Excel - Is CPU or GPU More Important for Big Data?

Dear johnny_899

Welcome to Microsoft Community.

Hi, I'm Linn and I'll do what I can to help you with the problems you're having.

This is a good question. In fact, we have all heard of the barrel effect, that is, the overall performance of a system will be decisively affected by the shortest plank. I will explain it in categories:

When processing large Excel files, the CPU is a more important factor. Although you mentioned the purpose of machine learning, it can be seen from your question that your main requirement is the speed of opening Excel files.

Let's first look at the difference between CPU and GPU. CPU is better at processing complex logical operations. The formulas in Excel obviously belong to this category, but the calculation scale is small, which is suitable for specialized and precise needs. Choosing a powerful multi-core processor can significantly improve the running speed of Excel, and it is recommended to choose a CPU with high main frequency and large cache. GPU was originally used for outputting pictures. The advantage is large-scale parallel computing, but the operation of complex logic is not ideal.

Excel relies more on the computing power of the CPU to run. Large cache and multi-core are the most critical indicators. For example, the Core i9 and Ryzen R9 series, the former has higher energy consumption but more cores, and the heat generation is also huge (yes, don't forget to consider the cooling system); the latter has a greater advantage in productivity, and there are large cache versions to choose from. More importantly, the energy efficiency ratio is higher and energy saving.

In addition, the importance of RAM lies not only in size but also in frequency. High-frequency memory will significantly speed up the data exchange rate between the CPU and memory, which is reflected in the faster user experience. Of course, considering the horrific size of your Excel file, the memory capacity should start at 32G.

These are some of my preliminary conclusions, which only represent my own views. You are welcome to reply to me to discuss together. The technical details are the most fascinating.

If you have any updates on this issue, please reply below and I will continue to help you.

Note: Be sure to block personal information (email/phone/name, etc.). (If not, please ignore it)

Sincerest greetings

Linn |Microsoft Community Support Specialist

The Xeon 6710E you mentioned is the latest product, Intel 3 process, which is only used in server products. Current information indicates that this process performs well. Its biggest disadvantage seems to be that it is more expensive. Its corresponding competitor product series is AMD EPYC series.

By the way, remember the memory I mentioned in my reply at the beginning? Its frequency and capacity will also affect the Excel work scenario with a large amount of data.

However, the main frequency of the server CPU will be much lower than the top consumer-grade CPU (AMD 9950X is recommended here), which will cause a great debuff to the Excel operation speed. Compared with the number of cores, high main frequency has more advantages, not to mention that even the most expensive consumer-grade CPU is much cheaper than the server CPU. Considering the compatibility of different motherboards? Compared with the increased cost of server-grade CPU, the amount of replacing the motherboard is negligible.

Sincerest greetings

Linn

Thank you for your evaluation, I am glad I could help you. I want to correct the part about CPU in my previous reply, that is, Intel's Core 13th and 14th generations have now been exposed to process quality problems, and any processors of these two generations are not recommended. A few 13th and 12th generations with C0 stepping can be chosen under the condition of price advantage. Here is the response from Intel's official community to this matter: July 2024 Update on Instability Reports on Intel Core 13th and 14th Gen Desktop Processors - Intel Community.

Sincerest greetings

Linn