NEC will deliver and deploy a direct-liquid cooled HPC solution comprising 2,300 Gigabyte compute nodes based on the AMD EPYC 7601 processor, totaling up to 147,200 cores within a budget of more than 20 million Euros. The new DLR cluster will be hosted and administrated by the Center for Information Services and High Performance Computing (ZIH) at the new Datacenter of the Technische Universitat Dresden (TU Dresden), according to the strategic scientific co-operation between NEC and DLR.

SEATTLE and FRANKFURT, Germany, June 18, 2019 — Global supercomputer leader Cray Inc. today announced that the Singapore Centre for Environmental Life Sciences Engineering (SCELSE), a renowned Research Center of Excellence at Nanyang Technological University, Singapore (NTU Singapore), has awarded the company a contract for a Cray CS500 cluster supercomputer. SCELSE will use the supercomputer for fast tracking its environmental meta-omics projects.

The CS500 was selected for its superior price-performance, scalability and functionality – surpassing all other solutions in an open tender. Cray’s price/performance matrix appealed to SCELSE, and the core count of the system will contain more than 12,000 AMD EPYC processor cores.

How a Big U.S. Chip Maker Gave China the ‘Keys to the Kingdom’

June 27, 2019 10:46 am ET

Advanced Micro Devices Inc. transformed itself from a financially struggling company to an investor’s dream in just three years, a turnaround that began with a decision to help Chinese partners develop advanced computer-chip technology.

That deal may have helped save the company, but it alarmed U.S. national-security officials, who saw it as a threat to their goal to rein in China’s supercomputing industry. Last week, after years of friction, the Commerce Department issued an order that effectively bars several Chinese entities—including AMD’s partners—from obtaining American technology.

It looked as though the U.S. had succeeded in stanching the flow of cutting-edge computer technology to China. In reality, it was too late. Chinese versions of AMD chips already have been rolling off production lines. That technology is helping China in its race with the U.S. to build the first next-generation supercomputer—an essential tool for advanced civilian and military applications.

AMD’s Chinese partner, a military contractor, already used those chips to build what may prove to be the world’s fastest supercomputer, according to high-performance computing experts briefed on the project.

The partnership with the Santa Clara, Calif.-based chip maker was a game changer for China, which has long been unable to match the U.S.’s supercomputing power because of its inferior chips, one product the country has so far struggled to master. The AMD deal gave China access to state-of-the-art x86chips, which are made by only two companies in the world: AMD and Intel Corp. They are the most dominant processor technology in use today.

“It’s the keys to the kingdom,” said retired Brig. Gen. Robert Spalding, who served on the National Security Council in 2017 and 2018 and discussed strategies to stop the AMD deal with officials at other agencies. “Everything today is built on x86.”

AMD said in a written statement that it “has and will continue to comply with all U.S. laws,” and that the technology transferred to China in the deal wasn’t as high-performing as other U.S. products commercially available there at the time.

Commerce Department officials said last week’s action was made in consultation with other agencies. It followed weeks of inquiries by The Wall Street Journal about AMD’s Chinese partnership and the belief of some government officials that AMD had plotted a sophisticated end-run around regulators.

This account of the protracted battle over the deal between AMD and the government is based on interviews with more than a dozen current and former government officials familiar with AMD’s China deal, senior chip-industry executives, lawyers and company and government filings in the U.S. and China.

When Lisa Su was named AMD’s chief executive in October 2014, the company was desperate for cash, its debts were mounting and its revenue was declining. Its stock had dipped to around $3 a share. Some analysts predicted it would seek bankruptcy protection.

Three weeks after getting the top job, Ms. Su, a Taiwan-born New Yorker, jetted to Beijing to meet officials at China’s Ministry of Industry and Information Technology. A Chinese vice minister urged her to partner with China “to achieve mutual benefits based on AMD’s technological strength,” according to a ministry press release at the time.

