« April 2022 | Main | April 2023 »
October 3, 2022
Was Intel 11th generation "Tiger Lake" the milestone we thought it was?
In the wonderful world of technology, there are few things where progress is as mind-blowingly fast as in electronics. Today you can get an iPhone with 100,000 times the storage capacity of the hard disk in an early IBM PC. And the clockspeed of the CPU in that early PC was a thousand times slower than that in a modern PC. That incredible pace of technological progress has revolutionized the world and our lives, opened up new opportunity and made things possible that weren't even dreamed of just a few decades ago. But that progress also means greatly accelerated obsolescence of the computing products we're using.
That electronic obsolescence would't be so bad if the software on our computers remained more or less static, but that's never the case. Every increase in computing power and storage capacity is quickly soaked up by more complex and more demanding software. Loaded with current software, a state of the art PC of five years ago is slow, and one built ten years ago is barely able to boot.
While the rapid advance is good news for many consumers who, even at a cost, love to have the latest and greatest and don't mind getting new gear every year or two, it's bad news for commercial, industrial and government customers that count on significantly longer life cycles. For them, rapid obsolescence either means extra cost to stay up to date, or fall behind, sometimes hopelessly so.
I wrote about that predicament some time ago in a post called "Intel generations" -- how the rapid succession of Intel Core processor families has made obsolescence an ongoing, costly problem.
And one that isn't going away anytime soon. That's why for manufacturers and consumers alike it's always great when a technology comes along that won't quickly be obsolete or replaced by something different. But that doesn't happen often, and so it's good to at least have milestones -- things that, while not representing the be-all and end-all, at least are here to stay for a while, giving customers a temporary reprieve from rapid obsolescence.
With Intel processors, that's a "generation" that's particularly good, particularly solid, and unlikely to be rendered obsolete at least for a bit longer. Instead of a year, maybe three years or four.
Intel's 6th generation "Skylake" was such an example. Skylake was both the last generation of Core processors that still supported earlier versions of Microsoft Windows, as well as having a microarchitecture that remained in use until the 11th generation. 8th generation "Coffee Lake" was another such milestone, being the first that brought quad-core processor to mobile computers -- a big step forward.
But it was the "Tiger Lake" 11th generation that (so far) trumped them all with the first new microarchitecture since Skylake, the long awaited switch to 10nm process technology, scalable thermal design power, Intel Iris Xe integrated graphics, Thunderbolt 4 support, and more. What made Tiger Lake so special? Well, for the first time ever, Intel allowed manufacturers to "tune" processors to optimally match their hardware as well as the requirements of their target customers.
Whereas prior to the 11th generation, mobile Core processors were delivered with a set default TDP -- Thermal Design Power -- and could only be tweaked via the Power Plans in the OS, "Tiger Lake" allowed OEMs to create power plans with lower or higher TDP with Intel’s Dynamic Tuning Technology. No longer was the thermal envelope of a processor a fixed given. It was now possible to match the processor's behavior to device design and customers' typical work flows.
Most manufacturers of rugged mobile computers took advantage of that. Here at RuggedPCReview.com we began, for example, seeing a significant difference in device benchmark performance when plugged in (battery life not an issue, emphasis on maximum performance) versus when running on battery (optimized battery life and sustained performance over a wider ambient temperature range). Likewise, device design decisions such as whether to use a fan or rely on less effective passive cooling could now be matched and optimized by taking that into consideration.
As a result, virtually all of the leading providers of rugged laptops and tablet computers switched to Tiger Lake as upgrades or in entirely new designs, leading to unprecedented levels of both performance and economy. A true milestone had been reached indeed.
But not all was well. For whatever reasons, the new technologies baked into Intel's 11th gen Tiger Lake chips seemed either considered too classified and proprietary to reveal, or too complicated to be properly implemented. Even the hard-core tech media seemed mostly at a loss. Our extensive benchmarking showed that tweaking and optimizing seemed to have taken place, but not always successfully, and most not communicated.
A second technology integrated into Tiger Lake fared even worse. "Thunderbolt" had been a joint efforts between Intel and Apple to come up with a faster and more powerful data transfer interface. In time Thunderbolt evolved to use the popular reversible USB Type-C connector, combining all the goodness of the PCIe and DisplayPort interfaces, and also supporting the super-fast USB 4 with upstream and downstream power delivery capability.
In theory that meant that a true "one wire" solution became possible for those who brought their mobile computer into an office. Tiger Lake machines would just need one USB 4 cable and a Thunderbolt 4 dock to connect to keyboards, mice, external drives, two external 4k displays as well as charging via Thunderbolt 4, eliminating the need for a bulky power brick.
In practice, it sometimes worked and sometimes didn't. Charging, especially, relied on carefully programmed "power contracts" that seemed hard to implement and get to work. Sometimes it worked, more often it didn't.
Then there was the general confusion about which USB port was which and could do what and supported which USB standards. End result: Thunderbolt 4's potential was mostly wasted. We contacted Intel both with 11th gen power mode and Thunderbolt 4 questions, but, after multiple reminders, got nothing more than unhelpful boilerplate responses. And let's not even get into some Iris Xe graphics issues.
So was Intel's "Tiger Lake" 11th generation the milestone it had seemed? For the most part yes, despite the issues I've discussed here. Yes, because most rugged mobile computing manufacturers lined up behind it and introduced excellent new 11th gen-based products that were faster and more economical than ever before.
It's still possible that power modes and Thunderbolt 4 become better understood and better implemented and explained, but time is against it.
That's because rather than building on the inherent goodness of Tiger Lake and optimizing it in the next two or three generations as Intel has done in the past, the future looks different.
Starting with the "Alder Lake" 12th generation of Intel Core processors, there are "P-cores" and "E-cores" -- power and economy cores, just like in the ARM processors that power most of the world's smartphones. How that will change the game we don't know yet. As of this writing (October 2022) there are only very few Alder Lake-based rugged systems and we haven't had one in the lab yet.
But even though "Alder Lake" is barely on the market yet, Intel is already talking about 13th generation "Raptor Lake," available any day now, 14th generation "Meteor Lake" (late 2023), 15th generation Arrow Lake (2024) and 16th generation "Lunar Lake." All these will be hybrid chips that may include a third type of core, and with Lunar Lake Intel is shooting for "performance per watt leadership."
It'll be an interesting ride, folks, but one with very short life cycles.
Posted by conradb212 at 8:28 PM