• 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.

Ryzen 3900X thread

anybody running a 3900x in a Gigabyte x570 board?
been through CTR guide a few times and for the life of me cant find the default bios settings for using CTR.

I know AMD cool & Quiet should be enabled and also PBO turned off, but what were the other settings???.

Ran CTR in its 1st edition and setup the bios correctly beforehand. now I`ve updated to F30 and want to re-apply above settings (and the rest).

These were the results with CTR 1.0. (Beta)
CPU Silver Sample.
Recomended values for overclocking.
Ref Freq 4275 Mhz
Ref Volt 1250 mV
Recomended values for underclock.
Ref Freq 4150 Mhz
Ref Volt 1175 mV

Thanks Folks.
 


Temps seem OK, and it's a B350 MSI Krait Gaming. So not great, not terrible.
Bare in mind it will peg all the cores just fine on the multi core run.

It just scored a 511 which seems to be where it should be.
 
So I guess something is bottle necking the CPU, memory not feeding it fast enough maybe?
Oh, but then how could it max all cores just fine on the multi core run..

I just set the memory to XMP 2 in the bios, it was never able to boot @ 3200 before - but it wont post with the CTL at 14 or 15

Corsair 16GB White Vengeance LPX DDR4 3200MHz RAM/Memory Kit 2x 8GB
 
YoungBlood said:
Morning guys,
So I dropped in my 3900XT to replace a 1600

Is it normal when running Cinebench single core, for the core to stutter like this?
I thought it would peg the core at 100%?
Click to expand...

What is the final score?

Edit

Just seen it, 511 is just about right.. Nothing wrong with it..

NathG79 said:
anybody running a 3900x in a Gigabyte x570 board?
been through CTR guide a few times and for the life of me cant find the default bios settings for using CTR.

I know AMD cool & Quiet should be enabled and also PBO turned off, but what were the other settings???.

Ran CTR in its 1st edition and setup the bios correctly beforehand. now I`ve updated to F30 and want to re-apply above settings (and the rest).

These were the results with CTR 1.0. (Beta)
CPU Silver Sample.
Recomended values for overclocking.
Ref Freq 4275 Mhz
Ref Volt 1250 mV
Recomended values for underclock.
Ref Freq 4150 Mhz
Ref Volt 1175 mV

Thanks Folks.
Click to expand...

Those options depends a lot from bios to bios. Disabling SVM is optional, and not all BIOSes have this option. LLC should be set to Turbo as your board is Gigabyte. Phase mode and current capability is just for Asus boards. I'd say just change you LLC and try. I don't find out the tool to be useful to me, much time before the tool had been released I'd already found my max clocks for each ccx, same voltage as CTR indicates (that was my limit anyway, 1.25V) but better clocks. CTR always gives me lower clocks. Plus every version gives me a different rating for the cpu, gold, silver, even bronze. Not that I would care about the CTR rating but "we know" that 4.3GHz all core with 1.25 would be a good sample and I'm able to do 4.3GHz all core with less than that, and stable. So the CTR rating is not to be trusted.
Not to mention you will lose the boost, therefore single core performance. In my testing with games (my usage scenario) I did not benefit a lot from CTR to let me a software deal with the clocks and voltage. 3DMark ended up with gains within margin of error. BF5 and Warzone did not show a good increase in FPS, hence the fact that when on auto the cores hoover around 4100-4300MHz and CTR suggested 4325/4375/4375/4325. If you do a lot of rendering, heavy tasks perhaps the tool is useful once the clock set will be higher than the auto clock under heavy load. But in most of those other scenarios it won't. My personal opinion, of course.
 
Last edited:
Thomasck said:
What is the final score?

Edit

Just seen it, 511 is just about right.. Nothing wrong with it..



Those options depends a lot from bios to bios. Disabling SVM is optional, and not all BIOSes have this option. LLC should be set to Turbo as your board is Gigabyte. Phase mode and current capability is just for Asus boards. I'd say just change you LLC and try. I don't find out the tool to be useful to me, much time before the tool had been released I'd already found my max clocks for each ccx, same voltage as CTR indicates (that was my limit anyway, 1.25V) but better clocks. CTR always gives me lower clocks. Plus every version gives me a different rating for the cpu, gold, silver, even bronze. Not that I would care about the CTR rating but "we know" that 4.3GHz all core with 1.25 would be a good sample and I'm able to do 4.3GHz all core with less than that, and stable. So the CTR rating is not to be trusted.
Not to mention you will lose the boost, therefore single core performance. In my testing with games (my usage scenario) I did not benefit a lot from CTR to let me a software deal with the clocks and voltage. 3DMark ended up with gains within margin of error. BF5 and Warzone did not show a good increase in FPS, hence the fact that when on auto the cores hoover around 4100-4300MHz and CTR suggested 4325/4375/4375/4325. If you do a lot of rendering, heavy tasks perhaps the tool is useful once the clock set will be higher than the auto clock under heavy load. But in most of those other scenarios it won't. My personal opinion, of course.
Click to expand...

