• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

5000 Series Undervolting

The Asgard

The Asgard

The Asgard

Soldato
Joined
27 Jan 2003
Posts
5,962
Location
Chesterfield, UK
I have had my 5900X for a couple of weeks now. I am a big fan of undervolting but my previous experience was just with Intel. So I have been playing with my chip and have been very surprised on how much I can push it.

I am currently running with a CO on all cores of -30. PBO disabled. Stable no problems. So tweaked the LLC to Mode 5 to bring the voltage overshoot down. Again stable. I then tried some negative offset -0.1 with a 50 Mhz boost to get a 5 Ghz boost. Again stable. I then removed the boost and tried for a bigger offset so I am now -0.1125. Machines been up all day building with Visual Studio with some Satisfactory at lunch, again stable. Amazing, no crashes, no WHEA errors.

I do have a custom loop but the rad fans are off. Idle temps of 34'c, Cinebench R23 multi core test in the 50-60, fans are quiet.

Here is a link to my Hwinfo for today https://photos.app.goo.gl/zRinUcrp9bnnrdFL8

How are you guys finding these chips?
 
The Asgard said:
I have had my 5900X for a couple of weeks now. I am a big fan of undervolting but my previous experience was just with Intel. So I have been playing with my chip and have been very surprised on how much I can push it.

I am currently running with a CO on all cores of -30. PBO disabled. Stable no problems. So tweaked the LLC to Mode 5 to bring the voltage overshoot down. Again stable. I then tried some negative offset -0.1 with a 50 Mhz boost to get a 5 Ghz boost. Again stable. I then removed the boost and tried for a bigger offset so I am now -0.1125. Machines been up all day building with Visual Studio with some Satisfactory at lunch, again stable. Amazing, no crashes, no WHEA errors.

I do have a custom loop but the rad fans are off. Idle temps of 34'c, Cinebench R23 multi core test in the 50-60, fans are quiet.

Here is a link to my Hwinfo for today https://photos.app.goo.gl/zRinUcrp9bnnrdFL8

How are you guys finding these chips?
Click to expand...
Would be interested to see you Cinebench scores if possible please?
 
cliffy said:
Would be interested to see you Cinebench scores if possible please?
Click to expand...
I have not tweaked the memory and it’s got lots installed because it’s a development machine. Like SQL ... So not tuned to break records. I did run it last night with HWinfo up to monitor the multi core boost which was 4650 and single 5Ghz.

The score on R23 multi was 22,872 and Single 1580. Nothing special
 
The Asgard said:
I have not tweaked the memory and it’s got lots installed because it’s a development machine. Like SQL ... So not tuned to break records. I did run it last night with HWinfo up to monitor the multi core boost which was 4650 and single 5Ghz.

The score on R23 multi was 22,872 and Single 1580. Nothing special
Click to expand...
I do not know. Based on your R23 multi and the R23 thread you are running very competitively with other 5900x
 
i didnt think 5900x default boost is 5Ghz+. tis spec boost is 4.8GHz but i think it can boost to 4.9GHz out of box. if you have switched off PBO i dont know how it is managing 5GHz.

using LLC and Curve optimiser really doesnt make any sense, as one counter the effects of the other. in your case you are aggressively allowing the Vcore to droop even further.

Have you tested your individual cores and threads throughly with light single thread workload such as Prime95 Large with no AVX or Aida64 system memory test with affinity set to single thread at a time?

my 5800X managed to boot and under load with CO -30 on all cores but i only found the WHEA errors by using Prime95 Large FFT with no AVX and set affinity to individual cores and let it run for like 20min at a time then will the WHEA error show up.
 
Undervolting Ryzen 5000 is always one answer, Negative Curve Optimiser, i'm running -15 and +150Mhz, runs 5Ghz on all 16 threads all day in games.

Its about time AMD was firing on all cylinders again and these chips are just ####### brilliant, its also nice to see new innovative ways to tweak CPU's, Curve Optimiser seems like such a simple thing but its very clever in what it does, "Take off 40mv" is so 2008, this knows how much or how little voltage to remove or add for everything that it does and dynamically makes those adjustments literally from 1 milliseconds to the next to give you the most optimal results.
 
pc-guy said:
i didnt think 5900x default boost is 5Ghz+. tis spec boost is 4.8GHz but i think it can boost to 4.9GHz out of box. if you have switched off PBO i dont know how it is managing 5GHz.

using LLC and Curve optimiser really doesnt make any sense, as one counter the effects of the other. in your case you are aggressively allowing the Vcore to droop even further.

