Intel unveiled its first Ice-Lake processors. With a very powerful integrated graphics chip that, in the ultrabooks, should match the performance of discrete graphics chips. More: new manufacturing process, new architecture and new graphics engine associated with "new" communication techniques and I / O.
Intel introduced its first 10th generation processors (code name: Ice Lake). We summarize the most important information: Eleven new processors, most of them are quad-threads with eight threads and a TDP of 9, 15 or 28 watts. Standard clock frequencies range from 700 MHz to 2.3 GHz, with a maximum turbo clock of 4.1 GHz. The clock frequency of the integrated graphics chip is still around the 1 GHz market with 32, 48 or 64 EU and increasing performance. We already have the first comparison of graphics performance.
Core Specifications of Intel Ice Lake
Since the old 14 nm architecture is still available in the market, it's not easy to navigate between product names. Intel introduces a new nomenclature with the 10th generation. There are still Core i3, i5 and i7 processors to differentiate performance levels. The following string starts with a 10 to illustrate the 10th generation. The previous names of the U and Y series are omitted, which specified the TDP level. This is now the fourth number: "5" represents the previous U series and "0" the previous Y series. The graphical level is indicated by the designations G1, G4 and G7 of the entrance to the high-end G4 and G7. represent the new graphics chips of the generation Iris Plus.
Programming in detail
All new Ice-Lake processors rely on Wi-Fi 6 and Thunderbolt 3 support. You can install up to four Thunderbolt ports, two on each side of the device. Intel's 11th generation graphics chips are divided into three performance groups based on the number of installed US (32, 48 or 64). The US is Intel's programmable shader units for 3D rendering, computing tasks or multimedia functions. The performance level of the graphics chips therefore differs primarily by the number of EU installed and less by the clock frequency of the GPU.
Detailed specifications on Ice Lake
The Y series processors (0 in the name) offer a maximum power rating of 9 watts, with a maximum margin of 12 watts. The U series (5 in the name) can normally consume 15 watts, the maximum limit here is 25 watts. The only exception is the i7-1068G7 with a TDP of 28 watts. The 3_Cache level is always determined by the CPU level: i3 = 4MB, i5 = 6MB and i7 = 8MB. According to an Intel representative, these eleven processors span the entire Ice Lake range, which will be available this year.
Although we do not know how many device manufacturers have to pay for each processor, we still got the first Intel price data:
In comparison, the 8th generation Whiskey-Lake processors cost $ 409 for the i7-8565U and $ 297 for the i5-8265U.
Differences between Whiskey Lake and Amber Lake
The most notable change is the increase in TDP value from the low power 5-watt Y-series at Amber Lake to 9 watts at Ice Lake. This may be due to the fact that very few devices have been configured on the Amber Lake Y-series processors and you want to use the increased performance level for more end customers. The standard clock rates are falling, unlike the 10th generation Whiskey Lake predecessor, the boost rate is much higher. This means that the increase in power depends mainly on how long the processor can keep the boost clock, and therefore directly cooling performance of the device. With the new processors, the number of supported Thunderbolt ports increases from two to four, and the WLAN transfer rate increases via the Wi-Fi 6 module. The control of external monitors has also been improved. It is possible to connect three 4K and 120 Hz monitors, two 5K and 60 Hz monitors or an 8K monitor at 30 Hz.
Intel has provided its PC-WORLD colleagues with a Software Development System (SDS) for this test. An Intel 7600P 256GB SSD, an 8GB dual mode LPDDR4X / 3733 RAM, and a Core i7-1065G7, which can be used in 15W or 25W mode, are installed. The processor has four cores with Hyper-Threading, a 1.3 GHz base clock and a maximum clocked clock of 3.9 GHz. The graphic unit has a frequency of 1.1 GHz and counts 64 EU. The L3 cache is 8 MB. For cooling, a simple heat pipe solution with two fans is installed in the test model. The test model is compared to a Dell XPS 13 9380 and an HP Specter x360 13 – both with an i7-8565U.
Software Development System Cooling Solution
In the Cinebench R15 multi-core benchmark, the Core i7-1065G7 is about six percent faster than the i7-8565U in the Dell XPS 13 9380 and 24 percent faster than the i7-8565U in the HP Specter x360 13. You can see here very well how strongly depends on the performance of the cooling capacity. In the Cinebench R20 Multi Core, IT performance is 5% slower than Dell's, but the HP Specter has dropped 41%. Here, the i7-1065G7 with 15W TDP also surpasses the HP Specter i7-8565U with an additional 8% power. With single-threaded performance, the analysis sample is a TDP of 15 W (184 points), a TDP of 25 W (181 points) and the Dell XPS 13 8380 (183 points).
One of the biggest benefits of the Ice Lake generation is the greatly increased bandwidth of the main memory. For example, clock frequencies increase from 2133 MHz for LPDDR3 to 3733 MHz for LPDDR4X. In Geekbench, we get a 34% increase in performance. First and foremost, the built-in graphics unit also benefits from faster RAM because it does not have its own graphics memory but uses the main memory.
The game performance
Intel has announced a significant increase in gaming performance with Gen-11 graphics chips. We test performance not only in synthetic tests such as 3DMark, but also in two games. We compare the results with those of the previous generation and with Nividia dedicated graphics chips in the form of the MX150 with TDP of 10 and 25 watts.
The Ice Lake GPU is on average 75% faster than the Dell XPS 13 9380. The comparison with the Huawei Matebook's MX 150 is even more remarkable. The Intel Iris Plus graphics processor is on a par with the MX 150 (10 watts). Even though the 25 – watt version of the MX150 is still set back, it 's a truly remarkable feat. The values are almost identical in both the 3DMark Sky Diver and 3DMark Time Spy tests. This not only attacks Nvidia in the field of low-budget players, but also comes before Qualcomm, which introduced the Snapdragon 8cx to Computex. Intel Iris Pro calculates to know 45% faster than the Qualcomm GPU.
To illustrate the actual performance of the game, two game tests have been proposed with World of Tanks Encore and Counter Strike: Global Offensive. The results are similar here as in the synthetic benchmarks compared to the previous generation. However, Nvidia's MX150s still have an advantage in terms of performance. Nevertheless, it is remarkable that the i7-1065G7 displays an impressive 64 frames per second on the screen at 25W TDP in CS: GO at 720p and at high settings.
First of all, it must be said that the test results are a pre-production model and that the performances here depend a lot on the whole product, in particular the cooling. In other words, for a true performance analysis, we must wait until the first manufacturer's products are available on the market.
The tenth generation of Intel 's focus is on coding, AI and deep improvements in learning. In these areas, Ice-Lake processors represent a significant advance. In applications beyond these areas, the new processors are only slightly faster than their predecessors in terms of computing power.
In the graphics area, this looks very different. The integrated graphics unit is extremely powerful and constitutes the main argument in favor of the purchase of a new processor of the 10th generation. Intel takes full account of what was promised in advance.
This article is based on two articles by colleagues Mark Hachman and Gordon Mah Ung from PC-World. There you will get more information on the
Ice Lake processors.