Post your Cinebench R23 Score

[cowboy44mag]

Just registered to post my Ryzen 3800X Cinebench R20 score and have it added to the chart. Have only had the processor for a couple days so still tweaking it.

 

[Roddey]

Just registered to post my Ryzen 3800X Cinebench R20 score and have it added to the chart. Have only had the processor for a couple days so still tweaking it.

View attachment 132608

That multi core score is quit a bit higher than mine has ever done. Would you mind divulging what heatsink your using?

 

[biffzinker]

That multi core score is quit a bit higher than mine has ever done.

Looks like a OC of 4.4 GHz on all cores no matter the load.

 

[Roddey]

Looks like a OC of 4.4 GHz on all cores no matter the load.

Ya that's it. I haven't tried with the new abba bios but the prior bios's I was never able to hit 4.4 all core. Maybe I should give it a try. Well, on second thought, maybe someday but I doubt I will hit it. I am tired right now of messing with it.
It does run great. Had a nvme drive failure but that was an easy return, swap and it wasnt my boot drive. My last computer ran for so long with no issues that it gave me a bit of apprehension about upgrading and experiencing problems that can be hard to track down. That didn't happen so all's good.

 

[cowboy44mag]

That multi core score is quit a bit higher than mine has ever done. Would you mind divulging what heatsink your using?

I am running an unconventional air cooled system. I have 5 140mm high performance RGB case fans as intake and one 140mm fan as exhaust creating overcharged positive pressure in the case but blowing a lot of cool air across everything. I am using a Noctua NH-U14S with two 140mm high performance RGB case fans in a push pull configuration. For the cost of the NH-U14S and the custom fans I could have gone with an AIO, but opted for the air cooling setup as there are no pumps to fail and as an added bonus the heat sink fans pull a lot of air across the RAM keeping it very cool.

As biffzinker stated, I am utilizing an all core multiplier overclock, however I use a voltage offset rather than static voltage so the processor can still enter a "rest" state of 0.5V. Under full all core load the voltage is 1.37V. I have noticed that single core stress will have voltage spikes up to 1.4V (which is common for Ryzen).

With my cooling and voltage offset my system keeps nice and cool. Attached below I have a pic of my system at idle, a pick of the system at the end of a Cinebench R15 run and a pic of it at the end of a Cinebench R20 run. Ambient room temp 72F (22C).

 

[cowboy44mag]

Updated Cinebench R20 score after CCX overclocking:

 

[speedgoat]

managed to pass 5400 on CB 20 possibly because of 3800C15 on ABBA

4450 MHz at 1.33V

 

[cowboy44mag]

Tried all day, but just can't get 4.445Ghz all core stable. I have managed to push a little further with CCX overclocking and have hit a wall. I'm running one CCX at 4.4Ghz and the other at 4.425Ghz and have tightened my timings further but am still @ 3600Mhz on the RAM.

 

[speedgoat]

Tried all day, but just can't get 4.445Ghz all core stable. I have managed to push a little further with CCX overclocking and have hit a wall. I'm running one CCX at 4.4Ghz and the other at 4.425Ghz and have tightened my timings further but am still @ 3600Mhz on the RAM.

how do you manage to do this OC with offset mate ? can you give me a couple of combinations of V and MHz you are locking ? looks like a really good idea !
I am running a 24/7 offset but every time i tried to lock the freq it just wont boot.. im on C7H btw

 

[phanbuey]

Did some more tweaking on the memory. Would be nice to break 530 sc...

edit: got it

 

[potato580+]

my 2nd gaming pc, good for playing on 720p
i just realize that mine software outofdate, pardon

 

[cowboy44mag]

how do you manage to do this OC with offset mate ? can you give me a couple of combinations of V and MHz you are locking ? looks like a really good idea !
I am running a 24/7 offset but every time i tried to lock the freq it just wont boot.. im on C7H btw

Every CPU and every motherboard is different, so my settings probably wouldn't be the same as yours, and in fact could be dangerous. Just a warning, but with my motherboard (Strix X470-F) and CPU the offset I needed was a positive offset of 0.28750V. If you just input that and your motherboard is supplying more voltage by default than mine you may end up greatly overvolting your system and may cause serious damage. Therefore what is more important than the end result is how I derived at that offset.

