| Both sides previous revision
Previous revision
|
|
ai_timelines:current_flops_prices [2023/07/20 22:38]
rickkorzekwa [Summary] fixed dollar signs |
ai_timelines:current_flops_prices [2023/07/20 22:39] (current)
rickkorzekwa fixed dollar signs |
| |
| <HTML> | <HTML> |
| <p>In November 2017, we estimate the price for one GFLOPS to be between \$0.03 and \$3 for single or double precision performance, using GPUs (therefore excluding some applications). Amortized over three years, this is $1.1 x 10<sup>-5</sup> -$1.1 x 10<sup>-7</sup> /GFLOPShour.</p> | <p>In November 2017, we estimate the price for one GFLOPS to be between \$0.03 and \$3 for single or double precision performance, using GPUs (therefore excluding some applications). Amortized over three years, this is \$1.1 x 10<sup>-5</sup> -\$1.1 x 10<sup>-7</sup> /GFLOPShour.</p> |
| </HTML> | </HTML> |
| |
| |
| <HTML> | <HTML> |
| <p>For CPUs and GPUs, we include only the original recommended retail price of the CPU or GPU, and not other computer components (i.e. we do not even include the cost of CPUs in the price of GPUs). In 2015 we compared prices between <a href="https://drive.google.com/file/d/1xTA4LoooCuHhCWLhkJZ2ZbOWDxso3a0UILO2JxLn5CweadJVJUziux7CMiumtITXTOjXfttY5zNHzCee/view?usp=sharing">one complete rack server</a> and the set of four <a href="http://www.ebay.com/itm/like/351337480917?lpid=82&chn=ps">processors</a> inside it, and found the complete server was around 36% more expensive ($30,000 vs. $22,000). We expect this is representative at this scale, but diminishes with scale.</p> | <p>For CPUs and GPUs, we include only the original recommended retail price of the CPU or GPU, and not other computer components (i.e. we do not even include the cost of CPUs in the price of GPUs). In 2015 we compared prices between <a href="https://drive.google.com/file/d/1xTA4LoooCuHhCWLhkJZ2ZbOWDxso3a0UILO2JxLn5CweadJVJUziux7CMiumtITXTOjXfttY5zNHzCee/view?usp=sharing">one complete rack server</a> and the set of four <a href="http://www.ebay.com/itm/like/351337480917?lpid=82&chn=ps">processors</a> inside it, and found the complete server was around 36% more expensive (\$30,000 vs. \$22,000). We expect this is representative at this scale, but diminishes with scale.</p> |
| </HTML> | </HTML> |
| |
| |
| <HTML> | <HTML> |
| <p>We have not included the costs of energy or other ongoing expenses in any prices. Non-energy costs are hard to find, and we suspect a relatively small and consistent fraction of costs. In 2015 we estimated energy costs to be around 10% of hardware costs.<span class="easy-footnote-margin-adjust" id="easy-footnote-1-477"></span><span class="easy-footnote"><a href="#easy-footnote-bottom-1-477" title='The Intel Xeon E5-2699 uses 527.8 watts and costs $5,190. The processor can be bought <a href="http://www.serversupply.com/products/part_search/pid_lookup.asp?pid=229567&amp;gclid=Cj0KEQjwi-moBRDL4Omf9d_LndMBEiQAQtFf83x42USI1XM-_KXMDkkDbi8NyYzbLFemeykjBuQfPrYaAlew8P8HAQ">here</a> for $5,190 as of April 1 2015. Its energy consumption is <a href="http://www.tomshardware.com/reviews/intel-xeon-e5-2600-v3-haswell-ep,3932-9.html">527.8 watts</a> under load, or 90.9 watts idle. Over three years, with $0.05/kWh this is $694, or 13% of the hardware cost.</p> <p>Titan also uses 13% of its hardware costs in energy over three years. Titan <a href="http://en.wikipedia.org/wiki/Titan_(supercomputer)">cost</a> about $4000 dollars per hour amortized over 3 years, and consumes about 10<span style="font-size: 13.3333330154419px; line-height: 20px;">M</span> watts, at a cost of $500 per hour (assuming $0.05 per kWh), which is also 13% of its hardware cost.'