In February 2016, AMD reached a joint-venture deal involving a leading Chinese supercomputer developer, a state-backed military supplier called Sugon Information Industry Co., to make chips licensing AMD’s x86 processor technology.

“Making contributions to China’s national defense and security is the fundamental mission of Sugon,” its website read as recently as December 2016. Sugon also makes computers for civilian use.

In exchange, the Chinese government gave AMD a lifeline: $293 million in licensing fees plus royalties on the sales of any chips developed by the venture.

That April, AMD got another boost from Beijing. It said it would get a $371 million payout for selling an 85% stake in two of its semiconductor factories in China and Malaysia to an entity controlled by China Integrated Circuit Industry Investment Fund Co., a state-backed financier known as “the Big Fund.” Its mission is to develop China’s indigenous chip industry.

The U.S. and China are competing to develop the world’s first exascale computer, a next-generation supercomputer that would be capable of doing one quintillion—or one billion billion—calculations per second. While supercomputers are used in tasks such as weather forecasting and cancer research, they also are integral to the development of nuclear weapons, encryption, missile defense and other systems. The chips American companies produce to power supercomputers, including AMD’s x86, are superior to any China can make on its own.

“Semiconductors are a space where the U.S. still leads China and the rest of the world,“ says William Evanina, the U.S. government’s top counterintelligence official.

Shortly before AMD announced the Sugon deal, Defense Department officials learned of a presentation the joint-venture partners made in China talking up the deal’s potential to transform the country into a leader in processor technology.

Pentagon officials quickly began seeking ways to unwind the deal, according to people familiar with the matter. They questioned company representatives and repeatedly tried to get them to submit the deal to a review by the Committee on Foreign Investment in the U.S., or Cfius, according to those people.

Companies routinely seek approval from the committee for deals that raise national-security issues. The committee is led by the Treasury Department and includes the Defense, Commerce, Justice and Energy departments, among others. It can recommend that the president block foreign investments in U.S. assets for national-security reasons.

AMD didn’t submit the deal for committee review, arguing Cfius didn’t have jurisdiction to review that type of joint venture, according to people familiar with the matter. The company also claimed it wasn’t turning over any state-of-the-art technology. Pentagon officials found that response at odds with how the joint venture had portrayed itself in China.

Treasury officials, who have the final say on the consensus-driven Cfius panel, ultimately agreed with AMD’s assessment that the deal fell outside its remit. That left AMD and Defense officials at a stalemate through the end of the Obama administration and the first months of the Trump administration.

Commerce Department officials also were investigating the deal for compliance with export controls. In June 2017, following a series of inquiries, they sent AMD an “is informed letter” that alerted the company they suspected the China deal violated export controls. AMD replied that it was complying with all regulations, according to people familiar with the matter.

In its statement to the Journal, AMD said that starting in 2015, it “diligently and proactively briefed the Department of Defense, the Department of Commerce and multiple other agencies within the U.S. government before these joint ventures were entered into, and we received no objections to their formation or the transfer of technology.”

Before the transfer of any technology, AMD said, the Commerce Department notified the company that it wasn’t restricted or otherwise prohibited from being transferred.

Current and former national-security officials said in interviews they believe AMD designed the deal’s complex structure, which involved the creation of two interlinked joint ventures, to sidestep U.S. regulations. AMD said the deal was structured for business and technological reasons and to comply with regulations, not to evade them.

AMD controls the first joint venture, which licenses the U.S. chip maker’s x86 intellectual property and manages production of the chips. The second venture, controlled by AMD’s Chinese partner, designs the devices that use the chips and sells the final products.

The arrangement enables AMD to share technology with China while retaining control over the entity working with its intellectual property. The creation of the second, China-controlled joint venture allowed the parties to claim that the resulting product was indigenously developed in China, a key goal of the Chinese government.

While Cfius has jurisdiction to review foreign purchases of U.S. chip assets, it doesn’t have clear authority to review overseas joint ventures that don’t grant a foreign entity control over a U.S. business.

