• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Definitive guide to configuring the Ryzen 3900X/3950X and all other 3000 Series CPUs

Status
Not open for further replies.
Joined
May 3, 2018
Messages
75 (0.03/day)
Negative voltage offset is not working with ZEN2 as it does with ZEN+. Not in the same manner anyway. ZEN2 is a whole new story and I've been saying it for months now that it needs to be treated from almost zero point, from scratch. Pretty match forget what we know about past CPUs. Just like static OC/voltage.

@Michael Nager has done a lot of testing, and found a way to setup these chips that suit his needs. Feel free to test it as I dont see anything that can compromise silicon longevity. Is it the best way? Is Buildzoid's best way? Probably not and that depends on what everyone is seeking for. There is no best way IMHO. I'm actually peeking a different approach from both of them.
It took me two months of utter frustration with the 3600X and my attempts to configure it, "the right way", that I realised that I pretty much had to toss out my 37 years of prior experience with computers (I am 60 years old) and start again from scratch.

The unlearning part was harder than the learning part :)

But this old dog can still learn new tricks.
 

eidairaman1

The Exiled Airman
Joined
Jul 2, 2007
Messages
42,218 (6.64/day)
Location
Republic of Texas (True Patriot)
System Name PCGOD
Processor AMD FX 8350@ 5.0GHz
Motherboard Asus TUF 990FX Sabertooth R2 2901 Bios
Cooling Scythe Ashura, 2×BitFenix 230mm Spectre Pro LED (Blue,Green), 2x BitFenix 140mm Spectre Pro LED
Memory 16 GB Gskill Ripjaws X 2133 (2400 OC, 10-10-12-20-20, 1T, 1.65V)
Video Card(s) AMD Radeon 290 Sapphire Vapor-X
Storage Samsung 840 Pro 256GB, WD Velociraptor 1TB
Display(s) NEC Multisync LCD 1700V (Display Port Adapter)
Case AeroCool Xpredator Evil Blue Edition
Audio Device(s) Creative Labs Sound Blaster ZxR
Power Supply Seasonic 1250 XM2 Series (XP3)
Mouse Roccat Kone XTD
Keyboard Roccat Ryos MK Pro
Software Windows 7 Pro 64
The thing is that I have been experimenting with this motherboard for four months or so now.

The other thing is, that when I revert the settings and just do it the way that I know WORKS then there are absolutely no problems or anomalous behaviour.

It has gotten to the stage now that I can take one look at the temp the CPU is reaching when running CineBench and I can tell you what score is going to result, and I won't me more than one or two points off.


Hell yeah I tried that.

It was a shitshow.

Utilities like Ryzen Master or HWMonitor or HWInfo or AIDA were showing me that I was using 0 (Zero) Volts going into the CPU. The temps were higher to the point where I had to stop CineBench runs and Windows would hard crash.

From some of the replies to my original post I get the feeling that people think that I WANT to do it the way I documented and that I am too much of a wimp to do the "Manly man" thing and configure it in the BIOS.

The thing is that the CPU is listening to the AGESA portion of the BIOS and NOT the GigaByte portion of the BIOS.

The other problem I suspect lies in the fact that this time around AMD itself is producing the chipset - which as far as I know will change with the next chipset where it will be manufactured by ASMedia again.

I will be testing out the 3900X my friend loaned me on the GigaByte X470 Gaming 7 WiFi motherboard I have to see what, if any, differences there are, because the chipset for that board was manufactured by ASMedia.

I am watching the Gamers Nexus factory tour video and Steve said something that really made me laugh, he said:

This just simplified my board Picks, Either Asus Crosshair or AsRock Taichi (w/e their top end is)

And Ryzen/Core I are way different than AM3+/775 or older.
 
Joined
Nov 11, 2019
Messages
20 (0.01/day)
I'm going to try the negative offset with a 3600 in a few days, hopefully I have better luck than you had Michael. The stock voltages seem like massive overkill, but I don't want to lose single thread performance.
 
Joined
Sep 3, 2019
Messages
3,515 (1.84/day)
Location
Thessaloniki, Greece
System Name PC on since Aug 2019, 1st CPU R5 3600 + ASUS ROG RX580 8GB >> MSI Gaming X RX5700XT (Jan 2020)
Processor Ryzen 9 5900X (July 2022), 220W PPT limit, 80C temp limit, CO -6-14, +50MHz (up to 5.0GHz)
Motherboard Gigabyte X570 Aorus Pro (Rev1.0), BIOS F39b, AGESA V2 1.2.0.C
Cooling Arctic Liquid Freezer II 420mm Rev7 (Jan 2024) with off-center mount for Ryzen, TIM: Kryonaut
Memory 2x16GB G.Skill Trident Z Neo GTZN (July 2022) 3667MT/s 1.42V CL16-16-16-16-32-48 1T, tRFC:280, B-die
Video Card(s) Sapphire Nitro+ RX 7900XTX (Dec 2023) 314~467W (375W current) PowerLimit, 1060mV, Adrenalin v24.10.1
Storage Samsung NVMe: 980Pro 1TB(OS 2022), 970Pro 512GB(2019) / SATA-III: 850Pro 1TB(2015) 860Evo 1TB(2020)
Display(s) Dell Alienware AW3423DW 34" QD-OLED curved (1800R), 3440x1440 144Hz (max 175Hz) HDR400/1000, VRR on
Case None... naked on desk
Audio Device(s) Astro A50 headset
Power Supply Corsair HX750i, ATX v2.4, 80+ Platinum, 93% (250~700W), modular, single/dual rail (switch)
Mouse Logitech MX Master (Gen1)
Keyboard Logitech G15 (Gen2) w/ LCDSirReal applet
Software Windows 11 Home 64bit (v24H2, OSBuild 26100.2161), upgraded from Win10 to Win11 on Jan 2024
The thing is that the CPU is listening to the AGESA portion of the BIOS and NOT the GigaByte portion of the BIOS.

