"The power of the PC is that it’s an open ecosystem while lot of other platforms like to do walled gardens in their controlled ecosystems."
Rocket Lake represents Intel’s first “new desktop architecture” in about 5 years.
Intel is officially launching its 11th Gen Intel Core S-series desktop processors (codenamed “Rocket Lake-S”) worldwide today. But unless you’ve been living under a rock or something, you will most probably know some units of these processors have already been sold at retail, purchased, and reviewed “extensively” before the sales date.
“Do I think that all these reviews are putting Rocket Lake in the best light? Well, no,” Intel’s Marcus Kennedy tells Financial Express Online. “Anything that you have read prior to this article coming out on March 16, isn’t going to be showcasing the best of Rocket Lake.”
Because, as Kennedy puts it, “it’s not standardized configuration and not really how people are going to be using it. When the product actually comes out, and people play with the enhanced BIOS and all the updates, the right chipsets, and the right motherboards, the truth of the product will come out and we think that this is a really exciting product.”
Rocket Lake represents Intel’s first “new desktop architecture” in about five years. The codename of this core architecture is Cypress Cove. By re-building two different backported technologies, the Sunny Cove core from (10nm) Ice Lake and Xe graphics from (10nm) Tiger Lake on 14nm, Kennedy says Intel has hit the “sweet spot” bringing up to 19 percent gen-over-gen instructions per cycle (IPC) improvement for the highest frequency cores and up to 50 percent better integrated graphics performance. The flagship processor in the lineup, the Intel Core i9-11900K (8 core, 16-thread), can reach speeds of up to 5.3GHz with Intel Thermal Velocity Boost.
FE: Why did it take five years between architectures?
Marcus Kennedy: With the previous architecture and with our world class engineers, we were able to really stay ahead of the pack and continue to bring generation-over-generation enhancements that really led the industry. The reason it took five years was because we didn’t need to change our architecture in order to continue to deliver the improvements (with our tick-tock) that gamers really wanted in this space. Now that we have a new architecture that actually gives that step function performance in addition to the Xe integrated graphics (which was just released earlier this year), we felt that now was the time to make that next leap, to continue onto the next performance step function.
FE: What’s your take on competition, AMD in particular—has it pushed you into switching gears?
Marcus Kennedy: Competition is always good. It is good for the market, but more than anything, it is good for us. We as a company just do better when we have better competition. Better competition makes us more focused. We are a competitive company so it absolutely does play a part. That said, our main focus always comes back to our customers. No matter who is playing in what market, no matter what competitors decide to do, we stay focused on delivering for the gamers and creators and the rest of our community. They want world-class performance all the time, they want great price-perf ratio, they want the ability to overclock, they want the ability to squeeze out as much performance as they possibly can.
FE: Could you take us through the new architecture and some of its major advantages?
Marcus Kennedy: It lets us take advantage of the new IP performance while maintaining the same kind of manufacturing efficiency that we have been able to gain since we have been on 14nm for a while. It’s a nice sweet spot of new core and graphics IP plus taking advantage of all of the high frequencies we have been able to tune in on the 14nm process. So, what you end up with is a part that is able to hit frequencies up to 5.3GHz with new integrated graphics and we are also able to bring in advanced AI features. Beyond the IPC and integrated graphics improvements, this brings PCIe Gen 4.0 to the market on desktop (it’s up to 20 CPU PCIe Gen 4 Lanes). It has up to DDR4-3200 support. There is new enhanced media 10-bit AV1, 12-bit HEVC and enhanced display with integrated HDMI 2.0 and HBR3. Then of course there are all the upgraded connectivity options like Wi-Fi 6E and Thunderbolt 4.
And then the chipsets that come along with that, that we’re also launching at the same time is the 500 Series. With the 500 Series you get double the USB connectivity speed with a USB 3.2 Gen 2X2 20 Gigs instead of the 10 Gigs that we had in the last generation. A couple of things that we are really excited about with the chipsets is that these actually help enable memory overclocking to the H and B-series. So, it goes beyond the Z-series which we have historically limited our memory overclocking to.
FE: Could you elaborate on this “overclocking’’ bit?
