Silicon, Glue, & NUMA Too

In the Ryzen family, AMD designed an 8 core silicon die known as a Zeppelin die. This consisted of two core complexes (CCX) of four cores each, with each CCX having access to 8 MB of L3 cache. The Zeppelin die had access to two DRAM channels, and was fixed with 16 PCIe lanes for add-in cards. With Threadripper, AMD has doubled up the silicon.

If you were to delid a Threadripper CPU, you would actually see four silicon dies, similar to what an EPYC processor would have, making Threadripper a Multi Core Module (MCM) design. Two of these are reinforcing spacers – empty silicon with no use other than to help distribute the weight of the cooler and assist in cooling. The other two dies (in opposite corners for thermal performance and routing) are basically the same Zeppelin dies as Ryzen, containing eight cores each and having access to two memory channels each. They communicate through Infinity Fabric, which AMD lists as 102 GB/s die-to-die bandwidth (full duplex bidirectional), along with 78ns to reach the near memory (DRAM connected to the same die) and 133ns to reach the far memory (DRAM on another die). We confirmed those numbers on DDR4-2400 memory, also achieving 65 ns and 108 ns respectively using DDR4-3200. 


Despite this AMD slide showing two silicon dies, there are four units of silicon in the package. Only two of the dies are active, so AMD has 'simplified' the diagram'

By comparison, EPYC lists die-to-die bandwidth as 42.6 GB/s at DDR4-2666. This is because EPYC runs fabric links to three dies internally and one die externally (on the next socket), which maximizes all the links available. The dies in Threadripper only have to communicate with one other die, so has more flexibility. To that extent, we’re under the impression that Threadripper is using two of these links at 10.4 GT/s using the following method:

  • Die to Die for EPYC is quoted as 42.6 GB/s at DDR4-2667
  • Die to Die for Threadripper is quoted as 102.2 GB/s at DDR4-3200
  • 42.6 GB/s * 2 links * 3200/2667 = 102.2 GB/s
  • 42.6 GB/s * 3 links * 3200/2667 at 8.0 GT/s = 115.8 GB/s (too high)
  • 42.6 GB/s * 3 links * 3200/2667 at 6.4 GT/s = 92.6 GB/s (too low)

This configuration for AMD is essentially what the industry calls a NUMA configuration: non-uniform memory access. If left as it, it means that code cannot rely on a regular (and low) latency between requesting something from DRAM and receiving it. This can be an issue for high-performance code, which is why some software is designed NUMA-aware, so that it can intelligently pin the memory it needs to the closest DRAM controller, lowering potential bandwidth but prioritizing latency.

NUMA is nothing new in the x86 space. Once CPUs began shipping with on-die memory controllers rather than using an off-die memory controller in the Northbridge, NUMA became an inherent part of multi-socket systems. In this respect AMD was the leader here right from the start, as they beat Intel to on-die memory controllers for x86 CPUs by years. So AMD has been working with NUMA for years, and similarly NUMA has been the state of affairs for Intel's multi-socket server systems for almost a decade.

What's new with Threadripper however is that NUMA has never been a consumer concern. MCM consumer CPUs have been few and far between, and we'd have to go all the way back to the Core 2 Quad family to find a CPU with cores on multiple dies, which was a design that predates on-die memory controllers for Intel. So with Threadripper, this is the very first time that consumers – even high-end consumers – have been exposed to NUMA.

But more importantly, consumer software has been similarly unexposed to NUMA, so almost no software is able to take its idiosyncrasies into account. The good news is that while NUMA changes the rules of the game a bit, it doesn't break software. NUMA-aware OSes do the heavy lifting here, helping unaware software by keeping threads and memory accesses together on the same NUMA node in order to ensure classic performance characteristics. The downside to this is that much like an overprotective parent, the OS is going discourage unaware software from using other NUMA nodes. Or in the case of Threadripper, discouraging applications from using the other die and its 8 cores.


