Zen: New Core Features

Since August, AMD has been slowly releasing microarchitecture details about Zen. Initially it started with a formal disclosure during Intel’s annual developer event, the followed a paper at HotChips, some more details at the ‘New Horizon’ event in December, and recently a talk at ISSCC. The Zen Tech Day just before launch gave a chance to get some of those questions answered.

First up, let’s dive right in to the high-level block diagram:

In this diagram, the core is split into the ‘front-end’ in blue and the rest of the core is the ‘back-end’. The front-end is where instructions come into the core, branch predictors are activated and instructions are decoded into micro-ops (micro-operations) before being placed into a micro-op queue. In red is the part of the back-end that deals with integer (INT) based instructions, such as integer math, loops, loads and stores. In orange is the floating-point (FP) part of the back-end, typically focused on different forms of math compute. Both the INT and FP segments have their own separate execution port schedulers

If it looks somewhat similar to other high-performance CPU cores, you’d be correct: there seems to be a high-level way of ‘doing things’ when it comes to x86, with three levels of cache, multi-level TLBs, instruction coalescing, a set of decoders that dispatch a combined 4-5+ micro-ops per cycle, a very large micro-op queue (150+), shared retire resources, AVX support, and simultaneous hyper-threading.

What’s New to AMD

First up, and the most important, was the inclusion of the micro-op cache. This allows for instructions that were recently used to be called up to the micro-op queue rather than being decoded again, and saves a trip through the core and caches. Typically micro-op caches are still relatively small: Intel’s version can support 1536 uOps with 8-way associativity. We learned (after much asking) at AMD’s Tech Day that the micro-op cache for Zen can support ‘2K’ (aka 2048) micro-ops with up to 8-ops per cache line. This is good for AMD, although I conversed with Mike Clark on this: if AMD had said ‘512’, on one hand I’d be asking why it is so small, and on the other wondering if they would have done something different to account for the performance adjustments. But ‘2K’ fits in with what we would expect.

Secondly is the cache structure. We were given details for the L1, L2 and L3 cache sizes, along with associativity, to compare it to former microarchitectures as well as Intel’s offering.

In this case, AMD has given Zen a 64KB L1 Instruction cache per core with 4-way associativity, with a lop-sided 32KB L1 Data cache per core with 8-way associativity. The size and accessibility determines how frequently a cache line is missed, and it is typically a trade-off for die area and power (larger caches require more die area, more associativity usually costs power). The instruction cache, per cycle, can afford a 32byte fetch while the data cache allows for 2x 16-byte loads and one 16-byte store per cycle. AMD stated that allowing two D-cache loads per cycle is more representative of the most workloads that end up with more loads than stores.

The L2 is a large 512 KB, 8-way cache per core. This is double the size of Intel’s 256 KB 4-way cache in Skylake or 256 KB 8-way cache in Broadwell. Typically doubling the cache size affords a 1.414 (square root of 2) better chance of a cache hit, reducing the need to go further out to find data, but comes at the expense of die area. This will have a big impact on a lot of performance metrics, and AMD is promoting faster cache-to-cache transfers than previous generations. Both the L1 and L2 caches are write-back caches, improving over the L1 write-through cache in Bulldozer.

The L3 cache is an 8MB 16-way cache, although at the time last week it was not specified over how many cores this was. From the data release today, we can confirm rumors that this 8 MB cache is split over a four-core module, affording 2 MB of L3 cache per core or 16 MB of L3 cache for the whole 8-core Zen CPU. These two 8 MB caches are separate, so act as a last-level cache per 4-core module with the appropriate hooks into the other L3 to determine if data is needed. As part of the talk today we also learned that the L3 is a pure victim cache for L1/L2 victims, rather than a cache for prefetch/demand data, which tempers the expectations a little but the large L2 will make up for this. We’ll discuss it as part of today’s announcement.

AMD is also playing with SMT, or simultaneous multi-threading. We’ve covered this with Intel extensively, under the heading ‘HyperThreading’. At a high level both these terms are essentially saying the same thing, although their implementations may differ. Adding SMT to a core design has the potential to increase throughput by allowing a second thread (or third, or fourth, or like IBM up to eight) on the same core to have the same access to execution ports, queues and caches. However SMT requires hardware level support – not all structures can be dynamically shared between threads and can either be algorithmically partitioned (prefetch), statically partitioned (micro-op queue) or used in alternate cycles (retire queue).

We also have dual schedulers, one for INT and another for FP, which is different to Intel’s joint scheduler/buffer implementation. 

CPUs, Speeds, Pricing: AMD Ryzen 7 Launch Details The Ryzen Die
Comments Locked

574 Comments

View All Comments

  • EchoWars - Thursday, March 2, 2017 - link

    No, apparently the failure was in your education, since it's obvious you did not read the article.
  • Notmyusualid - Friday, March 3, 2017 - link

    Ha...
  • sharath.naik - Thursday, March 2, 2017 - link

    I think you missed the biggest news in this information dump. The TDP is the biggest advantage amd has. Which means that for 150watt server cpu. they should be able to cram a lot more cores than intel will be able to.
  • Meteor2 - Friday, March 3, 2017 - link

    ^^^This. I think AMD's strength with Zen is going to be in servers.
  • Sttm - Friday, March 3, 2017 - link

    Yeah I can see that.
  • UpSpin - Thursday, March 2, 2017 - link

    According to a german site, in games, Ryzen is equal (sometimes higher, sometimes lower) to the Intel i7-6900K in high resolution games (WQHD). Once the resolution is set very low (720p) the Ryzen gets beaten by the Intel processor, but honestly, who cares about low resolution? For games, the probably best bet would be the i7-7700K, mainly because of the higher clock rate, for now. Once the games get better optimized for 8 cores, the 4-core i7-7700K will be beaten for sure, because in multi-threaded applications Ryzen is on par with the twice expensive Intel processor.

    I doubt it makes sense to buy the Core i7-6850K, it has the same low turbo boost frequency the 6900K has, thus low single threaded performance, but at only 6 cores. So I expect that it's the worst from both worlds. Poor multi-threaded performance compared to Ryzen, poor single threaded performance compared to i7-7700K.

    We also have to see how well Ryzen can get overclocked, thus improving single core performance.
  • fanofanand - Thursday, March 2, 2017 - link

    That is a well reasoned comment. Kudos!
  • ShieTar - Thursday, March 2, 2017 - link

    Well, the point of low-resolution testing is, that at normal resolutions you will always be GPU-restricted. So not only Ryzen and the i7-6900K are equal in this test, but so are all other modern and half-modern CPUs including any old FX-8...

    The most interesting question will be how Ryzen performs on those few modern games which manage to be CPU-restricted even in relevant resolutions, e.g. Battlefield 1 Multiplayer. But I think it will be a few more days, if not weeks, until we get that kind of in-depth review.
  • FriendlyUser - Thursday, March 2, 2017 - link

    This is true, but at the same time this artificially magnifies the differences one is going to notice in a real-world scenario. I saw reviews with a Titan X at 1080p, while many will be playing 1440p with a 1060 or RX480.

    The test case must also approximate real life.
  • khanikun - Friday, March 3, 2017 - link

    They aren't testing to show what it's like in real life though. The point of testing is to show the difference between the CPUs. Hence why they are gearing their benchmarking to stress the CPU, not other portions of the system.

Log in

Don't have an account? Sign up now