AMD Zen 2 Microarchitecture Analysis: Ryzen 3000 and EPYC Rome
by Dr. Ian Cutress on June 10, 2019 7:22 PM EST- Posted in
- CPUs
- AMD
- Ryzen
- EPYC
- Infinity Fabric
- PCIe 4.0
- Zen 2
- Rome
- Ryzen 3000
- Ryzen 3rd Gen
Building a core like Zen 2 requires more than just building a core. The interplay between the core, the SoC design, and then the platform requires different internal teams to come together to create a level of synergy that working separately lacks. What AMD has done with the chiplet design and Zen 2 shows great promise, not only in taking advantage of smaller process nodes, but also driving one path on the future of compute.
When going down a process node, the main advantages are lower power. That can be taken in a few ways: lower power for operation at the same performance, or more power budget to do more. We see this with core designs over time: as more power budget is opened or different units within the core get more efficient, that extra power is used to drive cores wider, hopefully increasing raw instruction rate. It’s not an easy equation to solve, as there are many trade-offs: one such example in the Zen 2 core is the relationship between the reduced L1 I-cache that has allowed AMD to double the micro-op cache, which overall AMD expects to help with performance and power. Going into the minutae of what might be possible, at least at a high level, is like playing with Lego for these engineers.
All that being said, Zen 2 looks a lot like Zen. It is part of the same family, which means it looks very similar. What AMD has done with the platform, enabling PCIe 4.0, and putting the design in place to rid the server processors of the NUMA-like environment is going to help AMD in the long run. The outlook is good for AMD here, depending on how high it can drive the frequency of the server parts, but Zen 2 plus Rome is going to remove a good number of questions that customers on the fence had about Zen.
Overall AMD has quoted a +15% core performance improvement with Zen 2 over Zen+. With the core changes, at a high level, that certainly looks feasible. Users focused on performance will love the new 16-core Ryzen 9 3950X, while the processor seems nice an efficient at 105W, so it will be interesting so see what happens at lower power. We're also anticipating a very strong Rome launch here over the next few months, especially with features like double FP performance and QoS, and the raw multithreading performance of 64 cores is going to be an interesting disruptor to the market, especially if priced effectively. We’ll be getting the hardware on hand here soon to present our findings when the processors are launched on July 7th.
216 Comments
View All Comments
nandnandnand - Tuesday, June 11, 2019 - link
Shouldn't we be looking at highest transistors per square millimeter plotted over time? The Wikipedia article helpfully includes die area for most of the processors, but the graph near the top just plots number of transistors without regard to die size. If Intel's Xe hype is accurate, they will be putting out massive GPUs (1600 mm^2?) made of multiple connected dies, and AMD already does something similar with CPU chiplets.I know that the original Moore's law did not take into account die size, multi chip modules, etc. but to ignore that seems cheaty now. Regardless, performance is what really matters. Hopefully we see tight integration of CPU and L4 DRAM cache boosting performance within the next 2-3 years.
Wilco1 - Wednesday, June 12, 2019 - link
Moore's law is about transistors on a single integrated chip. But yes density matters too, especially actual density achieved in real chips (rather than marketing slides). TSMC 7nm does 80-90 million transistors/mm^2 for A12X, Kirin 980, Snapdragon 8cx. Intel is still stuck at ~16 million transistors/mm^2.FunBunny2 - Wednesday, June 12, 2019 - link
enough about Moore, unless you can get it right. Moore said nothing about transistors. He said that compute capability was doubling about every second year. This is what he actually wrote:"The complexity for minimum component costs has increased at a rate of roughly a factor of two per year. Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years. "
[the wiki]
the main reason the Law has slowed is just physics: Xnm is little more (teehee) than propaganda for some years, at least since the end of agreed dimensions of what a 'transistor' was. couple that with the coalescing of the maths around 'the best' compute algorithms; complexity has run into the limiting factor of the maths. you can see it in these comments: gimme more ST, I don't care about cores. and so on. Mother Nature's Laws are fixed and immutable; we just don't know all of them at any given moment, but we're getting closer. in the old days, we had the saying 'doing the easy 80%'. we're well into the tough 20%.
extide - Monday, June 17, 2019 - link
"The complexity for minimum component costs..."He was directly referring to transistor count with the word "complexity" in your quote -- so yes he was literally talking about transistor count.
crazy_crank - Tuesday, June 11, 2019 - link
Actually the number of cores doesn't matter AFAIK, as Moores Law originally only was about transistor density, so all you need to compare is transistors per square millimeter. Looked at it like this, it actually doesn't even look that badchada - Wednesday, June 12, 2019 - link
Moore's law specifically talks about density doubling. If they can fit 6 cores into the same footprint, you can absolutely consider 6 cores for a density comparison. That being said, we have been off this pace for a while.III-V - Wednesday, June 12, 2019 - link
>Moore's law specifically talks about density doubling.No it doesn't.
Jesus Christ, why is Moore's Law so fucking hard for people to understand?
LordSojar - Thursday, June 13, 2019 - link
Why it ever became known as a "law" is totally beyond me. More like Moore's Theory (and that's pushing it, as he made a LOT of suppositions about things he couldn't possibly predict, not being an expert in those areas. ie material sciences, quantum mechanics, etc)sing_electric - Friday, June 14, 2019 - link
This. He wasn't describing something fundamental about the way nature works - he was looking at technological advancements in one field over a short time frame. I guess 'Moore's Observation" just didn't sound as good.And the reason why no one seems to get it right is that Moore wrote and said several different things about it over the years - he'd OBSERVED that the number of transistors you could get on an IC was increasing at a certain rate, and from there, that this lead to performance increases, so both the density AND performance arguments have some amount of accuracy behind them.
And almost no one points out that it's ultimately just a function of geometry: As process decreases linearly (say, 10 units to 7 units) , you get a geometric increase in the # of transistors because you get to multiply that by two dimensions. Other benefits - like decreased power use per transistor, etc. - ultimately flow largely from that as well (or they did, before we had to start using more and more exotic materials to get shrinks to work.)
FunBunny2 - Thursday, June 13, 2019 - link
"Jesus Christ, why is Moore's Law so fucking hard for people to understand?"because, in this era of truthiness, simplistic is more fun than reality. Moore made his observation in 1965, at which time IC fabrication had not even reached LSI levels. IOW, the era when node size was dropping like a stone and frequency was rising like a Saturn rocket; performance increases with each new iteration of a device were obvious to even the most casual observer. just like prices in the housing market before the Great Recession, the simpleminded still think that both vectors will continue forevvvvaaahhh.