You have to look at what voltage your processor is using under full load to run stable. For me my processor needed 1.373V to be stable at 4.4Ghz all core and is still stable with that voltage when overclocking the fastest CCX to 4.425Ghz. When trying to figure out a voltage offset I always start with the lowest possible positive offset possible, and you will want to ensure C-state is enabled. The system may not even boot at this setting, and if it doesn't it means you need more voltage so you have to increase the offset a notch at a time till the system will boot. Once the system boots into windows launch your chosen monitoring app (I use hwinfo64- the voltage offset will be displayed under the motherboard readings and VDDCR CPU) and launch something like Cinebench R15 that will put a full load on your system but can be stopped quickly if needed. Make sure you are using either balanced or Ryzen balanced power plan and once R15 is running you can see what voltage is being supplied under load to your CPU, if its lower than your static stable voltage then you will have to raise your offset, if it is higher you will have to lower your offset. Once the offset matches your static stable voltage then you have to test it for stability but you will now have an offset which will ramp up to ~1.4 for single core (just like at stock) and will run all core at your stable static voltage, but will also scale down when you don't have a load on your system.

My idle voltage will go ~0.5V, single core loads will have spikes as high as 1.4V (for like a second) and under full load I have a steady 1.373V. My system runs cooler with the offset than with a static voltage because when the system is at rest and cores a "sleeping" it will scale down to only 0.5V. I've noticed that during a full day of gaming/ rendering/ web surfing/ YouTube/ walking away for a few minutes my average overall voltage is usually 1.02V - 1.14V as the system doesn't need much to web surf or watch YouTube, but uses a constant 1.373V while gaming and rendering. I am manually overclocking with a multiplier and have a voltage offset, so the cores are "locked" at 4.4Ghz (with Ryzen Master I have the fastest CCX overclocked to 4.425Ghz). I know that with previous processors when I would enable C-state and balanced or Ryzen Balanced in Windows the cores would down clock themselves at rest, not sure if Ryzen does this or not, but the monitoring software shows the cores are constantly locked at 4.425Ghz and 4.4Ghz respectively.

I think I have pushed my system as far as I can, thinking about overclocking my RAM to 3733Mhz from 3600Mhz, but would have to loosen the timings then- currently running 3600Mhz CL14. I get great performance with 4 cores at 4.425Ghz and 4 at 4.4Ghz, but am always trying to squeeze just a little more out of it. I am really close to breaking 5400 after overclocking my fastest CCX to 4.45Ghz with a small voltage bump in CB20. I'm not stable with 4.45Ghz all core but I can run the fastest CCX at 4.45Ghz and think with better RAM I could breach the 5400 barrier. I am running 3200Mhz Samsung B die and already have it overclocked 400Mhz but maybe I can get it up to 3733Mhz... Realistically for 24/7 use the 4.4Ghz all core overclock is plenty and with my NH-U14S I am constantly below 70C.

 

[derp]

max safe voltage for silicon in zen2 is 1.32V, just saying.
mine 3900x already degraded in auto under load, cuz zen2 crazy overvolted out of the box (1.4V all cores load with 1.49V single core load) and positive offset is...

 

[Voltaj .45 ACP]

max safe voltage for silicon in zen2 is 1.32V, just saying.
mine 3900x already degraded in auto under load, cuz zen2 crazy overvolted out of the box (1.4V all cores load with 1.49V single core load) and positive offset is...

max safe voltage for silicon in zen2 is 1.32V

that's bullshit.show one valid source for that.

 

[speedgoat]

max safe voltage for silicon in zen2 is 1.32V

that's bullshit.show one valid source for that.

this max safe voltage myth is really funny.. it started from here page 1 :

Strictly technical: Matisse (Not really)

07/08/2019 6:33 PM (GMT) - Update on the bios issue on Crosshair VIII Hero motherboard ("the thing"). Earlier today I received a response to my inquiries from ASUS. The response was rather technical and I cannot go into the specifics of what exactly it involved. However, it confirmed my...

www.overclock.net

and then on page 47 The Stilt says he actually degraded a 3700 and never went above 1.33

there is no such thing as a max safe 24/7 fixed voltage

 

[TheLostSwede]

Looks like the new AGESA and what not has kicked up the performance a tiny bit.

