AMD Ryzen For Simracing?

I just looked at both my machines using PBO and let the voltage do its thing. After running 5x R20 all threads in a row the average voltages settled down at: 3800x 1.358v. 3950x 1.385v.

Are you sure this FIT module has a set value baked into the chip? and not simply dependant on the quality of the MB VRM, and your cooling? It's a good method to find out a safe value for your system instead of a random agreed voltage on the internet. So surely on a mental OC setup with say LN2 the voltages would automatically go even higher.
 
With PBO you will see high voltages but this only for microseconds as the load is shuffled around to prevent heat build up/degradation.

Hmmm those numbers I posted above were not spikes they were the lowest voltage numbers sustained with PBO on whilst doing a few Cinebench R20 Runs, both were spiking for split second at the start of the runs at around 1.42v. So with PBO on its sustaining and exceeding the safe numbers we are talking about anyway on both my systems. Voltage number as reported in Ryzen Master. Scratching my head, I guess PBO isn't safe to use long term either in that respect.
 
Overclocking ryzens are easy just up the voltage and change the multiplier,it’s the ram that can be dodgy,especially 4 sticks.

I got mine to 2800 ,will not do 3200 with 4 sticks.

2 sticks of 8 is ideal for ryzen.

I have the old 1700 cpu though.

I think the 3700x has better instructions,it may be better on ram speeds.
 
Have you followed the last few posts? Watch any of the video? So If I do like I did with my 1600, up the multiplier and set 1.36v on my 3700x you'll offer a replacement if it goes wrong?

Cheers for posting that also the degradation video he did which is linked to the one above. Very tricky subject in detail as there are many variables, but everyone should watch. So yeah I'll think twice before suggesting manual overclocking Ryzen 3000 again.

You shouldn't feel responsible for policing this though, it's AMD and the early adopter youtuber videos fault which have made this whole thing a bit of a mess. Plus standard practices in overclocking coming from Intel and previous generations.
 
I do apologise if that's how I come across.

Cheers!

Oh don't worry it's not a criticism.

Didn’t know the ryzen 3000 series were a bit sketchy with voltage.

It's not sketchy, there just isn't advice from AMD on manual overclocking Zen 2 safely, and the voltage / degradation depends on how much load you put through them, but as there is no way of checking if its degrading or not until its too late it's a risk not worth taking. All CPUs can degrade if you abuse them.

So the advice is put it in PBO with a non static voltage and let the chip overclock automatically. Or you can look into manual undervolting which is also worth doing, if you have a good chip.

Link to the Buildzoid degradation video, dry subject but worth knowing:
 
If you want to know what your CPU will do 3900X as example

4.2GHz allcores at 1.2v and stress test

Majority fall over at that and you are not stressing your CPU

I run 1.25v for sims and just clock for that, 5-10 fps with 1440p is not worth worrying about imo

I already have 3600 C14 @ stock

I do however understand chasing the dragon if you have triples or VR

I can do 4.35GGHz @1.3v easy but it is simply not worth it in my situation
 
Last edited:
And this is why I do above .........

@ 20C ambient it ony just hits 60C

You want to talk about degradation what does a extra 20+C do ? ')

4200.jpg



and 4.350GHz will run 24/7 as well but ... is it worth it ?

IT DEPENDS entirely on your PERSONAL situation ;)
I just don't see the advantage to extra fps IF you already have more then you need

rfactor2 physics sync is all I used to worry about endless when I had 2500K and GTX460 and 60Hz monitor

With current CPU/GPU I can just forget about it

4350.jpg
 
I was looking through my overclock settings yesterday as I've been hearing lots of different things about high or constant voltage and 3rd gen ryzen. Although it is set to run 1.28v fixed in the BIOS, according to ryzen master it is dropping voltage at idle/low useage still as per the Ryzen balanced plan. Quite happy with that myself, might drop it to 1.27v and see if it is still happy at 4350mhz all core.
 
CPU-Z heat is virtually same as my sims
my aim is to run 60C in sims or better @ 20C ambient
Not hard if you lower voltage to 1.25v ...still wish I could of got 1.2v ( silicone lottery) :(

I Disable sleep, hibernation, fast boot etc.
Yes I still use balanced plan myself on both 3200G and 3900X
although I think W10 "power throttling" is still active unless you use one
of the High Performance modes ?
 
P.S.

Mind you I have 280mm NXZT cooling on 3900X and the 3900X cooler on a 3200G so they both get
better then average temps

Silicone lottery would be 55C simming for my 3900X

Prime and other more stressful tests don't interest me anymore
I don't see how extra 10 fps is going to be any smoother with Gsync
I also run GTX2080 Gaming OC @stock except for initial testing

If I had VR I would be hang the 60C and push for everything, lucky I don't have to ;)

For new gen sims even rf2 a RTX2080 is only 1440p material in my book sure you can tweak this and that down and get acceptable performance but that is not me

I don't buy $2,000AU in gsync and gpu to look at anything less then Max gfx ;)
 
Last edited:
Here is my case in point with memory

I have identical sticks with same CPU and X570 Pro Wifi motherboard as opposed to X570 Master in review so results should be almost the same


To get those gains running @3733 24/7 it says you need a ram cooler ?

I have Phanteks Evolv X case has 1/2 decent airflow so at the 22c ambient they quoted ?

Dram calculator quotes 1.43v max where article needs 1.45v to get quoted results

I think i just better to run XMP ?

Also they only mention 3733@C14 " tight timings" no specifics
 
Here is my case in point with memory

I have identical sticks with same CPU and X570 Pro Wifi motherboard as opposed to X570 Master in review so results should be almost the same


To get those gains running @3733 24/7 it says you need a ram cooler ?

I have Phanteks Evolv X case has 1/2 decent airflow so at the 22c ambient they quoted ?

Dram calculator quotes 1.43v max where article needs 1.45v to get quoted results

I think i just better to run XMP ?

Also they only mention 3733@C14 " tight timings" no specifics

Tight timings means tuning your secondary and tertiary timings. Takes a while but it nets way more performance than just buying a better xmp kit as the performance impacting timings on xmp are so loose that the performance left on the table is massive.

Here's a frametime plot from Assassin's Creed which perfectly demonstrates the difference between xmp and tuning RAM (this is a 9900k btw)

xmp:
Wgsqq7y.png


manual tuning of all timings:
qUv7uTB.png


Same kits. Same everything else. Not just the framerate difference but notice the smoothness between the two graphs.
 
The Ryzen 3000 series responds best by setting the RAM frequency manually with tighter timings and letting the CPU self boost.
The cooler you keep it, obviously the better it performs.
I don't use any of those other features....no PBO ...no XMP.
I've set my cheap TimeTec 3600 to a fixed 3600 and tightened timings to 15, 19,19 19, 36 from the stock 19, 22, 22, 22, 42.
The performance difference is remarkable...especially in smoothness.
No temperature or stability issues whatsoever...and extremely smooth reliable gaming in everything.
Not sure why guys are still manually 'hard' overclocking these processors for indiscernible gains.
 
Last edited:

Latest News

To join the OverTake Racing Club races I want them to be: (multiple choice)

  • Free to access

    Votes: 64 88.9%
  • Better structured events

    Votes: 10 13.9%
  • Better structured racing club forum

    Votes: 12 16.7%
  • More use of default game content

    Votes: 6 8.3%
  • More use of fixed setups

    Votes: 19 26.4%
  • No 3rd party registration pages

    Votes: 22 30.6%
  • Less casual events

    Votes: 7 9.7%
  • More casual events

    Votes: 21 29.2%
  • Other, specify in thread

    Votes: 3 4.2%
Back
Top