[ad_1]
If you have been following developments in the semiconductor industry, especially over the past few weeks, it would be easy to see how some people may think that Intel (INTC) is in serious trouble. After all, several signs are not promising.
First, although Intel reported strong second-quarter earnings a few weeks ago, the stock fell when the company revealed another issue and delay for its 7nm manufacturing process. Second, in this context, Intel recently lost its crown of the largest and most valuable semiconductor company – a place it held for decades – to Nvidia (NVDA). In fact, by the market close on Aug. 17, the heavily loaded Nvidia had exceeded a market cap of $ 300 billion, compared to Intel’s $ 208 billion. Finally, from a processor performance perspective, it was AMD Inc. (AMD) (which recently and briefly broke the $ 100 billion capitalization mark) that has beaten Intel almost everywhere on desktop, mobiles and even servers. For a company that has traditionally dominated in all of these areas, this has been a big change.
However, just when you start to think Intel is slipping, the company may surprise you. And that’s what he did, both on Architecture Day last week and this week’s Hot Chips conference. The Architecture Day event, in particular, offered a comprehensive overview of a number of innovations that Intel is working on in the areas of process technology, chip packaging, chip architectures basic, high-speed chip-to-chip interconnection, security and more. Above all, while some of these developments fall within the more theoretical realm, the vast majority will bring practical and real performance improvements which should translate into offers that will be very competitive compared to the latest innovations from AMD.
Intel has gone back to its roots in semiconductor manufacturing to deliver what appear to be some very promising improvements in core transistor technology. Until just a year or two ago, Intel was able to use its long history of chipmaking to its advantage, making steady improvements that other chipmakers often took 18-24 months to catch up on. . As process delays and other manufacturing challenges began to pile up, however, it began to appear as if the company’s in-house manufacturing was becoming more and more of an albatross around its neck. With the launch of 10nm SuperFin technology, the company is a clear reminder to the rest of the industry that it has a number of different tricks up its sleeve when it comes to basic semiconductor capabilities. More importantly, these advancements can still be used to make meaningful progress in areas that companies that don’t manufacture their own chips cannot access until their manufacturing partners develop them.
By harnessing new metal capacitor technology in conjunction with the company’s improved FinFET transistors, Intel claims that the 10nm SuperFin technology allows it to conduct higher currents, reduce resistance, and thus significantly improve performance compared to to traditional transistor designs. In practical terms, the first Intel party to use the new technology (a chip named Tiger Lake which is expected to be unveiled in early September) is expected to have a performance boost that the company claims to be the biggest “intranode” (which means on the same process size) the improvement it never had In fact, the company says it would be “comparable to a full node transition,” which is a pretty big statement – although that is also intended to counteract somewhat continuous delays up to node 7 nm. Real-world performance remains to be seen, but early commentary and analysis from those who fully understand the details of chip architectures is very promising.
At the same time, the company also recognizes that, as transistors get smaller and smaller and processing technologies get harder and harder, it makes sense to separate or ‘break up’ the advancements in design. chips from manufacturing upgrades so both elements can move forward. separate calendars. Combining that with the fact that Intel recently openly discussed the idea of using outside chip manufacturing plants, or foundries, to build some of its parts, highlights a very practical approach, arguably even more so. mature, of the situation. It also reflects a very different attitude for Intel – one that should help the company move forward with advanced chip designs, regardless of where (or by whom) they are ultimately built.
Intel has also provided a lot more information on its GPU Xe architecture, with a particular focus on how it can go from better integrated graphics – the first iteration will also be part of Tiger Lake – to several different types of discrete GPUs targeted at different markets. Its first stand-alone GPU, named DG1, is due out later this year, as is a new server-focused GPU. These two chips, along with the integrated GPU, will be based on what Intel calls the Xe-LP (low power) architecture. The company also unveiled a new Xe-HP (high performance) GPU architecture for other discrete chips focused on more advanced data center applications, such as AI acceleration, along with an optimized part. for the game that will support ray tracing.
Both at HotChips and Architecture Day, the company has also talked a lot about its software efforts, especially its OneAPI architecture, which is designed to make the programming process for any type of non-CPU accelerator (from GPUs to FPGAs in going through other AI-focused accelerators and beyond) much easier to do. It is clearly a monumental goal and a difficult task. However, the company appears to be making solid progress and has pointed out that software improvements alone can often lead to significant performance improvements.
The bottom line is that it is clearly too early to count Intel. If nothing else, the significantly more competitive environment he finds himself in seems to have inspired him to do some of his best work in years. The final details are yet to come, but the good news is that the company is clearly up for competitive challenges, and whatever vendors you choose to support, we’ll all benefit in the end.
Warning: Some of the author’s clients are technology industry suppliers.
Disclosure: No.
Editor’s Note: The bullet points for this article were chosen by the editors of Seeking Alpha.
[ad_2]
Source link