 

[derp]

that's bullshit.show one valid source for that.

read here why safe voltage decreasing with smaller node

speedgoat already posted useful info for you too. 1.32v numbers were explained from the start of zen2, not remember by whom, but stilt confirmed it.
ppl like stilt and other engineers are much more valid source, than some intel/amd/nvidia marketing teams or reviewers, who put CPU into rig for 2 days, run tests and concluded that's all fine, i put 1.4V - all good and stable... and get rid of the CPU next day

there is no such thing as a max safe 24/7 fixed voltage

that's not true too. You can set fixed voltage at 0.8V for example and it will be pretty safe for years.
Also 1.32V fixed is more safe then auto (where 1.4V under full load and ~1.5V in ST, depends on mobo too, but CPU requesting tons of vcore, so strong VRM mobo gives lots of voltage). Me personally use 3900x at 1.28v now
even for 14nm zen1 safe voltage was around 1.4V, now we have much smaller 7nm

 

[speedgoat]

read here why safe voltage decreasing with smaller node

speedgoat already posted useful info for you too. 1.32v numbers were explained from the start of zen2, not remember by whom, but stilt confirmed it.
ppl like stilt and other engineers are much more valid source, than some intel/amd/nvidia marketing teams or reviewers, who put CPU into rig for 2 days, run tests and concluded that's all fine, i put 1.4V - all good and stable... and get rid of the CPU next day


that's not true too. You can set fixed voltage at 0.8V for example and it will be pretty safe for years.
Also 1.32V fixed is more safe then auto (where 1.4V under full load and ~1.5V in ST, depends on mobo too, but CPU requesting tons of vcore, so strong VRM mobo gives lots of voltage). Me personally use 3900x at 1.28v now
even for 14nm zen1 safe voltage was around 1.4V, now we have much smaller 7nm

yeah ok i sort of agree that such a low voltage would be safe but its lacking obvious usefulness

i am not sure about the 2nd point that 1.28 is safer than auto but i feel we clearly know too little about zen2 to have definite opinions, it might be true but then again any fixed voltage would keep averages higher than auto on any daily operation, despite avoiding the extremes. Btw i despise auto and all the 1.5V rubbish, im running an offset since day 1.

 

[derp]

yeah ok i sort of agree that such a low voltage would be safe but its lacking obvious usefulness

you are not right here also set some 3.5GHz 0.8V and you will get low power but still strong CPU for mini rig, HTPC etc.

but then again any fixed voltage would keep averages higher than auto on any daily operation, despite avoiding the extremes

i tested it. Day with auto and day on fixed with not much use of PC (daily usage without rendering both cases). Got some 66W average power consumption and 1.39V avg voltage on auto. But on fixed 1.28V got some 40W average power consumption and (suddenly) 1.28V average voltage.
Problem with auto is that it's pushing voltages almost all time to 1.4-1.5V, even with just 1-3% CPU load in idle with boost over 4.3GHz onto 2-3 cores.
Full load auto stress CPU much more too (both voltage and power consumption). Also temps is ~20C lower avg in fixed mode.

 

[Voltaj .45 ACP]

this max safe voltage myth is really funny.. it started from here page 1 :

Strictly technical: Matisse (Not really)

07/08/2019 6:33 PM (GMT) - Update on the bios issue on Crosshair VIII Hero motherboard ("the thing"). Earlier today I received a response to my inquiries from ASUS. The response was rather technical and I cannot go into the specifics of what exactly it involved. However, it confirmed my...

www.overclock.net

and then on page 47 The Stilt says he actually degraded a 3700 and never went above 1.33

there is no such thing as a max safe 24/7 fixed voltage

oh fixed voltage?no i'm aganist it because when i set some voltage even below 1.33 my ups going crazy and says overcapacity which is 480w but when i using auto voltage there is no beeepp beeeep you know.everytime in ac: odyssey game i don't know why power hungry game i guess.auto set 1.5v sometimes but under 1v most of the time so giving low voltage and sometimes high voltage is one thing but fixed 1.3xx+ another.you have to rest the cpu because apart from this it would be china torture small water drop but constant.

did you know manuel 4.35ghz oc'd 3700x power draw peak 164w when pcmark10 first test.

 

[hzy4]

3700X with PBO

and now stock values with all auto

 

[cowboy44mag]

max safe voltage for silicon in zen2 is 1.32V, just saying.
mine 3900x already degraded in auto under load, cuz zen2 crazy overvolted out of the box (1.4V all cores load with 1.49V single core load) and positive offset is...