The other problem I suspect lies in the fact that this time around AMD itself is producing the chipset - which as far as I know will change with the next chipset where it will be manufactured by ASMedia again.

I will be testing out the 3900X my friend loaned me on the GigaByte X470 Gaming 7 WiFi motherboard I have to see what, if any, differences there are, because the chipset for that board was manufactured by ASMedia.
This just simplified my board Picks, Either Asus Crosshair or AsRock Taichi (w/e their top end is)

And Ryzen/Core I are way different than AM3+/775 or older.
I think that this is for every board. No UEFI can by pass or ignore AGESA microcode. Its the core function and behaviour of the CPU that needs to listen. Unless static OC is intoduced. That is a different story.

I'm going to try the negative offset with a 3600 in a few days, hopefully I have better luck than you had Michael. The stock voltages seem like massive overkill, but I don't want to lose single thread performance.
You can try but 99% you wont see any clock increase, and from point after you will see clock reducing. I tried it a lot with my 3600...
 

eidairaman1

The Exiled Airman
Joined
Jul 2, 2007
Messages
42,218 (6.64/day)
Location
Republic of Texas (True Patriot)
System Name PCGOD
Processor AMD FX 8350@ 5.0GHz
Motherboard Asus TUF 990FX Sabertooth R2 2901 Bios
Cooling Scythe Ashura, 2×BitFenix 230mm Spectre Pro LED (Blue,Green), 2x BitFenix 140mm Spectre Pro LED
Memory 16 GB Gskill Ripjaws X 2133 (2400 OC, 10-10-12-20-20, 1T, 1.65V)
Video Card(s) AMD Radeon 290 Sapphire Vapor-X
Storage Samsung 840 Pro 256GB, WD Velociraptor 1TB
Display(s) NEC Multisync LCD 1700V (Display Port Adapter)
Case AeroCool Xpredator Evil Blue Edition
Audio Device(s) Creative Labs Sound Blaster ZxR
Power Supply Seasonic 1250 XM2 Series (XP3)
Mouse Roccat Kone XTD
Keyboard Roccat Ryos MK Pro
Software Windows 7 Pro 64
I think that this is for every board. No UEFI can by pass or ignore AGESA microcode. Its the core function and behaviour of the CPU that needs to listen. Unless static OC is intoduced. That is a different story.


You can try but 99% you wont see any clock increase, and from point after you will see clock reducing. I tried it a lot with my 3600...

I'm not worried about that just from the amount of trouble he had with the ga board just to do a bios flash, no thanks.

Also I research before i start any overclocking adventure.
 
Joined
Mar 10, 2015
Messages
3,984 (1.12/day)
System Name Wut?
Processor 3900X
Motherboard ASRock Taichi X570
Cooling Water
Memory 32GB GSkill CL16 3600mhz
Video Card(s) Vega 56
Storage 2 x AData XPG 8200 Pro 1TB
Display(s) 3440 x 1440
Case Thermaltake Tower 900
Power Supply Seasonic Prime Ultra Platinum
It was a shitshow.

Utilities like Ryzen Master or HWMonitor or HWInfo or AIDA were showing me that I was using 0 (Zero) Volts going into the CPU. The temps were higher to the point where I had to stop CineBench runs and Windows would hard crash.

So, I don't care how you setup your system, I won't lose sleep over it. However, what you are describing is something I have read no where else. From reading this, I believe you have additional problems.

I'm going to try the negative offset with a 3600 in a few days, hopefully I have better luck than you had Michael. The stock voltages seem like massive overkill, but I don't want to lose single thread performance.

I lost clocks with an underclock.
 
Joined
Dec 29, 2010
Messages
3,809 (0.75/day)
Processor AMD 5900x
Motherboard Asus x570 Strix-E
Cooling Hardware Labs
Memory G.Skill 4000c17 2x16gb
Video Card(s) RTX 3090
Storage Sabrent
Display(s) Samsung G9
Case Phanteks 719
Audio Device(s) Fiio K5 Pro
Power Supply EVGA 1000 P2
Mouse Logitech G600
Keyboard Corsair K95
Ways to run your Zen2...

Stock w/ PBO disabled - as default as it gets
Stock with PBO - as random as it gets
All core - single thread perf limited to all core frequency
Ratio overclock - best all around perf combining all core perf with higher single thread perf close to stock boost
 
Joined
May 3, 2018
Messages
75 (0.03/day)
Ways to run your Zen2...

Stock w/ PBO disabled - as default as it gets
Stock with PBO - as random as it gets
All core - single thread perf limited to all core frequency
Ratio overclock - best all around perf combining all core perf with higher single thread perf close to stock boost
This would be true EXCEPT in my original post I said:

Because of my back problems (I have had two spine operations and have spinal arthritis) I have to keep the room temperature pretty warm.

The ambient temperature in my room is 28 - 29 °C and you should keep that in mind when I show you the following benchmark results of my system:

The comfortable room temperature for normies like yourselves is around 20 °C (or 68 °F for those of you who have not quite made it into the 20th Century yet :))

When Buildzoid was doing his testing it was at a far colder ambient temperature than I have (he was doing the tests in a non heated room in the middle of Winter) and how temperature affects attainable clockspeed can be seen here in the tests that Gamers Nexus did with the 3950X and cold scaling:

1584270026035.png


The Infinity Fabric does not only become unstable at very low temperatures, but also at temperatures above 85 °C

Another thing to consider is that for every 10 °C rise in temperature the CPU requires 4% more energy to achieve the same performance.

The testing that Gamers Nexus did was only with PB2 and not with PBO.

In my circumstance it means that I am fighting for every degree of temperature I can shave off the CPU temp to get a result given the ambient temperature of my environment. It is for this reason that I have ordered the cooling kit from Der8auer which I linked to before and will be testing it out and if you guys are interested I can share the results of that with you.

Running CineBench R20 on a loop with an ambient temperature of 28 °C at 4.3 GHz on all cores with SMT On at 1.3 Volts I was getting a maximum CPU temp of 84 °C gives me the following score:

Capture.PNG


This score is not as high as what I previously had; however I upgraded the BIOS to version F12e the day before yesterday and I have recently upgraded Ryzen Master but have not tweaked the RAM settings as much as I had before.

These are the current settings for Ryzen Master and the temp you see is my idle temp, which is quite high as you will notice due the the ambient temperature being 28 °C

Ryzen Master settings.PNG


I am measuring the ambient temperature with a temperature sensor located just in front of the intake fans of my AIO radiator.
 
Joined
Nov 7, 2017
Messages
1,875 (0.73/day)
Location
Ibiza, Spain.
System Name Main
Processor R7 5950x
Motherboard MSI x570S Unify-X Max
Cooling converted Eisbär 280, two F14 + three F12S intake, two P14S + two P14 + two F14 as exhaust
Memory 16 GB Corsair LPX bdie @3600/16 1.35v
Video Card(s) GB 2080S WaterForce WB
Storage six M.2 pcie gen 4
Display(s) Sony 50X90J
Case Tt Level 20 HT
Audio Device(s) Asus Xonar AE, modded Sennheiser HD 558, Klipsch 2.1 THX
Power Supply Corsair RMx 750w
Mouse Logitech G903
Keyboard GSKILL Ripjaws
VR HMD NA
Software win 10 pro x64
Benchmark Scores TimeSpy score Fire Strike Ultra SuperPosition CB20
Sorry, but thats kind of a waste in my view. unless i missed something, your gpu isnt under water, biggest gain after WC cpu (with rad set as exhaust)
to lower cpu/case/vrm/ram temps, if gpu rad is setup as exhaust.
Not looking at it from a thermal transfer point, which is the biggest problem cooling zen 2, and where you would need to swap for a different block to see any improvement.
Going thru reviews for basically every corsair/arctic/alphacool 240/280 AIO and major brands for blocks,
the XPC raystorm pro seems to be the best to see any gains over stock/other aio/blocks.

XPC Raystorm Pro water block
 
Last edited:
Joined
Jun 28, 2014
Messages
2,388 (0.63/day)
Location
Shenandoah Valley, Virginia USA
System Name Home Brewed
Processor i9-7900X and i7-8700K
Motherboard ASUS ROG Rampage VI Extreme & ASUS Prime Z-370 A
Cooling Corsair 280mm AIO & Thermaltake Water 3.0
Memory 64GB DDR4-3000 GSKill RipJaws-V & 32GB DDR4-3466 GEIL Potenza
Video Card(s) 2X-GTX-1080 SLI & 2 GTX-1070Ti 8GB G1 Gaming in SLI
Storage Both have 2TB HDDs for storage, 480GB SSDs for OS, and 240GB SSDs for Steam Games
Display(s) ACER 28" B286HK 4K & Samsung 32" 1080P
Case NZXT Source 540 & Rosewill Rise Chassis
Audio Device(s) onboard
Power Supply Corsair RM1000 & Corsair RM850
Mouse Generic
Keyboard Razer Blackwidow Tournament & Corsair K90
Software Win-10 Professional
Benchmark Scores yes
Joined
May 3, 2018
Messages
75 (0.03/day)
Sorry, but thats kind of a waste in my view. unless i missed something, your gpu isnt under water, biggest gain after WC cpu (with rad set as exhaust)
to lower cpu/case/vrm/ram temps, if gpu rad is setup as exhaust.

I don't have water cooling on my EVGA 1080 Ti SC2.

There is a good reason for this. I had modded my original 1080 Ti SC2 and under full load for a number of hours a VRM went BANG! and that was end of the card. And no I was not running Furmark

I managed to put 100 MHz OC on the GPU and 250 MHz OC onto the RAM. I was testing out the card for stability with an eye to further pushing the OC, because the temps were pretty good.

EVGA replaced the card under warranty, but the lesson has been learned.

Sure you can get better thermals on the GPU with water cooling, but I would certainly NOT recommend running the graphics card on full load for many hours on end as I did.

Now if you take a look at my current main machine:

20190119_034143.jpg


The case itself is a Phanteks Evolv X case.

Unlike others who are stupid enough to think that a normal case fan is enough to punt air into the case, I have replaced the case fans with static pressure fans, in fact all the fans you see are the Noctua NF-A12x25.

Why on earth people think that normal case fans are good enough to draw in air through mesh which is a higher impediment to airflow than a radiator than static pressure optimised fans is beyond me.

The rear fan is moved down right above the back of the graphics card and the backplate of that graphics card is connected to the hotspots by thermal pads. It is drawing air over the back of the graphics card and at the same time drawing exhaust air away from graphics card as well.

Effectively I am isolating the GPU from the CPU and as the load on the system increases, the fans at the front spinning up will actually push air into the case instead of just being normal case fans which are spinning up and because they cannot pull the air through the mesh are just making a noise for next to no positive effect beyond a certain RPM.

The thing is, that I have been a techie for 38 years now and I have had a lot of experience with various computer systems over the years.

From the way you are talking it is as if you think that you are speaking with a n00b who is cerebrally challenged by the power switch of his computer and who doesn't know that the way to save a file on an Etch-A-Sketch is to super-glue it to a desk.
 
Joined
Mar 10, 2010
Messages
11,878 (2.21/day)
Location
Manchester uk
System Name RyzenGtEvo/ Asus strix scar II
Processor Amd R5 5900X/ Intel 8750H
Motherboard Crosshair hero8 impact/Asus
Cooling 360EK extreme rad+ 360$EK slim all push, cpu ek suprim Gpu full cover all EK
Memory Corsair Vengeance Rgb pro 3600cas14 16Gb in four sticks./16Gb/16GB
Video Card(s) Powercolour RX7900XT Reference/Rtx 2060
Storage Silicon power 2TB nvme/8Tb external/1Tb samsung Evo nvme 2Tb sata ssd/1Tb nvme
Display(s) Samsung UAE28"850R 4k freesync.dell shiter
Case Lianli 011 dynamic/strix scar2
Audio Device(s) Xfi creative 7.1 on board ,Yamaha dts av setup, corsair void pro headset
Power Supply corsair 1200Hxi/Asus stock
Mouse Roccat Kova/ Logitech G wireless
Keyboard Roccat Aimo 120
VR HMD Oculus rift
Software Win 10 Pro
Benchmark Scores 8726 vega 3dmark timespy/ laptop Timespy 6506
I don't have water cooling on my EVGA 1080 Ti SC2.

