In a sign that Intel's efforts to rebuild its engineering corps might be swift under the new incoming CEO, Glenn Hinton, the lead architect of Intel's Nehalem processors and several other programs, announced that he is returning to the company after a three-year retirement. In a LinkedIn post announcing his decision (opens in new tab), Hinton cited the return of Pat Gelsinger, Intel's new CEO, as a prime motivator for rejoining the company.
Hinton also remarked that he will be working on a high performance CPU project, and "If it wasn't a fun project I wouldn't have come back. As you know, retirement is pretty darn nice." (Hat tip to @DylanLJMartin for the spot.)
Intel has been criticized for losing much of its top engineering talent over the last several years, with activist hedge fund Third Point recently penning a letter to Intel decrying its loss of semiconductor leadership, saying, "The Company has lost many of its most inspiring and talented chip designers and leaders, and our sources indicate that those who remain (several of whom are highly regarded in the industry) are becoming increasingly demoralized by the status quo."
The majority of Intel's slow erosion of talent hasn't been as outwardly visible as public-facing figures like Jim Keller and Murthy Rendunchintala that have recently left the company. Rebuilding the company's technical backbone will be among incoming CEO Gelsinger's most pressing priorities as he seeks to re-establish the company's dominance in the semiconductor market, and bringing back experienced hands like Hinton is a good start.
Intel CEO Bob Swan, who was roundly criticized for his lack of engineering experience, will leave the company next month. Ex-VMware CEO Pat Gelsinger, who previously had a 30-year stint at Intel, will replace Swan, marking the return of an engineer to Intel's top spot. Gelsinger served as Intel's first Chief Technical Officer, developed the company's tick-tock methodology, and helmed the creation of 14 generations of Xeon and Core processors.
Hinton, an Intel Senior Fellow, also has an impressive track record with the company, having been one of three senior architects for the P6 design that led to the Pentium I, II, and III processors. He later led the development of the Pentium 4 CPU and was one of two lead architects of the i960 CA, the first super-scalar processor. Hinton was also the lead architect of the Nehalem processors, and after 34 years at Intel, has more than 100 patents spanning 8 CPU microarchitectures. Hinton holds bachelor's and master's degrees in electrical engineering from Brighton Young University.
As with any CPU design, the development and validation cycles often take several years, meaning it could be some time before we see Hinton's next design come to market. It's also possible that Hinton could intercept a design already in flight, so we could see his next project come to market sooner.
It really took all the fun out of it.
Nehalem was decent but not fantastic. IPC was better than like Conroe, sure. But their first attempt at an IMC baked onto the CPU was a bit "half baked" since they were pretty weak. Also the core voltage was tied to the memory voltage, so overclocking your CPU could cause you to overvolt the already weaker imc.
Plus, as already said, the issue isn't design, its the node.
I would say a lot of the reason I don't think Nehalem was very good was what came after it. My PFP is an Bloomfield system with an i7 940 and X58 sabretooth. Compared to my i5 2500k system, the 940 is honestly worse, despite the hyperthreading support. The IPC is noticeably worse, and the i7 940 does not overclock anywhere close to as high as my 2500k. Now an i7 980X vs a 2500k, that would be an interesting comparison...
I thought that's what people hated about intel and if the leaks are true intel is already doing this with RL even without a new node.
The 2500K was a new uArch and it was also 32nm vs 45nm. Of course it was going to clock and overclock better. The only benefit for Nehalem was the extra memory channels and bandwidth.