AMD didn’t have to seek an export license from the Commerce Department because it stripped out the parts of its x86 chip that would have required licenses, such as encryption technology, which China didn’t need anyway.

The x86 chips under development for China, code-named Dhyana, are similar to AMD’s own EPYC chips, minus U.S. encryption technology that AMD omitted, say people familiar with the deal.

The joint venture’s U.S. managers stressed to employees that the Dhyana chip was being developed for commercial purposes, such as providing processing technology to Chinese tech giants such as Baidu Inc. or Tencent Holdings Ltd. , according to one of the people familiar with the deal.

Sugon, however, told state-run media that the x86 technology could serve China’s bid to build the world’s first exascale supercomputer. The joint venture’s job postings in Chinese implore candidates to help strengthen China’s domestic chip ambitions.

Sun Ninghui, head of the computing-technology institute at the Chinese Academy of Sciences, which works closely with Sugon, likened its chip development strategy to what China did with high-speed trains—introduce a foreign technology to the market, absorb it, and then innovate to make China a leader.

“This gradually advances our ability to comprehend their core technologies,” Mr. Sun told a government-run newspaper. “That way, we no longer can be pulled around by our noses.”

By mid-2017, concerns about AMD’s China deal had reached the Trump White House. Retired Gen. Spalding, who left the National Security Council last year, said of AMD: “They’re using the letter of the law to violate the spirit of the law.”

In November 2017, Sen. John Cornyn (R., Texas) and then-Rep. Robert Pittenger (R., N.C.) introduced legislation to expand Cfius’s authority, including broader powers to review joint ventures overseas.

There was heavy opposition from many companies and trade organizations, which feared Cfius interfering in their overseas activities, and the provision didn’t make it into the final version of the legislation that passed in August 2018.

Defense Department officials decided to unilaterally submit AMD’s Sugon deal to Cfius for review, despite Treasury’s earlier interpretation that it fell outside the panel’s jurisdiction. Only rarely in Cfius’s 44-year history has the committee been asked to review deals without the cooperation of either party involved, according to lawyers who track the confidential cases.

As the Cfius filing sat in limbo, Lisa Porter, the Defense Department’s deputy undersecretary for research and engineering, criticized AMD’s China deal in front of industry executives. AMD officials hired Beacon Global Strategies, a Washington-based consulting firm that employs former top national-security officials, to try to make peace.

On June 21, however, the Commerce Department announced the new export restrictions banning Sugon and its affiliates on the AMD deal from accessing U.S. technology without a license. The move, which followed the imposition of similar export restrictions last month targeting Chinese telecommunications giant Huawei Technologies Co., effectively forces AMD to unwind the deal.

In its decision, the Commerce Department said the Chinese entities were determined “to be acting contrary to the national security or foreign policy interests of the U.S.” It added: “Sugon has publicly acknowledged a variety of military end uses and end users of its high-performance computers.”

In a statement on its WeChat account, Sugon said the decision would severely disrupt its cooperation with U.S. partners. It said it had complied with all U.S. laws and would seek to discuss the issue with U.S. officials. “We believe there is a large gap in the understanding of Sugon’s corporate circumstances on the part of relevant U.S. authorities,” it said.

The Commerce action will make it hard for China to make future generations of the x86 chip, and it’s unclear if AMD’s partners will be able to continue producing the current version without the U.S. company’s technical assistance, according to experts in semiconductor technology. But China gained significant technical know-how through the deal, which has already yielded chips currently powering supercomputers.

For AMD, pushing back against U.S. national-security officials while its China partnership gained traction paid off. The chip maker used the cash infusion to get back on its feet and has since introduced an array of competitive new products.