There is a good reason for this. I had modded my original 1080 Ti SC2 and under full load for a number of hours a VRM went BANG! and that was end of the card. And no I was not running Furmark

I managed to put 100 MHz OC on the GPU and 250 MHz OC onto the RAM. I was testing out the card for stability with an eye to further pushing the OC, because the temps were pretty good.

EVGA replaced the card under warranty, but the lesson has been learned.

Sure you can get better thermals on the GPU with water cooling, but I would certainly NOT recommend running the graphics card on full load for many hours on end as I did.

Now if you take a look at my current main machine:

View attachment 148201

The case itself is a Phanteks Evolv X case.

Unlike others who are stupid enough to think that a normal case fan is enough to punt air into the case, I have replaced the case fans with static pressure fans, in fact all the fans you see are the Noctua NF-A12x25.

Why on earth people think that normal case fans are good enough to draw in air through mesh which is a higher impediment to airflow than a radiator than static pressure optimised fans is beyond me.

The rear fan is moved down right above the back of the graphics card and the backplate of that graphics card is connected to the hotspots by thermal pads. It is drawing air over the back of the graphics card and at the same time drawing exhaust air away from graphics card as well.

Effectively I am isolating the GPU from the CPU and as the load on the system increases, the fans at the front spinning up will actually push air into the case instead of just being normal case fans which are spinning up and because they cannot pull the air through the mesh are just making a noise for next to no positive effect beyond a certain RPM.

The thing is, that I have been a techie for 38 years now and I have had a lot of experience with various computer systems over the years.

From the way you are talking it is as if you think that you are speaking with a n00b who is cerebrally challenged by the power switch of his computer and who doesn't know that the way to save a file on an Etch-A-Sketch is to super-glue it to a desk.
My GPU and CPU have been on full load their whole lives ,Vega bought day one, a full coverage waterblock covers the Vrm's too , I wouldn't say you can't blow them but it shouldn't be because of overheating if cooled right.
 

bug

Joined
May 22, 2015
Messages
13,779 (3.96/day)
Processor Intel i5-12600k
Motherboard Asus H670 TUF
Cooling Arctic Freezer 34
Memory 2x16GB DDR4 3600 G.Skill Ripjaws V
Video Card(s) EVGA GTX 1060 SC
Storage 500GB Samsung 970 EVO, 500GB Samsung 850 EVO, 1TB Crucial MX300 and 2TB Crucial MX500
Display(s) Dell U3219Q + HP ZR24w
Case Raijintek Thetis
Audio Device(s) Audioquest Dragonfly Red :D
Power Supply Seasonic 620W M12
Mouse Logitech G502 Proteus Core
Keyboard G.Skill KM780R
Software Arch Linux + Win10
This just simplified my board Picks, Either Asus Crosshair or AsRock Taichi (w/e their top end is)

And Ryzen/Core I are way different than AM3+/775 or older.
Having just acquired the X470 Taichi, I'm not too impressed with Asrock either. No more dual-BIOS and the BIOS is soldered to the motherboard (no easy replacement if your CPU is unsupported). God knows where else they cheaped out. Asus doesn't have diagnostic leds, but a lot of good those did when on the Asrock mobo they showed an undocumented code...
It's a really lousy time to be buying a mobo :(
 
Joined
May 3, 2018
Messages
75 (0.03/day)
My GPU and CPU have been on full load their whole lives ,Vega bought day one, a full coverage waterblock covers the Vrm's too , I wouldn't say you can't blow them but it shouldn't be because of overheating if cooled right.
Are you saying that your CPU and GPU are running at 100% load all the time??

When I sit down for a gaming session it lasts 12-16 hours with pauses only to make fresh coffee or go for a pee.
 
Joined
Mar 10, 2015
Messages
3,984 (1.12/day)
System Name Wut?
Processor 3900X
Motherboard ASRock Taichi X570
Cooling Water
Memory 32GB GSkill CL16 3600mhz
Video Card(s) Vega 56
Storage 2 x AData XPG 8200 Pro 1TB
Display(s) 3440 x 1440
Case Thermaltake Tower 900
Power Supply Seasonic Prime Ultra Platinum
The comfortable room temperature for normies like yourselves is around 20 °C (or 68 °F for those of you who have not quite made it into the 20th Century yet :))

I feel your pain there. My ambient is usually 26-28C.

the XPC raystorm pro seems to be the best to see any gains over stock/other aio/blocks.

I have the Raystorm Pro. Overall, it does it pretty well.
 
Joined
Mar 10, 2010
Messages
11,878 (2.21/day)
Location
Manchester uk
System Name RyzenGtEvo/ Asus strix scar II
Processor Amd R5 5900X/ Intel 8750H
Motherboard Crosshair hero8 impact/Asus
Cooling 360EK extreme rad+ 360$EK slim all push, cpu ek suprim Gpu full cover all EK
Memory Corsair Vengeance Rgb pro 3600cas14 16Gb in four sticks./16Gb/16GB
Video Card(s) Powercolour RX7900XT Reference/Rtx 2060
Storage Silicon power 2TB nvme/8Tb external/1Tb samsung Evo nvme 2Tb sata ssd/1Tb nvme
Display(s) Samsung UAE28"850R 4k freesync.dell shiter
Case Lianli 011 dynamic/strix scar2
Audio Device(s) Xfi creative 7.1 on board ,Yamaha dts av setup, corsair void pro headset
Power Supply corsair 1200Hxi/Asus stock
Mouse Roccat Kova/ Logitech G wireless
Keyboard Roccat Aimo 120
VR HMD Oculus rift
Software Win 10 Pro
Benchmark Scores 8726 vega 3dmark timespy/ laptop Timespy 6506
Are you saying that your CPU and GPU are running at 100% load all the time??

