Jump to content
AIDA64 Discussion Forum
Droiyan7

Stress test for L3 Cache (L3 only) would be nice.

Recommended Posts

Ryzen infinity fabric for ZEN and ZEN+ (Ryzen 1X00 series and 2X00 series) make memory overclock more useful but complicated.

Sometimes, overclock fails even if both memory modules and memory controller works well but infinity fabric fails.

 

There are claims that AIDA64 Cache & Memory Benchmark is useful to see if that happens.

1.png.2576a27c3cd88c3aecf8893ae2c43607.png

If L3 performance significantly lower before overclock, it is fail even if every other test(Testmem5, HCI Memtest, Linpack, etc...) succeed. Because Ryzen (exclude APU) 1X00 and 2X00 series have distributed L3 cache across 2 CCX which connected by infinity fabric, infinity fabric fails affect L3 cache performance.

So......how about allow us repeat L3 bench only (N times or time basic) ?

 

 

Or, maybe adding options to see L3 performance results in already exist cache stress test ?

2.png.2fd4da0f9a89d728ff9a27fa18728d2a.png

 

Edited by Droiyan7
A little typos...

Share this post


Link to post
Share on other sites
13 hours ago, Droiyan7 said:

Ryzen infinity fabric for ZEN and ZEN+ (Ryzen 1X00 series and 2X00 series) make memory overclock more useful but complicated.

Sometimes, overclock fails even if both memory modules and memory controller works well but infinity fabric fails.

 

There are claims that AIDA64 Cache & Memory Benchmark is useful to see if that happens.

 

If L3 performance significantly lower before overclock, it is fail even if every other test(Testmem5, HCI Memtest, Linpack, etc...) succeed. Because Ryzen (exclude APU) 1X00 and 2X00 series have distributed L3 cache across 2 CCX which connected by infinity fabric, infinity fabric fails affect L3 cache performance.

So......how about allow us repeat L3 bench only (N times or time basic) ?

 

Or, maybe adding options to see L3 performance results in already exist cache stress test ?

The Cache subtest of the AIDA64 System Stability Test stresses all cache levels.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

  • Similar Content

    • By jwj850jj@gmail.com
      Hello everybody!
      First time posting to the forum, indeed it is my first time viewing the forum.  Ive been using the Engineer version of Aida64 (registered) for about 3 months now, and overall Im impressed with the software.  Its nice to just throw together a few different benchmarks and have them run how and when I want in the order I want.  I was doing all this with macros before
      Ive recently gotten back into building PCs, took a long hiatus back in the late 90s my last system having been assembled was a 440BX based chipset on an Abit motherboard, both of which no longer exist at least in a meaningful way lol.  I was an OC neophyte back then, as I am now, although I do enjoy the hobby/sport(?)
      I decided to wade into the dark and murky waters of RAM OCing, and went down that rabbit hole for a few weeks.  Long story short, I bought 2 x 16GB sticks of G.Skill 3600mhz XMP mem.  Stock timings sucked at 19-20-20-40-500 but I didnt know or care much about timings at the point I bought and assembled the system and so didnt realize their huge impact (relatively speaking) on ram performance.
      I was able to get them to post at 4000mhz 17-22-22-36-485 and ran that for a few months, then last night got the idea to try and tweak them some more, thinking maybe all these on off cycles might have trained them to do better (i dont know if that logic is incorrect or not in regards to training, but its how I justified another sweaty night of swearing out loud and smashing keys to my wife)
      Im now resting at 4000mhz 16-22-22-40-500 but I can feel they have more to give in the timings department.  Frankly, im already amazed at how well these Hynix CJR chips are doing on the OC.  My question is basically, Aida shows me as being first place in Latency, and near top tier in every memory bench, with only quad channel or server boards (mostly) beating me.  Are these results typical for these chips and my rig?  What can I do to push it further if anybody has any recommendations?  And am I correct in feeling a sense of accomplishment at being in the top spot for latency, or is my joy tragically misplaced?
      Thanks a lot guys, and to the Devs, great product and I will continue to use and support you!
      Relevant Screens attached.  






    • By vlad
      A big hello to the developers, followed by a question:
      Any chance to add LCD support for Elgato Stream Deck?
      https://www.elgato.com/en/gaming/stream-deck
       
      That would be nice...
    • By cplchillaxton
      Hello everyone,
      It is my first post in the forum and I want to ask about something that bothers me, thanks for the help in advance.
      Specs, to begin with:
      Case: Corsair 100R
      Mobo: MSI B450M Bazooka AM4
      CPU: Ryzen 5 2600X
      CPU Cooler: Xigmatek Windpower 1266 (pretty sure it's not common in the global market atm)
      GPU: Sapphire Radeon RX 580 Nitro+ 8GB
      RAM: Adata XPG 2X8GB 3000 mHz
       
      The thing is that when I'm performing the stress test, CPU (always on stock values) hovers around 75 - 80 celsius which I think is ok considering it is under 100% load while MOBO temps go as far as 80-81 degrees and that kinda bothers me as I think it should not reach that far.
      Idle temps are not that bad but still higher than CPU at all times, current temps on regular desktop browsing are as attached.
      Do you think it's fine, or should I start looking for a more decent mobo e.g. something like B450 Tomahawk?

    • By nanohead
      Was wondering if and when AIDA will add support for these new APUs.   
    • By xixou
      gppu test fail on 1950x CPU

       
       

       
      On second run CPU test is passing but GPU benchmark looks like values are only for 1 gtx1070 instead of 4 .
×
×
  • Create New...