Jump to content


 


Register a free account to unlock additional features at BleepingComputer.com
Welcome to BleepingComputer, a free community where people like yourself come together to discuss and learn how to use their computers. Using the site is easy and fun. As a guest, you can browse and view the various discussions in the forums, but can not create a new topic or reply to an existing one unless you are logged in. Other benefits of registering an account are subscribing to topics and forums, creating a blog, and having no ads shown anywhere on the site.


Click here to Register a free account now! or read our Welcome Guide to learn how to use this site.

Photo

Increase in slower cache, huh?


  • Please log in to reply
3 replies to this topic

#1 Rocky82

Rocky82

  • Members
  • 43 posts
  • OFFLINE
  •  
  • Gender:Male
  • Local time:11:47 PM

Posted 11 May 2017 - 08:36 AM

Can someone explain to me why Intel and the like, are not increasing their FASTER L1 cache, and infact DECREASING the L1 cache to bring us slower L3? Yet bragging about all this extra L3-4 cache? Wouldn't it be better to have 3 megs of L1 instead of 3 megs of L3.. What's next, 10 megs of L8? From the little I know, L1 is as fast as it gets aside from the CPU itself. So what gives?

 

Can we not just stick to high'er' levels of L1 and L2 cache and just raise that to "megabyte god-like levels"? I'm sure there is an intelligent and reasonable explanation.. I just don't understand.

 

 



BC AdBot (Login to Remove)

 


#2 Platypus

Platypus

  • Global Moderator
  • 15,208 posts
  • OFFLINE
  •  
  • Gender:Male
  • Location:Australia
  • Local time:02:47 PM

Posted 11 May 2017 - 08:48 AM

It's not simply the outright speed of a cache that makes it effective, it's the hit rate - the likelihood of the instruction or data you want now being found in the cache. The largest size of cache that will be effective is set in turn by when the average time it takes to search the cache becomes greater than the time to fetch the required data from its actual location if it's not found in the cache.

What the size(s) is/are will be determined by the design architecture of the processor. But whatever the architecture, it's an inherent characteristic of a cache that there will be a size beyond which more cache doesn't increase performance, and can actually hamper performance in some circumstances.

With multi-level caches, each will have its own optimum size to achieve the best hit rate.


Edited by Platypus, 11 May 2017 - 08:54 AM.

Top 5 things that never get done:

1.

#3 Drillingmachine

Drillingmachine

  • Members
  • 2,454 posts
  • OFFLINE
  •  
  • Gender:Male
  • Local time:06:47 AM

Posted 11 May 2017 - 09:16 AM

Can we not just stick to high'er' levels of L1 and L2 cache and just raise that to "megabyte god-like levels"? I'm sure there is an intelligent and reasonable explanation.. I just don't understand.


Because bigger cache makes it also slower. So megabyte L1 cache would be very slow (compared to smaller caches) and also it would make CPU design very difficult.

Pentium 4 had 8KB L1 data cache (yep, 8 kilobytes). While it was small, it's latency is only 2 cycles. Skylake and Ryzen has 32KB and 4 cycle latency for comparison.

Edited by Drillingmachine, 11 May 2017 - 09:17 AM.


#4 Rocky82

Rocky82
  • Topic Starter

  • Members
  • 43 posts
  • OFFLINE
  •  
  • Gender:Male
  • Local time:11:47 PM

Posted 12 May 2017 - 01:48 PM

Thanks for the answers guys, I somewhat understand it now.. More than before at least :)






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users