><sup>1</sup></a></span></p> | <p>We have not included the costs of energy or other ongoing expenses in any prices. Non-energy costs are hard to find, and we suspect a relatively small and consistent fraction of costs. In 2015 we estimated energy costs to be around 10% of hardware costs.<span class="easy-footnote-margin-adjust" id="easy-footnote-1-477"></span><span class="easy-footnote"><a href="#easy-footnote-bottom-1-477" title='The Intel Xeon E5-2699 uses 527.8 watts and costs \$5,190. The processor can be bought <a href="http://www.serversupply.com/products/part_search/pid_lookup.asp?pid=229567&amp;gclid=Cj0KEQjwi-moBRDL4Omf9d_LndMBEiQAQtFf83x42USI1XM-_KXMDkkDbi8NyYzbLFemeykjBuQfPrYaAlew8P8HAQ">here</a> for \$5,190 as of April 1 2015. Its energy consumption is <a href="http://www.tomshardware.com/reviews/intel-xeon-e5-2600-v3-haswell-ep,3932-9.html">527.8 watts</a> under load, or 90.9 watts idle. Over three years, with \$0.05/kWh this is \$694, or 13% of the hardware cost.</p> <p>Titan also uses 13% of its hardware costs in energy over three years. Titan <a href="http://en.wikipedia.org/wiki/Titan_(supercomputer)">cost</a> about \$4000 dollars per hour amortized over 3 years, and consumes about 10<span style="font-size: 13.3333330154419px; line-height: 20px;">M</span> watts, at a cost of \$500 per hour (assuming \$0.05 per kWh), which is also 13% of its hardware cost.'><sup>1</sup></a></span></p> |
| </HTML> | </HTML> |
| |
| |
| <HTML> | <HTML> |
| <p>We collected performance and price figures from Wikipedia<span class="easy-footnote-margin-adjust" id="easy-footnote-3-477"></span><span class="easy-footnote"><a href="#easy-footnote-bottom-3-477" title='Wikipedia pages: <a href="https://en.wikipedia.org/wiki/Xeon_Phi">Xeon Phi</a>, <a href="https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units">List of Nvidia Graphics Processing Units</a>, <a href="https://en.wikipedia.org/wiki/List_of_AMD_graphics_processing_units">List of AMD Graphics Processing Units</a></p> <p>Other sources are visible in the last column of <a href="https://docs.google.com/spreadsheets/d/1yqX2cENwkOxC26wV_sBOvV0NxHzzfmL6tU7StzrFXRc/edit?usp=sharing">our dataset (see ‘Wikipedia GeForce, Radeon, Phi simplified’ sheet)</a>'><sup>3</sup></a></span>, which are available <a href="https://docs.google.com/spreadsheets/d/1yqX2cENwkOxC26wV_sBOvV0NxHzzfmL6tU7StzrFXRc/edit?usp=sharing">here</a> (see ‘Wikipedia GeForce, Radeon, Phi simplified’). These are theoretical performance figures, which we understand to generally be between somewhat optimistic and ten times too high. So this data suggests real prices of around $0.03-$0.3/GFLOPS. We collected both single and double precision figures, but the cheapest were similar.</p> | <p>We collected performance and price figures from Wikipedia<span class="easy-footnote-margin-adjust" id="easy-footnote-3-477"></span><span class="easy-footnote"><a href="#easy-footnote-bottom-3-477" title='Wikipedia pages: <a href="https://en.wikipedia.org/wiki/Xeon_Phi">Xeon Phi</a>, <a href="https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units">List of Nvidia Graphics Processing Units</a>, <a href="https://en.wikipedia.org/wiki/List_of_AMD_graphics_processing_units">List of AMD Graphics Processing Units</a></p> <p>Other sources are visible in the last column of <a href="https://docs.google.com/spreadsheets/d/1yqX2cENwkOxC26wV_sBOvV0NxHzzfmL6tU7StzrFXRc/edit?usp=sharing">our dataset (see ‘Wikipedia GeForce, Radeon, Phi simplified’ sheet)</a>'><sup>3</sup></a></span>, which are available <a href="https://docs.google.com/spreadsheets/d/1yqX2cENwkOxC26wV_sBOvV0NxHzzfmL6tU7StzrFXRc/edit?usp=sharing">here</a> (see ‘Wikipedia GeForce, Radeon, Phi simplified’). These are theoretical performance figures, which we understand to generally be between somewhat optimistic and ten times too high. So this data suggests real prices of around \$0.03-\$0.3/GFLOPS. We collected both single and double precision figures, but the cheapest were similar.</p> |
