Ryzen 5 1600 – The small 6-core with great ambitions
A little more than a month has passed since AMD, after a long absence and with a drumroll, managed to settle a processor in the high-end segment.
Yet the middle class segment should not be neglected either, although the words “middle class” are referred to only the financial aspect. With an estimated MSRP of 250 € a Ryzen 5 1600 is priced in between Intel’s i5-K and i7-non-K series, as well as AMD’s previous top models of the FX series, but in terms of performance, we are looking rather to a Core i7 5830K or i7 6800K. To what extent values on paper reflect reality and, above all, how much power do we get for the same money with AMD is something we will now take a closer look at.
Technical Data and Features
Basic clock speed: 3,2 GHz
Boost clock speed: 3,6 GHz (with max 2 stressed cores)
TDP: 65 Watt
L1 Cache (Data): 6 x 32 KiB
L1 Cache (Instructions): 6 x 64 KiB
L2 Cache: 6 x 512 KiB
L3 Cache: 2 x 8 MiB
Thanks to Precision Boost, thanks Precision Boost one moves almost permanently above the base clock. Even if the maximum frequency of 3.6 GHz is reached only when a maximum of two cores are used, the clock rates are also raised, resulting in a real frequency of 3400 MHz, which is achieved permanently for all cores. With these values, a core voltage of 1.25 volts is required under full load and a maximum of 71.706 A is drawn, which means that the Ryzen 5 1600 draws just under 90 watts without overclocking. (Please note: this is not equivalent to TDP).
Furthermore, even if none of the previously published information about the 1600 AMD XFR is mentioned as an existing feature, this obviously does not mean that this function of the Zen architecture is not present. Otherwise, the maximum clock rates of individual cores up to 3.7 GHz can not be explained with the default settings.
At the moment, Precision Boost and XFR are not yet perfectly working with Windows energy saving options. XFR will only work if the “minimum processor performance” is not set to 100%. Otherwise, the peak performance for individual cores is never achieved. The same applies, of course, even if this setting option should not be present at all, as is the case with a manual overclocking and setting of the core voltage.
Mainboard: ASRock x370 Professional Gaming
Processor: AMD Ryzen 5 1600
RAM: 2 x 8 GB G.Skill Trident-Z RGB F4-3200C14D
Graphics Card: NVidia GTX 780Ti
Hard Disk: Samsung 850 Pro 256 GB
Power Supply: Corsair VS650
Cooling: Custom Loop (Raijintek RAI-PM5, CWB-C1, GWB-C1, RAI-R10, Alphacool NexXxos XT45 X-Flow 360mm, Alphacool NexXos ST30 240mm)
Operative System: Windows 10 Professional
At this point be aware the ram modules from G.Skill used during this test (with the Version 2.0 of the BIOS) could not work at their fullest potential. The maximum clock rate that we were able to reach was 2933 MHz (without touching BCLK settings), and the best timings achieved were 14 14 20 34. This problem should, though, disappear in the next weeks little by little, since motherboard manufacturers will improve the compatibility and performance through regular updates of the BIOS. Indeed there is a major update already programmed for May, that should enable a clock rate of the memory of up to 3200 MHz and thus increase the performance as a whole..
To overclock the AMD Ryzen 5 1600 you have two choices: either you go the old route and do it through BIOS, or you install the software Ryzen Master. The latter works not only for the CPU, but also for the memory directly from the operative system, but in reality it is practical to use only if you are not overclocking CPU and RAM at the same time, since in the actual version (at the time of this review – and not always but randomly without any known cause) it is necessary to restart the PC so that the modifications can take effect.
To be considered stable, an overclocked system should be able to go through a stress test for at least 30 minutes with Prime95 (Small FFTs) without any issue whatsoever. Only then we run benchmark tests with the stable settings that we were able to reach. All the values that are here reported are to be considered then with 100% stability of the system. I was able to achieve the highest clock rate of 4160,12 MHz (39.25 x 105.99) with 1.55 Vcore, and a memory frequency of 3109 MHz with Timings at 14-14-20-34. Without raising the base clock, the stable limit was 4125 MHz, at 1.6V.
All tests were run five times and the result reported here represent the average value of all tests. Specified reference values result from publicly available benchmark results of comparable systems. For several results the best ones were used. In order not to distort the results during overclocking, the boost for individual cores was deactivated on the Ryzen 5 1600, since this would not be active at the higher, manually set voltages and clock rates anyway. In some reference systems, this cannot be made, so, for example, turbo is included in the case of an i7.
In many tests, I could notice an unusually high scaling of the memory clocks either with AMD FX or Intel CPUs, so I included these values as well. However there is no gaming benchmark result included in this review, since non actually can be considered as representative for the whole lot of CPUs. In general however it can be said that there is no real difference between the Ryzen 5 1600 and any Core i7 7700K, 6800K or similar CPU when it comes to many graphics-limited games and benchmarks (for example on very high settings above Full HD). In some games, specially under DX11, CPU frequency can be the only limiting factor, so higher the clock (and IPC), better results are to be expected in specific gaming benchmarks, as reflected in some of the rankings below. Moreover, many optimizations will still be needed, to bring performance on par with its potential. We could see what such optimization can do in the last week with Ashes of Singularity, and in the future more and more games will receive similar treatment. Last but not least, the actual issues with NVidia drivers led me to exclude gaming benchmarks and use 3D-Mark as the go-to example of this strange behaviour.
And now, without further ado, the test results.
