No Teraflops Will Compensate Larrabee Functionality – Game Developer

Tim Sweeney Disappointed with Larrabee Delay

by Anton Shilov
12/07/2009 | 10:37 PM

Tim Sweeney, the head of Epic Games and one of the most influential game developers in the world, said that he disappointed with the cancellation of the first-generation Larrabee graphics processing unit (GPU) by Intel Corp. and the delay of the whole Intel Larrabee GPU project.

 

“To focus on Teraflops misses a larger point about programmability: today's GPU programming models are too limited to support large-scale software, such as a complete physics engine, or a next-generation graphics pipeline implemented in software.  No quantity of Teraflopscan compensate for a lack of support for dynamic dispatch, a full C++programming model, a coherent memory space, etc,” said Tim Sweeney, reports Bright Side of News web-site.

Mr. Sweeney has been advocating various software rendering techniques, claiming that fixed-function GPUs will not be efficient going forward and that Epic Software will concentrate on developing rendering engine for general-purpose, not graphics specific, hardware. Therefore, it is not surprising that he is disappointed with the cancellation of the first-generation Larrabee graphics processing unit.

It is not a secret that peak performance of hardware is increasing very fast these days. However, in order to utilize the additional performance and feature-set, developers of software, namely video games, have to increase development time and budgets. Although the market of video games has been growing rather rapidly in the recent years, game budgets have not been increasing that rapidly. Mr. Sweeney claims that although performance increase may be 20 times, game budgets will only increase less than two times. As a result, it is vital for game developers to take advantage of the modern hardware capabilities and performance at the lowest possible cost. The head of Epic Games claimed earlier this year that using general purpose computing on graphics processing unit (GPGPU) technologies was not efficient from the cost standpoint and implied that flexibility of Larrabee would be more competent.

“I see the instruction set and mixed scalar/vector programming model of Larrabee as the ultimate computing model, delivering GPU-class numeric computing performance and CPU-class programmability with an easy-to-use programming model that will ultimately crush fixed-function graphics pipelines. The model will be revolutionary whether it's sold as an add-in card, an integrated graphics solution, or part of the CPU die,” concluded Mr. Sweeney.