AMD Response to a Recent Article in The Wall Street Journal

1. Google announced it has deployed 2nd Gen AMD EPYC processors in its internal infrastructure production datacenter environment and in late 2019 will support new general-purpose machines powered by 2nd Gen AMD EPYC processors on Google Cloud Compute Engine as well;
2. Twitter announced it will deploy 2nd Gen AMD EPYC processors across its datacenter infrastructure later this year, reducing TCO by 25%;
3. Microsoft announced the preview of new Azure virtual machines for general purpose applications, as well as limited previews of cloud-based remote desktops and HPC workloads based on 2nd Gen AMD EPYC processors today;
4. HPE announced continued support of the AMD EPYC processor family with plans to triple their AMD-based portfolio with a broad range of 2nd Gen AMD EPYC processor-based systems, including the HPE ProLiant DL385 and HPE ProLiant DL325 servers;
5. Cray announced The Air Force Weather Agency will be using a Cray Shasta system with 2nd Gen AMD EPYC processors to provide comprehensive terrestrial and space weather information to the U.S. Air Force and Army;
6. Lenovo announced new solutions that are specifically built to take advantage of the full range of enhanced capabilities found in the 2nd Gen AMD EPYC processors. Available today, the ThinkSystem SR655 and SR635 are ideal solutions for use cases such as video infrastructure, virtualization, software-defined storage and more, with exceptional energy efficiency;
7. Dell announced the upcoming availability of newly designed servers optimized for 2nd Gen AMD EPYC processors;
8. VMware and AMD announced a close collaboration to deliver support for new security and other features of the high-performance 2nd Gen AMD EPYC processors within VMware vSphere.

• Dell Technologies announced five new Dell EMC PowerEdge platforms powered by the 2nd Gen AMD EPYC processor. These platforms were designed from the ground up and optimized to support the features of the new AMD EPYC processor including PCIe 4.0.
• IBM Cloud detailed how 2nd Gen EPYC processors can support IBM Cloud customer needs in specific areas including helping improve cloud security, better memory bandwidth for big data and analytics workloads and core scaling and breakthrough performance for container workloads. IBM plans to have more to share in 2020 about its performance offerings for clients.
• Nokia highlighted how 2nd Gen EPYC processors significantly accelerate its Cloud Packet Core system which helps service providers deliver converged broadband, IoT, and machine-type communication services for 5G. In testing, Nokia found its Cloud Packet Core system with 2nd Gen AMD EPYC provided a 2X increase in packet throughput compared to previous systems.
• ATOS, a global leader in digital transformation, announced Genci is using the 2nd Gen AMD EPYC processors to expand the use of supercomputing for the benefit of French scientific communities. Genci and ATOS are using the 2nd Gen AMD EPYC processor due to its breakthrough performance, efficiency and TCO.
• OVHcloud, a global cloud provider specializing in delivering industry-leading performance and cost-effective solutions, announced a new high-end hosting instance based on the AMD EPYC 7402P processor. This instance will be available at the end of 2019.
• TSMC announced its adoption of 2nd Gen AMD EPYC helping power its next generation research and leading process technology.

We are focusing on meeting the strong demand for our 3rd generation AMD Ryzen processors in the market and now plan to launch both the AMD Ryzen 9 3950X and initial members of the 3rd Gen AMD Ryzen Threadripper processor family in volume this November. We are confident that when enthusiasts get their hands on the world’s first 16-core mainstream desktop processor and our next-generation of high-end desktop processors, the wait will be well worth it.

Micron stock falls as forecast disappoints, earnings continue to plunge

Sept 26, 2019

Micron Technology Inc. shares fell in the extended session Thursday after the memory-chip maker reported another large earnings decline and predicted a disappointing holiday season.

Micron said it expects adjusted earnings of 39 cents to 53 cents a share on revenue of $4.8 billion to $5.2 billion for its first fiscal quarter. Analysts surveyed by FactSet had forecast earnings of 53 cents a share on revenue of $4.8 billion.

The company reported fiscal fourth-quarter net income of $561 million, or 49 cents a share, compared with $4.33 billion, or $3.56 a share, in the year-ago period. Adjusted earnings were 56 cents a share.

