DL Boost

<h2 id="features">Features</h2>
<p>DL Boost consists of two sets of features:
</p>
<ul><li><a href="/facts/AVX-512/2KpGqI8t">AVX-512 VNNI</a>, 4VNNIW, or <a href="/facts/Advanced_Vector_Extensions/znDpc6Vt">AVX-VNNI</a>: fast multiply-accumulation mainly for <a href="/facts/Convolutional_neural_network/kHEdnmGU">convolutional neural networks</a>.</li>
<li>AVX-512 BF16: lower-precision <a href="/facts/Bfloat16_floating-point_format/wo1Q6IJY">bfloat16 floating-point</a> numbers for generally faster computation. Operations provided include conversion to/from <a href="/facts/Float32/e6Fez7BM">float32</a> and <a href="/facts/Dot_product/tNz8MLjT">dot product</a>.</li></ul>
<p>DL Boost features were introduced in the <a href="/facts/Cascade_Lake/EEALp9jD">Cascade Lake</a> architecture.
</p><p>A <a href="/facts/TensorFlow/kDwfBeNd">TensorFlow</a>-based benchmark run on the <a href="/facts/Google_Cloud_Platform/lPTNk7l6">Google Cloud Platform</a> <a href="/facts/Google_Compute_Engine/tJDHgQke">Compute Engine</a> shows improved performance and reduced cost compared to previous CPUs and to GPUs, especially for small batch sizes.<a class="footnote-ref" id="fnref:2" href="#fn:2"><sup>2</sup></a>
</p>
<h2 id="notes">Notes</h2>

<h2 id="external-links">External links</h2>
<ul><li><a href="https://ai.intel.com/intel-deep-learning-boost">Deep Learning Boost</a> at Intel</li>
<li>Andres Rodrigues et al., "Lower Numerical Precision Deep Learning Inference and Training", Intel White paper <a href="https://www.intel.com/content/www/us/en/artificial-intelligence/solutions/lower-numerical-precision-deep-learning-inference-and-training.html">[3]</a></li>
<li><a href="https://indico.cern.ch/event/595059/contributions/2499304/attachments/1430242/2196659/Intel_and_ML_Talk_HansPabst.pdf">Intel and ML</a> (2017), from Intel's Developer Relations Division</li></ul>

<h2 id="references">References</h2>

<ol>
<li id="fn:1"><p>"Intel Deep Learning Boost" Product Overview [1], p. 3 <a href="https://www.intel.com/content/dam/www/public/us/en/documents/product-overviews/dl-boost-product-overview.pdf" target="_blank">https://www.intel.com/content/dam/www/public/us/en/documents/product-overviews/dl-boost-product-overview.pdf</a> <a href="#fnref:1" class="footnote-back-ref">↩</a></p></li>
<li id="fn:2"><p>Samantha Gurriero, "Machine Learning Optimisation: What is the Best Hardware on GCP?", Datatonic, [2] <a href="https://datatonic.com/insights/machine-learning-optimisation-gcp-intel/" target="_blank">https://datatonic.com/insights/machine-learning-optimisation-gcp-intel/</a> <a href="#fnref:2" class="footnote-back-ref">↩</a></p></li>
</ol>

DL Boost open-in-new

DL Boost