Beijing on January 3, 2018, last Wednesday, Intel introduced the rumored processor for Intel CPU + AMD GPU to a handful of media (that's right, from February 2017 onwards CPU).
Intel & AMD jointly create this CPU officially code-named KBL-G, belonging to Intel's eighth-generation Core Duo processor family, but also Intel's first G suffix processor.
To be honest, from last February when I heard Intel and AMD jointly create the news of the CPU is not too dared to believe that, after all, Intel and AMD are two competitive relationship, and imagine who will walk with competitors?
But the fact is the case, two chip competition for many years the chip oligarch finally walk together in 2018, can be considered a milestone in the CPU.
Since Intel, AMD alliance has become a fact, let's take a look at this mystery KBL-G what exactly it.
First of all, KBL-G belongs to the eighth-generation Core Duo processor family, but KBL-G's code prefix is the seventh generation Kaby Lake, I guess the KBL-G should still be used in the previous generation of 14nm + process (and 8 generations of core Kaby Lake Refresh U is a routine), this conjecture has also been confirmed by Intel technicians.
In fact, KBL-G processor is not the first time to meet with us, before it has technical interpretation related to it, such as technology manufacturing day March 28 last year, Intel introduced the EMIB technology (used to connect Radeon RX Vega M graphics card and HBM2 memory), followed by the November 6 GPU & HBM introduction.
KBL-G's target population is gamers, content creators, and VR / MR users, and these target audiences seem to have been valued by Intel all along.
On the KBL-G processor positioning Intel also gave a more clear description, G series processor is between U and H, U series is the mainstream mobile terminal (more common in thin and light), H series is high performance (more common in the game), G series both U series of thin, low power and H series of high-performance features, of course, thanks to AMD's Vega GPU.
KBL-G processor chip layout are also exposed before almost, and from the PPT point of view, On the left is the AMD Radeon RX Vega M graphics card, 4GB of HBM2 memory on the left of the graphics card and 8 generations of Intel Core CPUs on the right.
Intel also mentioned several firsts, including the first 8 generations of Core 2 Duo H Series (KBL-G should not be the G Series?), The first consumer-level solutions using EMIB technology, the first consumer-level mobile solutions used HBM2 memory and so on.
CPU, GPU, HBM2 memory are provided overclocking, and 65W and 100W two specifications, equipped with KBL-G processor devices will meet in 2018 Q1, PPT Mentioned Dell, Hewlett-Packard and Intel's NUC.
AMD Radeon RX Vega M graphics card and HBM2 memory through the EMIB technology part of the connection, GPU and CPU through 8 PCIe 3.0 connection, The height of the CPU and GPU parts are controlled at 1.7mm.
Benefit from the use of HBM2 memory, KBL-G processor compared to the traditional CPU + GPU + GDDR5 memory layout can save more board area, Intel figures are 1900m ㎡.
Intel also mentioned a key technology, The reason why the KBL-G processor can be so thin is because it uses Dynamic Power Sharing, a technology that allows OEMs to adjust the power consumption of the CPU and GPU sections based on product positioning.
Such as product positioning is to play the game, the GPU part of the power consumption can be higher, if the positioning is content creation, CPU part of the power consumption can be higher, of course, the total processor power consumption is unchanged.
At the same time, Intel's dynamic regulation can also reduce processor power consumption, the same performance up to 18% efficiency.
KBL-G processor integrated AMD Radeon RX Vega M graphics card while retaining the CPU's HD 630 nuclear display section, Intel technicians give the explanation is not to redesign the CPU.
In this GPU PPT, Intel introduced the Radeon RX Vega M and HBM2 memory in more detail, in addition to the above mentioned 4GB HBM2 memory also gives some key features, such as Memory bandwidth of 1024bit, GPU part of the highest 24 groups of computing units, Radeon display engine provides six video output interface, Intel Gfx display engine provides three.
KBL-G processor is divided into i7 / i5 two, PPT is given logo printed 8 generations of Core and Radeon RX Vega M graphics card, this is the first time the CPU and GPU are printed on the same label, and The two sides are competitors.
Intel also compared the laptop with the KBL-G processor to the notebook three years ago, and the notebook with the KBL-G processor clearly showed better performance in terms of lightness, weight, battery life, and 4K video support (3 years ago, the device is a heavy notebook, PPT pictures hung up ...)
