On Tuesday, Intel introduced a new microprocessor chip for enterprise servers and cloud data centers, said its performance is expected to have a greater increase than previously expected.
As a veteran chip business leader, in recent years, Intel's lucrative server chip business has also been a number of competitive manufacturers encirclement and suppression. Intel said its Xeon Scalable processor runs average server tasks at an average speed of 1.65 times the previous generation of processors. In order to improve the hardware speed, Intel also made some adjustments to the chip software. But the chip's performance jump rate is really so big, yet to be confirmed by external experts. However, according to Wall Street and the industry is expected, Xeon Scalable processor performance is indeed a very good upgrade, but its amplitude is not so much self-declared Intel's exaggeration.
"It represents the highest level of our 20-year history of data center innovation," said Lisa Sperman, vice president of Intel's Data Center Group, at Fortune on Tuesday. "She said that the previous generation The chip was launched in March last year. "This is quite great, and this will continue to bring breakthroughs to our customers."
However, from the Intel conference on Tuesday, several customer companies to speak, the release of the new chip to bring the performance improvement may not be so big. For example, AT & T chief strategy officer John Donovan pointed out that the use of a new generation of Xeon chips, their company's system performance increased by about 30%.
Intel in the enterprise server chip and cloud data center chip market share of more than 90%, but it is still facing fierce competition is fierce. Intel CEO Brian Cranick has said that even at the expense of profit margins, but also faster to improve the performance of the server chip.
Why the PC market is no longer diving
With the artificial intelligence and machine learning technology more popular, NVIDIA has also developed a more suitable for these technologies graphics processing chip. Last month, AMD also introduced a chip compatible with Intel hardware, named Epyc. It is said that it not only better performance, but also with energy saving and cost savings and other advantages. And some self-operation of the data center technology companies (such as Google) is also developing their own custom chip.
Investors are beginning to worry, especially in the first quarter of this year, Intel server chip business revenue growth of only 6%. On Monday, Jeffrey Ripesi, an analyst at Jefferies Group, gave a "sell" rating on Intel's stock. "As a leading player with a market share, we think (Intel) is the most likely to go downhill," he said.
Intel's share price has fallen by 7% this year. Intel released its new Xeon chip on Tuesday, its share price basically did not change. While Intel's main rival, NVIDIA's stock, has risen 44% this year, and the release of the Xeon chip has not had any impact on the stock. AMD shares rose 1%, from the beginning to the present, the stock has risen by 22%.
Intel said the performance of its new chips has increased so much, partly because of the addition of more fast connectivity between the different components of each chip - such as in processing cores and memory. This new "mesh architecture" and AMD's "infinite connection" architecture is also quite similar.
Although Intel did not officially launch the Xeon chip last Tuesday, but before that, it has about 50 million of the electronics chip sold to Google and some other large customers, and allows multiple companies to test the performance of the chip. In an event held in New York City last Tuesday, there are some companies that have tried this new chip to "come to the fore". Which Oracle's 12c database performance increased by 1.3 times, Baidu search engine performance increased by 1.7 times, Ericsson's video encoding speed increased by 1.5 times. All of these improvements are due to these new chips, thanks to the associated software upgrades.
Although Intel did not get the latest AMD Epyc chip, but Intel said that in the AMD last month revealed the performance test, Intel's new Xeon Scalable chip version of at least with the performance of Epyc chip equivalent or even stronger. AMD has said it will provide Epyc chip single slot products, which compared to Intel's dual-slot chip has a greater cost advantage.
Intel said it does not believe that AMD's single slot chip can fully meet the needs of the task. "The reason why they do so is to spare a single slot, not because the single slot is best suited to the customer's interest, nor is it because it can meet all the needs of the data center, but because it is in the benchmark performance test can get a good result. "
In the face of NVIDIA's aggressive on the graphics card, Intel has recently no longer call the gun. AMD has also recently launched a more powerful Vega graphics card. Although the Xeon chip can perform a number of tasks in the field of artificial intelligence, but for how strong back to the field of the challenges of the card, Intel seems to have not yet given the final answer. At present, Intel is consolidating its acquisition of last year's only two years of artificial intelligence chip maker Nervana Systems technology to develop a new chip called Lake Crest.
"Intel is investing in research and development for more optimized chips," said Skelman.