Revenue fell to $4.87 billion from $8.44 billion in the year-ago quarter. Analysts had forecast earnings of 51 cents on revenue of $4.59 billion on average, according to FactSet.

What would you do with 1.5 million threads of computational horsepower? The UK Research and Innovation (UKRI) has an idea.

The Archer supercomputer system based in Edinburgh was first introduced in 2013 based on the Cray XC30 design and has since been in use by researchers in need of serious compute power. Where it was once among the top 20 supercomputers in the world, today, that’s no longer the case, and the time has come to replace it.

As such, the center announced that it has contracted with Cray to build the Archer 2 supercomputer, which will be based on AMD EPYC Rome processors.

The center will build the system out of 5,848 Shasta compute nodes (deep dive here). Each node will contain two second-gen AMD EPYC Rome processors with 64 cores apiece, with each clocked at 2.2GHz (likely the base frequency).

This adds up to 11,969 CPUs, and a grand total of 748,544 CPU cores and about 1.5 million threads (we rounded that number up simply because we’ve always wanted to).

Texas Instruments Stock Tumbles on Gloomy Outlook

Oct. 22, 2019

Texas Instruments stock fell in late trading Tuesday after the chip maker’s third-quarter results slightly missed estimates. Investors are likely to be more focused on the company’s fourth-quarter outlook, though, which missed Wall Street expectations by a wide margin.

Late Tuesday, Texas Instruments reported third-quarter earnings per share of $1.40, excluding certain items, versus the $1.42 analyst consensus. It generated $3.77 billion of revenue in the quarter, lower than the $3.82 billion average analyst estimate and down from $4.3 billion in the year-ago quarter.

“Revenue decreased 11% from the same quarter a year ago, as most markets weakened further,” CEO Rich Templeton said in the earnings release.

The chip maker also forecast an earnings-per-share range of $0.91 to $1.09 for its fourth quarter, which includes an estimated $5 million tax benefit. Analysts have been expecting earnings of $1.28 a share for the fourth quarter, according to FactSet.

Texas Instruments, a broad-based semiconductor bellwether, also said sales for the December quarter will be in range of $3.07 billion to $3.33 billion, versus the $3.6 billion average forecast.

XLNX reports tomorrow too

> Third-quarter revenue of $19.2 billion set a new record and exceeded July guidance, driven by record data-centric revenue*, which grew 6 percent year-over-year (YoY). PC-centric revenue declined5 percent YoY,consistent with guidance.

> Third-quarter earnings-per-share (EPS) exceeded July guidance. GAAP EPS of $1.35 declined 2 percent YoY;non-GAAP EPS of $1.42 was up1 percent.

> Raising full-year revenue outlook to $71 billion, up $1.5 billion from July guidance. Now expecting full-yearGAAP EPS of $4.42 and raising full-year non-GAAP EPS outlook to $4.60.

> Intel today announced its board of directors has approved a $20 billion increase in its stock repurchase programauthorization.

Q3 2019 Results
.

• Revenue was $1.80 billion, up 9 percent year-over-year and 18 percent quarter-over-quarter due to higher revenue in the Computing and Graphics segment, partially offset by lower revenue in the Enterprise, Embedded and Semi-Custom segment.
• Gross margin was 43 percent, up 3 percentage points year-over-year and 2 percentage points quarter-over-quarter, primarily driven by increased RyzenTM and EPYCTM processor sales.
• Operating income was $186 million compared to $150 million a year ago and $59 million in the prior quarter. Non-GAAP operating income was $240 million compared to $186 million a year ago and $111 million in the prior quarter. The year-over-year and sequential increases were primarily due to higher revenue in the Computing and Graphics segment.
• Net income was $120 million compared to $102 million a year ago and $35 million in the prior quarter. Non-GAAP net income was $219 million compared to $150 million a year ago and $92 million in the prior quarter.
• Diluted earnings per share was $0.11 compared to $0.09 a year ago and $0.03 in the prior quarter. Non-GAAP diluted earnings per share was $0.18 compared to $0.13 a year ago and $0.08 in the prior quarter.
• Cash, cash equivalents and marketable securities were $1.2 billion at the end of the quarter.