The NUC was also taken in comparison with a desktop from three years ago, and the new NUC outperformed desktops in size, power, quietness and lightness. (It feels meaningless to compare, even the ITX host In these few items is not an opponent of NUC)
KBL-G processor in the Radeon RX Vega M graphics section also made a division, One is Radeon RX Vega M GL and the other is Radeon RX Vega M GH.
Equipped with Radeon RX Vega M GL graphics CPU frequency is the highest 4.1GHz, 4 core 8 threads, 8MB L3 cache, the video frequency range is 931-1011MHz, 20 CU units, memory is 4GB, memory bandwidth 179GB / s, TDP is 65W.
In productivity, games, transcoding, content creation, the KBL-G processor powered by the Radeon RX Vega M GL leads the i7-4720HQ + GTX 950M portfolio three years ago, And in comparison with the i7-8550U + GTX 1050 is still the leader in performance up to 1.4 times the GTX 1050 4GB, which is up to 40% higher than the GTX 1050 4GB.
Processor equipped with Vega M GH graphics card The CPU frequency is 100MHz higher than the Vega M GL at 4.2GHz, with the same core / thread size and cache size as the processor with the Vega M GL graphics card.
Vega M GH is higher than Vega M GL in GPU frequency, computing unit, memory bandwidth, The frequency range of Vega M GH is 1063-1190MHz, 24 groups CU unit, memory bandwidth is 204GB / s, TDP is 100W.
Equipped with Vega M GH graphics CPU and 3 years ago, the same notebook compared to this contrast to the configuration upgraded to a GTX 960M graphics card, but the KBL-G processor is still hanging, Including the killer, Total War: Warhammer, Tomb Raider, including masterpiece can get more than 60 performance, and in the lower right corner of PPT impressively labeled 1080P / 60FPS label.
Even worse, The KBL-G processor with Vega M GH graphics is still at the edge of the i7-7700HQ + GTX 1060 Max-Q with up to 1.13 times the performance of the GTX 1060 Max-Q, or up to the GTX 1060 Max-Q High 13%.
This is a little scary, the original need two separate chips to complete the thing, now only need a KBL-G can do it, and in terms of product size, power consumption is better than the former. In other words Smaller, but stronger.
KBL-G processor also supports VR / MR, Oculus and Windows Mixed Reality can get a good experience.
Another interesting fact is that the KBL-G processor supports AMD's latest Radeon Adrenalin Edition driver and Intel's graphical control panel driver. It is not surprising that the adrenaline driver, including Chill, ReLive and FreeSync, is working.
Intel's drivers here not only support overclocking, but also this belongs to Radeon WattMan power management tools moved over, it's a bit messy.
Many players look forward to the Skeleton Canyon will usher in the update, the processor used KBL-G certainly did not run, the specific model is i7-8809G, not only play the game cool, productivity and content creation is also easily competent, and scalability It is also very powerful, including 2 lightning 3 interfaces, 2 M.2 interfaces, 2 HDMI interfaces, 2 DP interfaces, 2 Ethernet ports, 7 USB and so on.
The last is the KBL-G processor models and parameters, KBL-G total of 5 models are 4-core 8-thread design, memory support for DDR4-2400MHz, and are BGA package.
among them, i7-8809G highest specifications, clocked at 3.1GHz, maximum Core frequency 4.2GHz, equipped with Radeon RX Vega M GH graphics, is the only one of the 5 processors to support overclocking models.
In addition, i7-8809G and i7-8709G use are 24 groups of Radeon RX Vega M GH CU graphics card, the number of stream processors is 24 * 64 = 1536, i7-8706G, i7-8705G, i5-8305G Radeon RX Vega M GL graphics with 20 CU units with 20 * 64 = 1280 stream processors.
From the number of stream processors, the Radeon RX Vega M GH / GL is between RX 560 (16 CU and 1024 stream processors) and RX 570 (32 CU and 2048 stream processors).
Written in the last:
It is crazy to encapsulate Intel CPUs and AMD GPUs in a single chip. The reason we do not know it (let's not forget that I did not really understand why AMD should integrate Vega GPUs and RTGs Boss Raja handed over to Intel).
But for the player, a mini PC that is slim enough and powerful enough to play mainstream games at 1080P / 60FPS (such as the upcoming Skeleton Canyon) is enough to make us excited.
