A software update required a restart so the driver is not installed. 'Sensors' only returns minimal output. Would you mind refreshing me with the steps from this point so that I don't inadvertently cause a SNAFU please.
jsm
(Currently running max 80% and will jump down to 40% next weekend)

Navigate to where your asus-wmi-sensors directory is and drop into a Terminal.

sudo modprobe asus-wmi-sensors

Assuming you haven't deleted that directory with the compiled driver. If you have you will have to download it again and run the make and make install commands.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Congratulations!

I see your Threadripper 2990wx has passed the 25,000+ RAC plus mark.

And I see a lot more 1 hour tasks. And a fast review didn't show any 6+ hour cpu tasks which I do remember from when I was running 60 threads on my 2990wx.

Excellent!!!

Tom

---

A proud member of the OFA (Old Farts Association).

But still a lot of 3+ hour tasks. Could be much better and closer overall to 1 hour for ALL tasks.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

But still a lot of 3+ hour tasks. Could be much better and closer overall to 1 hour for ALL tasks.

Yup. I think I remember it getting down to that kind of performance at 26 threads. But even 30 threads were "much faster".

Tom

---

A proud member of the OFA (Old Farts Association).

Curious to see whether any of his test data sets ends up at 36 cpu counts. From what I remember in the online reviews of the platform when released, that was a trip point where performance degraded rapidly with any higher usage counts.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I thought you might like to see the latest from ASUS support re 'quiet n cool'.


Jim, after performing further research, please see the following update below:
​We hide this selection for some special bugs from AMD. It cannot be disabled even we select disable. We are unsure whether AMD solved this particular problem.

We are currently coordinating to our senior product manager for more details.

Please be advised that we can not enable/disable, because of AMD.

If you have any questions or comments, you can simply reply to this email to contact us again. Thank you for choosing ASUS Products and Services.

Please take note of your case number for future reference N190376588.

Regards,
Olga O.
ASUS Product Support

http://www.asus.com/us/support/

I thought you might like to see the latest from ASUS support re 'quiet n cool'.


Jim, after performing further research, please see the following update below:
​We hide this selection for some special bugs from AMD. It cannot be disabled even we select disable. We are unsure whether AMD solved this particular problem.

We are currently coordinating to our senior product manager for more details.

Please be advised that we can not enable/disable, because of AMD.

Darn. That means you are losing productivity because of Cool 'n Quiet on that brand of MB. Maybe on all brands even.

Tom

---

A proud member of the OFA (Old Farts Association).

Yes, that means that AMD moved the Cool 'n Quiet into the AGESA code which only they have control over. The OEM vendors have no access. So, in AMD's opinion, that bit of code is off limits to anyone but them and if they don't want to expose it, then tough luck.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

yeap but i kind of dont blame them. it was a running joke for a long time with amd mobo. it work or it simple did not.

---

The RAC on the 2990wx Threadripper is still climbing.

I wonder how many threads jsm is running now?

Tom

---

A proud member of the OFA (Old Farts Association).

I thought he should be down to 48 or so by now. He said his intervals were going to be a week before dropping the task count. His APR is still lousy at 15 even though he is turning in over 300 tasks a day.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Confirm as promised now running at 40% preference. Will see how that plays out over a week.
jsm

Confirm as promised now running at 40% preference. Will see how that plays out over a week.
jsm

Good to hear. So you can establish the baseline at 26 concurrent cpu tasks with the stock cpu app. Still seeing some pretty poor times though. Have you checked to make sure your memory is still running at the correct XMP ratings? If you run

lshw -C memory

from root terminal it will print out the memory clock speed for each installed stick. Also not sure whether you are in NUMA or UMA mode or the interleaving mode or channel mode for the memory.

You could get the Intel Memory Latency Checker and run it to see where the bandwidth, transfer rate and latency falls out for interleaved versus channel mode. Interleaved will have higher throughput at higher latency and channel mode will have lower throughput and better latency. The gaming people are always begging for lower latency because it improves the frames per second rate. For someone doing computational work, the higher throughput is more beneficial and the latency is not as important. Of course reducing latency via the BIOS settings is still better overall. and independent on the method the cpu is accessing the memory. https://software.intel.com/en-us/articles/intelr-memory-latency-checker But since I have not tested a 2990WX I am not sure how it behaves with four memory nodes. All I have been able to test is my 2920X and it only has two memory nodes so probably not applicable as an example.

After you have baselined the stock Linux cpu app, you really should then test the optimized Linux cpu apps from the Lunatics site. http://lunatics.kwsn.info/index.php?action=downloads;cat=48

Best candidates would be the SSE41 or the AVX apps for testing. The SSE41 is around 30-50% faster than the stock cpu app in my opinion. You would have to write your own app_info for those applications or use the new TBar All-in-One application which has the SSE41 as the default cpu app. The BOINC version level is 7.14.2. You can edit out the parts of the app_info for the gpu application since you aren't running any gpu tasks. http://www.arkayn.us/lunatics/BOINC.7z

