logo SilverBulletPC.com
The world's most practical PC.

PC Tips

Is Intel's iGPU Enough?

We see 3rd party GPU's and Intel's integrated graphics (aka iGPU, aka IGP) all advancing rapidly from year-to-year, but shouldn't there be a line that we can draw, to take some of those 3rd party GPU's out of consideration because they won't do any better than Intel's iGPU? Well, the answer is more of a twisted side-ways type answer...

It depends on what you're doing with it.

Basically, what Intel's iGPU is great at is encoding 4K video (yes, its way better at this than most of those 3rd party GPU's), 4K still graphics will look great (even vibrant over HDMI or DisplayPort), and slow paced moving graphics will work just fine. Where Intel's iGPU doesn't measure up is in displaying high performance graphics. Only the latest Intel CPU's include 60Hz 4K over DisplayPort. None of them yet include Adaptive Sync (a very important feature for those watching performance graphics). Calculating textures is not yet up to par. These are things a 3rd party GPU will do better. One may also want more screens. Adding a 3rd party GPU is great for this. In fact, you can use the iGPU for two of the screens, and a good 3rd party GPU for 3 more screens.

Which "Sync" Do I Need?

Adaptive-Sync is a standard created by VESA to help eliminate image artifacts caused by signal timing issues between GPU and monitor, which is more an issue with fast-paced graphics. A very high refresh rate compensates for most of the problem.

FreeSync is AMD's branded support for the Adaptive-Sync standard. In other words, if it says "FreeSync" then it is a product that supports the Adaptive-Sync standard.

G-Sync is a costly hardware based solution invented by NVIDIA. Diehard gamers insist it is better and will shell out the big bucks.

Intel's Quick Sync Video has to do with something else altogether: video encoding and decoding. But because people look for some sort of "sync" in Intel's brochure, it can easily be confused as the same thing.

Currently none of the Intel iGPU's support Adaptive-Sync, but a top Intel executive did leak news that they plan on adding Adaptive Sync to their gen10 and gen11 CPU's. More and more monitors and 3rd party GPU's are adding Adaptive Sync, making it the right standard to plan on. If clean fast moving graphics are important to you then be sure you buy a monitor with Adaptive Sync (or it may say "FreeSync"), and be sure it has a DisplayPort connection, because the HDMI implementation is not solid even when they say it has it.

Intel or AMD?

In short: given all points of value to the consumer, Intel is king, and will likely stay that way for some time.

It is easy to debate details endlessly. And right now the web is filled with these debates. However, what's important to take away from all this is the fact that each time AMD comes up with something that benchmarks better than Intel on some point, that Intel then takes action to improve things at their end. This is an effect of capitalism that accelerates the creation of more value for the consumer. It keeps Intel in check. If it weren't for AMD then Intel would leave pockets of untapped potential untended, and prices would be much higher. Just think of these debates as a reminder that this aspect of capitalism is working the way we want it to.

But how do we know for sure that Intel is better than AMD?

The right answer: "It doesn't matter."

The reason we say this is because no matter what, AMD and Intel will both spend billions trying to beat the other in value. And this will go on forever, or until one of them is way better than the other and the marketplace tips their vote (their purchases) in that direction. Or at least that's true in theory.

Deceptive marketing also plays a role.

According to a July/2019 analysis by Digital Trends, Intel and AMD were at that moment neck-and-neck overall, where Intel was ahead in single-core speed, and AMD was ahead in on-chip GPU capability. This is a case where AMD just released their latest chip and Intel was about to, so the comparison of course unfairly favored AMD. Now that it is switched around (Intel has since launched Ice Lake CPU's), the numbers now unfairly favor Intel. But Intel does not leverage this with marketplace deceptions.

Looking back at last year, AMD added their previously missing integrated graphics, and managed to leap-frog Intel's integrated graphics. There's a lot of this going on between them.

But what matters most? When looking at a cross-section of all the software packages people use today, its the core speed that matters most. This has always been true in the past (on average), and will likely remain true for a number of years into the future. Its really quite amazing that Intel's aging Coffee Lake Refresh platform, a 14nm platform, is still better at raw core speed than all AMD's advances in miniaturization (all the way down to 7nm), even after all these years of intense effort. This is a flag that something is up with their 7nm technology.

But what about AMD's 64 Core ThreadRipper 3990X? This is yet another example of deceptive marketing. Intel also has higher-end CPU's that cost thousands of dollars, going into servers and other special-purpose machines. What AMD did was take one of their high-end CPU's, then added a desktop chipset and shoved it into the consumer marketplace with a hefty price tag (about $4,000). And it worked! Extreme hobbyist's are emptying their pockets for this high-end CPU. Intel could learn from this tactic. This tactic isn't just making more money for AMD, it has duped the masses into thinking that AMD has passed up Intel. The thing to remember, is that these really aren't consumer grade CPU's. AMD pulled this trick to make it look as though they have the fastest consumer grade CPU. (If they didn't, then Intel's aging Coffee Lake Refresh i9 would continue to reign supreme over AMD's 7nm Ryzen, so they did this to save face.) But Intel could easily pull the same trick with one of their high-end CPU's. All they have to do is wrap it up in the same manner and market it to top-tier consumers under a cool sounding name. But Intel generally doesn't play this type of game.

