Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Memory data width and performance #33

Open
trepmag opened this issue May 2, 2021 · 7 comments
Open

Memory data width and performance #33

trepmag opened this issue May 2, 2021 · 7 comments

Comments

@trepmag
Copy link

trepmag commented May 2, 2021

When I switched from my previous laptop (ideadpad 710s 13ikb, i7-7500u) to the yoga slim 7 are05 it means going from:

  • 8 to 16 GB of ram
  • intel 7500u to ryzen 4800u

Since then, I didn't made any benchmarks to confirm I was switching to a better performing device, but I always felt that the general desktop behaviors was less reactive with the new device; weirdly!?!?...

So, lately I took some time to grab back my previous device to digg into this and here are some quick findings (ubuntu 20.04):

  • gnome shell feels more reactive on previous laptop which I can confirm side by side: window effects and app launch time (but still a feeling)
  • libreoffice.write launchs in ~ 1.30 seconds with the previous device while the yoga slim 7 takes ~ 3.20 seconds (hot launched)
  • sysbench memory test (https://wiki.gentoo.org/wiki/Sysbench) reveals that 'memory worload' performance of the yoga slim7 is under the previous laptop (magnitude of ~ 1.5 in favor of the intel device in terms of execution time threads fairness)

So, given the above I tried to figure any memory differences btw the 2 devices and while most parameters are in favor or the ryzen device the following warns me:

  • Data Width: 32 bits for the yoga slim 7 are05
  • Data Width: 64 bits for ideapad 710s 13ikb

At the moment I'm unsure of the Data Width parameter meaning; but I suspect it might be equivalent to something like bus width which could means the amount of bits transfered by cycles?!?...

Any thought on this?

Annexes: Yoga slim7 ar05 output for dmidecode memory and sysbench memory workload: https://gist.github.com/trepmag/bafc4370d0c21d483a73323bcbe15a6f

@sarvasana
Copy link

sarvasana commented May 2, 2021

Ubuntu 20.04 with which kernel?

I am on manjaro with kernel 5.12.
No performance issues here.
LibreOffice Write is on my screen in ~1.5 seconds.

If you let me know what Sysbench tests to run, I could post the results for comparison.

@trepmag
Copy link
Author

trepmag commented May 2, 2021

Ubuntu 20.04 with kernel 5.10.23... Just tried with kernel 5.12 but ~ same result...

Here is the sysbench test:

$ sysbench memory --threads=4 run

Result:

sysbench 1.0.18 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 4
Initializing random number generator from current time


Running memory speed test with the following options:
  block size: 1KiB
  total size: 102400MiB
  operation: write
  scope: global

Initializing worker threads...

Threads started!

Total operations: 95855763 (9583950.31 per second)

93609.14 MiB transferred (9359.33 MiB/sec)


General statistics:
    total time:                          10.0001s
    total number of events:              95855763

Latency (ms):
         min:                                    0.00
         avg:                                    0.00
         max:                                    0.09
         95th percentile:                        0.00
         sum:                                30129.92

Threads fairness:
    events (avg/stddev):           23963940.7500/84370.59
    execution time (avg/stddev):   7.5325/0.01

@sarvasana
Copy link

❯ sysbench memory --threads=4 run
sysbench 1.0.20 (using system LuaJIT 2.0.5)

Running the test with following options:
Number of threads: 4
Initializing random number generator from current time


Running memory speed test with the following options:
  block size: 1KiB
  total size: 102400MiB
  operation: write
  scope: global

Initializing worker threads...

Threads started!

Total operations: 91613835 (9159910.26 per second)

89466.64 MiB transferred (8945.22 MiB/sec)


General statistics:
    total time:                          10.0001s
    total number of events:              91613835

Latency (ms):
         min:                                    0.00
         avg:                                    0.00
         max:                                    0.41
         95th percentile:                        0.00
         sum:                                30431.69

Threads fairness:
    events (avg/stddev):           22903458.7500/76231.68
    execution time (avg/stddev):   7.6079/0.01

Yours outperforms mine.

@sarvasana
Copy link

I am not on Gnome 40.

@trepmag
Copy link
Author

trepmag commented May 2, 2021

(gnome 3.38.4)

Result for the ideapad i7-7500u; much better:

$ sysbench memory --threads=4 run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 4
Initializing random number generator from current time


Running memory speed test with the following options:
  block size: 1KiB
  total size: 102400MiB
  operation: write
  scope: global

Initializing worker threads...

Threads started!

Total operations: 104857600 (13456271.61 per second)

102400.00 MiB transferred (13140.89 MiB/sec)


General statistics:
    total time:                          7.7910s
    total number of events:              104857600

Latency (ms):
         min:                                    0.00
         avg:                                    0.00
         max:                                    7.70
         95th percentile:                        0.00
         sum:                                19510.72

Threads fairness:
    events (avg/stddev):           26214400.0000/0.00
    execution time (avg/stddev):   4.8777/0.01

@trepmag
Copy link
Author

trepmag commented May 2, 2021

What is you $ sudo dmidecode --type memory; here it is: https://gist.github.com/trepmag/bafc4370d0c21d483a73323bcbe15a6f#dmidecode-memory

@jrandiny
Copy link
Owner

jrandiny commented May 6, 2021

Interesting, I have also noticed the problem of slower apps opening, however I always think it must be the ssd

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants