10nm SuperFin, Willow Cove, Xe, and new SoC

Inside Intel’s 11th Generation ‘Tiger Lake’ processor are four ‘Willow Cove’ cores, 96 Execution Units of graphics built with the Xe-LP microarchitecture, all on top of Intel’s latest 10nm SuperFin manufacturing technology. Because each part of this processor has had updates, even trying to explain it in a single sentence can be cumbersome. For the most part, Intel has covered a number of details relating to all these features across the events in August and September – at Intel’s Architecture Day 2020 and the 11th Gen Core announcement, which we will recap here before going into our own testing.

Our full articles on this information can be found at:

10nm SuperFin

Intel’s Tiger Lake design has been enabled through its latest manufacturing process node update. This is a new version of Intel’s 10nm process, and Intel claims that this update is akin to a ‘full node transition’, as if Intel was offering a new manufacturing process altogether. To back up this claim, Intel says that the new generation 10nm offers 17-18% better transistor performance compared to the previous generation 10nm, equivalent to the four updates it applied to 14nm over several years but at the same time. This is a good thing – better performance, a wider frequency and voltage window, and potentially better density, all while using the same manufacturing hardware.

Intel is calling its new process node enhancement ‘SuperFin’, based two of the key design changes that the company has used.

At the lower layers of the stack, Intel is introducing a new set of barrier materials to enable thinner barriers, which also helps reduce resistance of vias by up to 30% by enabling the metal each via to be a bigger proportion of the fixed size. Reducing the resistance enhances the performance of the interconnect between the metal layers

At the higher levels, Intel is introducing a new SuperMIM (metal-insulator-metal) capacitor. Intel states that this new design gives a 5x increase in capacitance over an industry standard MIM cap within the same footprint. This drives a voltage reduction that ultimately leads to drastically improved product and transistor performance. Intel states that this is an industry first/leading design, enabled through careful deposition of new Hi-K materials in thin layers, smaller than 0.1nm, to form a superlattice between two or more material types.

On top of this, a new high-performance third generation FinFET design, enabled through a better gate-generation process and lower resistance source/drain, is what Intel has stated will enable future products. All combined, against a base 10nm process (Intel doesn’t specify which, although most analysts assume this is Cannon Lake, the OG 10nm), Intel is promoting 17-18% better transistor performance with the SuperFin.

Going forward, all products built on this process are going to be called ‘10nm SuperFin’. In the original 10nm parlance, this was 10++. Note that 10++ was renamed 10+ back at CES 2020, and so this is the second renaming. To avoid any confusion, Intel has provided a decoder ring to keep everyone up to date:

  1. (Pre-10nm: Cannon Lake)
  2. 10nm: Ice Lake Notebook, Ice Lake Xeon, Snow Ridge
  3. 10nm SuperFin: Tiger Lake, Xe-LP Products (DG1, SG1)
  4. 10nm Enhanced Superfin: Sapphire Rapids Xeon, Xe-HP Products (1-4 Tile)

After 10ESF we should see Intel move into its 7nm process and products.

Tiger Lake: Willow Cove

The new microarchitecture of the main CPU cores is called Willow Cove, and is an update over the previous generation Sunny Cove. Intel claims a 10-20% performance improvement generation on generation, however it is important to understand that this is not a raw clock-for-clock gain but rather a peak performance claim.

In terms of a clock-for-clock gain, Intel by and large has said in our technical briefings that Willow Cove will only have minor improvements due to the cache structure (more on that below). Aside from the cache, there are a few changes to deal with security and side-channel attacks, but no other changes to the core design to note. We did discover a few performance updates to specific instructions, which we’ll go into later in the review, but essentially unless the core is L2/L3 cache sensitive, users should not expect to see a clock-for-clock performance difference.

Where Tiger Lake’s 10-20% performance uplift comes from is both performance efficiency and peak performance. Because of the SuperFin design, and a new high-performance transistor library, Intel claims to have enabled the Willow Cove core to be more efficient at every frequency compared to the previous generation, but also extend the frequency range. Where the previous generation was limited to around 4.0 GHz, the new Tiger Lake as launched can go to 4.8 GHz. Based on our discussions with Intel’s engineers, it can go even higher.