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Output from lshw looks OK to me:
*-firmware
description: BIOS
vendor: American Megatrends Inc.
physical id: 0
version: 1701
date: 01/09/2019
size: 64KiB
capacity: 15MiB
capabilities: pci apm upgrade shadowing cdboot bootselect socketedrom edd int13floppy1200 int13floppy720 int13floppy2880 int5printscreen int9keyboard int14serial int17printer acpi usb biosbootspecification uefi
*-memory
description: System Memory
physical id: 3c
slot: System board or motherboard
size: 32GiB
*-bank:0
description: [empty]
product: Unknown
vendor: Unknown
physical id: 0
serial: Unknown
slot: DIMM 0
*-bank:1
description: DIMM DDR4 Synchronous Unbuffered (Unregistered) 3000 MHz (0.3 ns)
product: CM4X8GD3000C16K4D
vendor: Unknown
physical id: 1
serial: 00000000
slot: DIMM 1
size: 8GiB
width: 64 bits
clock: 3GHz (0.3ns)
*-bank:2
description: [empty]
product: Unknown
vendor: Unknown
physical id: 2
serial: Unknown
slot: DIMM 0
*-bank:3
description: DIMM DDR4 Synchronous Unbuffered (Unregistered) 3000 MHz (0.3 ns)
product: CM4X8GD3000C16K4D
vendor: Unknown
physical id: 3
serial: 00000000
slot: DIMM 1
size: 8GiB
width: 64 bits
clock: 3GHz (0.3ns)
*-bank:4
description: [empty]
product: Unknown
vendor: Unknown
physical id: 4
serial: Unknown
slot: DIMM 0
*-bank:5
description: DIMM DDR4 Synchronous Unbuffered (Unregistered) 3000 MHz (0.3 ns)
product: CMK16GX4M2D3000C16
vendor: Unknown
physical id: 5
serial: 00000000
slot: DIMM 1
size: 8GiB
width: 64 bits
clock: 3GHz (0.3ns)
*-bank:6
description: [empty]
product: Unknown
vendor: Unknown
physical id: 6
serial: Unknown
slot: DIMM 0
*-bank:7
description: DIMM DDR4 Synchronous Unbuffered (Unregistered) 3000 MHz (0.3 ns)
product: CMK16GX4M2D3000C16
vendor: Unknown
physical id: 7
serial: 00000000
slot: DIMM 1
size: 8GiB
width: 64 bits
clock: 3GHz (0.3ns)
*-cache:0
description: L1 cache
physical id: 3f
slot: L1 - Cache
size: 3MiB
capacity: 3MiB
clock: 1GHz (1.0ns)
capabilities: pipeline-burst internal write-back unified
configuration: level=1
*-cache:1
description: L2 cache
physical id: 40
slot: L2 - Cache
size: 16MiB
capacity: 16MiB
clock: 1GHz (1.0ns)
capabilities: pipeline-burst internal write-back unified
configuration: level=2
*-cache:2
description: L3 cache
physical id: 41
slot: L3 - Cache
size: 64MiB
capacity: 64MiB
clock: 1GHz (1.0ns)
capabilities: pipeline-burst internal write-back unified
configuration: level=3
threadripper:~/Downloads/asus-wmi-sensors-master>

Which memory access mode are you in, Channel or Auto? IOW, NUMA versus UMA. Also have you had a look at your memory transfer rates yet with MLC?

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

A good article at TechPowerUp magazine detailing the new features and changes to the memory architecture in upcoming "Matisse" or Ryzen 3000 series processors this year.

https://www.techpowerup.com/253954/amd-ryzen-3000-zen-2-bios-analysis-reveals-new-options-for-overclocking-tweaking

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

A good article at TechPowerUp magazine detailing the new features and changes to the memory architecture in upcoming "Matisse" or Ryzen 3000 series processors this year.

https://www.techpowerup.com/253954/amd-ryzen-3000-zen-2-bios-analysis-reveals-new-options-for-overclocking-tweaking

I am hoping for a reduced price on a 16c/32t AM4 chip :)

---

A proud member of the OFA (Old Farts Association).

Update after a week running with 28 threads. I am sorry to have to report that - on my system with this hardware and software - this is NOT the sweet spot. My criterion is to monitor the daily credit on the three AMD machines (two 1800X and 2990wx) because I believe that eliminates extraneous factors such as internet or BOINC problems as all three are on the same network. When running at 50 threads then the comparison averaged 1.25 * ryzen1 + ryzen2. After seven days at 28 threads this has fallen to less than 1.00. I am conscious of advice re other operators but that is my finding.
I am now going to repeat the test before trying anything else including changing the software by running at max threads for a week. It will be interesting to see if the throughput increases or falls.
However, personally I am not doing badly in the overall scheme of things. I have broken through 3000 myself and driven my team isle of man towards 800 while I am steadily overhauling the top spot in Isle of Man as a country. And all that in 2 1/2 years as I lost all the original credit when SETI first started.
jsm
P.S. After reading the tech article on zen3 a couple of times I realise that my 'geekiness' has all but evaporated as I found it very hard going ):

P.S. After reading the tech article on zen3 a couple of times I realise that my 'geekiness' has all but evaporated as I found it very hard going ):

Nah, your geekiness hasn't changed. They just moved the goal posts!

Tom

---

A proud member of the OFA (Old Farts Association).