Intel's Kaby Lake generation marks the company's abandonment of its famed tick-tock cadence. As expected, the launch finds us testing the latest and greatest top end chips in the Intel Kaby Lake Core i7-7700K, i7-7700, i5-7600K, i5-7600 Review. Intel's Kaby Lake moves forward with refined 14nm+ transistors that provide a few hundred extra MHz of base and turbo clocks.
Intel also increased the media capabilities of the HD Graphics integrated GPUs, but the enhancements don't come as an architectural overhaul. Instead, the company increased video encode and decode through optimizations to its existing fixed function hardware. The enhancements increase performance while reducing overhead, and thus power consumption, for various video encoding/decoding tasks. This incremental step forward won't make us abandon discrete GPUs any time soon, but it does provide a tangible benefit for the mainstream audience, particularly in the mobile segment.
The Kaby Lake generation, as with Intel's Skylake CPU offerings, scales from low-power 4.5W offerings up to 91W, and configurable TDPs help address everything in between. Intel altered the m7 and m5 branding it used with the Skylake generation and shifted to the i5 and i7 branding for Kaby Lake SKUs. The new series of chips brings the 200-series chipset to bear, which promises expanded connectivity and Intel Optane Ready status. We have full coverage of the new chipsets, and we've covered the new S-Series desktop chips in the review. We've also dug in on the existing Y- and U-Series launches, but Intel added a few new U-Series SKUs and a range of H-Series processors.
Intel changed branding for all processors with eDRAM to "Iris Plus Graphics." The U-Series addresses the mid-to high-performance mobile segment with both 15W and 28W processors. The U-Series features soldered BGA packages and either HD Graphics 620 GT2 with a 2+2 configuration and 24 EUs (Execution Units) or Iris Plus Graphics 640 GT3e with 48 EU and 64MB of eDRAM.
All of the U- and H-Series products feature a configurable TDP (cTDP) range that allows OEMs to tailor mobile products with lower TDP settings to boost battery life. The lower cTDP settings also allow vendors to cram powerful processors into thinner and lighter devices by easing the thermal dissipation requirements.
|Base Frequency (GHz)||2.8||2.7||2.6||2.5||2.4||2.5||2.4||2.3||2.2|
|Single Core Turbo (GHz)||3.9||2.7||2.6||2.5||2.4||4.0||3.8||3.6||3.4|
|Dual Core Turbo (GHz)||3.9||3.5||3.5||3.1||N/A||3.8||3.7||3.6||3.4|
|Intel Graphics||HD Graphics 620||HD Graphics 620||HD Graphics 620||HD Graphics 620||HD Graphics 620||Iris Plus Graphics 640||Iris Plus Graphics 640||Iris Plus Graphics 640||Iris Plus Graphics 640|
|Base/Max. Graphics Freq (MHz)||300/1150||300/1050||300/1100||300/1000||300/1000||300/1050||300/1050||300/1000||300/950|
|LPDDR3 Support (MHz)||1866||1866||1866||1866||1866||1866||1866||1866||1866|
|DDR4/DDR3L Support (MHz0||2133/1600||2133/1600||2133/1600||2133/1600||2133/1600||2133/1600||2133/1600||2133/1600||2133/1600|
|1K Unit Pricing||$393.00||$393.00||$281.20||$281.20||$281.20||$415.00||$415.00||$304.00||$304.00|
The U-Series consists of dual-core processors that expose themselves as four cores through hyperthreading. The 14nm+ process provides a few more bins of performance for base clocks, but we see larger gains to the Turbo Boost frequencies. The higher Turbo frequencies allow mobile processors to complete tasks quickly, thus entering into low power states faster to conserve battery life. The faster single-threaded and Turbo Boost clocks are the key to Intel's faster SpeedShift response time.
The 28W U-Series i5 and i7 processors feature 4MB of L3 cache, whereas several of the 15W i5 models come with only 3MB. The i3 also makes an appearance with 3MB of cache but retains hyperthreading capabilities, though with its nonexistent Turbo Boost, it also features the lowest maximum dual-core frequencies of the entire U-Series range.
