rebuilt html

docs/features.html

@@ -53,14 +53,15 @@
 </li>
 <li><a class="reference internal" href="#highlighted-features" id="id7">highlighted features</a><ul>
 <li><a class="reference internal" href="#disk-caching" id="id8">disk caching</a></li>
-<li><a class="reference internal" href="#network-buffers" id="id9">network buffers</a></li>
-<li><a class="reference internal" href="#piece-picker" id="id10">piece picker</a></li>
-<li><a class="reference internal" href="#merkle-hash-tree-torrents" id="id11">merkle hash tree torrents</a></li>
-<li><a class="reference internal" href="#customizable-file-storage" id="id12">customizable file storage</a></li>
-<li><a class="reference internal" href="#easy-to-use-api" id="id13">easy to use API</a></li>
+<li><a class="reference internal" href="#high-performance-disk-subsystem" id="id9">high performance disk subsystem</a></li>
+<li><a class="reference internal" href="#network-buffers" id="id10">network buffers</a></li>
+<li><a class="reference internal" href="#piece-picker" id="id11">piece picker</a></li>
+<li><a class="reference internal" href="#merkle-hash-tree-torrents" id="id12">merkle hash tree torrents</a></li>
+<li><a class="reference internal" href="#customizable-file-storage" id="id13">customizable file storage</a></li>
+<li><a class="reference internal" href="#easy-to-use-api" id="id14">easy to use API</a></li>
 </ul>
 </li>
-<li><a class="reference internal" href="#portability" id="id14">portability</a></li>
+<li><a class="reference internal" href="#portability" id="id15">portability</a></li>
 </ul>
 </div>
 <div class="section" id="introduction">
@@ -133,9 +134,6 @@ download speed is high enough from that particular peer.</li>
 <li>supports gzipped tracker-responses</li>
 <li>can limit the upload and download bandwidth usage and the maximum number of
 unchoked peers</li>
-<li>implements fair trade. User settable trade-ratio, must at least be 1:1,
-but one can choose to trade 1 for 2 or any other ratio that isn't unfair
-to the other party.</li>
 <li>possibility to limit the number of connections.</li>
 <li>delays have messages if there's no other outgoing traffic to the peer, and
 doesn't send have messages to peers that already has the piece. This saves
@@ -143,8 +141,12 @@ bandwidth.</li>
 <li>selective downloading. The ability to select which parts of a torrent you
 want to download.</li>
 <li>ip filter to disallow ip addresses and ip ranges from connecting and
-being connected</li>
+being connected.</li>
 <li>NAT-PMP and UPnP support (automatic port mapping on routers that supports it)</li>
+<li>implements automatic upload slots, to optimize download rate without spreading
+upload capacity too thin. The number of upload slots is adjusted based on the
+peers' download capacity to work even for connections that are orders of
+magnitude faster than others.</li>
 </ul>
 </div>
 </div>
@@ -194,6 +196,25 @@ better with piece sizes.</p>
 algorithm. It clearly shows an increased utilization, which means higher read hit rates
 or smaller caches with maintained hit rate.</p>
 </div>
+<div class="section" id="high-performance-disk-subsystem">
+<h2>high performance disk subsystem</h2>
+<p>In some circumstances, the disk cache may not suffice to provide maximum performance.
+One such example is high performance seeding, to a large number of peers, over a fast
+up-link. In such a case, the amount of RAM may simply not be enough to cache disk
+reads. When there's not enough RAM to cache disk reads, the disk throughput  would
+typically degrade to perform as poorly as with no cache at all, with the majority
+of the time spent waiting for the disk head to seek.</p>
+<p>To solve this problem, libtorrent sorts read requests by their physical offset on the
+disk. They are processed by having the disk read head sweep back and forth over the drive.</p>
+<p>This makes libtorrent very suitable for large scale, high-throughput seeding.</p>
+<img alt="disk_access_no_elevator.png" src="disk_access_no_elevator.png" style="width: 49%;" />
+<img alt="disk_access_elevator.png" src="disk_access_elevator.png" style="width: 49%;" />
+<p>These plots illustrates the physical disk offset for reads over time. The left plot
+is of a run where disk operation re-ordering is turned off and the righ is when it's
+turned on. The right one has a relatively smooth sine wave shape whereas the left
+one is more random and involves much longer seeks back and forth over the disk.</p>
+<p>True physical disk offset queries are only supported on newer linux kernels and Mac OS X.</p>
+</div>
 <div class="section" id="network-buffers">
 <h2>network buffers</h2>
 <p>On CPUs with small L2 caches, copying memory can be expensive operations. It is important