| </HTML> | </HTML> |
| |
| |
| <HTML> | <HTML> |
| <p>We looked at a small number of popular CPUs on Geekbench from the past five years, and found the cheapest to be around $0.71/GFLOPS.<span class="easy-footnote-margin-adjust" id="easy-footnote-5-477"></span><span class="easy-footnote"><a href="#easy-footnote-bottom-5-477" title='See <a href="https://docs.google.com/spreadsheets/d/1yqX2cENwkOxC26wV_sBOvV0NxHzzfmL6tU7StzrFXRc/edit?usp=sharing">&#8216;Geekbench 4 History&#8217; tab</a>'><sup>5</sup></a></span> However there appear to be 5x disparities between different versions of Geekbench, so we do not trust these numbers a great deal (these figures are from the version we have seen to give relatively high performance figures, and thus low implied prices).</p> | <p>We looked at a small number of popular CPUs on Geekbench from the past five years, and found the cheapest to be around \$0.71/GFLOPS.<span class="easy-footnote-margin-adjust" id="easy-footnote-5-477"></span><span class="easy-footnote"><a href="#easy-footnote-bottom-5-477" title='See <a href="https://docs.google.com/spreadsheets/d/1yqX2cENwkOxC26wV_sBOvV0NxHzzfmL6tU7StzrFXRc/edit?usp=sharing">&#8216;Geekbench 4 History&#8217; tab</a>'><sup>5</sup></a></span> However there appear to be 5x disparities between different versions of Geekbench, so we do not trust these numbers a great deal (these figures are from the version we have seen to give relatively high performance figures, and thus low implied prices).</p> |
| </HTML> | </HTML> |
| |
| |
| <HTML> | <HTML> |
| <p>Amazon <a href="http://en.wikipedia.org/wiki/Amazon_Elastic_Compute_Cloud">Elastic Cloud Compute</a> (EC2) is a major seller of virtual computing. Based on their <a href="https://aws.amazon.com/ec2/pricing/" rel="nofollow">current pricing</a>, as of October 5th, 2017, renting a <a href="https://aws.amazon.com/ec2/instance-types/">c4.8xlarge</a> instance <a href="https://aws.amazon.com/ec2/pricing/" rel="nofollow">costs</a> $0.621 per hour (if you purchase it for three years, and pay upfront).</p> | <p>Amazon <a href="http://en.wikipedia.org/wiki/Amazon_Elastic_Compute_Cloud">Elastic Cloud Compute</a> (EC2) is a major seller of virtual computing. Based on their <a href="https://aws.amazon.com/ec2/pricing/" rel="nofollow">current pricing</a>, as of October 5th, 2017, renting a <a href="https://aws.amazon.com/ec2/instance-types/">c4.8xlarge</a> instance <a href="https://aws.amazon.com/ec2/pricing/" rel="nofollow">costs</a> \$0.621 per hour (if you purchase it for three years, and pay upfront).</p> |
| </HTML> | </HTML> |
| |
| |
| <HTML> | <HTML> |
| <p>This implies that a GFLOPShour costs $6.3 x 10<sup>-3</sup> , or optimistically as little as $3.2 x 10<sup>-4</sup> . This is much higher than a GPU, at $3.4 x 10<sup>-6</sup> for a GFLOPShour, if we suppose the hardware is used over around three years. Amazon is probably not the cheapest provider of cloud computing, however the difference seems to be something like a factor of two,<span class="easy-footnote-margin-adjust" id="easy-footnote-7-477"></span><span class="easy-footnote"><a href="#easy-footnote-bottom-7-477" title='We wrote in 2015: &#8220;Other sources of virtual computing seem to be similarly priced. An <a href="http://www.infoworld.com/d/cloud-computing/ultimate-cloud-speed-tests-amazon-vs-google-vs-windows-azure-237169?page=0,2">informal comparison</a> of computing providers suggests that on a set of &#8220;real-world java benchmarks&#8221; three providers are quite closely comparable, with all between just above Amazon&#8217;s price and just under half Amazon&#8217;s price for completing the benchmarks, across different instance sizes. This analysis also suggests Amazon is a relatively costly provider&#8230;&#8221;'><sup>7</sup></a></span> which is not enough to make cloud computing competitive with GPUs.