If you have had the opportunity to read our analysis of the Ryzen microprocessors that AMD has placed on the market until now, you will remember that our assessment of this family of chips has always been positive. Your tricks? A very interesting global performance, a margin to practice quite broad overclocking, and, in addition, an undoubtedly competitive price if we face the Intel microprocessors of similar benefits.
However, there is a scenario in which, until now, AMD’s Ryzen family chips fail to match the capabilities of Intel’s equivalent offerings : game execution. As we increase the resolution the difference in performance between the Intel and AMD chips is significantly reduced, but at 1080p, which is still the most popular resolution, it is Intel that takes “the cat to the water”. This is one of the handicaps that AMD engineers have tried to solve in this iteration of the Ryzen family. If they have succeeded or we will not discover it throughout this article.
AMD Ryzen 7 2700X and Ryzen 5 2600X: Technical specifications
The table below shows the characteristics of the four new microprocessors with which AMD “baptizes” the second generation of Ryzen chips. Two of them, belonging to the Ryzen 7 family, incorporate 8 physical cores and have, like the previous proposals of this range, with SMT (Simultaneous MultiThreading) technology, so they are capable of processing a maximum of 16 threads of execution (threads) simultaneously.
The other two processors, the Ryzen 5, integrate 6 physical cores and also have SMT technology, so they can process up to 12 threads of execution. The other two relevant differences between these four microprocessors reside in the base and maximum clock frequencies to which they can work, which are greater in the models that carry the “last name” X, and the multilevel cache subsystem, which has more capacity in the Ryzen 7 that in the Ryzen 5. As far as the micro architecture is concerned, in which we will delve into the following section, these chips are identical.
|MODEL||NUCLEI||EXECUTION THREADS||MAXIMUM / NATIVE FREQUENCY||CACHE SMART PREFETCH (L2 / L3)||TDP||REFRIGERATION||PRICE||AVAILABILITY|
|Ryzen 7 2700X||8||16||4.3 / 3.7 GHz||4/16 MB||105 W||Wraith Prism (LED)||324.90 euros||04/19/2018|
|Ryzen 7 2700||8||16||4.1 / 3.2 GHz||4/16 MB||65 W||Wraith Spire (LED)||294.90 euros||04/19/2018|
|Ryzen 5 2600X||6||12||4.2 / 3.6 GHz||3/16 MB||95 W||Wraith Spire||225.90 euros||04/19/2018|
|Ryzen 5 2600||6||12||3.9 / 3.4 GHz||3/16 MB||65 W||Wraith Stealth||€ 196.90||04/19/2018|
Lithography of 12 nm and microarchitecture “Zen +”
The two pillars of AMD’s Zen architecture are efficiency and multithreaded SMT technology, which is present in the entire range of microprocessors. About this technology we have talked long and hard in other analyzes that we suggest you read if you could not do it at the time, so this time we will stick to the innovations that the engineers of the American company have introduced in the chips that belong to the second generation of the Ryzen family.
To note the magnitude of the changes introduced in this implementation of the micro architecture, AMD has modified its name by adding a “+” sign. For this reason, the new Ryzen 7 and 5 processors show off the Zen + architecture, which is based on two key premises: the increase in the work carried out by the CPU in each clock cycle, and, on the other hand, the reduction in latency associated with the multilevel cache subsystem and main memory.
As for the IPC ( Instructions Per Cycle ), which, as you know, is a measure of performance that reveals the instructions that a processor is capable of performing in a single clock cycle, the new Ryzen chips, according to AMD, benefit of an increase of 3% compared to the previous generation . This improvement, together with the increase in the clock frequency, should yield a clearly appreciable performance increase (we will check later, in the sections that we have dedicated to our test bench).
Regarding the reduction of latency, AMD claims that the Zen + microarchitecture manages to reduce Zen architecture to its predecessor, 11% of the main memory, 13% of the level cache 1, a very remarkable 34% latency of the level 2 cache, and finally, 16% of the level 3 cache. Latency is nothing other than the time that elapses since the CPU requires the content from a memory position until the moment in which that data or instruction is at your disposal. The smaller this time, the better.
