Dynabook X40-K with i5-1240p reading 1333mhz ram (2666) instead of 1600mhz (3200)

[adam_b16]

Hi,

Long time user of Hwinfo, thank you very much for such a great program for those who like details =)

Dynabook X40-K with i5-1240p reading 1333mhz ram (2666) instead of 1600mhz (3200).

Report and debug file attached, along with screenshot showing Hwinfo open with Cpuz - Cpuz showing 1600mhz (3200 c22 JEDEC), where Hwinfo only shows 1333mhz (2666 c22).

Odd thing is, other programs are showing mixed speeds too (unless its from idle ram speed reductions)
Firestrike shows correct mem speed at the bottom, but 2666 in the integrated Iris Xe graphics - https://www.3dmark.com/3dm/89270640
Aida64 shows correct memory speed in the memory section, but odd speeds during benchmarks.

Task manager shows memory speed fixed as 3200.

I do have 2x16gb 3200 on order for dual rank 2Rx8 and dual channel setup.

Thanks in advance!
-Adam

 

[Martin]

Well, there are two methods how to determine current memory clock and they return different values.
Anyway, the memory clock should be changing depending on load/usage. Try to put some load on the CPU (e.g. benchmark) if the memory clock will ramp up in HWiNFO.

 

[adam_b16]

Hello Martin!

Thank you for the quick reply! While I had only just turned on debug mode for requirements of this post, I have made heaps of data records over various tests like CPUz stress, Geekbench 3 + 5, 3dmark etc. Each one shows 1333 and doesn't move except for +/- 0.1 mhz. The timings do change though suggesting clocks would be changing to suit jedec profiles.

Would you like me to post one of those logs without debug on? Otherwise I can record another this evening after work.

Cheers!
-Adam

 

[Martin]

Thank you, no need to post further Debug Files now. Problem is that I'm not sure who's right here.
So regardless of idle/load HWiNFO reports 1333. Does the memory clock change in CPU-Z when idle/load?

 

[adam_b16]

Yes, CPUz does change frequency reported on memory tab. I had to wait a few moments to time the screenshot with some background load for it to show 1600

 

[Martin]

And in idle what clock does CPU-Z show, 800 MHz perhaps?

 

[adam_b16]

Hi Martin,

seems to go through 3 speeds - 1066, 1333 (both attached) and 1600 (original post) - edit, added extra for 1600 to this post for easy reference

edit 2 - just spotted another step - 1466 (now also attached)

 

[Martin]

Thank you. Can you please make a HWiNFO Debug File that would capture such clock changes?

 

[adam_b16]

Well, there are two methods how to determine current memory clock and they return different values.
Anyway, the memory clock should be changing depending on load/usage. Try to put some load on the CPU (e.g. benchmark) if the memory clock will ramp up in HWiNFO.

Also to reply to this with a log file now that I'm home - Geekbench 5 benchmark at 500ms shows constant 1333 +/- 0.4, while showing the Tcas jumping up and down the JEDEC speed steps during the stages of the benchmark. Debug mode was not on for this bench, but can happily do it again with it on if you'd like.

Cheers!

 

[adam_b16]

Thank you. Can you please make a HWiNFO Debug File that would capture such clock changes?

ahh, you read my mind. 5 minutes I'll do it now

 

[adam_b16]

Thank you. Can you please make a HWiNFO Debug File that would capture such clock changes?

Attached as requested, with csv log file as well.

 

[Martin]

Thanks. This is quite strange as both methods to read memory clock report a static value for HWiNFO..

 

[adam_b16]

No worries, hope it doesn't cause any headaches for you!

I found the issue when the stock 1600mhz (3200) memory was showing 1333mhz (2666) in a review of the laptop (Notebookcheck showed HWinfo screenshots), and after purchasing, my laptop was doing the same. Latest bios used the whole time.

Cheers!
-Adam

 

[Martin]

No problem, I'm still looking into this.
Please run this build and create a new Debug File while the memory clock is changing: www.hwinfo.com/beta/hwi64_737_4991.zip

 

[adam_b16]

Hi Martin,

Memory clock remaining constant +/- tiny fractions. Debug file and csv log attached.

Cheers!

 

[Martin]

Thanks. Now I have used 3 different methods to read memory clock and they all report a constant value all the time. This is a big mystery...

 

[adam_b16]

If it helps at all, I was also chasing with Toshiba support why 3dmark and GPUz are reporting 2666 for integrated video memory. Attached snips from those.

So far, correct reporting of variable memory speed is only CPUz.

Windows task manager shows 3200 constantly, Userbench picks up ram "running @ 3200", your HWinfo report screen shows memory speed 1600mhz (ddr4-3200).

Aida64 picks up variable memory speed, but readings vary a lot, and I've only seen up to 2666 - might run a few more of their benchmark runs to see if I can catch a 3200.

Cheers!

 

[adam_b16]

Well! First run after mentioning above, and it logged 1600! Attached snip. Hope that helps!

edit: oddly I just noticed it reads DDR4-1600... meaning it would be 800mhz... it also has SDRAM after that when I extend the column.... odd

 

[Martin]

Update: I think I finally figured out what the problem is here. Will try to fix this in the next version.

 

[adam_b16]

Update: I think I finally figured out what the problem is here. Will try to fix this in the next version.

Ahh very nice. Let me know if you want me to test another beta for you. Happy to be of assistance.

Cheers!