Have you tested your individual cores and threads throughly with light single thread workload such as Prime95 Large with no AVX or Aida64 system memory test with affinity set to single thread at a time?

my 5800X managed to boot and under load with CO -30 on all cores but i only found the WHEA errors by using Prime95 Large FFT with no AVX and set affinity to individual cores and let it run for like 20min at a time then will the WHEA error show up.
Click to expand...

Last night was with a 50 Mhz boost. Out of the box with the setting above boost's to 4950, light load. You can see from the Hwinfo.

Testing wise. Its used for development with a light and medium loads all day. At night I'm gaming. Just had a couple of hours in Satisfactory with temps of 48'c and boosts to 4950. I removed the 50 Mhz boost when I dropped the voltage to -0.1125. So testing is actually using it. No WHEA errors over different loads.

Currently running Prime 95 with Large FFT, All core boost 4550.2. 1.087 load with a temp of 56'c, No WHEA Errors, 15 minutes in going to let out the dogs and will report back. Crazy chip!
 
This thing sips power. I have a 1080ti, Aquaero, 3x 140mm, 1x 120, 2x NVME, 2x SSD, 2x Mecahnical, D5 Pump. The 3 fans on the rad never come on. Sits on the desktop drawing while idle 76 Watts according to my HX1000i. Power draw running Prime looking at the image is 172 watts

This setup uses less than my 8700K. Pretty amazing.
 
Last edited:
The Asgard said:
This thing sips power. I have a 1080ti, Aquaero, 3x 140mm, 1x 120, 2x NVME, 2x SSD, 2x Mecahnical, D4 Pump. The 3 fans on the rad never come on. Sits on the desktop drawing while idle 76 Watts according to my HX1000i. Power draw running Prime looking at the image is 172 watts

This setup uses less than my 8700K. Pretty amazing.
Click to expand...

And twice the number of core / threads
 
Its been a while since I used an AMD chip. The last one was a dual core Athlon 64. So far I have been extremely pleased with this setup. Very impressed with Tomahawk, build quality is up there with the Asus Hero's. The fan never comes on. No USB issues. Totally solid. Love the BIOS. Best upgrade I have had in a long time. AMD are onto a winner for sure.
 
Dropped the override down to -0.1250. I have done a full R23 run multi and single. Previous was with a 50 Mhz boost. 21682, 1561, Multi Core boost 4,650

Prime 95 Large, 23 minute run lowest load voltage was 1.050 volts. 168 watt total power draw, down from 172 Watts

No WHEA errors. Just going to leave it on the desktop.
 
The Asgard said:
Its been a while since I used an AMD chip. The last one was a dual core Athlon 64. So far I have been extremely pleased with this setup. Very impressed with Tomahawk, build quality is up there with the Asus Hero's. The fan never comes on. No USB issues. Totally solid. Love the BIOS. Best upgrade I have had in a long time. AMD are onto a winner for sure.
Click to expand...


Pentium III
Skt 478 Pentium IV 3.2Ghz HT (Good Grief... switched to AMD)

Athlon 3800+ (oh wow.....)
Athlon X2 5200+
Phenom II X6 1090T (one of my favourites)
FX 9590 (Embarrassed to admit it, switched back to Intel)

Core i7 930 (loved that chip, regretted selling it on)
Core i5 4690K (Was quite happy with that for a couple of years but it started to come over all 4 core, switched back to AMD again)

Ryzen 1600
Ryzen 3600
Ryzen 5800X (blows my mind, still)
 
If that 1561 is the single core score, then it strikes me that you have clock stretching happening (and that is why power is going down). On Zen 3 with sustained boost at ~4.7-4.8 GHz I see scores just into the low 1600s, so mid 1500s should very much like ~4.7 GHz or a touch below for the real clocks.
 
Yes 1561 is a single core score. Basically I just have C0, PBO Disabled, LLC and a negative offset. Default single core boost is 4950.2, The chips doing its thing given how cool its running along with my custom loop. I'm assuming
 
You must log in or register to reply here.
Back
Top Bottom