We Are on a Path
To a New Horizon
A New Form of Strength
A New Determination
Powerful in Purpose
Efficient in Design
The New Era
Okay, that’s the intro Ryzen. Now let me go the background introduction of Zen, the architecture behind Ryzen processor. AMD started to develop Zen 4 years ago, from scratch, starting with a clean sheet of paper. That is according to Dr Lisa Su anyway. There are several things serve as the main focus of AMD when developing Zen.
1. Great performance, targeting 40% increase of IPC over the previous generation.
2. Efficiency, able to perform much faster at the same power consumption
3. Smart & Intelligent, able to sense its environment and adapt to the applications so it will get better and better over time.
AMD managed to met and even beat 40% IPC improvement goal! I was calculating at that time, 40% IPC improvement will take AMD to Ivy Bridge’s IPC, but Zen managed to land at Broadwell-E’s IPC which is more than 11% faster compared to Ivy Bridge! So it’s clear that AMD really managed to beat its goal.
Dr. Lisa Su also presented the key technology inside Ryzen CPU which is called AMD SenseMI technology. AMD SenseMI technology is a key enabler of AMD’s landmark increase of greater than 40 percent in instructions per clock, and is comprised of five components:
- Pure Power – more than 100 embedded sensors with accuracy to the millivolt, milliwatt, and single degree level of temperature enable optimal voltage, clock frequency, and operating mode with minimal energy consumption;
- Precision Boost – smart logic that monitors integrated sensors and optimizes clock speeds, in increments as small as 25MHz, at up to a thousand times a second;
- Extended Frequency Range (XFR) – when the system senses added cooling capability, XFR raises the Precision Boost frequency to enhance performance;
- Neural Net Prediction – an artificial intelligence neural network that learns to predict what future pathway an application will take based on past runs;
- Smart Prefetch – sophisticated learning algorithms that track software behavior to anticipate the needs of an application and prepare the data in advance.
Pure Power and Precision Boost accounted for the efficiency of Ryzen processor. Both Neural Net Prediction and Smart Prefetch accounted for one quarter of performance uplift of Ryzen. XFR is a new boost mode, different that AMD Turbo Core. It relies on sensor to sense the cooling capability to overclock itself. This is a very cool feature for gamers.
In terms of compute performance, AMD demonstrated how Ryzen performs in heavy workload tasks like rendering and video encoding compared to Intel Core i7-6900K which is based on Broadwell-E architecture. Blender and Handbrake are the chosen applications. In Blender scenario, Zen managed to performed roughly equal to Broadwell-E while in Handbrake, it managed to outperform it by almost 10%. Ryzen is clocked at 3.4 GHz in this test while Core i7-6900K is clocked at 3.2 GHz to 3.7 GHz boost depending on core load.
For gaming, AMD demonstrated how Ryzen perform as good as, if not better than Core i7-6900K in Battlefield 1 when paired with Nvidia Titan X on 4K Resolution. AMD also demonstrated how CPU handle DOTA2 maximum settings at 1080p while streaming live to twitch. This workload includes gaming and encoding at the same time, putting a lot of stress to the CPU. Ryzen has no trouble handle the task without frame drops. For comparison, there are two other systems, one with Intel Core i7-6900K running at stock speed and the other one packed Intel Core i7-6700K running overclocked to 4.5 GHz. Intel Core i7-6900K offers similar performance to Ryzen so it performs smoothly without any trouble as well. Core i7-6700K on the other hand seems to be struggling with the streaming even when overclocked to 4.5 GHz. Simply because the CPU has half amount of core counts despite the higher clock speed.
For the closing, AMD also show us a brief performance of Vega GPU performance by showing it running Star Wars : Battlefront with the soon to be released Rogue One : Scarif DLC. The system was powered by Ryzen CPU and the display resolution is 4K. Both Ryzen and Vega were said to able to pull over 60 FPS on this game.
With all of that great features, Ryzen packs 95 W TDP versi 140 W TDP on Core i7-6900K. Roughly 30% less power but still performs similarly. Ryzen will start at 3.4 GHz. There will be a boost mode, relative to it’s own XFR feature. The boost mode however, is not used in this New Horizon demo but Dr Lisa Su mentioned that it will be announced the CPU is actually released. Dr Lisa Su also mentioned that they are on track to release Ryzen on Q1 2017 which should mean somewhere between January and March. But for now, I can only see that February is the earliest possible timeline. Pricing is unknown but I expected the top SKU will sell around $500. For the performance, if it can really be on par with $1,100 Core i7-6900K, I have no problem with that. It is still much cheaper, especially if you take motherboard price into the account. X99 Motherboards are always expensive after all.
My Personal thoughts :
4 years ago when Zen was still the concept, AMD best CPU architecture at that time was Piledriver. Surely, 40% IPC improvement over Piledriver is nothing interesting. It’ll end up on par with Ivy Bridge, something that Intel already has 4 years ago and Zen was still pretty much a concept at that time. But Zen is not offering 40% IPC improvement from Piledriver, it offers 40% IPC improvement over Excavator which lands in Broadwell’s IPC territory.
The good thing for AMD is Intel stuck with Kaby Lake with zero IPC improvement over Skylake. Even better, Skylake IPC improvement over Haswell is only 5.7% on average tests done by Anandtech. Zen managed to land in Broadwell’s territory and that means around 3% faster than Haswell and 2% slower than Skylake. So, AMD will be able to compete with Intel this time in terms of IPC. Even better if AMD could clock the CPU Core faster to make up for the lost performance.
There are three more things that I still want to know. First, the price. Second, the overclocking capability. Third, Power Consumption. Well, I just need to be patient and see until professional reviewers are allowed to review and publish the results. Hopefully soon.