Ryzen 3900X can't clock 1800MHz Infinity Fabric with 3600 CL14. Disappointed with Ryzen.

[Roddey]

Bank Group Swap

Bank Group Swap

From googling its sopposed to be in AMD

Bank Group Swap
View attachment 131583

I don't see those settings in the Prime x570 Pro Asus bios.

From googling its sopposed to be in AMD

I don't see those settings in the Prime x570 Pro Asus bios.

Not under the AMD settings that I can see.

 

[evilhf]

Here I get 3733 cl14 mem / 1867mhz IF at 3900x on an MSI X370 Gaming pro carbon card.
Set the IF voltage in OVERRYDE MODE to 1.115v.
With RyzenDramCalculator set the timings according to the memory frequency you will use.
I got incredible results with 3733mhz cl14.

UploadDeImagens.com.br - 7.4_original.jpg

I lovelly Lisa Su ❤

 

[Chomiq]

Bank Group Swap
View attachment 131583

My first thought was "sounds like a sex thing". No wonder they're using the acronym instead.

 

[Roddey]

Here I get 3733 cl14 mem / 1867mhz IF at 3900x on an MSI X370 Gaming pro carbon card.
Set the IF voltage in OVERRYDE MODE to 1.115v.
With RyzenDramCalculator set the timings according to the memory frequency you will use.
I got incredible results with 3733mhz cl14.

UploadDeImagens.com.br - 7.4_original.jpg

I lovelly Lisa Su ❤

The 3800x does ok in the benchmarks at 3600mhz c14.

 

[Shou Miko]

If no one have linked it lookand listen to Buildzoid @ Actually Hardcore Overclocking talk about Ram and FCLK settings vs performance on Ryzen 3000:

I think it's really interesting to listen to Buildzoid talking about different hardware.

 

[Roddey]

The 3800x does ok in the benchmarks at 3600mhz c14.View attachment 131664View attachment 131665

I do notice that I have to mess with the cpu fan curve to get it to boost that high. If I start the fans spinning up too late(benefit: it does keep the noise down)I can see the thermal limits engage and multi-core scores decrease. If I start the fans spinning up early the benches go up as I get ahead of the heat. This is the first time that I have seen fan speeds being so important on cpu performance. My previous system was a 4 core 4 thread 3.4 ghz. cpu with a noctua heatsink were I could just about turn the cpu fans off(5 volt) and it would cool it fine.

 

[TheLostSwede]

I do notice that I have to mess with the cpu fan curve to get it to boost that high. If I start the fans spinning up too late(benefit: it does keep the noise down)I can see the thermal limits engage and multi-core scores decrease. If I start the fans spinning up early the benches go up as I get ahead of the heat. This is the first time that I have seen fan speeds being so important on cpu performance. My previous system was a 4 core 4 thread 3.4 ghz. cpu with a noctua heatsink were I could just about turn the cpu fans off(5 volt) and it would cool it fine.

It's nothing to do with that really, I hit 4.6GHz with my fans barely running. Wait for the new AGESA it really changes the behaviour of the CPU.
Obviously, I have an AiO cooler, which makes a bit of a difference, but trust me, it has a lot more to do with the AGESA than your cooler.

HWinfo has been running the background for a couple of hours and it's like night and day compared to what it was like two AGESA versions ago. I had a hard 4.4GHz ceiling that simply couldn't be broken. Now I have two cores that hits 4.6GHz.

 

[Roddey]

It's nothing to do with that really, I hit 4.6GHz with my fans barely running. Wait for the new AGESA it really changes the behaviour of the CPU.
Obviously, I have an AiO cooler, which makes a bit of a difference, but trust me, it has a lot more to do with the AGESA than your cooler.

HWinfo has been running the background for a couple of hours and it's like night and day compared to what it was like two AGESA versions ago. I had a hard 4.4GHz ceiling that simply couldn't be broken. Now I have two cores that hits 4.6GHz.

View attachment 131675

Late reply. Took a drive through the Redwood forest of Northern California and took a walk along the beach of the pacific. Sept. is a nice time of year to be there. Slightly cool breeze coming off the ocean and the scenery is rare to find. But back now and I must say I cant hit anything like the core speeds your showing. Should get the new bios in a week or two. The core speeds look good and I can turn my fans down. My last PC was silent and I miss that.

