Intel likewise provided a terrific deal more info about its Xe GPU architecture, with a specific focus on how it will be able to scale from better integrated graphics – the first version will likewise become part of Tiger Lake – to several different kinds of discrete GPUs targeted at various markets. Its first standalone GPU, codenamed DG1, is anticipated later this year, as is a new server-focused GPU. Both of these chips, as well as the integrated GPU, will be based on what Intel is calling the Xe-LP (low power) architecture. The company likewise revealed a brand-new Xe-HP (high-performance) GPU architecture thats destined for other discrete chips concentrated on advanced datacenter applications, such as AI acceleration, in addition to a gaming-optimized part that will support ray tracing.
Editors Note: The summary bullets for this post were chosen by Seeking Alpha editors.
Simply when you begin to believe that Intels slipping, the company can surprise you. The Architecture Day occasion, in specific, provided a thorough look at a number of innovations that Intel is working on across procedure technology, chip product packaging, core chip architectures, high-speed chiplet-to-chiplet interconnect, security and much more.
Though Intel revealed strong second-quarter incomes a couple of weeks back, the stock tanked when the company revealed yet another problem and hold-up for its 7nm manufacturing process. Second, related to that, Intel just recently lost its crown as the biggest and most important semiconductor business – an area it held for years – to Nvidia (NVDA).
The bottom line is that its clearly way too early to count Intel out. The substantially more competitive environment in which it finds itself seems to have influenced it to be doing some of its finest work in years if absolutely nothing else. Final details are still to come, but fortunately is that the company is plainly up for competitive difficulties, and despite what vendors you pick to support, we will all benefit in the end.
Disclaimer: Some of the authors customers are suppliers in the tech market.
Disclosure: None.
At the very same time, the business likewise recognizes that as transistors get smaller sized and smaller sized and procedure innovations get harder and harder, it makes good sense to separate, or “disaggregate”, chip design advances from manufacturing improvements so that the 2 components can progress on different timelines. Combining this with the truth that Intel just recently honestly talked about the concept of utilizing outdoors chip manufacturing plants, or foundries, to develop a few of its parts highlights a really useful, perhaps a lot more mature, method to the circumstance. It likewise shows a very different attitude for Intel – one that need to assist the business progress forward on advanced chip designs, regardless of where (or by whom) theyre ultimately built.
Though Intel revealed strong second-quarter profits a couple of weeks back, the stock tanked when the business revealed yet another issue and delay for its 7nm production process. Second, related to that, Intel recently lost its crown as the biggest and most important semiconductor company – a spot it held for years – to Nvidia (NVDA). By leveraging a new metal capacitor innovation in combination with the businesss improved FinFET transistors, Intel declares that the 10 nm SuperFin technology allows it to drive greater currents, decrease resistance, and consequently considerably enhance efficiency over traditional transistor styles. In practical terms, the very first Intel part to utilize the brand-new technology (a chip codenamed Tiger Lake thats anticipated to be unveiled in early September) is anticipated to have an efficiency boost that the business declares is the biggest “intranode” (significance on the same procedure size) improvement it has ever had. It likewise reflects a really different attitude for Intel – one that need to help the business progress forward on innovative chip designs, regardless of where (or by whom) theyre ultimately constructed.
If youve followed developments in the semiconductor industry, particularly over the past couple of weeks, it would be easy to see how some individuals may think Intel (INTC) remains in major difficulty. After all, several indications have actually not been guaranteeing.
Intel returned to its roots in semiconductor manufacturing to deliver what looks to be really promising enhancements in basic transistor technology. Till simply a year or two earlier, Intel was able to utilize its long history of expertise in chip making to its clear advantage, bringing out routine improvements that other chip makers would typically take 18-24 months to catch up to.
Both at HotChips and at Architecture Day, the company likewise talked a lot about its software application efforts, especially its OneAPI architecture, which is developed to make the procedure of shows for any type of non-CPU accelerator (from GPUs to FPGAs to other AI-focused accelerators and beyond) a lot easier to do. This is plainly a huge goal and a difficult job. Nevertheless, the company appears to be making strong development and highlighted how software improvements alone can frequently result in significant performance enhancements.
By leveraging a brand-new metal capacitor technology in conjunction with the businesss boosted FinFET transistors, Intel claims that the 10 nm SuperFin innovation enables it to drive higher currents, minimize resistance, and thus substantially improve performance over conventional transistor styles. In practical terms, the very first Intel part to utilize the brand-new technology (a chip codenamed Tiger Lake thats expected to be unveiled in early September) is expected to have an efficiency increase that the company declares is the largest “intranode” (significance on the exact same procedure size) enhancement it has ever had. In fact, the company states it would be “comparable to a full-node shift”, which is a pretty big declaration – though its likewise indicated to rather counter continued delays to the 7nm node. Real-world efficiency remains to be seen, however initial commentary and analysis from those who thoroughly understand the details of chip architectures are extremely appealing.