At the beginning of November, Intel dropped the bomb, a core processor with AMD graphics becomes reality and now the time has come. The Las Vegas-based CES 2018 is also the age of thinner yet powerful gaming or graphics workstations, Intel claims, up to 17mm thin, yet up to 40% faster than current discrete graphics systems. In addition to gamblers, content creators and VR fans are likely to be pleased about the upcoming generation of thinner notebooks, 2-in-1 devices and mini-PCs. Intel is pushing forward with its two new NUC PCs, Dell and HP will also present new devices based on this combo technology at CES in the next few days. In the trade, the new hardware will appear in the spring, details of availability have not yet been made.
Kaby Lake G processor Family
The new processor / graphics modules are part of the 8th Intel Core generation, which has already provided in thin and light notebooks with the increase in the core number of 2 to 4 for more power (U series, Kaby Lake Refresh) and also available in variants for the desktop and potent laptops (H Series, Coffee Lake). In today’s G-Series processors (i7-8809G, i7-8709G, i7-8706G, i7-8705G and i5-8305G), Intel combines 45W Core i CPUs with 2 different Radeon RX Vega M GPUs (GL or GH) and 4 GB High Bandwidth Memory Gen 2 (HDMB2) graphics memory. In addition, there is the integrated Intel Graphics HD630 directly in the chip. As you can see on the picture above, all this is nicely packaged in a thin and small module, which not only saves space in the case but also in terms of efficiency benefits.
The Intel Kaby Lake G processor family consist of 5 fast quad-core processors for notebooks based on the Kaby Lake architecture, which was launched in January 2018. It integrates four CPU cores with 2.8 – 4.1 GHz and as a special feature an AMD Radeon graphics card based on the current Vega architecture (24 CUs = 1536 shader with 1190 MHz). Furthermore, there are 4 GB of HBM2 memory in the package. Since a standard Kaby Lake processor is included in the package, the 8709G offers the same features as the Core i7-7920HQ, ie an HD Graphics 630 graphics unit and a dual-channel memory controller (DDR3L-1600 / DDR4-2133). Manufacturing is done in a 14 nanometer process with FinFET transistors.
Intel subdivides the five new processors as mentioned in principle into two groups, each differing in relation to the integrated Radeon RX Vega variant and consuming either a maximum of 65 watts or 100 watts. In the benchmarks provided by Intel (see pictures below), the smaller version competes with the Vega M GL-GPU with conventional systems (Intel Core i7-8550U) with the Nvidia Geforce GTX 1050 and performs up to 40 percent better, depending on the benchmark. The 100 watt modules with Vega M GH graphics positioned Intel against a system with i7-7700HQ and Nvidia GeForce GTX 1060 Max-Q and still creates up to 13 percent more power.
Kaby Lake G Comparison table
i7-8809G vs i7-8709G vs i7-8706G vs i7-8706G vs i5-8305G
| i7-8809G | i7-8709G | i7-8706G | i7-8705G | i5-8305G | |
|---|---|---|---|---|---|
| Architecture | Kaby Lake | Kaby Lake | Kaby Lake | Kaby Lake | Kaby Lake |
| Cores/Threads | 4/8 | 4/8 | 4/8 | 4/8 | 4/8 |
| Base Clock | 3.1 GHz | 3.1 GHz | 3.1 GHz | 3.1 GHz | 2.8 GHz |
| Max Turbo Clock | 4.2 GHz | 4.1 GHz | 4.1 GHz | 4.1 GHz | 3.8 GHz |
| Cache Size | 8MB | 8MB | 8MB | 8MB | 6MB |
| Video Performance | 3.7 TFLOPS | 3.7 TFLOPS | 2.6 TFLOPS | 2.6 TFLOPS | 2.6 TFLOPS |
| Video Memory | 4GB HBM2 | 4GB HBM2 | 4GB HBM2 | 4GB HBM2 | 4GB HBM2 |
| ROPs | 64 | 64 | 32 | 32 | 32 |