Thanks for the info.
 
YoungBlood said:
Morning guys,
So I dropped in my 3900XT to replace a 1600

Is it normal when running Cinebench single core, for the core to stutter like this?
I thought it would peg the core at 100%?
Click to expand...


That is normal, afaik it is to do with the work AMD did with Microsoft around how thread scheduling works in Windows.

It is bouncing the thread between cores to spread the thermal load, so as to help keep the single core boost higher but Windows is aware enough to keep the thread on the same CCD (maybe same CCX?) so as not to induce latency by having to move between caches.

If you overlaid the two graphs you should notice they are practically out of phase with each other, so the two of them added together would give you the solid 100% load.

Run it again with HWiNFO open and look at the effective clocks of the threads, you should see the clocks change there as the load moves around.
 
I've been experimenting with overclocking the 3900X and currently it's running all-core at 4.3Ghz with a VID of 1.35v.

I've only had it at this frequency for about a day but so far all seems fine - for general use and gaming. I've yet to see how it handles rendering or extended software use, though.

Out of interest, if I left it with an all-core overclock like this would the power consumption shoot through the roof, or doesn't it make all that much difference?
 
arc@css said:
I've been experimenting with overclocking the 3900X and currently it's running all-core at 4.3Ghz with a VID of 1.35v.

I've only had it at this frequency for about a day but so far all seems fine - for general use and gaming. I've yet to see how it handles rendering or extended software use, though.

Out of interest, if I left it with an all-core overclock like this would the power consumption shoot through the roof, or doesn't it make all that much difference?
Click to expand...

I won't get into a "Zen 2 safe voltage" debate, as that just causes arguments over what is deemed safe, but I will say 1.35v for 4.3GHz seems high for that speed, mine does 4.3GHz at 1.181v, for example. Whilst I know every chip is different, that voltage for that speeds seems a bit off.

In answer to your question, no, you'll likely see your power consumption go down on most tasks unless you're powering through a particularly heavy benchmark/stress test, like Prime95, but also bear in mind you're losing single core performance with a fixed OC.

I'm an avid OC'er, have been for many years, but even I don't see the point in OC'ing Zen 2 or 3, if I'm honest.
 
Spec-Chum said:
I won't get into a "Zen 2 safe voltage" debate, as that just causes arguments over what is deemed safe, but I will say 1.35v for 4.3GHz seems high for that speed, mine does 4.3GHz at 1.181v, for example. Whilst I know every chip is different, that voltage for that speeds seems a bit off.

In answer to your question, no, you'll likely see your power consumption go down on most tasks unless you're powering through a particularly heavy benchmark/stress test, like Prime95, but also bear in mind you're losing single core performance with a fixed OC.

I'm an avid OC'er, have been for many years, but even I don't see the point in OC'ing Zen 2 or 3, if I'm honest.
Click to expand...

Thanks. I'll look into trying a lower VID and see how I get on.

I'm trying to learn a bit more about the various software associated with Ryzen, and just generally get to grips with my processor since I haven't really spent much time looking into it since purchasing.
 
@Thomasck Cheers, I've just tried setting it to 1.275 (4.3Ghz) and will see how it goes.

Am I right in thinking that 4.3 - 4.4 all-core tends to be the most people generally get out of the 3900X?
 
arc@css said:
@Thomasck Cheers, I've just tried setting it to 1.275 (4.3Ghz) and will see how it goes.

Am I right in thinking that 4.3 - 4.4 all-core tends to be the most people generally get out of the 3900X?
Click to expand...

My old 3900x would do 4.4 at 1.2875v stable, but all core OC didn't suit my usage criteria. I believe that's generally about all you will get from these though, which is a bit of a shame. Maybe if Ryzen clock tuner matures and works for real world scenarios outside of benchmarking that might be the answer though.
 
KungFuSpaghetti said:
My old 3900x would do 4.4 at 1.2875v stable, but all core OC didn't suit my usage criteria. I believe that's generally about all you will get from these though, which is a bit of a shame. Maybe if Ryzen clock tuner matures and works for real world scenarios outside of benchmarking that might be the answer though.
Click to expand...

I see. Out of interest, how do you find the 5800X compared to the 3900X? Thanks.
 
arc@css said:
I see. Out of interest, how do you find the 5800X compared to the 3900X? Thanks.
Click to expand...

For my use which is basically gaming... pretty big improvement as I rarely utilised 12 cores outside of messing about streaming and ripping blue rays for my son's tablet.
The 5800x runs warmer when left unchecked, but the cores are significantly stronger, my worst one hits over 5ghz... to me it's been 100% worth the change and there's more to tinker and mess with, which I enjoy doing.
 
You must log in or register to reply here.
Back
Top Bottom