The Intel Iris Xe Graphics mystery

by conradb212

When things hidden in fine print can mean the difference between your computer being fast or sluggish

Until about a decade ago, most computer processors did not include graphics. Graphics were part of the chipsets that worked in tandem with CPUs, and if that wasn’t good enough, there were separate graphics cards. Around 2010, wanting to play a bigger part in graphics, Intel began integrating graphics directly into the CPU, and that’s the way it’s been ever since. Intel’s integrated graphics are plenty good enough for most work, but there are tasks and industries that need more graphics firepower. CAD (Computer-Aided Design), high-end imaging, 3D graphics are examples that need more graphics punch. And computer gamers will tell you that there’s never enough graphics performance for their ever more complex games.

That’s where modern-day graphics cards come. Some are big, heavy add-on cards with their own fans and power supplies, costing thousands of dollars. Others are simpler and more affordable, and some are just modules. Rugged laptops and high-end tablets often have “discrete” graphics options. These have their own memory and power source, and provide significantly higher performance than the graphics integrated into CPUs.

Being sensitive to the demand for more graphics power, Intel has been enhancing the graphics integrated into their processors for years, often using them to distinguish processors from one another. Higher-end integrated graphics often carried the “Iris” name, and the Intel “Tiger Lake” 11th generation processors we’re now seeing in high end rugged laptops and tablets come with integrated Intel Iris Xe graphics. Intel considers that a good step up from the Intel UHD Graphics in earlier processor generations. The “Xe” indicates a new and more powerful instruction set architecture.

Why this whole preamble? Because things are getting a bit confusing.

I noticed this when we recently reviewed some of the latest rugged laptops and tablets based on Intel 11th gen Tiger Lake chips. When we ran our usual suite of performance testing benchmarks, machines that should have yielded comparable graphics performance did not. We saw differences of up to 30% where there should have been none. This turned out to be not the only weirdness in “Tiger Lake” performance, but whereas we had clues about those, the discrepancy in graphics power was a puzzler.

Then we came upon a press release announcing an upgraded version of Panasonic’s Toughbook 55, long a favorite in the market for semi-rugged laptops. The release mentioned “optional Intel Iris Xe graphics.” That sounded weird because both of the Intel Tiger Lake chips available for the Toughbook 55 have integrated Iris Xe Graphics; it’s not like you get an optional module or something like that. So I inquired with Panasonic’s PR agency about that.

The response was that “Intel Iris Xe Graphics requires two memory cards installed. Any unit that has just one memory card installed will have just Intel UHD.”

That still seemed weird and we sent follow up questions. The response was a reference to fine print at the bottom of Intel’s spec sheet for the processors included in the Toughbook. The fineprint said:

“Intel® Iris® Xe Graphics only: to use the Intel® Iris® Xe brand, the system must be populated with 128-bit (dual channel) memory. Otherwise, use the Intel® UHD brand.”

Clear as mud. So I googled that. Wikipedia described the Xe architecture as the successor to the Intel UHD architecture. References indicated that 11th generation “Tiger Lake’ chips all used the new Xe core. A Lenovo document said “Intel Iris Xe Graphics capability requires system to be configured with Intel Core i5 or i7 processor and dual channel memory. On the system with Intel Core i5 or i7 processor and single channel memory, Intel Iris Xe Graphics will function as Intel UHD graphics.”

That still didn’t answer how it all fit together. Then I came across a YouTube video where a guy ran the same 3DMark Time Spy graphics benchmark side-by-side on two Dell laptops that were identical except that one had two 8GB RAM modules and the other a single 16GB RAM module. Well, the one with the two 8GB sticks was a full 30% faster than the one with the single 16GB module. So, same amount of memory, but one configuration is 30% faster, which, depending on the task, can make a world of difference.

More googling revealed other specs that said “Integrated Iris Xe graphics functions as UHD Graphics” when just one RAM slot is populated, but “Integrates Iris Xe graphics” when two slots are used.

Sooo, apparently much of the performance of Intel Tiger Lake systems depends on that “the system must be populated with 128-bit (dual channel) memory” fine print. What IS dual channel memory? From what I can tell, most modern motherboards have two 64-bit channels between the processor and memory. By using two (presumably identical) RAM modules, the bandwidth between memory and the CPU is doubled.

Apparently that doubling of bandwidth amounts to Iris Xe running as it should. And if that condition isn’t fulfilled, graphics can and will be a lot slower. Seems like something customers should be made aware, majorly. And not via fine print with weird statements like that “Otherwise, use the Intel® UHD brand” or that a “unit that has just one memory card installed will have just Intel UHD.”

What’s remarkable is that all my googling didn’t yield a single authoritative technical explanation as to why and how that rather important dual-channel mandate works and why. Even the smart reddit and Quora techies seemed mostly baffled.

It’s now clear to me why one of the 11th gen machines we tested had much slower graphics with the same chip: it only had one RAM slot; dual-channel is not possible with it.

Folks, there are things that you need to tell customers, because it’s important. And not in tiny print or nebulous statements.

Comments |278|

Category: Editor
Tags: , , ,