Aside from the frequency adjustment, the biggest change inside the core is the cache hierarchy.

The private L2 cache gets the biggest update, with a +150% increase in size. Traditionally increasing the cache size by double will decrease the miss rate by √2, so the 2.5x increase should reduce L2 cache misses by ~58%.  The flip side of this is that larger caches often have longer access latencies, so we would expect the new L2 to be slightly slower. After many requests, Intel said that its L2 cache was a 14-cycle latency, which we can confirm, making it only +1 cycle over the previous generation. It’s quite impressive to more than double a cache size and only add one cycle of latency. The cache is also now a non-inclusive cache.

The L3 also gets an update, in two ways. The size has increased for the highest core count processors, from 2 MB per core to 3 MB per core, which increases the L3 cache line hit rate for memory accesses. However, Intel has reduced the associativity from 16-way at 8 MB per 4C chip to 12-way at 12 MB per 4C chip, which reduces the cache line hit rate, but improves the power consumption and the L3 cache latency. There is some L3 latency cycle loss overall, however due to the size increase Intel believes that there is a net performance gain for those workloads that are L3-capacity bottlenecked.

Cache Comparison for Mobile CPUs
AnandTech Coffee
Lake 4C
Lake 4C
Lake 4C
Zen2 4C
L1-I 32 KiB
32 KiB
32 KiB
  32 KiB
L1-D 32 KiB
48 KiB
48 KiB
  32 KiB
L2 256 KiB
512 KiB
1.25 MiB
  512 KiB
L3 8 MiB
8 MiB
12 MiB
  8 MiB
*As measured new for this review

For most users, the adjustment in cache sizes and latency should not be specifically noticeable. In discussing with Intel, they recognize that some specific workloads will benefit, however these changes were made from a higher-level SoC design viewpoint. We are of the opinion that this might have been done to help with the manufacturing, or thermal density, or may lead into future products. Intel, like other companies, does perform a lot of customer workload analysis about where the cache bottlenecks are, ultimately designing the next generation of cores to serve what it assumes will be the major workload mix when it comes to market.

Tiger Lake: Xe-LP

A big part of the Tiger Lake/Ice Lake comparison will be the performance difference in graphics. Where Ice Lake has 64 Execution Units of Gen11 graphics, Tiger Lake has 96 Execution Units but of the new Xe-LP architecture. On top of that, there’s the new SuperFin transistor stack that promises to drive frequencies (and power windows) a lot higher, making Tiger Lake more scalable than before. Where Gen11 graphics saw frequencies up to 1100 MHz, the initial Xe-LP designs with Tiger Lake are at 1450 MHz. Intel has teased that future versions of Xe-LP (likely the discrete DG1 card) will go up to 1750-1800 MHz.

Within Intel’s graphics cores, we speak about ALUs, or Arithmetic Logic Units. These ALUs can often do a number of things, such as math on integers (whole numbers), math on floating point numbers (fractions), or extended math (such as trigonometry).

In the Ice Lake Gen11 graphics system, each one of the 64 execution units consisted of an 8 ALU design with one set of four for FP/INT, and the other set of four for FP/Extended Math.

For Xe-LP, each of the 96 execution units have 10 ALUs, and they are split differently: the first 8 are for FP/INT, and the last 2 are for extended math. The 8 FP/INT in the new graphics support both INT16 (at double rate) and INT32 data types as well as new DP4a instructions can accelerate INT8 inference workloads. The new execution units also now work in pairs – two EUs will share a single thread control block to help assist with coordinated workload dispatch.

The graphics has its own L3 cache as well, which has a junction on the ring interconnect used inside Tiger Lake. In order to help accelerate graphics workloads further, Intel has doubled the bandwidth of the ring interconnect by essentially giving the SoC two rings, supporting 2x32B/cycle transfers bidirectionally.