At a hardware level, Threadripper is natively two NUMA nodes

In an ideal world, all software would be NUMA-aware, eliminating any concerns over the matter. From a practical perspective however, software is slow to change and it seems unlikely that NUMA-style CPUs are going to become common in the future. Furthermore NUMA can be tricky to program for, especially in the case of workloads/algorithms that inherently struggle with "far" cores and memory. So the quirks of NUMA are never going to completely go away, and instead AMD has taken it upon themselves to manage the matter.

AMD has implemented BIOS switches and software switches in order to better support and control the NUMAness of Threadripper. By default, Threadripper actually hides its NUMA architecture. AMD instead runs Threadripper in a UMA configuration: a uniform memory access system where memory is sent to any DRAM and the return is variable in latency (e.g. ~100ns averaging between 78ns and 133ns) but focusing for a high peak bandwidth. By presenting the CPU to the OS as a monolithic, single-domain design, memory bandwidth is maximized and all applications (NUMA-aware and not) see all 16 cores as part of the same CPU. So for applications that are not NUMA-aware – and consequently would have been discouraged by the OS in NUMA mode – this maximizes the number of cores/threads they can use and the memory bandwidth they can use.


All 32 threads are exposed as part of a single monolithic CPU

The drawback to UMA mode is that because it's hiding how Threadripper really works, it doesn't allow the OS and applications to make fully informed decisions for themselves, and consequently they may not make the best decisions. Latency-sensitive NUMA-unaware applications that fare poorly with high core/memory latencies can struggle here if they use cores and memory attached to the other die. Which is why AMD also allows Threadripper to be configured for NUMA mode, exposing its full design to the OS and resulting in separate NUMA domains for the two dies. This informs the OS to keep applications pinned to one die when possible as previously discussed, and this mode is vital for some software and some games, and we’ve tested it in this review.

Overall, using a multi-silicon design has positives and negatives. The negatives end up being variable memory latency, variable core-to-core latency, and often redundancy in on-die units that don’t need to be repeated. As a result, AMD uses 400mm2+ of silicon to achieve this, which can increase costs at the manufacturing level. By contrast, the positives are in silicon design and overall yeilds: being able to design a single piece of silicon and repeat it, rather than design several different floor plans which multiplies up the design costs, and having the (largely) fixed number of wafer defects spread out over many more smaller dies.

By contrast, Intel uses a single monolithic die for its Skylake-X processors: the LCC die up to 10-core and HCC die from 12-core up to 18-core. These use a rectangular grid of cores (3x4 and 5x4 respectively), with two of the segments reserved for the memory controllers. In order to communicate between the cores, Intel uses a networking mesh, which determines which direction the data needs to travel (up, down, left, right, or accepted into the core). We covered Intel’s MOdular Decoupled Crossbar (MoDe-X) methodology in our Skylake-X review, but the underlying concept is consistency. This mesh runs at 2.4 GHz nominally. Prior to Skylake-X, Intel implemented a ring topology, such that data would have to travel around the ring of cores to get to where it needed to go.

With reference to glue, or glue-logic, we’re referring to the fabric of each processor. For AMD that’s the Infinity Fabric, which has to travel within the silicon die or out to the other silicon die, and for Intel that’s the internal MoDe-X mesh. Elmer’s never looked so complicated.

Feeding the Beast and CPU Top Trumps AMD’s Solution to Dual Dies: Creator Mode and Game Mode
Comments Locked

347 Comments

View All Comments

  • drajitshnew - Thursday, August 10, 2017 - link

    You have written that "This socket is identical (but not interchangeable) to the SP3 socket used for EPYC,".
    Please, clarify.
    I was under the impression that if you drop an epyc in a threadripper board, it would disable 4 memory channels & 64 PCIe lanes as those will simply not be wired up.
  • Deshi! - Friday, August 11, 2017 - link