Now, if AMD had a 64 Core CPU at a "consumer-affordable" price. Then that would be something to talk about. But they don't.

Its easy to point at AMD's progress in miniaturization, already launching their 7nm chip while Intel is still dabbling in 10nm and 14nm details. But the real challenge comes afterwards: all the supporting technologies that have to sync up to it. Intel's advancing more slowly with miniaturization because they are pulling with them a far bigger effort, and this results in more value.

Getting the whole picture together first (a full-featured 14nm Coffee Lake Refresh platform from Intel), is a lot more important than launching one part of the picture first (a working 7nm Ryzen platform with a couple features that top Intel's 14nm platform).

An example of something either CPU manufacturer could have easily done at any time, was to significantly boost the internal cache, which then significantly boosts the performance of just those programs that primarily process software code, not data. Such as a flight simulator. This capability is important here and there, but not very important overall. But, AMD was first. And they are making use of this in their marketing materials, as if it proves that their 7nm Ryzen 9 is better than Intel's 14nm i9-9900. This is just plain twisted nonsense. Don't fall for it. If they have to use this tactic to make their 7nm technology look better than Intel's 14nm technology, then you know that something is not at all optimal with their 7nm technology, and that this was likely a cover for some problem they're having. A good implementation of 7nm technology should simply scream past everything else, leaving no questions at all about anything challenging it. This is because the mathematical advantages to each level of miniaturization compound exponentially.

For now, we are convinced that the sum of all the details affecting value continue to favor Intel. However, we will watch AMD's 7nm progress closely, and if they do get the whole picture together (most notably a superior raw core speed) then our opinion might change. Meanwhile, what we're really looking forward to is the desktop version of Intel's Icelake CPU platform, and along with it the promised Adaptive-Sync, 1 TFLOP iGPU performance (w/512 concurrent pipelines), CPS (Intel's variation on VRS), the HEVC encoder, and other potential game-changer features.

Coffee Lake? Coffee Lake Refresh? Ice Lake? Comet Lake? Tiger Lake?

Ok then, which "Lake" is best? Intel has pushed hard to move forward into 10nm with its new Ice Lake platform, and finally in 2019 launched a major invasion of Ice Lake CPU's into the mobile markets. Dell and other big names have been selling laptops based on Ice Lake en masse. These Ice Lake CPU's are about double the capability of Coffee Lake in graphics, and about 30% better performing per equivalent ghz-rated-core. However, Ice Lake CPU's are capped at just 4 cores and 8 threads. Their primary advantage is low power and low heat production, making these ideally suited to mobile. The desktop counterparts continue to get post-poned and rumor suggests that they will skip these altogether and jump down to either 7nm like AMD, or an improved 10nm called Tiger Lake. And to tide people over they did a quick workover at 14nm to produce a line-up of CPU's they call "Comet Lake", and are calling these "10th generation" just like the Ice Lake CPU's that they put into mobile. These Comet Lake CPU's are about the same thing as Coffee Lake except that you get a few more threads here and there, and a bit more RAM speed. So, altogether you might see a 10% increase in performance over Coffee Lake, but not with any of the existing motherboards, because they switched to an entirely new CPU socket. It really doesn't make much sense to go after these CPU's when there's not much to gain over a Coffee Lake CPU. They are just trying to buy time. Buy time for what? Our best guess is Tiger Lake. This is a redesign to their 10nm process that uses improved core technology. Initial rumors hold that a modest 2.3Ghz Tiger Lake CPU can boost all the way to 5Ghz with turbo, which is a phenomenal performance level for such a low wattage CPU. Rumors suggest it may come out over the next few months, or up to 1.5 years at most. Its a ways off. So we stick with Coffee Lake? Not anymore. Quite recently Coffee Lake Refresh pricing dropped to match Coffee Lake pricing. And it is important to note that a defect in heat transmission between the actual CPU chip and the CPU cover-plate (a defect present in gen3 through gen8 core CPU's) was finally fixed in Coffee Lake Refresh, resulting in an improvement of about 15 degrees Celsius in worst case scenarios. This one small detail is actually quite important to value, resulting in our conclusion that Coffee Lake Refresh is now the highest value desktop CPU platform.

ECC or NON-ECC?

Most experts consider this matter settled. If you want your computer to not produce memory errors then you go with memory designed with Error Correction Coding (ECC). However, there is a lot more to this than a black & white answer. ECC may actually make things worse if not handled properly. And there's a software workaround that trumps it altogether. If your use for the PC is critical in nature then you may want to learn more here... The Truth About ECC

About Contact Privacy Disclaimer Warranty HOME