L3 Cache Explained [CPUs]

See full bio

Ali Rashid Khan

At HI Digital I test and write about different hardware configurations to show other PC users what they can expect from their build. My end goal is to educate the masses to make the right decision when they are looking for new PC parts to buy

See full bio

With AMD highlighting the performance impact of L3 cache through their 3D V-cache design, there’s no better time to learn about this critical component of CPUs. In this article, we’ll explain the L3 cache and how it works. 

Key Takeaways

• The L3 cache, the third level of the CPU cache, is larger but slower than the L2 cache and is situated farther from CPU cores.
• It is a shared pool among multiple cores, unlike L1 and L2 caches, where each core has its fixed capacity.
• Its capacity varies from 4 MB in entry-level CPUs to 128 MB in enthusiast-grade CPUs.
  

  ---

CPU Cache

Before diving into the main topic, you must know the CPU cache. A CPU cache is a temporary form of storage that is much faster than system memory or DRAM^[1]. Thus, the CPU cache reduces the latency of CPU-DRAM operations^[2] as it’s much closer to the CPU than DRAM is.

A CPU cache comprises high-speed memory called SRAM or Static RAM^[3]. This speed comes at a cost, as it’s much more expensive than DRAM^[4]. There are three CPU cache levels, one of which is L3. 

L3 Cache

The L3 cache is the third level or “type” of CPU cache that is much larger (but slower)^[5] than the L1 and L2 cache. At the same, it’s further away from the CPU cores than the higher-level caches. 

Each core has a specific amount of L1 and L2 cache memory. Meanwhile, it is a pool shared by multiple cores (usually up to 4). This is visualized in the diagram above. 

Although the L3 cache is the slowest of the three cache levels^[5], it is still way faster than system memory. In terms of size, the L3 cache varies from 4 MB of entry-level CPUs to 128 MB of enthusiast gaming CPUs (such as the Ryzen 9 7950X3D). 

How Does L3 Cache Work?

The CPU stores instructions that need quick access in the cache^[6]. Depending on the specific instructions, they may be stored in this cache level. This traffic of instructions moves from DRAM to the CPU cache through the system bus for this purpose.

When the CPU needs to process these instructions, it looks for them in the L1 cache^[7]. If it fails to find the instructions in the L1 cache, it searches in the L2 cache. If it does not see the instructions in the L2 cache, it will look for them in the L3 cache.

Assuming the CPU finds the necessary instructions in the L3 cache, it will use them to carry out a specific task. This is called a cache hit. It’s worth mentioning that if the CPU does not find the instructions in this, it proceeds to look in the system memory^[8]. 

Why Companies Are Focusing On More Cache

CPU manufacturers in Intel and AMD are actively focusing on fitting more cache into their CPUs^[9], along with more cores and higher clock speeds. This is because the cache is a significant operator in your CPU^[10], allowing the CPU to process workloads much faster—the more cache there is, the more room the CPU has to store instructions for quick access. In general, more cache is always better than less cache^[11].

AMD’s revolutionary 3D V-cache technology vertically stacks L3 cache on their CPU dies. This allows much more L3 cache to be stacked into a single die. The performance upgrade brought by this seemingly simple move is impressive. For example, the Ryzen 7 5800X3D outperformed the Ryzen 9 5900X by about 10% despite having lower clock speeds and fewer cores. 

Final Words

The L3 cache is a vital component of your CPU. It drastically reduces operation times by acting as a bridge between the CPU and the system memory^[12]. Still, it shouldn’t be your primary focus when looking for a CPU. Clock speeds and core/thread counts are much more critical factors.

Related Helpful Resources By Tech4Gamers:

• L2 Cache
• CPU Socket
• VRM
• PCIe
• DIMM Slots

References:

1. Computer Organization Chapter 12: Memory. (n.d.). Retrieved from https://alg.manifoldapp.org/read/computer-organization/section/3d477606-2c6e-4b2f-b6dc-6c5eb9afb56a
2. Improvement of Power-Performance Efficiency for High-End Computing. (n.d.). Retrieved from https://web.archive.org/web/20070205221318id_/http://fortknox.csc.ncsu.edu:80/proj/hppac/2005/papers/ge.pdf
3. Sheldon, R. (2022, May 3). SRAM (Static Random Access Memory). Retrieved from https://www.techtarget.com/whatis/definition/SRAM-static-random-access-memory
4. G. (2024, March 5). Difference between RAM and Cache. Retrieved from https://www.geeksforgeeks.org/difference-between-ram-and-cache/
5. How The Cache Memory Works. (n.d.). Retrieved March 8, 2024, from https://hardwaresecrets.com/how-the-cache-memory-works/
6. Parthasarathi, R. (n.d.). Basics of Cache Memory – Computer Architecture. Retrieved from https://www.cs.umd.edu/~meesh/411/CA-online/chapter/basics-of-cache-memory/index.html
7. Central Processing Unit. (n.d.). Retrieved from https://bob.cs.sonoma.edu/IntroCompOrg-x64/bookch6.html
8. FLUSH+RELOAD: a High Resolution, Low Noise, L3 Cache Side-Channel Attack. (n.d.). Retrieved from https://www.eecg.toronto.edu/~moshovos/000/lib/exe/fetch.php?media=wiki:aca2019:flush_and_reload.pdf
9. The Future of Microprocessors. (n.d.). Retrieved from https://dl.acm.org/doi/pdf/10.1145/1095408.1095418
10. Dive Into Systems: CPU Caches. (n.d.). Retrieved from https://diveintosystems.cs.swarthmore.edu/book/C11-MemHierarchy/caching.html
11. Buffer-Controlled Cache for Low-Power Multicore Processors. (n.d.). Retrieved March 8, 2024, from https://scholarworks.utrgv.edu/cgi/viewcontent.cgi?article=1117&context=etd
12. Dada, E. G. (2016, November 29). Analysis of the Effectiveness of a Third-Level Cache. Retrieved from https://www.academia.edu/30160903/Analysis_of_the_Effectiveness_of_a_Third_Level_Cache

Ali Rashid Khan

[How-To Guides Expert]

Laiba, our tech guru at HI Digital, simplifies and crafts How-To Guides with a Google IT Support Certificate. Beyond tech, she captures the city’s aesthetics through her lens. Join her journey where tech meets creativity!

Get In Touch: Laiba@hidgt.com