The AMD Zen and Ryzen 7 Review: A Deep Dive on 1800X, 1700X and 1700
by Ian Cutress on March 2, 2017 9:00 AM ESTZen: 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.
Facebook
Twitter
RSS
574 Comments
View All Comments
BurntMyBacon - Friday, March 3, 2017 - link
@Gothmoth: "gamer... as if the world is only full with idiotic people who waste their lives playing shooter or RPG´s."PC Gaming happens to be one of the few growing areas in the PC market. Not everyone games, but for those that do, the 7700K is still worth considering. Dropping $500 on the 1800X may not be the best call for those that don't take advantage of the parallelism. Of course, the 1800X wasn't really meant for people who can't take advantage of the parallelism. AMD will have lower cost narrower processors to address that gap. I'm curious as to how the performance/price equation will stand once AMD releases their upper end 6c/12t and 4c/8t processors.
Beany2013 - Friday, March 3, 2017 - link
Sod the 1800X - I need a new VM server, and if I want all the threads (sixteen), I can either drop £450 on a Xeon E5 2620 at 2.1-3ghz (cheapest Intel 16 thread option I can find), or I can spend £100 less, and get a Ryzen 7 1700 (3.0-3.7ghz) and put that extra money towards more RAM so I can run more VMs and get more work done.For those of us who aren't high end gamers - which is basically almost everyone, and a far more significant market - these chips may well give Intel a bloody nose in the workstation space; AMD have confirmed they'll use ECC RAM quite happily.
Photographers, videographers, CAD-CAM, developers etc are a bigger market in terms of raw units than high end gamers, and these chips look like being a pretty compelling option as it stands.
Steven R
Beany2013 - Friday, March 3, 2017 - link
(VM server for home, I should have noted - for work, I'll see how the Ryzen based opterons and supermicro mobos etc pan out - money is important in these factors, but I'm not a moron, and I'm not going to run production gear on gaming hardware, natch....)BurntMyBacon - Friday, March 3, 2017 - link
@Beany2013: "I need a new VM server, and if I want all the threads (sixteen), I can either drop £450 on a Xeon E5 2620 at 2.1-3ghz (cheapest Intel 16 thread option I can find), or I can spend £100 less, and get a Ryzen 7 1700 (3.0-3.7ghz) and put that extra money towards more RAM so I can run more VMs and get more work done."It is clear by this statement that you fall into the category of people that can take advantage of the parallelism. Therefore, my statement doesn't apply to your presented in the slightest.
I don't disagree that the Ryzen 7 series has a lot to offer to a lot of people (myself included). If I were in the market today, I'd be looking long and hard at an R7 1700X. The minor drop in gaming performance is less significant to me than the increase in performance for many other tasks I use my computer for. I do a little bit of dabbling in a lot of different things (most of which benefit from high thread count). I have noticed that for the set of applications I have open simultaneously and the tasks I have running, my computer is more responsive with more cores or threads, but single threaded performance is still important to the individual tasks.
In my workflow: (i3 < i5/FX-8xxx < i7 <? R7)
My point was that there is in fact a not so insignificant market of people putting computers together for the primary purpose of gaming. This market appears, by all metrics, to be growing. For this market, Intel's i7-7700K or better yet i5-7600K are still viable options that provide better performance/price than AMD's current options. I'll repeat: "AMD will have lower cost narrower processors to address that gap. I'm curious as to how the performance/price equation will stand once AMD releases their upper end 6c/12t and 4c/8t processors."
Cooe - Sunday, February 28, 2021 - link
"or better yet i5-7600K"Arguably the most short-sited statement in this entire comments section lol. The 4c/4t i5's had roughly equal gaming performance to Ryzen at launch but with ZERO headroom left for the future. This is why the i5-7600K gets absolutely freaking ROFLSTOMPED by the R5 1600 in modern titles/game engines.
JMB1897 - Friday, March 3, 2017 - link
Compelling, but I don't think it's totally there yet. I'd be worried about the memory issues. Increased latency as you add more DIMMs and dual vs quad channel. I'd spend that extra 100 on a Xeon personally.Sttm - Friday, March 3, 2017 - link
Thats who buys off the shelf CPUs thats cost $$$, Gamers. Thats who AMD needs to please with their product. GAMERS. Thats why AMD's stock has been tanking since Ryzen reviews went up, because GAMERS are the demographic that matters when it comes to performance CPU sales.deltaFx2 - Saturday, March 4, 2017 - link
@Sttm: You have an inflated opinion of the impact of gamers. No, AMD's stock isn't tanking because of gamers. I suggest you also look at Nvidia's stock, which is well down from its high of ~120, to ~98. Wed-Friday, Nvidia dropped from 105 to 98, and it dipped below that to ~96 at one point. That's roughly 7-8%. The two stocks are often correlated on drops, with AMD amplifying nvidia's drop. Both do GPUs, see? Some people make tonnes of money shorting AMD (and in recent times have lost their shirt doing so).Here's the truth: All Desktop, as per Lisa Su, is a 5 bn TAM market and gaming is part of this (let's say 50%). Nothing to scoff at, sure, but compared to laptop and server, it's a rounding error. There's NOTHING in these tests/reviews to suggest that AMD will suck in those markets; in fact, quite the opposite: power looks good, perf looks good. AMD's stock (long term) won't tank on the whims of gamers. They help get the mindshare, which is the only reason they're worth catering to (they tend to be a vocal, passionate, and sometimes irrational lot. You won't see datacenter gurus doing the stuff that gamers do. They certainly won't shoot each other over whose GPU is the best).
cmdrdredd - Saturday, March 4, 2017 - link
Believe it or not there are millions of people worldwide who pretty much use their PC for two things. The internet (web browsing, email etc) and gaming. You don't need 16 threads to check email and read forums either so gaming performance is going to be critical. It's not just the CPU performance, it's the entire platform that contributes to Gaming related performance.sans - Thursday, March 2, 2017 - link
Yeah, stick with Intel because Intel is the standard and its products are the best for each respective market. AMD is a total failure.