    No AMD have stated that won;t work. Its probably not hardware incompatible, but they probably put microcode on the CPUS so that if it doesn;t detect its a Ryzen CPU it doesn't work. There might also be differences in how the cores are wired up on the fabric since its 2 cores instead of 4. Remember, Threadripper has only 2 Physical Dies that are active. on Epyc all processors are 4 dies with cores on each die disabled right down to the 8 core part. (2 enabled on each physical die)
  • Deshi! - Friday, August 11, 2017 - link

    Wish there was an edit function..... but to add to that, If you pop in an Epyc processor, it might go looking for those extra lanes and memory busses that don;t exist on Threadripper boards, hence cause it not to function.
  • pinellaspete - Thursday, August 10, 2017 - link

    This is the second article where you've tried to start an acronym called SHED (Super High End Desktop) in referring to AMD Threadripper systems. You also say that Intel systems are HEDT (High End Desktop) when in all reality both AMD and Intel are HEDT. It is just that Intel has been keeping the core count low on consumer systems for so long you think that anything over a 10 core system is unusual.

    AMD is actually producing a HEDT CPU for $1000 and not inflating the price of a HEDT CPU and bleeding their customers like Intel was doing with the i7-6950X CPU for $1750. HEDT CPUs should cost about $1000 and performance should increase with every generation for the same price, not relentlessly jacking the price as Intel has done.

    HEDT should be increasing in performance every generation and you prove yourself to be Intel biased when something finally comes along that beats Intel's butt. Just because it beats Intel you want to put it into a different category so it doesn't look like Intel fares as bad. If we start a new category of computers called SHED what comes next in a few years? SDHED? Super Duper High End Desktop?
  • Deshi! - Friday, August 11, 2017 - link

    theres a good reason for that. Intel is not just inflating the cost because they want to. It literally cost them much more to produce their chips because of the monolithic die aproach vs AMDs Modular aproach. AMDs yeilds are much better than INtels in the higher core counts. Intel will not be able to match AMDs prices and still make significant profit unless they also adopt the same approach.
  • fanofanand - Tuesday, August 15, 2017 - link

    "HEDT CPUs should cost about $1000 "

    That's not how free markets work. Companies will price any given product at their maximum profit. If they can sell 10 @ $2000 or 100 at $1000 and it costs them $500 to produce, they would make $15,000 selling 10 and $50,000 selling 100 of them. Intel isn't filled with idiots, they priced their chips at whatever they thought would bring the maximum profits. The best way for the consumer to protest prices that we believe are higher than the "right" price is to not buy them. The companies will be forced to reduce their prices to find the market equilibrium. Stop complaining about Intel's gouging, vote with your wallet and buy AMD. Or don't, it's up to you.
  • Stiggy930 - Thursday, August 10, 2017 - link

    Honestly, the review is somewhat disappointing. For a pro-sumer product, there is no MySQL/PostgreSQL benchmark. No compilation test under Linux environment. Really?
  • name99 - Friday, August 11, 2017 - link

    "In an ideal world, all software would be NUMA-aware, eliminating any concerns over the matter."

    Why? This is an idiotic statement, like saying that in an ideal world all software would be aware of cache topology. In an actual ideal world, the OS would handle page or task migration between NUMA nodes transparently enough that almost no app would even notice NUMA, and even in an non-ideal world, how much does it actually matter?
    Given the way the tech world tends to work ("OMG, by using DRAM that's overclocked by 300MHz you can increase your Cinebench score by .5% !!! This is the most important fact in the history of the universe!!!") my suspicion, until proven otherwise, is that the amount of software for which this actually matters is pretty much negligible and it's not worth worrying about.
  • cheshirster - Friday, August 11, 2017 - link

    Anandtechs power and compiling tests are completely out of other rewiewers results.
    Still hiding poor Skylake-X gaming results.
    Most of the tests are completely out of that 16-core CPU target workloads.
    2400 memory used for tests.
    Absolutely zero perf/watt and price/perf analisys.

    Intel bias is over the roof here.
    Looks like I'm done with Anandtech.
  • Hurr Durr - Friday, August 11, 2017 - link

    Here`s your pity comment.

Log in

Don't have an account? Sign up now