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
lakerssuperman - Thursday, March 2, 2017 - link
People like me. I was previously running a 2600k overclocked. Nice chip. Still runs great, but I was looking for an upgrade about a year ago as one of the things I do a lot of is Handbrake conversion for my HTPC. Going to even the newest Intel 4 core got me maybe 20% improvement on one of my major workloads for insane amounts of money and going to the high end to get 8-10 cores was just not justifiable.I ended up buying a used Xeon/X79 motherboard combo for around $300 off ebay. 8 cores/16 threads and it works great for Handbrake. I lost some clock speed in the move so single thread performance took a bit of a hit, but was more than made up for in multi-thread performance. I can still game on this CPU just fine and I don't play the newest stuff right away anyway just because of time constraints.
The X79 platform is fine for what I'm doing with it. Would I like the new stuff? Sure. And if I was in the position I was last year looking for an upgrade I don't see how I wouldn't get an 1800x. It gives me the right balance of features for what I do with my computer.
If I was just gaming, I'd look at Intel currently because their 4 core i5 is the sweet spot for this. But I'm not just gaming so this chip is infinitely more attractive to someone like me. With the price and features I can't see how it isn't a winner and when the 4 and 6 core parts come out at likely higher frequencies, I think they are going to be the real winners for gaming.
rarson - Thursday, March 2, 2017 - link
Ryzen is clearly well-suited to anyone who values high performance in a multitude of usage scenarios over one single usage scenario, especially if one cares about how much money they need to spend to achieve those results.injurer - Friday, March 3, 2017 - link
1800X is definitely designed for enthusiast, and AMD fans, but when you go to 1700X this is a price killer targeting the mainstream. 1700 is on the same boat but at even lower price. All the 3 are 8 core chips and are quite close to the 6900K but at 2-4 times lower price.At the end I really believe AMD are still having to show us the real potential of their architecture. Those chips are just the start. Remember Ryzen design is a new from its core, so they definitely have room to ecpand and enhance it.
bill.rookard - Thursday, March 2, 2017 - link
Well, thing to remember is that for those looking for a new build, they now have a legitimate choice. I still do see in the future that things will only go more multithreaded, and even though the i7-7700k is still a great chip, having more physical cores and resources to throw at it will only help.To that end, again, anyone planning a NEW build from the ground up will be able to seriously consider a Ryzen system.
Worst case, think about it. In the deep dive they had mention of 'competitive resource sharing' with SMT enabled. If you were to disable SMT on Ryzen - it would give you 8 PHYSICAL cores versus the 4 physical/4 logical cores of the 7700k. Without those resources being partially used across 16 threads - all resources would be allocated to the physical cores instead, potentially allowing more processing power per physical core.
There's still quite a bit to be checked out and dug through.
lilmoe - Thursday, March 2, 2017 - link
This. I want 2 things dug deeper in follow ups:1) Single/multi threaded performance with SMT disabled VS SMT enabled.
2) Game comparisons with more sensible GPUs (which actually ship and sell in volume, IE: the ones people actually buy), like the GTX 1060 and/or RX 480.
BurntMyBacon - Friday, March 3, 2017 - link
@lilmoeI agree with 1). Intel had HT for several generations before it was universally better to leave it enabled (still needs to be disabled some times, but these are more the edge cases now).
Not so sure I'm onboard with 2). Pairing a $200 GPU with a $500 processor for gaming purposes seems a little backwards. I'd like to see that (GTX1060 / RX480) gaming comparison on a higher clocked R5 or R3 processor when they are released.
Meteor2 - Friday, March 3, 2017 - link
I'd rather see tests paired with a 1080 Ti. At RX480/1060 level, it's well known the bottleneck is GPU performance not CPU. A 1080 Ti should be fast enough to show up the CPU.lilmoe - Friday, March 3, 2017 - link
@BurntMyBacon @Meteor2Lots of people, like me, are more into CPU power. I'm OK with a mid-range GPU. Gaming is not my top priority, and when I do, It's never above 1080p.
It'd be interesting to see if there are differences. I wouldn't dismiss it, saying the GPU would be the bottleneck so fast.
bigboxes - Sunday, March 5, 2017 - link
I'm with you on that. Gaming is way down in my priority list. I do it occasionally just because I love to see what my hardware can do. I currently have a ultrawide 1080p monitor. When I move to 4K then hopefully midrange GPU will cover that. My CPU is a 4790K. It's great for most tasks. I've been wanting to go to 6/8 core for some time, but the cost for the platform was too high. I think in a couple of years I will seriously think about Ryzen when building a new workstation.rarson - Thursday, March 2, 2017 - link
I am interested in seeing potential improvement due to BIOS updates. Additionally, I'm interested in seeing potential improvement due to better multi-threaded software. My hunch is that AMD is either on-par or better than Intel, or maybe damn near that prediction, so I think the 4-core parts will compare well to the current Skylake SKUs. I also expect them to overclock better than the 8-core chips. I guess we'll just have to wait for them to release.8 physical cores is definitely better than 4 cores with SMT/HTT/whatever you want to call it.