Skip to main content

One of the internet’s largest networks chooses AMD CPUs over Intel’s

AMD just picked up a big win. Cloudflare, one of the largest networks powering the internet today, announced that it will continue using AMD’s Epyc server CPUs instead of Intel Xeon for its next generation of servers. For these new servers, Cloudflare is moving from second-generation Epyc Rome processors to third-generation Epyc Milan processors.

In a blog post describing the testing processor, Cloudflare’s Chris Howells said that the competing Intel chips “did not meet our requirements.” Cloudflare tested Intel’s Ice Lake Xeon processors against the latest AMD Epyc chips and found that the power consumption was just too high.

AMD Epyc processor in a server.
Image used with permission by copyright holder

“Although Intel’s chips were able to compete with AMD in terms of raw performance, the power consumption was several-hundred watts higher per server — that’s enormous,” Howells wrote. The blog post also described Intel’s performance per watt as “unattractive.”

Cloudflare switched to AMD chips with its 10th generation of servers. Previously, the company used dual-socket systems with two Xeon processors. In 2020, it switched to single-socket systems using AMD Epyc, which it’s continuing to do with the 11th-generation servers.

Although Cloudflare is announcing the switch now, the company has been working on the new servers since mid-2020. The network touches over 200 cities around the world, so as Howells writes, “it’s essential to get things right the first time.” The new servers are being updated with the 64-core AMD Epyc 7713 CPU, over the 48-core Epyc 7642.

Cloudflare noted a 29% increase over the last-generation chip, while consuming about the same amount of power. The new servers are getting a few other updates as well, including two 2TB SSDs instead of three 1TB SSDs, as well as 384GB of DDR4-3200 memory rather than 256GB of DDR4-2933.

Although AMD won this generation of servers, that could change. Intel’s Ice Lake Xeon platform is a few years old, and the company is working on its next-generation Sapphire Rapids platform for 2022. The upcoming chips will be built using Intel 7, which is the same manufacturing process behind Intel’s upcoming Alder Lake processors.

Intel has struggled to maintain its lead in the desktop and server market over the last few generations, largely due to high thermal and power requirements. Alder Lake and Sapphire Rapids feature a smaller manufacturing process, which should help with thermal and power demands.

However, numerous delays have brought the new manufacturing process into question. Sapphire Rapids chips are planned for use in the Aurora supercomputer at the Argonne National Laboratory. If they hold up, maybe we’ll see Intel show back up in the data center.

Right now, AMD is making inroads into that space. The Epyc processors largely mirror Threadripper processors seen on high-end desktops like the Lenovo P620, boasting high core counts with relatively low power demands.

Editors' Recommendations

Jacob Roach
Senior Staff Writer, Computing
Jacob Roach is a writer covering computing and gaming at Digital Trends. After realizing Crysis wouldn't run on a laptop, he…
The iPhone 15’s chip challenges Intel’s fastest desktop CPU — but there’s a catch
Intel Core i9-13900K held between fingertips.

Who would have thought that some of the best CPUs would face competition not from a desktop or laptop CPU, but from a mobile system-on-a-chip (SoC)? Well, the latest Geekbench 6 scores prove that it's possible. Apple's new A17 Pro chip, announced during the September 2023 Apple event and found in the iPhone 15 Pro and Pro Max, challenges AMD and Intel -- but there's a catch.

Yes, it's real. The Geekbench 6 test gives the A17 Pro chip a score of 2,914 in single-core operations, and that's mighty impressive for something that will end up in a smartphone. However, the generational leap is not that impressive -- the last-gen A16 Bionic chip is only around 10% behind in terms of single-threaded performance. The A17 Pro was built based on TSMC's 3nm technology, while the A16 Bionic is a 5nm chip, also made by TSMC.

Read more
AMD might still have some next-gen GPUs left in the tank — but I don’t buy it
AMD RX 7600 on a pink background.

AMD has said it's done with new GPU dies, but a filing with the European Economic Commission (EEC) suggests that Team Red could still launch graphics cards in its RX 7000 range.

The filing points to AMD releasing an RX 7600 XT sometime in the future, both in 12GB and 10GB variants. AMD's Scott Herkelman says the RDNA 3 lineup is "complete," so what gives? It comes down to the GPU dies.

Read more
I got a rare peek into Intel’s largest offshore manufacturing facility — here’s what I learned
An inside look of a test cell that goes into a sorting module at the Intel Penang facility in Malaysia.

It is rare to see a technology giant open its doors to showcase how it really operates. But that is exactly what Intel had in mind when it invited global press and media representatives to visit its factories in Malaysia. As a part of its Tech Tour 2023 event, the chip maker gave us a detailed insight into how exactly a CPU is made. I learned a lot, but before I share my experience, here’s a brief about the company’s history and future plans in Malaysia.

Intel established its first offshore facility in Malaysia in 1972 with a total of 100 employees. Today, there are two major facilities in Penang and Kulim that expand close to 900,000 sq. ft. with 15,000 employees.

Read more