First of all every motherboard is different and supplies different voltages to the CPU by "default". I worked on a X570 system that by default supplied a lot more voltage and to run in offset it required a negative offset. My motherboard, perhaps because its an X470, or possibly just the motherboard itself needed a positive offset as at the lowest positive offset it would only supply something like 1.1V to the CPU, which is no where near enough for a 4.4Ghz all core overclock. by methodically bumping up the offset one "peg" at a time and doing a lot of testing I found where my processor would be supplied with the needed 1.373V under full load for a stable 4.4Ghz overclock. I would never recommend just setting a positive offset of 0.287V on any motherboard/ CPU and expecting it to work. Some boards that are already supplying more voltage than necessary might just over volt it past 1.5V and insta-fry the system. That is why I explained my methodology in detail in my previous post.

Secondly there are a lot of guys out there that have had 3700X and 3800X from launch and have run a constant static voltage of 1.4, 1.44V to overclock to 4.4, 4.45Ghz all core and have seen no ill effects. There are other guys who have run 1.3V and now have a degraded processor, or so they think. There are far too many variables to take fully into account with Ryzen 3000 because everyone is running different boards with different cooling solutions and different power supplies and cases. There are people rocking 3700X on X370 and people comparing their 3800X on X570 to it. Its not a true comparison because the motherboards are so different. Same socket but just look at the VRMs between X370 and X570, no to mention all the "B" series boards. Some of the degraded performance may not be the processor but may actually be the motherboard or other supporting hardware. If it is the processor it could be the power supply or motherboard degraded the processor by supplying inconsistent voltages... I personally use a 80+ Platinum power supply, but some feel comfortable on much less. Another issue I've seen a lot is people will put high end AIOs on their processor and monitor the heck out of processor temp while not even giving their RAM or VRMs a single nearby case fan to help keep them cool. I always build with a ton of case fans to keep everything in the build as cool as is possible.

I'm not saying I'm 100% safe from degrading my CPU, as no one can really answer that. I am saying that there are a lot more variables to account for than just the Vcore. I also think it is a very good idea when manually overclocking with a multiplier to use an offset so you don't have constant high voltage going through the processor all the time. At rest I'm at ~0.5V. After a Cinebench R20 run my CPU temp is always under 70C and motherboard temps across are under 40C (usually ~35C). Saying I have read over and over about guys hitting over 80C running Cinebench, well heat has never been good for any part of a computer. The cooler you can keep everything the safer you are. Some of the degraded processors may actually be from running too hot rather than the Vcore utilized.

max safe voltage for silicon in zen2 is 1.32V, just saying.
mine 3900x already degraded in auto under load, cuz zen2 crazy overvolted out of the box (1.4V all cores load with 1.49V single core load) and positive offset is...

I can't find the article now, but there was an AMD spokesperson actually asking customers to not undervolt Ryzen 3000. The claim was that the 1.4 and 1.5V spikes are normal and only last a fraction of a second and are absolutely needed for Ryzen to be able to boost properly. If what you are saying is true then your CPU was damaged following AMD's advice on how to run Ryzen 3000 (ie leave it on auto, XFR2, PBO2, ect...). I would also be much more wary of running a 3900X @ 1.35 - 1.4 volts than I would a 3700X or 3800X as its has two more cores and four more threads and that voltage is going to create a lot of heat.

you are not right here also set some 3.5GHz 0.8V and you will get low power but still strong CPU for mini rig, HTPC etc.


i tested it. Day with auto and day on fixed with not much use of PC (daily usage without rendering both cases). Got some 66W average power consumption and 1.39V avg voltage on auto. But on fixed 1.28V got some 40W average power consumption and (suddenly) 1.28V average voltage.
Problem with auto is that it's pushing voltages almost all time to 1.4-1.5V, even with just 1-3% CPU load in idle with boost over 4.3GHz onto 2-3 cores.
Full load auto stress CPU much more too (both voltage and power consumption). Also temps is ~20C lower avg in fixed mode.

