Today, we will finally explore Kaby Lake G, one of the most unexpected and most interesting pieces of hardware coming from Intel in a moment. It's the very first chip that combines an Intel processor and an AMD GPU on a single piece of silicon, forming what is essentially the fastest APU-style processor on the market.
This is not quite the same as an APU or SoC, but it brings together a powerful processor and a high-performance GPU on a single compact chip.
You may have already seen some of Intel's Kaby Lake G NUC cover, otherwise known as Hades Canyon. This was the first product of Kaby Lake G to enter the market, but as NUC it is a niche product and most critics have access to the most upscale model with the Kaby Lake G SKU the faster.
Whereas Kaby Lake G is primarily designed for mobile devices like laptops, where high-end SKU is unlikely to be used, to test this chip, we really wanted to get their hands on the processor in a computer Real laptop as a more common SKU.
This is not quite the same as an APU or SoC, but it brings together a powerful processor and a high-performance GPU on a single compact chip.
You may have already seen some of Intel's Kaby Lake G NUC cover, otherwise known as Hades Canyon. This was the first product of Kaby Lake G to enter the market, but as NUC it is a niche product and most critics have access to the most upscale model with the Kaby Lake G SKU the faster.
Whereas Kaby Lake G is primarily designed for mobile devices like laptops, where high-end SKU is unlikely to be used, to test this chip, we really wanted to get their hands on the processor in a computer Real laptop as a more common SKU.
The processor we have on hand to review is the Core i7-8705G, which is one of the 65W Kaby Lake-G variants, as opposed to the 100W full-featured units seen in the NUC. There are already two laptops that use the 65W model, and if other laptops use Kaby Lake-G, it is much more likely that they will choose a 65W model like the one we are testing today. ; hui. So the benchmarks we'll have in a little bit should provide a good idea of Kaby Lake G's performance for most users.
We tested the 8705G inside the Dell XPS 15 2-in-1, a 15-inch thin and light laptop that also contains 16GB of dual-channel DDR4-2400. kinds of thin and light laptops, again, we will have a good idea of how the chip works in a typical scenario of use.
Let's explore Kaby Lake G, the line-up and its features before heading to performance numbers. The chip itself contains both an Intel processor and an AMD graphics processor on the same package, but it's not a combination of both in the single die. Instead, there are two dies on the package: one is an Intel quad-core processor with integrated HD Graphics and the other is a Radeon RX Vega M GPU with up to 24 computing units, with 4 GB of HBM2 memory. ]
Both matrices are connected via eight lines of PCIe 3.0, and the GPU is connected to its HBM2 via what Intel calls the Embedded Multi-Die Interconnect, or EMIB. You can see on this image exactly how the arrays are arranged on the chip: the Intel processor is on the right, while the graphics processor AMD and HBM2 sit on the left. The HBM2 chip is separate from the GPU, as you can see, and it's between these chips that the Intel EMIB comes in.
There are a total of five SKUs from Kaby Lake G, perfectly separated variants with a TDP of 100W and those with a TDP of 65W. The SKU 100Ws use the Vega M "GH" graphics, which is a fully unlocked version of the GPU with 24 computing units and a base clock of 1063 MHz that increases to 1190 MHz. The 65W variants use Vega M "GL" graphics, which reduce the GPU to 20 compute units and lower the clock frequency to 931 MHz and 1011 MHz. This is the main difference between 100W and 65W chips; additional power reserve gives you more CU and higher clocks for better GPU performance.
The processor is pretty much the same for all five references. We examine four cores and eight threads, with minor variations in clock speeds between models. The Core i7-8705G we are looking at today is clocked at a base of 3.1 GHz, with a maximum Turbo frequency of 4.1 GHz on a single core. This drops to 3.9, 3.8, and 3.7 GHz, respectively, in two, three, and four-core workloads.
The same processor is used in the 8709G, but this chip includes the high-end variant 24 CU Vega. The high-end 8809G slightly increases the clock speed of the processor from there. And at the lower end we have the Core i5-8305G, which offers lower CPU clock speeds and less L3 cache, but the same GPU 20 CU Vega as the 8705G we are looking at today. 39; hui.
The Core i7-8706G is the only one out, in that it is exactly the same as the 8705G from the point of view of specifications, except for the business features of Intel such as vPro and Trusted Execution Technology. The 8809G is the only overclockable chip.
For HBM2 configurations, all Kaby Lake G chips are equipped with 4 GB on a 1024-bit bus, but 100W SKUs have 800 MHz memory, compared to 700 MHz on 65W models. All Kaby Lake G SKUs also come with Intel's HD Graphics integrated into the CPU chip; it is the standard HD 630 GPU clocked at 1100 MHz. The idea is that when you're on battery, the laptop can completely turn off the AMD GPU, more energy-hungry, and simply use built-in graphics. When the discrete AMD GPU is required, it can be powered up for much better graphics performance.
The last thing we want to discuss before entering performance is the power split between the CPU and the GPU. The 8705G has a TDP of 65W in total, however the PL1 power limit for the CPU - in other words, the sustained power consumption limit for the CPU - is capped at 47W in the Dell XPS 15 2- in-1. That's quite logical, since Intel's four-core Kaby Lake H-series processors had a 45-Watt TDP and Kaby Lake G increased frequencies to about 300 MHz.
This gives the GPU, at the very least in extended workloads, a full 18W to work with. In practice, when the GPU and CPU are fully utilized, the GPU will operate above 18W and the CPU below 47W as needed. As with any SoC, the algorithms determine the amount of energy allocated to each part of the chip depending on the workload. But you can see just from the PL1 limit that there is still some room for the GPU even under a full CPU load.