Tag Archives: IBM

Google Bristlecone, The Race To Quantum Supremacy

On Monday, March 05, 2018, research scientists from the Google Quantum Al lab whose goal is to build a quantum computer that can be used to solve real-world problems, presented their latest quantum processor called Bristlecone at the annual American Physical Society meeting in Los Angeles.

Qubits or quantum bits are merely the quantum analogue of classical binary bits. Two of the most critical challenges researchers face in their journey to achieve quantum supremacies are error rules and subsequent scalability, this is because qubits are unstable and can be unfavorably affected by noise and can only maintain one state for less for one hundred of microseconds.

Researchers from Google have calculated that a system with 49 quantum bits, a circuit depth exceeding 40 and a two-qubit error below 0.5 percent can “comfortably demonstrate” quantum supremacy. Quantum supremacy is the point where quantum computers can run certain algorithms faster than a classical computer ever could. This has been the dream of many major tech startups and companies including Microsoft, IBM, and Intel.

Bristlecone is Google’s newest quantum processor

Every Bristlecone chip has 72 qubits which might significantly reduce the error rates associated with qubits; however, Google believes quantum computing is not all about qubits. The research team further backed this belief with what they wrote in a blog post:

Operating a device such as a Bristlecone at low system error requires harmony between a full stack of technology ranging from software and control electronics to the processor itself.

The guiding design principle for Bristlecone is to preserve the underlying physics of Google’s previous 9-qubit linear array technology which demonstrated low error rates for readout single-qubit gates to 0.1 percent and most importantly two-qubit gates to 0.6 percent as its best result. This device uses the same scheme for coupling, control, and readout, but is now scaled to a square array of 72 qubits. Therefore they chose a device of moderate size to be able to demonstrate quantum supremacy in the future, first investigate and secondly order error-correction using the surface code to facilitate quantum algorithm development on actual hardware (quantum computers).

Right now, Bristlecone has crowned Google – King of Quantum Computing, a title which previously belonged to IBM because of their 50 qubits chip. However Bristlecone did not just crown Google, it also shortened the race for quantum supremacy as we know it, which Google is “cautiously optimistic” about winning. Despite Google leading the race in Quantum Computing, the ultimate goal of Quantum Supremacy is still far off and might not be surprised if companies like IBM pull something up in the near future.

IBM just unveiled the ‘world’s smallest computer’

by @ theverge.com

The computer is 1mm x 1mm, smaller than a grain of fancy salt, and apparently costs less than ten cents to manufacture. To be clear, the picture above is a set of 64 motherboards, each of which hold two of this tiny computer.

IBM claims the computer has the power of an x86 chip from 1990. That puts it exactly on the edge of enough power to run the original Doom (the original README.TXT for Doom says a 386 processor and 4MB of RAM is the minimum). Hopefully IBM will be more forthcoming with benchmarks in the next five years, and I’m looking forward to repurposing this chip’s LED as a one pixel display.

ICECool – An Intra-Chip Cooling System That Is More Efficient

In the Moore’s Law race to keep improving computer performance, the IT industry has turned upward, stacking chips like nano-sized 3D skyscrapers. But those stacks have their limits, due to overheating. Researchers from IBM have solved this problem by developing an intra-chip cooling system as a contribution to ICECool program research project by the DARPA (Defense Advanced Research Projects Agency).

ICECool - intra-chip cooling system by IBM
ICECool – intra-chip cooling system by IBM

Today, chips are typically cooled by fans which blow air through heatsinks that sit on top of the chips to carry away excess heat. Advanced water-cooling approaches, which are more effective than air-cooling approaches, replace the heatsink with a cold plate that is fixed on the top of the chip.  But this approach requires extra protection and proper insulation of the chip because of the electrical conductivity of water. Neither of these technologies can cool down the chip much efficiently. Here comes the ICECool that cools the chip down from the inside rather than just from the upper surface.

ICECool uses a nonconductive fluid to bring the fluid into the chip. This completely eliminates the need for a barrier between the chip and fluid. It not only delivers a lower device junction temperature, but also reduces system size, weight, and power consumption significantly. The tests performed on the IBM Power 7+ chips demonstrated junction temperature reduction by 25ᵒ C, and chip power usage reduction by 7 percent compared to traditional air cooling. This is clearly a great achievement when the operating cost is much smaller than the conventional cooling technologies.

IBM’s ICECool intra-chip cooling system solves the problem of cooling the 3D “skyscraper” chips by pumping a heat-extracting dielectric fluid right into microscopic gaps, some no thicker than a single strand of hair, between the chips at any level of the stack. Being nonconducting, the dielectric fluid used in ICECool can come into contact with electrical connections without causing any short circuit, so is not limited to one part of a chip or stack. Based on the tests with IBM Power Systems, ICECool technology could reduce the cooling energy for a traditional air-cooled data center by more than 90 percent.

A 5nm GAAFET Chip By IBM, Samsung & GlobalFoundries

In less than two years since making a 7nm test node chip with 20 billion transistors, scientists have paved the way for 30 billion switches on a fingernail-sized chip. IBM with its Research Alliance partners, GlobalFoundries and Samsung, have unveiled their industry-first process that will enable production of 5nm chips.

The new 5nm technology is one of the first ICs based on GAAFET (Gate-All-Around) topology transistors and also probably the first serious application of EUV (Extreme UltraViolet) lithography.

5 nm GAAFET IC from IBM, Samsung & GlobalFoundries
5 nm GAAFET IC from IBM, Samsung & GlobalFoundries

Gate-all-around FETs are similar in concept to FinFETs except that the gate material surrounds the channel region on all sides. Depending on design, gate-all-around FETs can have two or four effective gates. Successfully, Gate-all-around FETs have been characterized both theoretically and experimentally. Also, they have been successfully etched onto InGaAs nanowires, which have a higher electron mobility than silicon.

IBM claims that it can fit in up to 30 Billion transistors on the chip using GAAFET on a 50 mm² chip. It’s a big move in the semiconductor world, as designs become increasingly complicated to apply. While comparing 5nm GAAFET to 10nm commercial chips, it will achieve a 40% performance boost and a 75% power consumption reduction, at similar performance levels. These are some big claims, so expect some big changes just around the corner.

“For business and society to meet the demands of cognitive and cloud computing in the coming years, advancement in semiconductor technology is essential,” said Arvind Krishna, senior vice president, Hybrid Cloud, and director, IBM Research. “That’s why IBM aggressively pursues new and different architectures and materials that push the limits of this industry, and brings them to market in technologies like mainframes and our cognitive systems.”

For more information you can visit the official announcement.