 

[TheLostSwede]

Late reply. Took a drive through the Redwood forest of Northern California and took a walk along the beach of the pacific. Sept. is a nice time of year to be there. Slightly cool breeze coming off the ocean and the scenery is rare to find. But back now and I must say I cant hit anything like the core speeds your showing. Should get the new bios in a week or two. The core speeds look good and I can turn my fans down. My last PC was silent and I miss that.

Well, that's what the new AGESA is supposed to fix. Let's see how quickly Asus gets it out there.

 

[Roddey]

Well, that's what the new AGESA is supposed to fix. Let's see how quickly Asus gets it out there.

They have a beta out there. The boost are better.

 

[Espumador]

bottom line....your pushing your rams to hard for a pair of rams rated at 2666mhz XMP! using them at 3600mhz is already 1Gz more than the rated already OC 2666mhz XMP profile......so you are pushing them really hard and that is your probleme........cant run a Fiat 500 like a La Ferrari......your rams are top of the line since they are B-die and that is why you are able to push them to 3600. So i suggest you stick to 3600 or you will eventualy break them! cheers m8

 

[oobymach]

Lots of good info in this thread, tested and re-tested a bunch of different frequencies and ended up going from 4400mhz @ 74.5ns to 3666 @ 64.4 with faster read speed on the latter (went from 50gbps to 54.5gbps). The max stable fclk I could get is 1833 (any higher and onboard audio would crackle really bad) so x2 = 3666 which is the optimal speed for my system apparently, write and copy speeds are the same at the lower speed as they were at max. Really happy with the patriot viper ram kit.

 

[Raulsanchez]

X570 Aorus ultra 4X 8GB G.skill 3600 16-16-16-36 @ 3800 16-16-16-36 FCLK 1900 3900X. Default voltages, 24X7stable. Beats 9900KS in single thread Cinebench and Cpu-z. 555 Cpu-z single thread. Over 7800 Cinebench r20 @ 4300 1.375V forgot single core Cinebench score but was higher than 9900ks using PBO up to 4.65ghz single core burst

 

[Zach_01]

X570 Aorus ultra 4X 8GB G.skill 3600 16-16-16-36 @ 3800 16-16-16-36 FCLK 1900 3900X. Default voltages, 24X7stable. Beats 9900KS in single thread Cinebench and Cpu-z. 555 Cpu-z single thread. Over 7800 Cinebench r20 @ 4300 1.375V forgot single core Cinebench score but was higher than 9900ks using PBO up to 4.65ghz single core burst

If you have a static OC 4.3GHz with 1.375V have a second thought about that because the chances you will degrade your CPU is high...

 

[Raulsanchez]

If you have a static OC 4.3GHz with 1.375V have a second thought about that because the chances you will degrade your CPU is high...

Don't run that 24x7, just did few Cinebench runs. Actually run with default voltages and CPU clock speeds. Actually all my voltages are stock. Just turned spread spectrum off and run ram at 3800 and FCLK at 1900. No problems 24x,7 stable.. no need to overclock CPU, PBO does fine job

Don't run that 24x7, just did few Cinebench runs. Actually run with default voltages and CPU clock speeds. Actually all my voltages are stock. Just turned spread spectrum off and run ram at 3800 and FCLK at 1900. No problems 24x,7 stable.. no need to overclock CPU, PBO does fine job

I would never run the system 24/7 that the higher voltage. I also have excellent cooling with the Thor V2 case And nhd15. I went for function over form I love all those 230 mm fans

 

[Zach_01]

I was just wanted to say that these CPUs are not for core OC.
For DRAM, UCLK and FCLK it’s ok up to a point.

 

[Raulsanchez]

Yes sir, true. I hope 7nm+ ryzen 4000 parts bring higher clock speeds. Maybe 4.5 to 4.6 all core

Yes sir, true. I hope 7nm+ ryzen 4000 parts bring higher clock speeds. Maybe 4.5 to 4.6 all core

Zen 2 ipc is already superior to Intel latest parts

 

[moproblems99]

Yes sir, true. I hope 7nm+ ryzen 4000 parts bring higher clock speeds. Maybe 4.5 to 4.6 all core


Zen 2 ipc is already superior to Intel latest parts

I am curious, but why do you quote yourself and hold a conversation? Or did posts get moved around?