When I sit down for a gaming session it lasts 12-16 hours with pauses only to make fresh coffee or go for a pee.
Yes in effect, though obviously 100% is never possible but as close to that as possible.
Folding@home.
World community grid.

Rare, very rare now I have a laptop, gaming sessions but I do most office work on the main rig while it's still working.
 
Joined
May 3, 2018
Messages
75 (0.03/day)
Yes in effect, though obviously 100% is never possible but as close to that as possible.
Folding@home.
World community grid.

Rare, very rare now I have a laptop, gaming sessions but I do most office work on the main rig while it's still working.
I have been a member of SETI@Home since 1999 (although I have slacked off over the past three or four years).
 
Joined
Mar 10, 2010
Messages
11,878 (2.21/day)
Location
Manchester uk
System Name RyzenGtEvo/ Asus strix scar II
Processor Amd R5 5900X/ Intel 8750H
Motherboard Crosshair hero8 impact/Asus
Cooling 360EK extreme rad+ 360$EK slim all push, cpu ek suprim Gpu full cover all EK
Memory Corsair Vengeance Rgb pro 3600cas14 16Gb in four sticks./16Gb/16GB
Video Card(s) Powercolour RX7900XT Reference/Rtx 2060
Storage Silicon power 2TB nvme/8Tb external/1Tb samsung Evo nvme 2Tb sata ssd/1Tb nvme
Display(s) Samsung UAE28"850R 4k freesync.dell shiter
Case Lianli 011 dynamic/strix scar2
Audio Device(s) Xfi creative 7.1 on board ,Yamaha dts av setup, corsair void pro headset
Power Supply corsair 1200Hxi/Asus stock
Mouse Roccat Kova/ Logitech G wireless
Keyboard Roccat Aimo 120
VR HMD Oculus rift
Software Win 10 Pro
Benchmark Scores 8726 vega 3dmark timespy/ laptop Timespy 6506
I have been a member of SETI@Home since 1999 (although I have slacked off over the past three or four years).
Not me I'm pondering putting the laptop to work, definitely for compos.
 

eidairaman1

The Exiled Airman
Joined
Jul 2, 2007
Messages
42,218 (6.64/day)
Location
Republic of Texas (True Patriot)
System Name PCGOD
Processor AMD FX 8350@ 5.0GHz
Motherboard Asus TUF 990FX Sabertooth R2 2901 Bios
Cooling Scythe Ashura, 2×BitFenix 230mm Spectre Pro LED (Blue,Green), 2x BitFenix 140mm Spectre Pro LED
Memory 16 GB Gskill Ripjaws X 2133 (2400 OC, 10-10-12-20-20, 1T, 1.65V)
Video Card(s) AMD Radeon 290 Sapphire Vapor-X
Storage Samsung 840 Pro 256GB, WD Velociraptor 1TB
Display(s) NEC Multisync LCD 1700V (Display Port Adapter)
Case AeroCool Xpredator Evil Blue Edition
Audio Device(s) Creative Labs Sound Blaster ZxR
Power Supply Seasonic 1250 XM2 Series (XP3)
Mouse Roccat Kone XTD
Keyboard Roccat Ryos MK Pro
Software Windows 7 Pro 64
It's a lot like a lightning strike for a chip.

Considering they are super sensitive and only powered by millivolts or microvolts sometimes...
 
Joined
Nov 7, 2017
Messages
1,875 (0.73/day)
Location
Ibiza, Spain.
System Name Main
Processor R7 5950x
Motherboard MSI x570S Unify-X Max
Cooling converted Eisbär 280, two F14 + three F12S intake, two P14S + two P14 + two F14 as exhaust
Memory 16 GB Corsair LPX bdie @3600/16 1.35v
Video Card(s) GB 2080S WaterForce WB
Storage six M.2 pcie gen 4
Display(s) Sony 50X90J
Case Tt Level 20 HT
Audio Device(s) Asus Xonar AE, modded Sennheiser HD 558, Klipsch 2.1 THX
Power Supply Corsair RMx 750w
Mouse Logitech G903
Keyboard GSKILL Ripjaws
VR HMD NA
Software win 10 pro x64
Benchmark Scores TimeSpy score Fire Strike Ultra SuperPosition CB20
@Michael Nager
sorry, but LOL.
1. the gpu air doesnt come out of the card like let's say like on a cpu cooler tower
that has the fan at its rear blowing air straight to the rear case fan and out the case. but blowing it into the case.

2. this will make your gpu re-breathing warmer (than needed) air.

3. unless your running a blower type card (which you dont according to pic), or the rear fan is a turbine,
it wont produce enough airflow to suck out all the gpu heat, this will make the cpu fans pull warmer air thru the rad, increasing cpu temps.

going from air cooled to water on a 2080S dropped cpu temps under full load (rig) from 85 to 75C,
not sure what else could cause that ase nothing else was changed (hw/sw/settings).

and not even talking about the fact that the 2080 (non S) i had first, was faster than the aircooled 2080S i had after,
and the main reason i swapped it for a LC card.
Boost clocks go down once you pass 40C on Nv cards, so having the card max out at about 50ish C keeps boost past 2Ghz on full load,
when the aircooled card would peak almost 100 mhz less, and drop another 50-100 mhz depending on load/temps for sustained boost.
and all that with the cpu being LC and mounted on the right side, leaving top/rear case with three 140mm fans just to exhaust board/gpu heat.
So even if i ignore the cpu temp increase, you will definitely have lower clocks on the ti, than if the card was LC.
 
Last edited:
Joined
May 3, 2018
Messages
75 (0.03/day)
@Michael Nager
sorry, but LOL.
1. the gpu air doesnt come out of the card like let's say like on a cpu cooler tower
that has the fan at its rear blowing air straight to the rear case fan and out the case. but blowing it into the case.

