AMD Ryzen's biggest performance shortcoming is confirmed: it is still difficult to compete with Intel
|
After the AMD Ryzen series processors were launched, they were highly praised by users, and many dealers were out of stock. However, is Ryzen really so perfect? The answer is no. Recently, there has been information showing that the performance of Ryzen 7 1800X is lower than expected in some application scenarios, and the main reason is believed to be the memory. There are also reports that AMD sacrificed memory compatibility and performance at the beginning of Ryzen development in exchange for improved core performance. Regarding this statement, foreign media recently verified it through AIDA 64 software, and the results were surprising. In the test of L3 cache performance, under the same memory configuration and the same 3GHz main frequency setting, the memory latency result of Ryzen 1800X was 90ns, which is higher than the 60ns of Intel i7-6900K and its own FX-8350K. In order to make the results more convincing, the detailed configuration information of the test platform was not announced. Among them, the Ryzen 1800X processor was limited to 3GHz. At the same time, the SMT and HT functions of all test objects were disabled. In terms of memory settings, the Ryzen 1800X and Intel i7-6900K platforms are equipped with DDR4-2400 memory with a timing of 15-15-15-35, while the AMD FX-8350K is equipped with DDR3-1600 memory with a timing of 9-9-9-24. The memory capacity is 16GB (4*4GB). From more test results, Intel's L1 cache performance is still ahead of AMD, but AMD has regained a position in L2 cache performance. However, in terms of L3 cache performance, all indicators of AMD are far behind Intel, and the delay is almost three times that of its opponent. The root cause of this problem can be explained by some high-load computing tasks. The L1, L2, and L3 of the Intel i7-6900K have clear division of labor. Data below 32KB will be processed in L1 first, and data exceeding the capacity will overflow into L2. When the data volume is greater than 256KB and less than 16MB, L3 will take on the computing task. Only when the data volume exceeds 16MB will the processor ask the system memory for help, resulting in increased latency, with the maximum value being 70ns. The i7-6900K will only seek help from the system memory when the amount of data being processed is greater than 16MB, resulting in a significant increase in latency. However, AMD's logic is not exactly the same. The allocation logic of L1 and L2 is no different from that of Intel, mainly L3. Analysts believe that AMD's logic for using L3 is completely different from Intel. Since Ryzen uses a modular design (each four cores has 8MB of L3 cache), the effective access limit of L3 is limited to 8MB at any time. Therefore, even if the Ryzen 1800X has the same 16MB L3 total as the i7-6900K, the utilization efficiency is halved. This causes the Ryzen 1800X to seek help from the system memory too early, which greatly increases latency. Ryzen 1800X starts to ask system memory for help when the amount of computing data exceeds 4MB and 8MB, and the latency increases sharply. It is worth mentioning that AMD officials are also well aware of the problem of memory latency in Ryzen processors, and have stated that they will launch optimization solutions as soon as possible to improve system performance. As a winner of Toutiao's Qingyun Plan and Baijiahao's Bai+ Plan, the 2019 Baidu Digital Author of the Year, the Baijiahao's Most Popular Author in the Technology Field, the 2019 Sogou Technology and Culture Author, and the 2021 Baijiahao Quarterly Influential Creator, he has won many awards, including the 2013 Sohu Best Industry Media Person, the 2015 China New Media Entrepreneurship Competition Beijing Third Place, the 2015 Guangmang Experience Award, the 2015 China New Media Entrepreneurship Competition Finals Third Place, and the 2018 Baidu Dynamic Annual Powerful Celebrity. |
<<: New battery technology emerges: iPhone battery life is expected to increase by 3 times
>>: IBM unveils 50-qubit computer Q, which beats all traditional computers
Recommend
Tree rings are the most commonly remembered traces of time
Humans are the longest-lived mammals, but we stil...
Rumor has it that the iPhone 7 will have a built-in projector and a liquid metal body
Although the iPhone 6 was released not long ago, ...
Internet TV is constrained by panels and the low-price strategy is no longer effective
LeTV TV, which has always been a disruptor with i...
A universal formula for increasing followers on Xiaohongshu!
In the process of analyzing nearly 100 bloggers, ...
Where did the sunflower, which is full of treasures, come from? Its domestication history began 3,000 years ago
As an ancient snack, sunflower seeds are widely l...
Apple delays watchOS 2 release due to thorny vulnerability
[[149425]] Apple has delayed the release of watch...
PiaPia slaps her in the face! The truth is revealed after disassembling Qin L and Galaxy L6. Do domestic cars really cut corners?
The automobile market has been quite lively durin...
Daniel Wu: How to allocate the budget for information flow advertising? Nick Cheung: Just do it without any limit!
Please don’t think that Ai Qijun is a clickbait t...
2022 General Event Program Planning!
If marketing is the top priority in operations, t...
What changes will the quality of server rental bring to the website?
What changes will the quality of server rental br...
How many days of holidays will Guangxi have on March 3rd in 2022? How many days of vacation are there? Attached is the latest holiday schedule
The third day of the third lunar month every year...
Apple recommends not sliding to close background apps: otherwise it will affect battery life
Apple said that swiping to close apps on your iPh...
Curious: Why do people like to work in coffee shops?
In life, there are always many people who carry n...
Mi Xiaoquan Animation Idiom Lesson Part 2
Introduction to the second part of Mi Xiaoquan An...
With only 20 people, how did they manage 20,000,000+ users?
In this article, Pocket founder Nate Weiner share...