RWT: Memory Bandwidth and GPU Performance

Tuesday 26th April 2011, 03:11:00 PM, written by Alex G

Our friends at RealWorldTech have put out a new article taking a look at memory bandwidth and its relevance to GPU performance:

"Memory bandwidth is a critical to feeding the shader arrays in programmable GPUs. We show that memory is an integral part of a good performance model and can easily impact graphics by 40% or more. The implications are important for upcoming integrated graphics, such as AMD’s Llano and Intel’s Ivy."

It's a pretty interesting analysis with a wide variety of GPUs referenced from AMD and nVidia. Check it out!
Discuss on the forums

Tagging

± memory, bandwidth, integrated, amd, nvidia, gpu, performance


Latest Thread Comments (2 total)
Posted by iwod on Thursday, 28-Apr-11 05:58:45 UTC
No one reply? Strange. To summarize, Assuming Drivers plays the +- 5% difference in test, an actual GFLOPS output / Bandwidth Performance Numbers actually gives a VERY accurate results of the GPU performance. But it doesn't take into account the amount of memory with GFX card. With 22nm and 3D Package, Intel could literally stack 64MB of L3 Cache offering ~300GB/s of bandwidth to its iGPU. Of coz since the iGPU has low number of GFlops, it doesn't require that much bandwidth.But would these excess bandwidth help with the lower amount of memory?

Posted by Malo on Friday, 29-Apr-11 21:23:00 UTC
Well it was already being discussed herehttp://forum.beyond3d.com/showthread.php?t=60046


Add your comment in the forums

Related News

RWT explores Haswell's eDRAM for graphics
ATI shoots a Bolt through its GPU compute stack
AMD releases CodeXL 1.0
Travelling in Style: Beyond3D's C++ AMP contest
Analysis of Ivy Bridge Graphics Architecture at RWT
RWT analyzes Kepler's architecture
Nvidia 680 GTX (Kepler) Released
Microsoft Releases C++ AMP Open Specification
Nvidia's 2x Guaranteed Program
It's Dead Jim - a debate about the future of the graphics API