3D Mark reveals the current largest problematic spot: as clearly shown, the scores, especially with NVidia cards, fall way behind the expectations, but this is not due to the pure computing power of the processor. Looking at the rank list, it is clear that Ryzen processors with NVidia cards always have a significantly greater gap in performance to Intel than it would be the case with the RX 480, for example. There is a difference of up to 10% if compared to CPUs like the 7700K at 4.2 GHz or the 6850K at 3.4 GHz, two processors that in 3D-Mark achieve results similar to each other. The difference can be slightly reduced by higher memory clocks, but it does not disappear completely, because it is caused by lower graphics scores which can not be even improved by overclocking the CPU. Interestingly, this difference can only be related to games that do not fully load the graphics card. Even if 5-10% is not a distance that would make a game suddenly unplayable, it must be fixed in any case, if possible, as happened with Ashes of Singularity. A precise cause is however here not officially yet known.
Infinity Fabric – Bottleneck?
As you can see, the memory clock is much more important in Ryzen processors, than for example, in Intel’s Core i5 / i7 and AMDs FX series. The reason for this is primarily the CCX design, or AMD’s Infinity Fabric. This on the one hand ensures that the modular design to be possible at all, but at the same time also causes the communication between the complexes as well as the shifting between them to scale according to the memory clock. Here we have a transfer rate of 32 bytes per memory cycle. With DDR4 2133 this is 33.3 GB / s, or 16.65 GB / s per direction. Another not negligible point is that each complex has its own L3 cache. This improves the access times of each individual complex when all threads are active, but ensures that a thread that is moved from the scheduler to another complex can not keep its data in the cache and, in the worst case, has to reload from the memory. The latter occasionally expresses itself in a very short fall of the picture rates in games.
But not only the performance of several parallel threads is influenced by this phenomenon. Also, individual threads can migrate between the complexes and thereby lose performance, which is impressive when you disable an entire complex and then perform benchmarks that measure only the performance of individual threads (e.g., Cinebench Single CPU). The same effect can be achieved by defining the processor affinity of programs so that wild changes between the complexes can be avoided. Here the values with 3.99GHz were then suddenly at 164 (memory clock 2240 MHz), or 168 (memory clock 3080 MHz), which would mean that the negative influence of the thread changes between the CCX complexes on the performance somewhere between DDR4 3200 And DDR 4000 could be negligible. At the same time, however, this means that we can not yet fully assess the true performance of Ryzen, since such clock rates are not yet usable, so you can only use the 4C / 8T series An active CCX complex. In any case, we get an idea of what we should expect when all the optimizations are made and all compatability problems are cleared.
With a 64% increase in performance per thread compared to the FX series, the development effort that has been dedicated to this CPU has already paid off. But that was obviously not enough. AMD’s current mid-class CPU is outperforming its most powerful consumer processors and predecessor at an identical introductory price, negating the in-house trend of the last generation to compensate for the lack of IPC performance by ever-higher clock rates. Not least by the fact that this middle class starts there, as has been the case with Intel’s enthusiast segment. 6 cores and 12 threads, a performance per core that has nothing to hide from the current market leader (this is even more than enough in some areas) and all of this at a price / performance ratio that you can only see on very special ofers, or buying from a friend. Of course, price is a declaration of war and a sign of recovered self-confidence, but this does not mean that this is likely to be a calculated move. AMD has at least remained faithful to the fact that not every chip has to be designed and manufactured separately, but simply produces a version with deactivated features, lower performance, or deactivated cores, which increases both the yield and production costs.
Indeed issues and problems are not excluded from such radical innovations, like a completely new architecture and die shrinking at the same time. Especially if the manufacturer, as done with the Ryzen, tries to prevent information leaks for as long as possible and also provides partners with only minimal information. But now most of the problems are resolved and the optimizations are running on high speed, so that the full potential can slowly shine through. And this is actually impressive. Thanks to excellent basic performance, features such XFR that bring even a little more out for each user and simple and easy overclocking features, coupled with an unbeatable price there are no wishes left behind. And we’ll see on the Intel Pro / Pro AMD camps that would either anyway never mount a Ryzen, or had done so in any case, Ryzen is a win for everyone. Either through the prices that make 6-core SMT processors affordable for almost everyone, or by the now existing competition on the market, which is indispensable for real innovation. However the future looks, the real winner will be the consumer.
Of course, everything has its shadows. The big drawback are currently the relatively limited clock rates for memory and we can only hope that some progress will be made in this area soon. The second problem, directly related to the first one, are the missing optimizations for the CCX design, so that the full performance can only rarely be displayed, as long as one does not hand itself. But even if with increasing clock rates, less power will be lost here, further optimizations and updates are in the pipeline.
The Ryzen 5 1600 is in any case worth more than just a glance and thanks to its excellent easy overclockability is a powerful beast for a wide range of users with a small budget. With such an unexpected strong performance (it is after all the “small” 1600 version) he knows how to score exactly where high-end enthusiasts with limited budget had always to capitulate. To reach a 6800K is missing only a few MHz and a Quad Channel Support. In relation to the technical performance, depending on the application, the differences do not exist at all, or even fall in favor of the 1600. This is rounded off by the good overall package price, since even high-quality x370 boards are cheaper than most X99 counterparts. After a successful return to the high-end market, AMD has now also entered the middle price segment to redefine the medium class. And the first impressions suggest that they can succeed.
Awards are always a serious matter and even if there are still compatibility problems and optimizations have to be made, the potential behind the Zen architecture is already apparent. Especially since the absolutely strong performances above expectations in other areas and the unbeatable price more than compensate for this temporary drawbacks. Therefore, I decided to give the AMD Ryzen 5 1600 the Gold Award, since there are hardly any other points that justify a lesser grade.