| HBM2 Bandwidth | 204.8GB/s | 204.8GB/s | 179.2GB/s | 179.2GB/s | 179.2GB/s |
| RAM | Dual-Channel DDR4-2400 | Dual-Channel DDR4-2400 | Dual-Channel DDR4-2400 | Dual-Channel DDR4-2400 | Dual-Channel DDR4-2400 |
| Unlocked | Yes | No | No | No | No |
| Video Card Architecture | Vega M GH | Vega M GH | Vega M GL | Vega M GL | Vega M GL |
| RX Strean Processors | 1536 | 1536 | 1280 | 1280 | 1280 |
| Video Base Clock | 1063 MHz | 1063 MHz | 931 MHz | 931 MHz | 931 MHz |
| Radeon Boost Clock | 1190 MHz | 1190 MHz | 1011 MHz | 1011 MHz | 1011 MHz |
| Video Memory bandwidth | 204.8 GB/s | 204.8 GB/s | 179.2 GB/s | 179.2 GB/s | 179.2 GB/s |
| TDP | 100 watts | 100 watts | 65 watts | 65 watts | 65 watts |
| Packaging | BGA | BGA | BGA | BGA | BGA |
| vPro Support | No | No | No | No | No |
Architecture
Compared to Skylake, Intel has adopted the underlying microarchitecture virtually unchanged, with no difference in per MHz performance. Only the speed-shift technique was revised for faster dynamic adaptation of voltages and clock rates, and the mature 14-nanometer process allows significantly higher frequencies and better energy efficiency than before.
Performance
If the clock speeds of the processor are correct, the performance should correspond to the Intel Core i7-7920HQ (3.1 – 4.1 GHz). If the HBM2 graphics memory can also be used for the CPU, the performance could be a bit faster (also due to the higher common TDP of 100 watts). For 65 watts processors such as i5-8305G, i7-8705G and i7-8706G performance will be lower; however, their energy-saving capabilities will be much higher, therefore they can be used in smaller devices.
Graphics unit
As a graphics unit, the Kaby Lake G processors integrate not only the HD Graphics 630 but also an AMD Radeon RX Vega M GH or AMD Radeon RX Vega M GL. With 24/20 CUs, it offers significantly more than the maximum 11 CUs of the Vega graphics card in the Ryzen APUs (eg Ryzen 7 2700U with 10 CUs).
Unlike Skylake, Kaby Lake can now also decode H.265 / HEVC in the 10-bit Main 10 profile and Google’s VP9 codec in hardware. Also HDCP 2.2 should be integrated. However, HDMI 2.0 is questionable and will likely only be integrated with DisplayPort adapters.
Manufacturing process
The manufacturing process of the processor should be done in the same 14-nanometer process with FinFET transistors of the Intel Core i7-7920HQ (14nm +). The TDP of 100 watts also includes the graphics card and the HBM 2 memory. Thus, the SoC is not suitable for very flat notebooks, but allows a smaller design than a comparable notebook with z.b. i7-7920HQ and dedicated graphics card (ala Radeon RX 560).
Is i7 8809G better than Geforce GTX 1060 Max-Q with i7 7700HQ
In could be. The specifications speak for themselves – 4GB HBM2 memory, 24 compute units, high clock speed, integration with intel processor) . In total, Intel plans to have five models with Vega graphics, three with a Radeon RX Vega M GL (65 Watts TDP, 20 compute units and slightly lower clock) and two with a RX Vega M GH (100 Watts TDP, 24 CUs and higher clock speeds ) to publish.
The i7 8809G with Vega M GL is said to outperform the comparison system by 1.1 to 1.4 times, the combination of the Intel processor with Vega M GH brings it to the 1.07 to 1.13-fold FPS Power. The Benchmark tests included 3DMark11, Hitman, Deus Ex: Mankind Divided, and Total War: Warhammer.
Hewlett Packard and Dell notebook computers with Kaby Lake G processors are expected to hit the market in the first quarter of 2018, with more specific statements about the overall performance of Intel / AMD hardware CPU / GPU combinations expected by then to let.