2. this will make your gpu re-breathing warmer (than needed) air.

3. unless your running a blower type card (which you dont according to pic), or the rear fan is a turbine,
it wont produce enough airflow to suck out all the gpu heat, this will make the cpu fans pull warmer air thru the rad, increasing cpu temps.

going from air cooled to water on a 2080S dropped cpu temps under full load (rig) from 85 to 75C,
not sure what else could cause that ase nothing else was changed (hw/sw/settings).

and not even talking about the fact that the 2080 (non S) i had first, was faster than the aircooled 2080S i had after,
and the main reason i swapped it for a LC card.
Boost clocks go down once you pass 40C on Nv cards, so having the card max out at about 50ish C keeps boost past 2Ghz on full load,
when the aircooled card would peak almost 100 mhz less, and drop another 50-100 mhz depending on load/temps for sustained boost.
and all that with the cpu being LC and mounted on the right side, leaving top/rear case with three 140mm fans just to exhaust board/gpu heat.
So even if i ignore the cpu temp increase, you will definitely have lower clocks on the ti, than if the card was LC.

First of all, do you know what kind of fans they are? No, obviously not.

The fans are balanced with regard to their RPM and I put a LOT of thought into it. All of the fans are the exact same make and model, they all come from the same batch of fans, and they all within ten or so RPM spinning at the same rate as each other.

I spent a hell of a lot of time balancing the fan profiles of the radiator fans, the fans that act as intake and the fan that acts as exhaust and the GPU fans.

I also gave a lot of thought to the case I was putting the system into.

The fans are all linked to the package temp of the CPU and the rear fan and middle fan of the radiator spin at a higher rate than the left and right fans on the radiator which spin with the same profile.

The front fans are all linked and run on one single profile.

What this looks like when my current main system (i7-4790K running at 4.4 GHz on all cores which is pictured above) is running AIDA64 Extreme is the following

First the CPU temp
aida running.JPG


This is how the fans are running during the stress test you see above:

aida.JPG


Fan#1 Are the three front fans
Fan#2 Is the fan on the radiator at the top nearest the front of the case
Fan#3 Is the middle fan of the radiator
Fan#4 Is the fan on the radiator nearest the rear of the case
Fan#5 Is the fan on the rear of the case blowing air out.

What you can't see is that I have put mesh above the rear fan thus lessening the amount of air that the rear fan might recycle but I didn't want to block it off entirely.

The maximum RPM of the fans is 2,000 RPM.

Running Furmark with the normal fan profile set for the CPU:

1584399472976.png


Now the same test with the case fans ramped up:

1584400418991.png


In the first test the 1080 Ti GPU was running at 1974 MHz the temp was 74 °C and the GPU fans ran at 64% after ten minutes

In the second test the 1080 Ti GPU was running at 1987 MHz the temp was 69 °C and the GPU fans ran at 50% after ten minutes

At the start of the test the GPU was running at 2008 MHz so I lost 34 MHz on the first run and 21 MHz on the second run.

I really don't have anything to complain about and looking at this, you don't have anything to complain about either.

I have plenty of headroom for cooling the GPU because, as you can see, the GPU fans still have a lot of potential left in them if I wanted to create a sharper profile for them.

As you can see I gave the configuration of my system a HELL of a lot of thought and Furmark is about as bad a load on the GPU as I am EVER going to get.

I could get even better temps if I took the plant out of the system, but nah, I like it there. I put it in there originally (likewise with the RGB) just for shits and giggles, and it has kinda grown on me :D

I don't know what all this has to do with configuring a Ryzen 3000 series CPU, but whatever.
 
Last edited:
Joined
Nov 7, 2017
Messages
1,875 (0.73/day)
Location
Ibiza, Spain.
System Name Main
Processor R7 5950x
Motherboard MSI x570S Unify-X Max
Cooling converted Eisbär 280, two F14 + three F12S intake, two P14S + two P14 + two F14 as exhaust
Memory 16 GB Corsair LPX bdie @3600/16 1.35v
Video Card(s) GB 2080S WaterForce WB
Storage six M.2 pcie gen 4
Display(s) Sony 50X90J
Case Tt Level 20 HT
Audio Device(s) Asus Xonar AE, modded Sennheiser HD 558, Klipsch 2.1 THX
Power Supply Corsair RMx 750w
Mouse Logitech G903
Keyboard GSKILL Ripjaws
VR HMD NA
Software win 10 pro x64
Benchmark Scores TimeSpy score Fire Strike Ultra SuperPosition CB20
Never said your rig is crap, but all "tweaks" to improve performance, will be less than what you gain by having the gpu LC.
that to me means, its not the "best possible" setup.

Because any gpu that is air-cooled WILL dump heat inside the case, and most will go thru your cpu rad,
leading to higher temps ergo lower clocks.
This has nothing to do with type/size/brand of fans your using.

just the fact that any Nv gpu will throttle clocks once you go past 50C, which yours clearly does, means you're also loosing gpu performance, not just on the cpu.

you gain about 100-200 mhz switching from air to water (gpu). you know how much you need to increase cpu clocks to make up for that?
 

Johnno

New Member
Joined
Mar 19, 2020
Messages
1 (0.00/day)
In the months that have passed since I started experimenting with the Ryzen 5 3600X on my X470 motherboard (GigaByte X470 AURUS Gaming 7 WiFi Rev. 1.1 and then helping my friend configure his Ryzen 3900X on his motherboard (ASUS X570 ROG Crosshair VIII Hero (WiFi).

I bought the X570 motherboard I wanted (at a price I was willing to pay) the GigaByte X570 AURUS XTREME and experimented with my 3600X in that until I managed to get the CPU I wanted, the Ryzen 9 3950X which I now have.

