10 nanometers are here: Intel’s 11th-generation Core is on the market, with a 20% performance increase, and AI computing power multiplied by 5 times

For notebook users, the end of this year seems to be a good time to switch phones.

The long-rumored 11th generation Core is officially released. Today’s launch of Tiger Lake heralds the official rollout of Intel’s 10-nanometer chips. This time, the first model of the new CPU is still a mobile version, and Intel positions it as “the best thin and light notebook processor in the world”.

In August of this year, Intel held a rare “Architecture Day” event full of dry goods, revealing some information about the upcoming Tiger Lake processors. Today, these products are announced for sale, and in the new generation of Core CPU, Intel provides faster speed and more powerful AI processing power. The performance of the notebooks that will be launched at the end of this year will surely improve a lot.

New Architecture: Towards Full Integration

Intel has told us before that Tiger Lake’s standard form is a quad-core, eight-thread processor with Intel’s latest manufacturing process and an upgraded graphics processing architecture.

The 11th-generation Core uses Intel’s latest-generation Willow Cove microarchitecture—a minor tweak compared to the Sunny Cove. Tiger Laker’s cores offer higher performance than ever before: the new CPU’s peak frequency reaches 4.8 GHz, a 20 percent increase compared to the previous generation’s 4.0 GHz. Originally, Intel’s strength was in single-core frequency, and now it seems to be stronger.

The frequency gain mainly comes from the improvement of the process. First of all, 10nm is finally here: it should be Intel’s 10nm++ process to be precise, which has just been renamed as SuperFin technology by Intel. This improvement in the manufacturing process is twofold: first, increasing efficiency and scale to achieve higher frequencies, but also ease of production. Therefore, the chips we finally see have higher frequency and performance than the previous generation, and the power consumption is the same as the previous generation.

In addition, these cores continue to support AVX-512 and Intel’s DL-Boost acceleration library.

In terms of graphics computing, Tiger Lake uses Intel’s next-generation graphics architecture, Xe-LP. In short, Xe-LP increases the raw number of threads and computation per execution unit (EU), and improves other details such as cache structure. On Ice Lake processors we can have 64 EUs running at 1100 MHz, while on Tiger Lake we will see 96 EUs (+50%) running at 1350 MHz (+22%). Combined with other improvements, we can see that Tiger Lake has doubled the graphics processing power of the previous generation.

In AI computing, Intel’s integrated graphics also support the inference of DP4A instructions at INT8 precision.

In addition to the speed boost, the graphics capabilities of Intel CPUs have also been improved, with Tiger Lake supporting AV1 decoding, and a Display pipeline for up to four 4K 60Hz monitors or a single 8K 60Hz monitor. Tiger Lake supports memory including LPDDR4X-4267 (up to 32 GB), DDR4-3200 (up to 64 GB), and LPDDR5-5400. For the acceleration of artificial intelligence inference, Intel has proposed Gaussian Neural Accelerator version 2.0, which improves efficiency by eliminating noise in inference tasks.

Other improvements to Tiger Lake include support for native Thunderbolt 4, the controller embedded into the CPU, and support for up to four Thunderbolt 4 ports per device. Intel also provides Wi-Fi 6 support through the CNVi interface. Intel’s power/frequency algorithms have also been updated to scale power and clock speeds across the CPU, GPU, and memory fabrics, respectively. Intel says the new CPU has doubled the internal fabric bandwidth, allowing more data to communicate between all these parts.

The 11th-generation mobile version of the Core processor further improves the level of chip integration, with a smaller PCB area and higher efficiency. That’s good news for thin and light laptops.

In addition, Intel claims that the Core i7-1185G7 Tiger Lake system leads the performance in the following areas compared to the AMD Ryzen 7 4800U system:

28% higher computing performance (SYSmark 25);

67% higher graphics performance (3DMark Fire Strike);

When using Adobe Premier Pro to encode HEVC video, the speed is 200% faster.

With Intel’s deep learning accelerated instruction set, Tiger Lake provides the first instruction set DP4a for neural network inference on integrated graphics and the first native support for the INT8 data type, which can improve AI performance by up to 5 times.

Power and Specifications

Intel’s Tiger Lake announcement today is different from the past by focusing more on end-user performance, which is the speed and benchmark numbers that end-users often use to compare with other products on the market. Additionally, Intel partners are now free to announce future Tiger Lake-based products.

Next, we will focus on analyzing the processor details. Intel first introduced two types of Tiger Lake chips, ranging from 7W to 28W. The picture below shows the UP3 part of Tiger Lake:

These mid-to-high-end 12-25W processors are technically called “UP3” processors, and we usually call them U-series. While these processors have a stated TDP of 15W, notebook manufacturers can also turn the power up or down depending on how they build their systems.