|Base Frequency (GHz)||3.5||3.3||3.1||2.8|
|Max. Single Core Turbo (GHz)||4.0||3.7||3.5||2.8|
|Max. Dual Core Turbo (GHz)||3.9||3.7||3.5||2.8|
|Intel Graphics||Iris Plus Graphics 650||Iris Plus Graphics 650||Iris Plus Graphics 650||Iris Plus Graphics 650|
|Base/Max. Graphics Freq (MHz)||300/1000||300/1100||300/1050||300/1000|
|LPDDR3 Support (MHz)||1866||1866||1866||1866|
|DDR4/DDR3L Support (MHz0||2133/1600||2133/1600||2133/1600||2133/1600|
|1K Unit Pricing||$415.00||$304.00||$304.00||$304.00|
The 28W U-Series bumps up graphics performance to the Iris Plus Graphics 650, which wields the same GT3e configuration with 64MB of eDRAM as the Iris Plus Graphics 640 that features higher graphics base/boost frequencies and an increase to 28W. Interestingly, we don't see a price increase over the 15W series with Iris Plus Graphics 640, and the Graphics 650-equipped models have lower base CPU frequencies.
45W Xeon H-Series
The H-Series features two Xeon v6 processors with HD Graphics P630, which is an eDRAM-less GT2 integrated GPU with 24 EU. Intel bumped the maximum HD Graphics frequency up from 1,050 to 1,100 over the previous-generation v5 products, but the graphics base clock remains the same.
|45W Xeon H-Series||Xeon E3-1535M v6||Xeon E3-1505M v6|
|Base Frequency (GHz)||3.1||3.0|
|Max. Single Core Turbo (GHz)||4.2||4.0|
|Max. Dual/Quad Core Turbo (GHz)||4.1/3.9||3.8/3.6|
|Intel Graphics||HD Graphics P630||HD Graphics P630|
|Base/Max. Graphics Freq (MHz)||350/1100||350/1100|
|LPDDR3 Support (MHz)||2133||2133|
|DDR4/DDR3L Support (MHz)||1600/2400||1600/2400|
|1K Unit Pricing||$623.00||$434.00|
The step up to quad cores and eight threads brings with it a 45W TDP, which should provide more power for the professional workstation-class products in the target market. Intel bumped the base frequency up 200MHz, and we note up to a 400MHz increase in Turbo Boost clocks over the v5 generation. The processors also feature an 8MB helping of L3 cache, along with a much higher price point than the comparable H-Series products listed below. Intel also brings vPro support into the picture for its mobile Xeon SKUs and increased DDR4 support up to 2,400MHz.
The quad- and dual-core mainstream H-Series features a 45W TDP with a cTDP down to 35W. The four i7 SKUs feature hyperthreading, and the value-centric i5 models feature four cores and four threads. The i3-7100H features two cores and four threads but lacks Turbo Boost functionality. All of the H-Series processors are BGA mounted.
|Base Frequency (GHz)||3.1||2.9||2.9||2.8||2.8||2.5||3.0|
|Max. Single Core Turbo (GHz)||4.1||3.9||3.9||3.8||3.8||3.5||3.0|
|Max. Dual/Quad Core Turbo (GHz)||3.9/3.7||3.7/3.5||3.7/3.5||3.6/3.4||3.6/3.4||3.3/3.1||-|
|Intel Graphics||HD Graphics 630||HD Graphics 630||HD Graphics 630||HD Graphics 630||HD Graphics 630||HD Graphics 630||HD Graphics 630|
|Base/Max. Graphics Freq (MHz)||300/1100||350/1100||350/1100||350/1100||350/1000||350/1000||300/950|
|LPDDR3 Support (MHz)||2133||2133||2133||2133||2133||2133||2133|
|DDR4/DDR3L Support (MHz0||2400/1600||2400/1600||2400/1600||2400/1600||2400/1600||2400/1600||2400/1600|
|1K Unit Pricing||$568.00||$378.00||$378.00||$378.00||$250.00||$250.00||$225.00|
As with all of the Kaby Lake series, we see a few hundred MHz of clock increases across the board, along with minor bumps to the maximum graphics frequency. The 45W H-Series also features the HD Graphics 630 with 24 EU but doesn't come with eDRAM. The upper-tier i7 models feature up to 8MB of L3 cache, whereas the i3 model steps all the way back to 3MB of L3 cache. Intel also boosted DDR4 support from 2,133MHz to 2,400MHz across the entire consumer-oriented 45W H-Series.
As expected, the Kaby Lake mobile SKUs come with unimpressive bumps in performance and incrementally lower power consumption, largely due to the increased clock speed and optimized media encoding and decoding capability. These products are fused inside OEM mobile products, and the incremental gains shouldn't spur anyone to upgrade from Skylake to Kaby Lake devices. We expect a flood of announcements from the various OEMs in the coming week as they debut their new Kaby Lake products.