Intel’s 8th-gen line of Core processors includes chips based on the company’s “Kaby Lake Refresh” and “Coffee Lake” architecture, and we’re expecting upcoming “Cannon Lake” chips to be branded as 8th-gen chips as well.
But rumor has it that Intel is on track to launch its first “Ice Lake” processors later this year, and they’ll be some of the first 9th-gen Intel processors manufactured using a 10nm+ process.
It seems like that might not be all that’s new though. Leaked benchmarks suggest that Ice Lake architecture will bring a big boost in integrated graphics performance to laptops.
According to a SiSoft Sandra listing uncovered by TechPowerUp it looks like someone’s testing an “Intel Ice Lake Client Platform” system with an Ice Lake U series processor, DDR4 RAM, and Intel UHD Gen11 graphics.
While Sandra reports the test machine as a desktop, U-series chips are typically 15 watt processors designed for laptops, 2-in-1s, and low-power desktops (like Intel’s NUC line of mini desktops). So it’s possible that the system being tested is an NUC-style computer, or maybe just a hardware sample.
But the listing still provides some interesting details about the Gen11 graphics associated with that Ice Lake-U chip:
- 600 MHz base clock (boost speed unknown)
- 48 execution units
- 383 unified shaders
- Support for up to 6GB of shared memory
- 768kB of L2 cache
By comparison, here’s what you get with the Intel UHD 620 integrated GPU that comes with 8th-gen Kaby Lake-U laptop-class chips:
- 300 MHz base clock (and 1.1 GHz boost)
- 24 execution units
- 192 unified shaders
- Support for up to 3.2 GB of shared memory
- 512kB of L2 cache
Will that be enough to help Intel stave off competition from AMD, which already offers a number of Ryzen Mobile processors with integrated Radeon Vega graphics? That remains to be seen. But Intel did recently hire AMD’s former GPU chief in order to beef up its own discrete graphics technology. So it looks like the company is looking to stay competitive in the GPU space while Intel tries to hold onto its lead in the desktop and notebook CPU space.
No free lunch here, folks. This extra processing power makes this a 30W (note that it explicitly reports “30.0W”) U-class CPU, double that of any of the current U-class processors, including the Iris Pro 15W models. With that said, I don’t expect to see this in many Surface Pro-style 2-in-1’s anytime soon given this 30W power draw being at odds with that form factor’s battery life and cooling constraints. Heck, with such power requirements, I could easily see Microsoft using a 7-nm AMD chip in 2019 instead.
Has not past performance demonstrated that the first iteration of a new architecture of Intel chips is always best avoided?
Has not Meltdown demonstrated that all iterations of Intel chips are always best avoided?
This is surprising. Will their 5W Y-series ones have the same number of execution units and shaders?
any chance for safe speculative execution this time?
Any answer I could give would just be speculation.
Well, I’ll be glad to throw some speculation for Evan.
The new 8.5th-gen chips are essentially running at double the speed, with double the memory, with double the shaders, and double the execution units. So essentially we should see a Doubling of Performance.
In a range of anywhere from 140% – 300% performance, depending on the application. Games are the most taxing, so expect gains at the smaller side of the equation.
With that said, this certainly is higher than the performance we see on the AMD Vega 8… and probably even the Vega 11. I mean, the Vega-architecture has demonstrated to not offer much performance but require a lot of heat. Great for mining, but not for Gaming. I think its still below Nvidia’s GT 1030/MX 150 graphics, at least in games.
However, it should be too much for the Hybrid Intel-AMD Core i7-8809G Vega-M which boasts a Vega 24 unit, and its ridiculous/incomparable to the much faster/efficient designs of Max-Q/Nvidia’s GTX 1050 Ti which is also a Mobile Graphics chip.
Intel Iris iGPU: 100% Slideshow
Intel Iris Pro UHD iGPU: 110% Slideshow
Vega 8: 115% Slideshow
Vega 11: 125% 720p/Low/30
MX150: 150% 720p/Low/60
Vega M: 155% 720p/Med/30
GTX 1050 Ti: 400% 1080p/Med/60
Intel <5th gen + Maxwell = Good Efficiency and Good Performance
Ryzen + Vega = Great Efficiency and Great Performance
Intel 6th gen + Pascal = Better Efficiency and Better Performance
8th gen + Pascal (MaxQ) = Best Efficiency and Best Performance
Isn’t competition great?
Comments are closed.