A YouTube Techie who I respect from a channel called "Actually Hardcore Overclocking" https://www.youtube.com/channel/UCrwObTfqv8u1KO7Fgk-FXHQ called "Buildzoid" who recently made three videos on the topic of configuring PBO on a the same GigaByte board I have, an ASUS board and then general thoughts on PBO:

1) The easy way to get a bit more performance out of Ryzen 3000 CPUs on Gigabyte X570 motherboards

2) The easy way to get a bit more performance out of Ryzen 3000 CPUs on Gigabyte X570 motherboards

3) The easy way to get a bit more performance out of Ryzen 3000 CPUs on Gigabyte X570 motherboards

In the course of his experimentation he has managed to degrade his Ryzen 7 3700X and I fear that with the way he is configuring his Ryzen 9 3950X he will be doing the same thing there as well.

In the video he has put up some benchmarks where, by configuring PBO in the BIOS he managed to get the results up by some amount, the problem is though, that he is doing so at higher voltages than I consider to be prudent and also at higher temps than I experience with my 3950X.

His maximum CineBench R20 result after configuring his 3950X was at 9,554.

The way I configure my Ryzen 9 3950X I get a CineBench R20 score of 10,170 and still remain within the specification as laid down by TSMC for their 7nm Node. This specifies a far lower voltage than AMD considers safe and personally I am going to go with the recommendations of the creator of the 7nm Node and constrain the voltage of my Ryzen 3000 CPUs to 1.3 Volts MAXIMUM

Because of my back problems (I have had two spine operations and have spinal arthritis) I have to keep the room temperature pretty warm.

The ambient temperature in my room is 28 - 29 °C and you should keep that in mind when I show you the following benchmark results of my system:

My R9 3950X with SMT On:

1) CineBench R20 all-core score of 10,170 and a single core score of 500

2) FireStrike EVGA 1080 Ti SC2 I have a Graphics Score of 28,213, a Physics Score of 33,848 and a Combined Score of 15,488
3) FireStrike Extreme EVGA 1080 Ti SC2 I have a Graphics Score of 14,130, a Physics Score of 33,821 and a Combined Score of 7,057
4) FireStrike Ultra EVGA 1080 Ti SC2 I have a Graphics Score of 7,180, a Physics Score of 34,089 and a Combined Score of 3,902

5) TimeSpy EVGA 1080 Ti SC2 I have a Graphics Score of 10,292 and a CPU Score of 15,390
6) TimeSpy Extreme EVGA 1080 Ti SC2 I have a Graphics Score of 4,791 and a CPU Score of 9,421

7) Ghost Recon Wildlands benchmark 1080p everything at max FPS 86.33, CPU 14.7% (Min. 9.8% Max. 23.2%) and GPU 96.7%

8) 7zip Compression Average 124.906 MB/s, Decompression 199.303 MB/s

My R9 3950X with SMT Off:

1) CineBench R20 all-core score of 7,817 and a single core score of 513

2) FireStrike EVGA 1080 Ti SC2 I have a Graphics Score of 28,295, a Physics Score of 30,052 and a Combined Score of 15,833
3) FireStrike Extreme EVGA 1080 Ti SC2 I have a Graphics Score of 14,170, a Physics Score of 30,168 and a Combined Score of 7,076
4) FireStrike Ultra EVGA 1080 Ti SC2 I have a Graphics Score of 7,186, a Physics Score of 30,164 and a Combined Score of 3,906

5) TimeSpy EVGA 1080 Ti SC2 I have a Graphics Score of 10,271 and a CPU Score of 15,340
6) TimeSpy Extreme EVGA 1080 Ti SC2 I have a Graphics Score of 4,788 and a CPU Score of 7,564

7) Ghost Recon Wildlands benchmark 1080p everything at max FPS 86.51, CPU 23.4% (Min. 17.2% Max. 48.6%) and GPU 97%

8) 7zip Compression Average 103.106 MB/s Decompression 129.844 MB/s

The cooler I am using is the AlphaCool EisBaer 360 LT which as the name suggests has a 360 rad and I am running it with three Noctua NF-A12x25 fans.

In the following I will be giving you a step-by-step guide to configuring your Ryzen 3000 series CPU.

This is based on the BIOS in my GigaByte X570 AURUS XTREME board, but the few values that you need to change can be found in the other BIOS's from ASRock, ASUS or MSI.

The following is the step-by-step guide to configuring the system whereby you will be able to maximise the performance without running the risk of frying your CPU and you don't have to live in a ice-locker to get a result:

I have to preface this with some info that is woefully lacking in the videos or articles you may see or read.

The thing is that as opposed to Intel CPUs that you have been used to (and that I was used to) the BIOS is supplied to the motherboard manufacturers as a binary and is called AGESA.

So what you see displayed as "The BIOS" is in effect just a configuration menu for the AGESA. The problem about editing the AGESA portion found under "Settings" under the headings "AMD CBS" and "AMD Overclocking" directly is that with some of the options, if you enter a wrong value, then your system will not boot. What is worse however is that some of the settings cannot be removed with a "Clear CMOS" and your mobo is effectively bricked.

So now to configuring your BIOS:

1) Go into Easy Mode (F2) and click on "Load Optimized Defaults (F7)

a) Switch to Advanced Mode (F2)

2) Under the heading "Tweaker" do the following:

a) Go down to the bottom of the page and open "CPU/VRM Settings"

i) Set "CPU Vcore Loadline Calibration to "Turbo" (third highest value)
ii) Set "SOC Loadline Calibration" to "High" (third highest value)
iii) Set "PWM Phase Control" to "eXm Perf" (eXtreme Performance)

3) Under the heading "Setting"

a) Go to "AMD CBS"

i) Go to "XFR Enhancement"
ii) Set the FCLK Freqency to the desired value (in the case of 3600 RAM that would be 1800 MHz)
iii) Set the "UCLK DIV1 MODE" to "UCLK ==MEMCLK"

b) Go to "AMD Overclocking" under "Settings"

i) Click on "Accept"
ii) Go to "DDR and Infinity Fabric Frequency/Timings"
iii) Go to "Infinity Fabric Frequency and Dividers"
iv) Set "Infinity Fabric Frequency and Dividers" to the desired value (in the case of 3600 RAM that would be 1800 MHz).