Here, the top end is the Intel i7-1185G7, a core processor with hyperthreading and 12MB of L3 cache. Additionally, the Intel i7-1185G7 has a base frequency of 3.0 GHz, a single-core turbo of 4.8 GHz, and an all-core turbo of 4.3 GHz. For the graphics card, it has 96 EU available and runs at a peak frequency of 1350 MHz. Memory supports LPDDR4X-4266 and DDR4-3200.

Intel divides processors by core count, graphics card, and frequency. Any processor listed as “G7” means having the highest level of graphics card, but it’s important to note that there’s still a difference here, 96EU for the Core i7 part and 80EU for the Core i5, and less L3 cache. Next is the G4 graphics card, which has only 48 execution units.

Only the top-of-the-line Core i7-1185G7 has the highest graphics frequency. The Core i3 processor has reduced memory support, and the lowest-end Core i3-1115G4 has only two cores.

Further down, the 7-15W processor is the previous “Y-series” processor, now called UP4.

The top end is the Core i7–1160G7, which has 4 cores with hyperthreading and 12MB of L3 cache. Additionally, the Core i7-1160G7 has a base frequency of 1.2 GHz, a single-core turbo of 4.4 GHz, and an all-core turbo of 3.6 GHz. The 96 operating units run at 1.1 GHz, but only support LPDDR4X memory up to 4266 MT/s. All processors support 4 PCIe 4.0 lanes.

There is only one Core i7 processor in this power range, the remaining Core i5 has fewer EUs and lower frequencies, and the Core i3 is divided into quad-core and dual-core processors that only support 48EU.

The power of these processors is set in the UP4 range mainly because the base frequency is very low, which has also been a major feature of Intel’s low-power processors for many years.

The new generation of processors is targeting fanless designs ranging from around 9W in the UP4 power envelope, to models up to 28W, and even with discrete GPUs. As with Intel’s 10th Gen processors, users want to see Tiger Lake designed in the 25-28W power range, with the highest performing processors and guaranteed higher power modes. This will allow products with Tiger Lake processors to provide better sustained performance after the turbo is over, such as rendering and video encoding.

Intel says Tiger Lake has a better turbo algorithm that allows the system to boost the CPU/GPU frequency as needed during the upscaling window. The entire system is a feedback loop and inputs the workload, on-system sensors, power limits and balancers to generate the correct frequency.

Intel’s “TDP” value, also known as Power Limit 1, keeps the power between 7W and 28W depending on the system, and the power level can be much higher in turbo mode. Over the years, the secondary power level PL2 has increased to +4W and +20W in TDP. In the case of the previous generation Ice Lake, the 15W processor peaked at 44W in turbo mode.

What does this mean for Tiger Lake? According to Intel’s own data, peak turbo power can reach about 50W when plugged in or out. What Intel wants to emphasize to customers is that its turbo algorithm enables a consistent experience when the system is connected to a power source or running on battery alone.

Unlike previous-generation processors, Intel is able to test its own mobile development platform ahead of product launch, but it’s uncertain which Tiger Lake products will be the first to be tested in the lab.

New Logo Released

With the mix of X^e graphics cards and Tiger Lake, Intel also introduced a new product logo. This is the third time Intel has completely changed its logo.

Some of the Logos are shown in the figure below. In the new design, Intel decided that “Core” and “Iris” in the new Logo need to use mixed case, but “intel” is still completely lowercase, “POWERED BY” is completely uppercase, and the e in “X^e” is still superscripted .

Evo Certification Program

The Intel Project Athena Validation Program was introduced with the release of the previous generation and includes a set of validation criteria such as minimum battery life, Thunderbolt and Wi-Fi 6 support, good display, AI support performance and a certain level of responsiveness.

While Intel is releasing Tiger Lake, it is also heavily promoting Project Athena, calling any device that meets the latest updated specifications the “Intel Evo” certification program. Evo is clearly meant to call for an evolution of a new generation of products.

Specifications for Intel Evo certification.

Therefore, as long as the OEM’s notebooks are affixed with this label, it means that their products have passed the Intel Evo platform index test, which can ensure the user’s experience.

When will it be available to buy?

Finally, when will Intel’s 10nm notebooks be available?

Intel says we’ll see PCs based on Pentium, Celeron, and the new vPro later this year. The DG1 discrete graphics platform may also surprise us.

The newly released Tiger Lake mobile processors will appear in laptops from various manufacturers in the next few weeks. Currently known manufacturers such as Acer, Asus, Dell, HP, Lenovo, LG, MSI, Razer, and Samsung are expected to launch more than 150 notebooks equipped with 11th-generation Intel Core processors.

However, Intel still has some high-end CPUs that have not been released. At the Architecture Day in August, the company said that Tiger Lake can reach 65W, which means that there will be an 8-core processor in the plan. But so far, no news has emerged.

The Links:   6RI75E-080 LB070WV3-SD02