of Larrabee is only part of the Intel graphics strategy. However, Larrabee represents a significant investment in expanding the parallel processing capabilities of the x86 architecture, which Intel Labs is also looking to advance through the development of its own 48 x86-core test chip code-named Rock Creek. If nothing else, it would appear that Intel is also highly committed to parallel processing solutions for the future.
Despite the efforts, however, this is still a major blow to Intel as the company looked to distance itself from both NVIDIA and AMD with whom it shares partner-competitor (or love-hate) relationships. The plan was to enter the discrete GPU market on the high-end and migrate the technology down to the integrated graphics solutions. Now, Intel will have to rely on the older graphics architecture for integrated solutions while it regroups and re-enforces the Larrabee development efforts. This does put Intel at a disadvantage in term of graphics technology, but with at least a twelve-month lead on rival AMD on introducing processors with integrated graphics, the upcoming Westmere processor generation should still provide Intel with a price-, power-, and performance-competitive offering for the value and mainstream PC segments. The strategy shift will, however, provide AMD and NVIDIA more breathing room in the higher margin discrete GPU space.
The change may also give Intel an opportunity to enhance its expertise in parallel computing and possibly reconsider its decision to not support OpenCL or Direct Compute for using the graphics cores for processing computation instructions, also called GPGPU, which is supported in the new operating systems from Apple (Snow Leopard) and Microsoft (Windows 7).
Research firm
In-Stat does not believe Intel will give up on developing graphics technology because this function is too critical as the world moves to heterogeneous computing solutions, something that In-Stat indicated in 2005 that AMD and Intel would both have to do by the end of the decade. Nor the company believes that this will open the door to an NVIDIA acquisition by Intel. AMD, Intel, and NVIDIA are all companies with significant expertise that are advancing the direction of computing in different areas, and in some cases, different directions. In addition, they face a growing threat of competition as the market changes and other technologies advance. With the rate of innovation increasing, this battle is far from over.