SMCANI to AVX512 talk

Author: tomforsyth1000

larrabee/LRBNI origins v4 full fat.pdf

@@ -44,6 +44,10 @@ <h2>Stanford EE820 course 2010</h2>
 <h2>International Supercomputing Conference 2010</h2>
 <p><a href="http://download.intel.com/pressroom/archive/reference/ISC_2010_Skaugen_keynote.pdf">Petascale to Exascale</a>: starting from slide 32 there's Pradeep Dubey's presentation on the current version of Larrabee codenamed "Knight's Ferry", including details on core counts (32) and frequency (1.2GHz)</p>
 
+<h2><a href="https://www.handmade-seattle.com/">Handmade Seattle 2019</a></h2>
+<a href="LRBNI origins v4 full fat.pdf">SMCANI to AVX512 - the life cycle of an instruction set</a>. How we went about creating the Larrabee instruction set and its path from a research project through to mainstream use as the AVX512 instruction set.<br>
+
+
 <h2>Misc</h2>
 
 <p><a href="http://en.wikipedia.org/wiki/Larrabee_%28microarchitecture%29">Larrabee at Wikipedia</a></p>

larrabee/larrabee.html

@@ -2,17 +2,19 @@
 <h1>TomF's talks and papers.</h1>
 In rough chronological order, newest first.<br>
 <br>
+<a href="LRBNI origins v4 full fat.pdf">SMCANI to AVX512 - the life cycle of an instruction set</a> presented at <a href="https://www.handmade-seattle.com/">Handmade Seattle 2019</a> conference. I talk about how we went about creating the 512-bit SIMD extensions to the 80x86 instruction set - starting from its inception as a research project, going through the <a href="../larrabee/larrabee.html">Larrabee</a> project, and into mainstream use as the AVX512 instruction set. This is a longer version than presented, with some details I had to skip in the time available.<br>
+<br>
 <a href="FDG2015_Oculus_Tom_Forsyth_v3_final.pdf">Virtual Reality as a Medium: Changing Mechanics</a> presented as a keynote talk to the <a href="http://www.fdg2015.org/">Foundations of Digital Games 2015</a> conference. This was my thoughts on how VR as a medium differs from existing 2D-screen-based games in the ways it can communicate with the player. So very few low-level details on coding, hardware or graphics (there's other presentations for that) but more of the higher-level mechanics and the channels of input and output available to designers.<br>
 <br>
-<a href="Software_Engineering_and_VR_TomForsyth_Stanford_v6_final.pdf">Software Engineering and VR</a> presented to Stanford CS ungrads in January 2015 - sorry, the talk wasn't recorded so no video or audio. There's a little bit about how VR works, but most of it is about development methodologies, so probably quite dry if you're not a full-time coder. Please note that much of it is my personal opinion, and by no means represents a company-wide consensus (if indeed such a thing can exist in any group of coders where N>1)<br>
+<a href="Software_Engineering_and_VR_TomForsyth_Stanford_v6_final.pdf">Software Engineering and VR</a> presented to Stanford CS undergrads in January 2015 - sorry, the talk wasn't recorded so no video or audio. There's a little bit about how VR works, but most of it is about development methodologies, so probably quite dry if you're not a full-time coder. Please note that much of it is my personal opinion, and by no means represents a company-wide consensus (if indeed such a thing can exist in any group of coders where N>1)<br>
 <br>
 <a href="http://www.gdcvault.com/play/1020714">Developing VR Experiences with the Oculus Rift</a> at GDC2014, and slides in <a href="GDC2014_Developing_Virtual_Reality_Games_and_Experiences__Tom_Forsyth_final_with_notes.pptx">PowerPoint</a> or <a href="GDC2014_Developing_Virtual_Reality_Games_and_Experiences__Tom_Forsyth_final_with_notes.pdf">PDF</a> format. Slightly tweaked <a href="https://www.youtube.com/watch?v=addUnJpjjv4">version given at Oculus Connect 2014</a>, with <a href="http://static.oculus.com/connect/slides/OculusConnect_Developing_VR_Experiences_with_the_Oculus_Rift.pdf">PDF slides</a>. This talk is not so much about how to use the Rift SDK, but more about the things games need to change for VR to maintain presence, be kind to the users, and let them be immersed and have a good time.<br>
 <br>
-<a href="http://www.stanford.edu/class/ee380/ay0910.html">The Challenge of Larabee as a GPU</a>, given at Stanford 2010. The first half is a very brief intro to programming for Larrabee - it was a nice high-level audience, so I could skim a bunch of stuff. The second half was about how we persuaded a mostly general-purpose CPU to "emulate" a GPU at full speed. Click on the icon to the right of the talk title for video. Some great questions afterwards - a boisterous and very knowledgeable audience. <a href="../larrabee/Standford%20Forsyth%20Larrabee%202010.zip">Locally-hosted slides are here.</a><br>
+<a href="http://www.stanford.edu/class/ee380/ay0910.html">The Challenge of Larrabee as a GPU</a>, given at Stanford 2010. The first half is a very brief intro to programming for <a href="../larrabee/larrabee.html">Larrabee</a> - it was a nice high-level audience, so I could skim a bunch of stuff. The second half was about how we persuaded a mostly general-purpose CPU to "emulate" a GPU at full speed. Click on the icon to the right of the talk title for video. Some great questions afterwards - a boisterous and very knowledgeable audience. <a href="../larrabee/Standford%20Forsyth%20Larrabee%202010.zip">Locally-hosted slides are here.</a><br>
 <br>
 <a href="../larrabee/Larrabee%20GDC%202009.zip">Rasterization on Larrabee and SIMD Programming With Larrabee</a>: GDC 2009, Michael Abrash and I doing our double-act. We both talk about the instruction sets, Michael talks about the hierarchical descent rasterisation algorithm, and I talk about how we do basic language structures such as conditionals and flow control with our 16-wide vector units. We were both absurdly proud to be able to finally talk about the architecture we'd worked on for so long - it's not every programmer that gets to design their own instruction set.<br>
 <br>
-<a href="../larrabee/Larrabee%20Siggraph%202008%20Forsyth.zip">Larrabee Graphics Architecture: Software is the New Hardware</a></b>: Siggraph 2008, part of the <a href="http://s08.idav.ucdavis.edu/">Beyond Programmable Shading Course</a>. The basics of how Larrabee does in software what everyone else does in hardware, and how it can be more flexible and general than a standard GPU.<br>
+<a href="../larrabee/Larrabee%20Siggraph%202008%20Forsyth.zip">Larrabee Graphics Architecture: Software is the New Hardware</a></b>: Siggraph 2008, part of the <a href="http://s08.idav.ucdavis.edu/">Beyond Programmable Shading Course</a>. The basics of how <a href="../larrabee/larrabee.html">Larrabee</a> does in software what everyone else does in hardware, and how it can be more flexible and general than a standard GPU.<br>
 <br>
 <a href="trilight/trilight.html">Trilights</a> - a very brief paper on a type of light that I've been using for a while that is a generalisation of a bunch of other light types that are common in games. Not exactly earth-shattering - it's hardly a paper at all - but people don't seem to document little tricks like this, so they never get passed along. So I thought I should. Also here are links to the <a href="trilight/trilight_vs_wraparound.xls">Excel spreadsheet</a> and the <a href="trilight/trilight_demo.zip">demo</a> mentioned in the paper.<br>
 <br>