Quantum Computers vs. Supercomputers vs. Mainframes
There has been a great deal of talk lately about supremacy. It’s been all over the news causing people in cities around the nation to become overwhelmed with fear. Are we on the cusp of an event that will forever change the course of our great nation, and affect nearly everyone on planet earth? We’re talking of course about quantum supremacy, that moment in which we know for certain that quantum computing has arrived. As investors, we’re mainly interested in how quantum computing may displace the world’s most powerful computers today. When quantum supremacy arrives this year as expected, and the dust settles, what’s going to happen to all the powerful computers that we’re using today? In order to better understand the impact of quantum supremacy, we decided to take a closer look at each type of powerful computer in existence and the companies that operate them. Mainframes – x86 vs. non-x86 Last week we wrote about mainframes, the IBM mainframe in particular. There were a few key takeaways from that article that investors should note. Firstly, mainframes are pervasive in the world’s biggest businesses, they’re incredibly reliable, and most importantly, they are secure. We wouldn’t expect to see them going away anytime soon which is largely due to the second takeaway. These things are dirt cheap, with things like encryption taking place at 18X the speed and at 5% of the cost. We also learned that not all mainframes are created equal. The reason why this processing power is so cheap can be found in the mainframe architecture being used. The “z/architecture” used by IBM has dominated x86 architecture which is why IBM commands somewhere around 75% market share in the “non-x86” mainframe space. That’s an important distinction to make here. Now that we know about the type of mainframes that still maintain a dominant position in computing today, let’s move on to talking about supercomputers. Supercomputers vs. Mainframes Supercomputers and mainframes are not competing solutions, as both address a particular type of need. The difference between #supercomputers and #mainframes is best described as follows: Mainframes – Computing power used to perform billions of small transactions Supercomputers – Used for focusing on solving a single big problem When it comes to measuring how powerful supercomputers are, we use a unit of measurement referred to as “floating-point operations per second” or “ #flops ”. The authority when it comes to keeping track of who has the fastest computer is the #TOP500 project. Ever since 1993, TOP500 has been using the #Linpack benchmark to measure relative performance among supercomputers to determine who has the fastest computer. The list they produce every six months contains the top-500 computers in the world, along with some really interesting metrics. Firstly, let’s look at the top-10 most powerful computers in the world as of June 2017: We can see that the fastest computer in the world is the Sunway TaihuLight, one of the greatest sources of Chinese national pride after the giant panda. With a price tag of $273 million, the TaihuLight is built entirely on Chinese engineered semiconductors as part of a program to decrease China’s reliance on foreign technology. That seems to have worked. When it comes to the country with the most supercomputers, the USA and China are now in a dead heat when measured by total GFlops (251 million vs. 235 million) and by number of machines (168 vs. 160). Moving on, 88% of the computing power in the TOP500 list is owned by just 10 companies: In third place we see the Sunway TaihuLight (NRCPC is the supercomputing center where this monster lives) and in first place we see none other than Cray Inc. (NASDAQ:CRAY). Second place is HPE which stands for Hewlett-Packard (NYSE:HPE). Along with other familiar names on the list are some outfits you may never have heard of: NUDT – The National University of Defense Technology is a Chinese military academy and research university Sugon – China’s leading supercomputer company with a $4.2 billion market cap which trades under the name Dawning Information Industry Co (SHA:603019) Bull – A French owned computing company which is a fully owned subsidiary of Atos (EPA:ATO) In addition to the above list of companies that hold ownership over all this computing power, we found an interesting excerpt from the TOP500 June list as follows: The new number three supercomputer is the upgraded Piz Daint, a Cray XC50 system installed at the Swiss National Supercomputing Centre (CSCS). The upgrade was accomplished with additional NVIDIA Tesla P100 GPUs, doubling the Linpack performance of the system’s previous mark of 9.8 petaflops in November 2016, which itself was the result of a significant upgrade. Piz Daint’s current Linpack result of 19.6 petaflops enabled the system to climb five positions in the rankings. As it turns out, NVIDIA (NASDAQ:NVDA) has been moving into the supercomputer space. In November of last year, they debuted their own supercomputer, the DGX SaturnV, which currently sits in 32nd position on the top-500 list. More importantly, it’s ranked the world’s most efficient supercomputer by a wide margin. An article published by NVIDIA announcing this accomplishment is quite telling: We’re also training neural networks to understand chipset design and very-large-scale-integration, so our engineers can work more quickly and efficiently. Yes, we’re using GPUs to help us design GPUs. That’s simply incredible, and it’s not just NVIDIA’s server using their chips. A quick look at the top-500 list shows that just over 14% of all supercomputers are now using NVIDIA accelerators. Even more notable is that IBM and NVIDA have teamed up with Oak Ridge National Laboratory to build the world’s most powerful supercomputer, Summit, which will use IBM POWER9 CPUs and NVIDIA Volta GPUs. Expected to be operational next year, the Summit supercomputer will be the most powerful computer in the world at 150-300 million GFlops (compared to the Sunway at 93 million GFlops today). This takes us back to our original question. Will the planned announcement of quantum supremacy this year by any one of the four main contenders be the catalyst that brings NVDA shares back down to earth? Will shares in CRAY also take a hit with the news? That’s a good segue into our next topic. Supercomputers vs Quantum Computers A while back we wrote an article about “The Definition of Quantum Everything” in which we attempted to understand quantum computing. When even the people who are designing quantum computers don’t know how they work, we don’t want to focus too much on the technical aspects. Instead, we can just assume that at some point we’ll achieve quantum supremacy and we want to try and anticipate what the impact will be on supercomputers. According to CRAY, there’s little to be concerned about at the moment. Here’s what Chief Strategy Officer at CRAY, Barry Bolden, had to say about the matter last year: This is certainly a new and exciting technology and the excitement it generates is good for the industry, but it still won’t be a truly productive technology in 2016, or 2020, or perhaps even 2025 for that matter. When quantum computers are complete, they will be expected to work on the same sort of problems that supercomputers work on today – everything from designing better airplanes to discovering new drug molecules, to figuring out why some people today believe that the earth is flat. If we are to believe all the hype, quantum computers are expected to be orders of magnitude more powerful than today’s supercomputers, the extent of which is simply mind boggling: In the early 2000s, for example, people thought it would take about 24 billion years to calculate on a quantum computer the energy levels of ferredoxin, which plants use in photosynthesis. Now, through a combination of theory, practice, engineering and simulation, the most optimistic estimates suggest that it may take around an hour. – Alex Bocharov, Nature Magazine Interview, October 2016. One big question remains. Will we see #quantumsupremacy announced this year and by whom? We gave you a list of the four main contenders in an article earlier this year titled “ #ArtificialIntelligence ( #AI ) and #QuantumComputing ” and it appeared at the time that @Google was the furthest ahead in the race. Then just last month, Google revealed their “blueprint for quantum supremacy” and said that “the first demonstration of quantum supremacy could be just months way”. Unless that means less than three months, we’re not going to get quantum supremacy before the year ends. The Final Verdict It’s probably reasonable to assume that with a long enough horizon, quantum computers will eventually displace supercomputers. When quantum supremacy is announced, what sort of long-term strategy will @CRAY adopt? Maybe they’ll be to quantum computing what the x86 mainframe is to IBM’s mainframe – a cheaper substitute that can’t compete on performance but still has a place in the world of computing. Regardless of the outcome, we’re going to take a closer look at Cray Inc. in a future article so we can better understand their business and see what a supercomputer business looks like from an investor’s perspective.