[Mesa-dev] Thank you, Roland Scheidegger!

[Dave Witbrodt]

I updated Mesa last night from my previous version (from git master on 
May 31, commit 869c5d43) to the latest HEAD (commit 213c207b of Jun 8). 
  The result was a huge boost to the performance of my system, based on 
a Radeon HD 5750.

After perusing the git log I found nothing directly related to my 
hardware, but when I filtered out the changes which were obviously 
unrelated to the speedup I found all of this:

     commit 458a9a0f85b93e3520c33dd3b502e26a31f08e4b
     Author: Roland Scheidegger <sroland at vmware.com>
     Date:   Sat Jun 1 02:17:09 2013 +0200

	gallivm: fix out-of-bounds access with mirror_clamp_to_edge
         address mode

     commit 6b53e2b0380ce1617402217cb77d5670a44c3fb5
     Author: Roland Scheidegger <sroland at vmware.com>
     Date:   Sat Jun 1 20:02:17 2013 +0200

	gallium: add support for layered rendering

     commit 44993c1808dc10da36cac156408a10e05b5dc140
     Author: Roland Scheidegger <sroland at vmware.com>
     Date:   Wed Jun 5 00:11:45 2013 +0200

	gallivm: enhance special sse2 4x4f and 2x8f -> 1x16ub conversion

     commit ef3e887084fbb36b9041faafc18fb75f6cfa7501
     Author: Roland Scheidegger <sroland at vmware.com>
     Date:   Mon Jun 3 16:01:05 2013 +0200

	llvmpipe: cleanup of generate_unswizzled_blend

     commit ffe2a1ca3c097661dd3f6e3ca5cfd72be184426c
     Author: Roland Scheidegger <sroland at vmware.com>
     Date:   Wed Jun 5 00:17:22 2013 +0200

	llvmpipe: reduce alignment requirement for 1d resources from
         4x4 to 4x1

     commit 008fd036007a9e71f669ca49c87767fd19fd77e1
     Author: Roland Scheidegger <sroland at vmware.com>
     Date:   Tue Jun 4 03:20:55 2013 +0200

	llvmpipe: improve alignment calculation for fetching/storing
         pixels

     commit d0518c4c69b32815cbf6e850d47190bb57d25a0c
     Author: Roland Scheidegger <sroland at vmware.com>
     Date:   Thu Jun 6 02:43:51 2013 +0200

	llvmpipe: bump 3d and cube map limits to 2048 and 8192
         respectively

     commit d8146f240e628e70d4c07f7e805a179f70c36e23
     Author: Roland Scheidegger <sroland at vmware.com>
     Date:   Fri Jun 7 21:03:40 2013 +0200

	llvmpipe: add support for layered rendering

     commit b47d13f4255b9708ff52fa80c11fabe53540cdea
     Author: Roland Scheidegger <sroland at vmware.com>
     Date:   Fri Jun 7 21:10:59 2013 +0200

	util: fix util_clear_render_target and util_clear_depth_stencil
         layer handling

     commit 213c207b3ac40ae769afe01b8578f566b5e7840d
     Author: Roland Scheidegger <sroland at vmware.com>
     Date:   Fri Jun 7 21:20:01 2013 +0200

	gallivm: work around slow code generated for interleaving
         128bit vectors

Wow!  One or more of those changesets really is making a difference on 
this system.

I just want to thank everyone working on Mesa -- especially the devs 
working on Radeon -- and in particular I would like to thank Roland for 
this recent work.


Thanks,
Dave W.