Intel’s 8-Core Tiger Lake H Pours on the Performance, Power Consumption

Intel’s 8-Core Tiger Lake H Pours on the Performance, Power Consumption

For the past few years, Intel has had to split its mobile stack between 10nm CPUs optimized for power consumption with best-in-class integrated graphics, and high-end 14nm CPUs optimized for maximum performance, with a weaker iGPU. This left the company hopping a bit from foot to foot when it came to explaining why some 10th Gen mobile CPUs were better for gaming while others were best for compute.

The eight-core Tiger Lake-H CPUs you can buy today are Intel’s solutions to this problem. Quad-core Tiger Lake CPUs, which launched last year, were lauded for their performance and battery life improvements relative to Ice Lake, but Intel hadn’t replaced the higher TDP parts until the launch of these eight-core and six-core Tiger Lake chips.

Intel’s 8-Core Tiger Lake H Pours on the Performance, Power Consumption

The new Tiger Lake-H CPUs offer a 45W TDP with an option to clock up to 65W. At 45W, these CPUs have a base frequency of ~2.3-2.6GHz, while the 65W TDP allows this to rise to as high as 3.3GHz for the Core i9-11980HK. All of these CPUs feature 32 EUs and 256 onboard Xe graphics cores. These chips are all expected to ship with external graphics cards; the GPUs inside the lower-end quad-core Tiger Lake chips are much better, despite their lower TDPs and smaller core counts.

Eight Core Tiger, Burning Bright

(With apologies to William Blake)

It’s clear that review coverage came in hot and fast; multiple publications report having less than 48 hours with test hardware before publishing results. Some publications are reporting on Intel reference platforms and some have notebooks. There’s also some variance in which CPUs are being tested — and, interestingly, in the conclusions the reviews come to.

Gordon Mah Ung at PCWorld has a full review of the Aorus 17G with a Core i7-11800H. It’s still an eight-core Tiger Lake, but it backs a few notches off the 65W Core i9-11980HK in terms of base and boost clock. The Core i7-11800H might not quite catch the Ryzen 9 5900HX, but that’s literally AMD’s highest-end CPU. It’s more than a match for the Ryzen 9 5980, for example:

Intel’s 8-Core Tiger Lake H Pours on the Performance, Power Consumption

Gordon’s tests show the Core i7-11800H is often 15-20 percent faster than the 10th Gen CPUs it replaces. That’s a very solid generational improvement and on par with what AMD has offered recently.

But the story around Tiger Lake isn’t quite as simple as this, as Anandtech’s experience indicates:

Reference Platform, Pants Ignite

(Did you know William Blake wrote many poems, some of which I haven’t ruined?)

Anandtech has data on the Core i9-11980HK, the top-end model inside an Intel reference system. Their system arrived poorly configured and would initially allow CPU temperatures up to 96C, with frequent severe instances of CPU throttling.

Intel’s 8-Core Tiger Lake H Pours on the Performance, Power Consumption

This graph is busy, but all three lines tell different parts of the same story: The CPU throttles up to full performance, slams into its thermal limit, and throttles back to cool down. Anandtech wound up testing the CPU in 45W mode due to this behavior at 65W.

While this sounds bad, the fact that we’re talking about an Intel-built reference platform complicates things. Intel’s reference platform either needs better UEFI tuning or a better cooling solution in order to showcase the potential of a 65W CPU in a laptop this thin. That doesn’t mean the boutique OEMs who will actually ship these chips won’t do a better job cooling them. At 65W, Anandtech notes the 11980HK was only 9 percent faster in multi-threaded SPEC than the 45W variant, despite having a 1.44x larger thermal envelope. Clearly, the existing cooling solution was already near its limit.

Reviewed at 45W, the Core i9-11980HK is said to trade blows with the Ryzen 9 5980S, except that “we’re comparing a 45W chip vs a 35W chip, and more often in compute heavy workloads such as rendering or encoding, the AMD chip comes ahead even though it has a lower TDP,” Anandtech wrote.

We may need to wait for shipping systems to see the Core i9-11980HK fairly, but it looks a bit as if Intel pushed TGL well out of its comfort zone in mobile to close the door on 14nm and move ahead with a unified 10nm product family. The company has previously stated that 10nm would not be the node it originally wanted in the first place.

It may be that we’ll see the strongest uplift for 11th Gen over 10th Gen in the relative middle of the product stack. This would echo what we see with Rocket Lake, where Intel’s Core i5 CPUs compare very well against the current Ryzen 5 stack. Similarly, we may find that Tiger Lake-H CPUs like the Core i7-11800H offer the best relative competition to AMD, not the Core i9-11980HK.

At present, it looks as though some of Intel’s eight-core Tiger Lake-H CPUs are a definite improvement on their predecessors, though the 65W picture is a bit murkier. For now, we’re chalking that up to an inadequate reference design and we’ll see how things evolve. Compared with AMD’s Ryzen 5000 series, Intel appears to be competitive as well, though AMD’s top-end CPUs draw less power. Lower power consumption is always an advantage in mobile, though it’s a bit less important at the top of the stack, where we verge into the desktop replacement end of things.

Continue reading

Intel’s Desktop TDPs No Longer Useful to Predict CPU Power Consumption
Intel’s Desktop TDPs No Longer Useful to Predict CPU Power Consumption

Intel's higher-end desktop CPU TDPs no longer communicate anything useful about the CPUs power consumption under load.

AMD Engineer Confirms RDNA3 GPU Power Consumption Will Increase
AMD Engineer Confirms RDNA3 GPU Power Consumption Will Increase

This is the trend, said the AMD fellow and veteran chiplet engineer.

Ethereum Switches to Proof-of-Stake, Will Reduce Global Energy Consumption
Ethereum Switches to Proof-of-Stake, Will Reduce Global Energy Consumption

There could be some rough times ahead as the market comes to terms with such a major change, but this seems like a win for almost everyone.

MIT Neural Network Processor Cuts Power Consumption by 95 Percent
MIT Neural Network Processor Cuts Power Consumption by 95 Percent

MIT has developed new neural network processing methods that could cut the power consumption of existing solutions by up to 95 percent. It's a sea change that could effectively reinvent the nature of AI — and where such workloads are performed.