<html> <head> <base href="https://bugs.freedesktop.org/" /> </head> <body> <div> <a class="bz_bug_link bz_status_UNCONFIRMED " title="UNCONFIRMED --- - Ubuntu12.04.2 LinearFramebuffer under extended mode broken by intel_gen4_compute_offset_xtiled()" href="https://bugs.freedesktop.org/show_bug.cgi?id=61152#c18">Comment # 18</a> on <a class="bz_bug_link bz_status_UNCONFIRMED " title="UNCONFIRMED --- - Ubuntu12.04.2 LinearFramebuffer under extended mode broken by intel_gen4_compute_offset_xtiled()" href="https://bugs.freedesktop.org/show_bug.cgi?id=61152">bug 61152</a> from <a class="email" href="mailto:chris@chris-wilson.co.uk" title="Chris Wilson <chris@chris-wilson.co.uk>"> Chris Wilson</a> <pre>Yes, the minimum access pattern of the render pipeline is a 2x2, and if the stride is such that cross a page boundary, you incur a TLB miss on every pixel. The BLT engine doesn't have the same access pattern and so is not as adversely affected. For the operations you want to perform the render pipeline is typically much faster (as that uses the L3 cache, which even has a notable impact for simple copies). Conversely the BLT has lower latency... Forcing a LinearFramebuffer is a hack. Using a linear intermediate is still a hack, but at least a step forward.</pre> </div> <hr> You are receiving this mail because: <ul> <li>You are the QA Contact for the bug.</li> </ul> </body> </html>