4) Under the heading "Boot" do the following

a) Set "Full Screen LOGO Show" to "Disabled"

Of course setting the boot drive etc. should be obvious and I don't think I need to explain that.

Do NOT set anything else, like "Extreme Memory Profie(X.M.P)" for instance.

There that's you done with the BIOS part of the configuration

Boot into Windows and install "Ryzen Master".

When Ryzen Master has loaded, click on "Creator Mode" on the left hand side.

1) Make sure "Control Mode" is expanded and under that heading click on "Manual"

2) Make sure that the section "Cores Section" is expanded

a) Expand "CCD0" and "CCD1"

b) Click on the red circle on the right hand side so that it changes to what looks like a Green "X"

i) Click in the first field beside "C 01" and change the clockspeed. You should have absolutely no problems setting it to "4250". When you have done the rest of the configuration then test it and increase it (in my case it is set to 4300 and I have no problems). When you set one field, because the Green X is activated, all the other values will change to what you set.

3) Make sure "Voltage Control" is expanded

a) Set "Peak Core(s) Voltage to 1.3 Volts

4) Make sure Memory Control is expanded and that it is "Included"

a) "Coupled Mode" should be "On"

b) Set your memory clock speed (in the case of 3600 RAM it would be 1800) remember this is the data rate. Infinty Fabric runs at the data rate and RAM runs at double data rate.

5) Make sure "Voltage Contols" is expanded

Unless otherwise stated, leave the values on "Auto"

a) MEM VDDIO should be set to 1.35

b) MEM VTT should be set to 0.675

c) VDDCR SOC should be set to 1.05

6) Make sure "DRAM Timing Configuration" is expanded

Now I have found that unless these values are set then every time you change something (like the voltage or the clockspeed) the system will want to reboot. If these are set then the values are just changed and you can continue

a) Change "CAS Latency" from "Auto" and you should see the correct value for your RAM

b) Change "Row Precharge Delay" from "Auto" and you should see the correct value for your RAM

c) Change "Read Row-Column Delay" from "Auto" and you should see the correct value for your RAM

d) Change "Write Row-Column Delay" from "Auto" and you should see the correct value for your RAM

e) Change "Row Cycle Time" from "Auto" and you should see the correct value for your RAM

Leave everything else on "Auto" and you can configure those sub-timings at your leisure.

7) Make sure that "DRAM Controller Configuration" is expanded

a) Change "Cmd2T" from "2T" to "1T". If you have good quality RAM then it should run at 1T. If not then change this back to 2T.

Now at the bottom click on "Save Profile" and then click on "Apply & Test" and the system will reboot.

As you will see, the CPU is limited to a maximum of 1.3 Volts and essentially you are just seeing how much clockspeed you can squeeze out of those 1.3 Volts. When the system is not under load then of course the operational voltage will decrease.

Now comes the best part about the 3950X.

If you are mainly gaming, then click on a different profile "Profile 1" for instance and do exactly the same as above EXCEPT:

1) Under the heading "Additional Control" turn "Simultaneous Multithreading" to "OFF". This will run your CPU as a straight 16 Core/16 Thread CPU.

2) Under "Cores Section" make sure the red circle is a green "X" and add 100 MHz to whatever was stable running 16 Cores /32 Threads with SMT ("Simultaneous Multithreading") ON

The one problem with the GigaByte BIOS is that this is not changed and you have to go into the BIOS and

1) In the Advanced Mode go to "Tweaker"

a) Under "Advanced CPU Settings"

i) Go down to "SMT Mode"
ii) Change from "Auto" to "Disabled"

Save and exit.

If you want to go back to using 16 Cores/32 Threads just choose the "Creator Profile" and then change this value back to "Auto" again.

That's it.

I know it looks like a lot, but it really isn't.

IMPORTANT!!

Every time you reboot the system you have to load Ryzen Master and apply the profile you want. Unfortunately there is no way as yet to automatically load a default profile, but I hope that option will be forthcoming in the future.

If you are applying the same Profile you had before you shut down then the system will not need a reboot.

After you have applied the profile you can close Ryzen Master.

Have fun.

Hi

I joined specifically to thank you for your terrific post.

I cannot believe all of the negative reaction to your post. It's disappointing and there is absolutely no need for it.

People should be offering additional information - not attacking a post that was fantastic and intended to help people.

Regards

Johnno
 
Joined
Jun 3, 2010
Messages
2,540 (0.48/day)
I concur. I hope we will see more of his example being spread around because techpowerup should be the intersection of expert opinion holders in contrast to marketing product placements.

I think that is the worst possible example if shills take the place of actual users in sharing their anonymous generalities. Nothing kills a forum debate quite like a false member.
 

eidairaman1

The Exiled Airman
Joined
Jul 2, 2007
Messages
42,218 (6.64/day)
Location
Republic of Texas (True Patriot)
System Name PCGOD
Processor AMD FX 8350@ 5.0GHz
Motherboard Asus TUF 990FX Sabertooth R2 2901 Bios
Cooling Scythe Ashura, 2×BitFenix 230mm Spectre Pro LED (Blue,Green), 2x BitFenix 140mm Spectre Pro LED
Memory 16 GB Gskill Ripjaws X 2133 (2400 OC, 10-10-12-20-20, 1T, 1.65V)
Video Card(s) AMD Radeon 290 Sapphire Vapor-X
Storage Samsung 840 Pro 256GB, WD Velociraptor 1TB
Display(s) NEC Multisync LCD 1700V (Display Port Adapter)
Case AeroCool Xpredator Evil Blue Edition
Audio Device(s) Creative Labs Sound Blaster ZxR
Power Supply Seasonic 1250 XM2 Series (XP3)
Mouse Roccat Kone XTD
Keyboard Roccat Ryos MK Pro
Software Windows 7 Pro 64
Can we just be humble and not arrogant and just help each other?

I mean there is a Zen Garden here too
 
Status
Not open for further replies.
Top