AMD's Ruilong 3000 series processor is currently the top model of the Ruilong 9 3950X, 16 core 32 threads, acceleration frequency of 4.7GHz. On the 7nm Ruilong 3000 processor, the problem of weaker single-core performance has been lost, IPC performance has increased by 15%, and the acceleration frequency has also increased by 4.6 or even 4.7GHz.
What about the overclocking performance? This is also a concern for many players, because in the case of Intel's Nine-generation Core has reached 5GHz frequency, if the AMD Ruilong 3000 processor is overclocked, then the player's own overclocking impact of 5GHz is also very attractive, after all, AMD The Ruilong processor promises not to lock the frequency and can be freely overclocked.
At the E3 event, AMD really did overclocking test, 16 core 32 threaded Ruilong 9 3950X overclocked to 5GHz in full core state, although the LN2 liquid nitrogen limit overclocking, the voltage also reached 1.6V, but 16 core 5GHz is still a very scary thing, you must know the same 16-core Threadripper 1950X Although the limit frequency reaches 5.5GHz, but the core number is shielded to 4 cores and 4 threads, the difficulty is completely different..
For AMD, they also know the lack of frequency in the first and second generations of the Ruilong. The acceleration frequency of the Ruilong processor is only 4.1GHz under the 14nm process, and the second-generation Ruilong has achieved 4.35GHz under the 12nm process. 4.6GHz was achieved on the 7nm node.
In addition, AMD also brought a new Pricision Boost Overdrive (PBO) precision acceleration overclocking and automatic overclocking on the Ruilong 3000, and the frequency can be increased by 200MHz.
Not only is the CPU core frequency increased, but the memory frequency has also increased dramatically.New memory controller, new IF bus/memory crossover mechanism, optimized X570 motherboard layout, can achieve ultra-high frequency of 5100MHz under air cooling, but note that the best performance frequency is DDR4-3733Because the Infinity Fabric bus frequency and the memory frequency are 1:1, the delay is low. After the IF bus and memory frequency will be 1:2, the delay will increase.
To the best of it, according to AMD's commitment, the AM4 slot will be used until 2020, which means that the 7nm EUV process and the Zen 3 architecture of the Yanlong 4000 series will be compatible with the existing motherboards. The existing motherboard players can at least enjoy the next. Compatible with a generation of processors.
When will AMD upgrade the slot? AMD responded to this question in an interview recently.Reiterating that the AM4 slot will continue to support until 2020,If something forces them to upgrade, it must be a major underlying technology change, such as next-generation memory or PCIe technology.
On the X570 platform, AMD has implemented PCIe 4.0 technical support, and it is impossible to upgrade to PCIe 5.0 in the foreseeable future, so it is impossible for PCIe technology to force AMD to upgrade the slot.
The most likely factor is DDR5 memory. It is now recognized that DDR5 memory will enter the market around 2020. At that time, AMD will also push the Zen4 architecture. It is also logical to support DDR5 memory. It is justifiable to upgrade the AM4 slot. It is.
Back in the X570 generation, Anadntech editor Ian Cutress has released new information.The original IO core of the Ruilong 3000 (codenamed Matisse) is different from the core of the Roman processor. The IO core of the Roman processor is the GF 14nm process, and the IO core of the Ruilong 3000 is the GF 12nm.It is a process optimized and optimized by GF based on the 14nm process. The second generation of the Zen+ architecture is the product of the 12nm process.
The most surprising thing is the X570 chipset. The X570 chipset was previously known to be 28nm. However, the X570 is actually equivalent to the IO core of the 14nm process version of the Ruilong 3000 processor. Its layout is related to Matisse. The IO core is the same, with more or less features.
Simply put, AMD does this by maximizing the use of cores. The CPU core is sensitive to transistor density, performance, and power consumption. It is helpful in the most advanced 7nm process, while the IO core and X570 Southbridge do not require advanced processes. Reusing mature processes can save money.