With my positive voltage offset in a normal day when the monitoring software is up all day I have average voltage usage of ~1.1V. In a normal day I do a few rendering projects, some video editing and video conversion, some web browsing, some gaming and some time when I'm out and the computer is left running. Usually if I have to go out and feed the horses, goats, ducks, ect or muck some stalls I'll just leave the system up so there are times its at "rest" for like an hour or so, but it usually runs from early morning till late evening every day. You get much better overall voltages using an offset than a fixed voltage, even when pushing a little higher offset for overclocking.

 

[speedgoat]

i think i am starting reaching the limit of what the frequency can offer me
4.475@1.388V
speaking about safe voltages that is not one

 

[cowboy44mag]

i think i am starting reaching the limit of what the frequency can offer me
4.475@1.388V
speaking about safe voltages that is not one
View attachment 133351

That's an awesome overclock for any Ryzen 3000 CPU, your only 25Mhz away from the top rated boost frequency of the processor. I have often regretted not getting the C7H and instead getting the Strix. Not only does the C7H have better VRMs but the bios also has more tools available for fine tuning overclocking than the Strix bios. Your chip is probably a little better binned than mine, but overall the 3800X is a great processor. You have really done a great job pushing it to the max.

AMD has just announced a new Agesa with over 100 fixes for Ryzen processors. It is being supplied to motherboard manufactures now and will be available in November as new bios updates. Hopefully it will allow us to get higher frequency at the same (or lower) Vcore.

 

[derp]

First of all every motherboard is different and supplies different voltages to the CPU by "default".

yes. But CPU the one, who requesting voltage from mobo. Better mobo can provide closer voltage to what CPU is asking.

Secondly there are a lot of guys out there that have had 3700X and 3800X from launch and have run a constant static voltage of 1.4, 1.44V to overclock to 4.4, 4.45Ghz all core and have seen no ill effects.

because those guys usually set high voltage, OC CPU, run benches, show screenshots and then never use their CPUs, just browsing or gaming, where very low load on CPU.

I personally use a 80+ Platinum power supply

me too

Another issue I've seen a lot is people will put high end AIOs on their processor and monitor the heck out of processor temp while not even giving their RAM or VRMs a single nearby case fan to help keep them cool.

im using just normal water and VRM temps at some 50C under load

At rest I'm at ~0.5V

at rest it 1.4-1.5V! I have never seen 0.5V. 0.8-0.9V is lowest i have seen, when killed most of software running. Close to any load makes CPU to work in boost mode with crazy voltage.

After a Cinebench R20 run my CPU temp is always under 70C

in fixed voltage mode my CPU temps is under 70C during cb20 run. At rest etc it's some 30-40 usually. When in auto it's 45-70 at rest and 70-85 under load.

The claim was that the 1.4 and 1.5V spikes are normal and only last a fraction of a second and are absolutely needed for Ryzen to be able to boost properly.

and this is total bullsh!t. Even speedgoat just proved it here with 4.475@1.38V all cores, not 1.5V for the same clock on only one core. Same i had on 1700, which did one core boost to 3750MHz on 1.44V or something like this. I set 1.2V fixed and 3.8GHz all cores - it worked fine with lower temps and much more performance. Another proof is that all their CPUs of same model requesting same voltage, when all of them have different quality. So auto voltages just broken mess. Also for different mobos requesting voltage should be different too. For example when CPU requesting 1.42V on weak mobo under load this voltage will drop to say 1.27V actual, but on strong mobo it will drop just to 1.4V actual in same conditions. Ofc CPU have own voltage regulator inside, but who know how it works with what limits.
And their 1.4-1.5v is not for fraction of a second, but constant. I degraded CPU when ran avx2 test in auto for day (b4 did 98-100% load of CPU for many hours) and voltage never dropped below 1.39V... under full load of 12 cores. With crazy temps also.
By setting fix voltage on 3900x i got less power consumption, lower average and lower max voltage, lower temps and more performance (actual clocks with lower voltage is higher under load, cuz CPU not throttle much). Also im afraid of doing any good load on CPU in auto anymore, because it can fry my CPU.
Ofc max voltage limit would be much better, but there is just no such a option
Tried with offsets, but in my case negative offset just gives lower clocks with about the same voltages

AMD has just announced a new Agesa with over 100 fixes for Ryzen processors. It is being supplied to motherboard manufactures now and will be available in November as new bios updates. Hopefully it will allow us to get higher frequency at the same (or lower) Vcore.

that's what I waiting for now