Marcus Kennedy: Besides the memory overclocking, we have a new integrated memory controller with Gear 2 with wider timing, we give the ability for real-time memory frequency. We are bringing AVX2 and AVX512 offsets and including an AVX disabled and enabled option. So, with all of those things and an updated extreme tuning utility, when you put all those things together, it allows overclockers to really have way more knobs in tuning ability to get the most performance out of their system.
FE: What are some of the tangible benefits you’d observe on a PC powered by Rocket Lake?
Marcus Kennedy: One is smoother gameplay experience, the second is greater performance efficiency and the third is increased workflow productivity. We are talking about games running in single threaded mode at the highest frame rates possible. We see double digit gains gen-over-gen on Microsoft Flight Simulator, which is an extremely complex game to run and a game like Total War: Three Kingdoms where you have a ton of characters on screen at the same time. With our integrated graphics, the workloads whether it’s gaming, content creation or whatever, it doesn’t necessarily have to travel over your PCIe and use that bandwidth and drive that latency as you move the workload. It can actually run locally, so, an example of this would be like 4K streaming on Hulu, Netflix, etc. You can’t actually do 4K streaming on a lower-end discrete graphics card, but you can in this Xe integrated graphics architecture, so it leads to greater performance efficiency because Rocket Lake then allows the workload to figure out the best place to run for the efficiency and the smoothness of experience.
With AI integration, we have Deep Learning Boost and VNNI commands and instructions. All this helps leverage AI inference in order to drive things like image classification, object detection, speech recognition, image and video upscaling. With Deep Learning Boost features, what that allows us to do (versus our previous generation), it’s like 88 percent higher video editing performance. When you talk about something like a Microsoft application, the higher single threaded performance in addition to some of those AI features can lead up to double digit performance gains generation-over-generation. We also have something called re-sizable BAR that allows the processor to access more of the discrete graphics memory, the VRAM on discrete graphics which reduces data transfer latency between the board and the VRAM. So, this is coming to market here obviously enabled through VBIOS and the motherboards but, assuming that you have an enabled system, this is actually going to be a really good feature to help reduce system latency as well.
FE: But what about the thermals?
Marcus Kennedy: The way we think about thermals is again depending on how you run real-world usages. There are people out there who want to turn the thermals up. If you run AVX512, plus overclocking, you are going to want an enhanced cooling solution in order to bring your system down. But people who are doing that, the technical overclockers, the experienced overclockers, people like that, they are going to know that, and they are going to have the right system. Outside of that, we are running within our standard thermal profiles that we talk about all the time.
FE: How relevant are benchmark scores?
Marcus Kennedy: We’ve been, over the last (little over a) year, been really moving towards what we call focusing on the real-world usages as opposed to synthetic benchmarks. Synthetic benchmarks have their usage. It’s a good way to try to look at standardized performance so that you can compare platforms to platforms with all the different features. You have to have a way to be able to compare and synthetic benchmarks have a place in that world. But at the end of the day, it really comes down to how people are using the products.
FE: How important is annual cadence of processors?
Marcus Kennedy: The two things that we do with all of our systems is, we try to make them future-proof and we know that our industry refreshes roughly between a 2- and 3-year cadence. So, we work with our developers to make sure that anything that they are bringing, whether it is ray tracing, whether it’s new and exclusive things that we try to bring to the market with them, that just has a 2 to 3-year window for us to make sure that we are optimized within on each platform. But that is always changing. That window is always moving. The reason we continue to drive an annual cadence of leadership products, which is extremely important to us, it is to make sure that we are moving with the market. Developers aren’t waiting for us. They are going to continue releasing new and better experiences and we have got to make sure that we offer the platforms that bring that experience to the market that demands it. So, is it necessary? We think it’s absolutely necessary in order to continue to keep up with the demands of the market.
*Intel is also launching a beta version of an anti-toxicity application with Rocket Lake that takes advantage of some of its platform features.
“We believe that the power and control of their own experience needs to be in the hands of the gamers and we think the PC is the best place to offer that. A lot of other platforms like to do walled gardens in their controlled ecosystems. The power of the PC is that it’s an open ecosystem, and it allows you the highest level of flexibility and customizability in addition to the highest levels of performance,” Kennedy reiterates–probably hinting at Apple, maybe?
And last but not the least, “We do take our embargo agreements seriously,” he quips.