Overall Intel is claiming up to 2x graphics throughput with the new Xe-LP compared to previous Gen11. Moving from 64 EUs to 96 EUs would immediately give +50% performance, and then the increase from 1100 MHz to 1450 MHz is another 32%. Together they would account for a 98% throughput increase, all other things scaling equally (memory bandwidth and such).

Tiger Lake: Media and Display

Tiger Lake will be Intel’s first official support for the AV1 codec in decode mode, and Intel has also doubled its encode/decode throughput for other popular codecs. This means a full hardware-based 12-bit video pipeline for HDR and 8K60 playback support.

Display Support for Tiger Lake is also extended with four 4K display pipelines. Connections over DP1.4, HDMI 2.0, Thunderbolt 4, and USB4 Type-C simultaneously is how Intel expects users to operate if all four outputs are needed at once. The display engine also supports HDR10, 12-bit BT2020 color, Adaptive Sync, and support for monitors up to 360 Hz.

Other SoC Adjustments

There are a few other elements about Tiger Lake that are known at an SoC level that should also be noted.

Tiger Lake now has support for LPDDR4X-4266 (32 GB max) as well as DDR4-3200 (64 GB max), which is a slight improvement over the previous generation Ice Lake on the LPDDR4X side. The silicon is also designed to support LPDDR5-5400, however we are unlikely to see any devices with LPDDR5 until the new year due to the extra validation requirements and the current extra cost that LPDDR5 brings. Intel discussed with its partners, and they preferred that Intel prioritized LPDDR4X validation to begin with. We have been told by Intel that LPDDR5 support on the current silicon, pending full validation, does not require a new silicon stepping.

Also related to memory, the new silicon now supports Total Memory Encryption.  TME has been a popular feature of new silicon designs of late, and enables mobile device users to have the data held in the memory on a system physically secure against hardware attacks. In other systems we’ve been told that a feature like TME, when implemented correctly, only gives a 1-2% performance hit in the most usual worst case – Intel has not provided equivalent numbers as of yet. Given the type of feature this is, we suspect TME might be more of a vPro-enabled product feature, however we will have to get clarity on that.

One question that everyone likes to ask is die size. We took out some calipers and measured 13.49 x 10.70 mm = 144.34 mm2. This is rather close to our estimate from CES from the wafer photographs, where we guessed 146.10 mm2 (~1.2% error).

Die Sizes
AnandTech x y Die Size Process Cores EUs/
AMD Zen 2 Chiplet 10.32 7.34 75.75 mm2 TSMC N7 8 -
Intel Ice Lake 11.44 10.71 122.52 mm2 Intel 10 4 64
Intel Tiger Lake 13.49 10.70 144.34 mm2 Intel 10SF 4 96
AMD Picasso 19.21 10.92 209.78 mm2 GF 12 4 11
AMD Renoir APU 13.59 10.98 149.22 mm2 TSMC N7 8 8

There are also security features, which we will cover in the next page.

Tiger Lake: Playing with Toe Beans New Instructions and Updated Security
Comments Locked


View All Comments

  • blppt - Saturday, September 26, 2020 - link

    Sure, the box sitting right next to my desk doesn't exist. Nor the 10 or so AMD cards I've bought over the past 20 years.

    1 5970
    2 7970s (for CFX)
    1 Sapphire 290x (BF4 edition, ridiculously loud under load)
    2 XFX 290 (much better cooler than the BF4 290x) mistakenly bought when I thought it would accept a flash to 290x, got the wrong builds, for CFX)
    2 290x 8gb sapphire custom edition (for CFX, much, much quieter than the 290x)
    1 Vega 64 watercooled (actually turned out to be useful for a Hackintosh build)
    1 5700xt stock edition

    Yeah, i just made this stuff up off the top of my head. I guarantee I've had more experience with AMD videocards than the average gamer. Remember the separate CFX CAP profiles? I sure do.