</p> | <p>This implies that a GFLOPShour costs \$6.3 x 10<sup>-3</sup> , or optimistically as little as \$3.2 x 10<sup>-4</sup> . This is much higher than a GPU, at \$3.4 x 10<sup>-6</sup> for a GFLOPShour, if we suppose the hardware is used over around three years. Amazon is probably not the cheapest provider of cloud computing, however the difference seems to be something like a factor of two,<span class="easy-footnote-margin-adjust" id="easy-footnote-7-477"></span><span class="easy-footnote"><a href="#easy-footnote-bottom-7-477" title='We wrote in 2015: &#8220;Other sources of virtual computing seem to be similarly priced. An <a href="http://www.infoworld.com/d/cloud-computing/ultimate-cloud-speed-tests-amazon-vs-google-vs-windows-azure-237169?page=0,2">informal comparison</a> of computing providers suggests that on a set of &#8220;real-world java benchmarks&#8221; three providers are quite closely comparable, with all between just above Amazon&#8217;s price and just under half Amazon&#8217;s price for completing the benchmarks, across different instance sizes. This analysis also suggests Amazon is a relatively costly provider&#8230;&#8221;'><sup>7</sup></a></span> which is not enough to make cloud computing competitive with GPUs.</p> |
| </HTML> | </HTML> |
| |
| |
| <HTML> | <HTML> |
| <p>A top supercomputer can perform a GFLOPS for around $3, in 2017. (See <em><a href="/doku.php?id=ai_timelines:hardware_and_ai_timelines:price-performance_trend_in_top_supercomputers">Price performance trend in top supercomputers</a>)</em></p> | <p>A top supercomputer can perform a GFLOPS for around \$3, in 2017. (See <em><a href="/doku.php?id=ai_timelines:hardware_and_ai_timelines:price-performance_trend_in_top_supercomputers">Price performance trend in top supercomputers</a>)</em></p> |
| </HTML> | </HTML> |
| |
| |
| <HTML> | <HTML> |
| <p><a href="https://en.wikipedia.org/wiki/Tensor_processing_unit">Tensor processing units</a> appear to perform a GFLOPS for around $1, in February 2018. However it is unclear how this GFLOPS is measured, which makes it somewhat harder to compare (e.g. whether it is single precision or double precision). Such a high price is also at odds with rumors we have heard that TPUs are an especially cheap source of computing, so possibly TPUs are more efficient for a particular set of applications other than the ones where most of these machines have been measured.</p> | <p><a href="https://en.wikipedia.org/wiki/Tensor_processing_unit">Tensor processing units</a> appear to perform a GFLOPS for around \$1, in February 2018. However it is unclear how this GFLOPS is measured, which makes it somewhat harder to compare (e.g. whether it is single precision or double precision). Such a high price is also at odds with rumors we have heard that TPUs are an especially cheap source of computing, so possibly TPUs are more efficient for a particular set of applications other than the ones where most of these machines have been measured.</p> |
| </HTML> | </HTML> |
| |
| <td class="column-3">Theoretical peak</td> | <td class="column-3">Theoretical peak</td> |
| <td class="column-4">.03-0.3</td> | <td class="column-4">.03-0.3</td> |
| <td class="column-5">$0.03/GFLOPS is given, but is underestimate</td> | <td class="column-5">\$0.03/GFLOPS is given, but is underestimate</td> |
| </tr> | </tr> |
| <tr class="row-3"> | <tr class="row-3"> |