As we have just seen, the improvements introduced by AMD in the Zen + micro architecture are quite remarkable. But, undoubtedly, the innovation that will most attract the attention of hardware fans is the new lithography used by GlobalFoundries, the company that manufactures AMD microprocessors, which was a part of this same company until the beginning of 2009. And is that the manufacturing technology of Ryzen second-generation chips is 12 nm.
This improvement in integration technology is responsible for increasing the clock frequency of second-generation Ryzen microprocessors compared to their predecessors. The new chips can now work between 200 and 250 MHz faster, and, in addition, the margin that we have the users when it comes to overclocking is broader thanks to their lower voltage requirements, which is always appreciated.
AMD wants to conquer the gamers once and for all
As I mentioned at the beginning of the analysis, the main outstanding debt of AMD in the previous microprocessors of the Ryzen family is its performance with the games, which, although very good, does not match that of the Intel equivalent chips in a large part of the Titles. To solve it, in addition to the improvement represented by the introduction of the 12 nm lithography and the Zen + micro architecture, the engineers of this company have polished the SenseMI technology with a very clear objective: to increase the efficiency of its chips.
On the one hand, the Precision Boost 2 algorithm monitors each of the CPU cores independently to optimize the clock frequency at which it works according to its temperature. This technology works “side by side” with the Pure Power algorithm, which uses hundreds of tiny sensors that collect information in real time of the temperature to adjust the rate of rotation of the fan to the needs of the CPU at a certain time. And, by the way, it reduces the noise level as much as possible by preventing the fan from turning at a rate higher than necessary.
According to AMD, its engineers have also improved the algorithms that are responsible for predicting which data should be previously loaded into the multilevel cache of the CPU. This “prefetch” seems to be smarter now, reducing the possibility of a cache failure, which occurs when the CPU needs a data that has not been previously stored in this memory, which inevitably implies that it is necessary to invest several clock cycles to carry out the load afterwards.
Not every day we analyze a new batch of microprocessors, so to prepare this article we have developed a test platform that is up to the requirements of the latest AMD chips. The motherboard is an X470 Aorus Gaming 7 WiFi manufactured by Gigabyte and with which, honestly, it is a pleasure to work for its great stability. The memory modules are a Sniper X DDR4-3400 from G.Skill, with its profile correctly adjusted from the BIOS and 8 GB capacity each (there are two modules in total).
The graphics card is a reference design of the NVIDIA itself and is equipped with a GeForce GTX 1070 Ti GPU well accompanied by 8 GB GDDR5. The hard drive, on the other hand, is a 3D NAND SATA SSD manufactured by Western Digital, and the ASUS power supply with a continuous output power of 500 watts. The fan that we have used with the CPU is the Wraith reference model proposed by the AMD itself. And, finally, we have also used an AOC Agon monitor with a native resolution of 2,560 x 1,440 points, a maximum refresh rate of 144 Hz and a response time of 1 ms.
Consumption and temperature
We start our test bench by evaluating the actual consumption of the test platform (the figures indicate the joint consumption of all components with the sole exception of the monitor) and also the maximum temperature reached by the CPU when we subject it to the stress that carries a load of work close to 100%.
In the graph that you have below these lines you can see that the consumption of the new processors Ryzen 7 2700X and Ryzen 5 2600X AMD does not differ much from the previous Ryzen chips of the brand. And this is good news because the clock speed at which the new chips work is greater, so we can already guess that the performance per watt of the second generation Ryzen will be greater than that of the first batch of Ryzen processors.
Regarding the maximum temperature reached by the new processors under intense stress, we again have a result similar to that of the previous generation of Ryzen chips, despite the fact that the new ones work at a higher clock frequency. In addition, during the tests we used the AMD reference fan, which is a conventional air cooling solution. This means that anyone who wants to practice extreme overclocking and decide to resort to liquid cooling will have a wide margin of maneuver when it comes to squeezing their CPU as much as possible.