    So please, tell me again how I'm only a Nvidia owner.
  • Santoval - Sunday, September 20, 2020 - link

    If the top-end Big Navi is going to be 30-40% faster than the 2080 Ti then the 3080 (and later on the 3080 Ti, which will fit between the 3080 and the 3090) will be *way* beyond it in performance, in a continuation of the status quo of the last several graphics card generations. In fact it will be even worse this generation, since Big Navi needs to be 52% faster than the 2080 Ti to even match the 3070 in FP32 performance.

    Sure, it might have double the memory of the 3070, but how much will that matter if it's going to be 15 - 20% slower than a supposed "lower grade" Nvidia card? In other words "30-40% faster than the 2080 Ti" is not enough to compete with Ampere.

    By the way, we have no idea how well Big Navi and the rest of the RDNA2 cards will perform in ray-tracing, but I am not sure how that matters to most people. *If* the top-end Big Navi has 16 GB of RAM, it costs just as much as the 3070 and is slightly (up to 5-10%) slower than it in FP32 performance but handily outperforms it in ray-tracing performance then it might be an attractive buy. But I doubt any margins will be left for AMD if they sell a 16 GB card for $500.

    If it is 15-20% slower and costs $100 more noone but those who absolutely want 16 GB of graphics RAM will buy it; and if the top-end card only has 12 GB of RAM there goes the large memory incentive as well..
  • Spunjji - Sunday, September 20, 2020 - link

    @Santoval, why are you speaking as if the 3080's performance characteristics are not already known? We have the benchmarks in now.

    More importantly, why are you making the assumption that AMD need to beat Nvidia's theoretical FP32 performance when it was always obvious (and now extremely clear) that it has very little bearing on the product's actual performance in games?

    The rest of your speculation is knocked out of what by that. The likelihood of an 80CU RDNA 2 card underperforming the 3070 is nil. The likelihood of it underperforming the 3080 (which performs like twice a 5700, non-XT) is also low.
  • Byte - Monday, September 21, 2020 - link

    Nvidia probably has a good idea how it performs with access to PS5/Xbox, they know they had to be aggressive this round with clock speeds and pricing. As we can see 3080 is almost maxed, o/c headroom like that of AMD chips, and price is reasonable decent, in line with 1080 launch prices before minepocalypse.
  • TimSyd - Saturday, September 19, 2020 - link

    Ahh don't ya just love the fresh smell of TROLL
  • evernessince - Sunday, September 20, 2020 - link

    The 5700XT is RDNA1 and it's 1/3rd the size of the 2080 Ti. 1/3rd the size and only 30% less performance. Now imagine a GPU twice the size of the 5700XT, thus having twice the performance. Now add in the node shrink and new architecture.

    I wouldn't be surprised if the 6700XT beat the 2080 Ti, let alone AMD's bigger Navi 2 GPUs.
  • Cooe - Friday, December 25, 2020 - link

    Hahahaha. "Only matching a 2080 Ti". How's it feel to be an idiot?
  • tipoo - Friday, September 18, 2020 - link

    I'd again ask you why a laptop SoC would have an answer for a big GPU. That's not what this product is.
  • dotjaz - Friday, September 18, 2020 - link

    "This Intel Tiger" doesn't need an answer for Big Navi, no laptop chip needs one at all. Big Navi is 300W+, no way it's going in a laptop.

    RDNA2+ will trickle down to mobile APU eventually, but we don't know if Van Gogh can beat TGL yet, I'm betting not because it's likely a 7-15W part with weaker Quadcore Zen2.

    Proper RDNA2+ APU won't be out until 2022/Zen4. By then Intel will have the next gen Xe.
  • Santoval - Sunday, September 20, 2020 - link

    Intel's next gen Xe (in Alder Lake) is going to be a minor upgrade to the original Xe. Not a redesign, just an optimization to target higher clocks. The optimization will largely (or only) happen at the node level, since it will be fabbed with second gen SuperFin (formerly 10nm+++), which is supposed to be (assuming no further 7nm delays) Intel's last 10nm node variant.
    How well will that work, and thus how well 2nd gen Xe will perform, will depend on how high Intel's 2nd gen SuperFin will clock. At best 150 - 200 MHz higher clocks can probably be expected.

Log in

Don't have an account? Sign up now