Quarterly Financial Segment Summary
.

• Computing and Graphics segment revenue was $1.28 billion, up 36 percent year-over-year and sequentially. Higher revenue was primarily driven by increased Ryzen client processor sales.
• Client processor average selling price (ASP) increased year-over-year primarily driven by Ryzen desktop processor sales and increased quarter-over-quarter driven by both Ryzen desktop and mobile processor sales.
• GPU ASP increased year-over-year driven by higher channel sales and decreased quarter-over-quarter due to a higher proportion of mobile sales.
• Operating income was $179 million compared to $100 million a year ago and $22 million in the prior quarter. The year-over-year and quarter-over-quarter increase was primarily due to higher revenue.
• Enterprise, Embedded and Semi-Custom segment revenue was $525 million, down 27 percent year-over-year and 11 percent sequentially. The year-over-year and quarter-over-quarter decreases were primarily due to lower semi-custom product revenue, partially offset by higher EPYC processor sales.
• Operating income was $61 million, compared to $86 million a year ago and $89 million in the prior quarter. The year-over-year and quarter-over-quarter decreases were due to lower revenue and higher operating expenses.

Current Outlook

For the fourth quarter of 2019, AMD expects revenue to be approximately $2.1 billion, plus or minus $50 million, an increase of approximately 48 percent year-over-year and approximately 17 percent sequentially. The year-over-year and sequential increases are expected to be driven by an increase in Ryzen, EPYC and Radeon product sales. AMD expects non-GAAP gross margin to be approximately 44 percent in the fourth quarter of 2019.

Rosenblatt Securities analyst Hans Mosesmann increased his price target on shares of Advanced Micro Devices Inc. AMD, +0.30% to $52 from $42 on Wednesday, with the new target representing the highest listed on FactSet by $8. The target bump comes after AMD reported in-line results on Tuesday afternoon and gave a forecast that was weaker than anticipated at the midpoint. Mosesmann remains confident about the company's potential. "AMD saw no pull-ins from 4Q to 3Q, is comfortable with CPU availability, and is seeing most of the demand at the high end of desktops and high-end of servers (48/64 core)," he wrote. "The setup for the company is for continued momentum into 2020 on broad based desktop and server 7-nanometer deployments, continued 12-nanometer notebook traction (7 nanometer next year), and a new game console cycle that is a bit stronger than we have been modeling." AMD shares are off 1.6% in Wednesday trading. They've gained 76% so far this year, as the S&P 500 SPX, +0.33% has risen 21%.

New Azure Dav4-series and Eav4-seriesvirtual machines are now available.

Posted on Monday, November 4, 2019


New Azure Dav4-series and Eav4-series virtual machines based on AMD EPYC™ are now available.

The Dav4-series and Das v4-seris are well suited for general purpose workloads. These new Azure VMs feature the latest AMD EPYC™ 7452 processor and provide up to 96 vCPUs, 384 GiBs of RAM and 2,400 GiBs of SSD-based temporary storage. The Dav4 and Das v4 VMs offer great performance at competitive price points and can run enterprise-grade applications, relational databases and application servers. The Das-series VMs support Azure Premium SSDs and will support Ultra Disks in the near future.

The Eav4-series and Eas v4-series are ideal for memory intensive workloads. Like their Da-series counterparts, these new Azure VMs feature the latest AMD EPYC™ 7452 processor and provide up to 96 vCPUs, 672 GiBs of RAM, and 2,400 GiBs of SSD-based temporary storage. The Eav4 and Eas v4 VMs offer great performance for large in-memory business critical workloads at competitive price points. The Eas-series VMs support Azure Premium SSDs and will support Ultra Disks in the near future.

To learn more about regional availability and overall capabilities, please read the documentation.