AMD Ryzen 7 5800X 3D Cache Eight Core 4.5GHz (Socket AM4) Processor - Retail - Go Go Go xD

[Guest2]

I'm about the same with my Corsair 150i.

29 idle, but using chrome and general use I'm at 50-60, bf2042 was I think 70.

Prime, on default hits 90 after a short time.

I think this is fairly normal right?

Way better than my 3900x for sure.

Late reply, sorry I missed this.
Yes, thats about right.

You guys who've played with limbo strike and pbo tuner. Has it made a difference to the boosts you're getting? Especially in games? I mean that's why we have these chips right?

I have tried Kombo strike. Mode 3 is the best and does about the same as PBOTune2 setting -25 on all cores. Multicore boost is higher but I found it only makes performance worse in games - stuttering / jerkiness / hitching whatever you want to call it.

Use kombo strike / PBO / other undervolt method to get slightly lower temps and running benchmarks but for gaming it's best left to do its thing. Gaming max temp i've seen mine his is 73c in Warzone 2. So still way under the max rated temp.
I'm using air cooling

I don't know why anyone is even bothered about running Cinebench, Prime and other tools with the 5800X3D and getting high all core boost and/or higher benchmark scores.
It's main use is gaming, if you want benchmarking chip, you should have probably bought a 5950X or 7950X instead. To play games smoothly, the 5800X3D at stock will work better than both those chips

For test purposes and nobbing around though, sure tweak the chip all you want. I enjoyed at least trying, only to find out it's better left alone

 

[martos123]

Late reply, sorry I missed this.
Yes, thats about right.

I have tried Kombo strike. Mode 3 is the best and does about the same as PBOTune2 setting -25 on all cores. Multicore boost is higher but I found it only makes performance worse in games - stuttering / jerkiness / hitching whatever you want to call it.

Use kombo strike / PBO / other undervolt method to get slightly lower temps and running benchmarks but for gaming it's best left to do its thing. Gaming max temp i've seen mine his is 73c in Warzone 2. So still way under the max rated temp.
I'm using air cooling

I don't know why anyone is even bothered about running Cinebench, Prime and other tools with the 5800X3D and getting high all core boost and/or higher benchmark scores.
It's main use is gaming, if you want benchmarking chip, you should have probably bought a 5950X or 7950X instead. To play games smoothly, the 5800X3D at stock will work better than both those chips

For test purposes and nobbing around though, sure tweak the chip all you want. I enjoyed at least trying, only to find out it's better left alone

I've not found any detrement in using PBO, i'm not experiencing any more jerkiness or stuttering than when it's on stock either.

 

[Guest2]

I've not found any detrement in using PBO, i'm not experiencing any more jerkiness or stuttering than when it's on stock either.

OK, that's good. I would say its hard to notice and might be game specific. For me and Warzone 2, there was definately something that felt off when I had PBO enabled instead of running stock.
Maybe try Warzone 2 if you have it, play a game or 2 in quads where it will be busy then reboot, turn off PBO and try again. To me it felt way more fluid, sort of like going from Vsync to Gsync.
You don't notice the difference until you use Gsync. Similar with this

 

[eeii]

I have the same cooler but i guess there are some other factors keeping your temps 10c lower. Your ambient is a few degrees cooler and my GPU and case size and cooling could also be a factor.

How high do you run your CPU and case fans when doing this benching, are they at 100%? Also which case (useful to add to your sig)?

So the case is a thermal take x71. I have 200mm fan in the roof extracting hot air out the roof. 140mm fan extracting out the top rear and 3 140 mm fans at the front. All fans are controlled by a fan controller and never go over 60%.

 

[Nexus18]

I've not found any detrement in using PBO, i'm not experiencing any more jerkiness or stuttering than when it's on stock either.

Can't say I have noticed any issues with PBO2 and -30 either.

 

[StephenJ]

Can't say I have noticed any issues with PBO2 and -30 either.

Also no issues with -30

 

[Guest2]

Could be game specific. Maybe Warzone 2 just doesnt like undervolt as Warzone 1 didnt. I had a complete system crash on the final game circle when I undervolted my 5800X.
I played for maybe 2 hours with and undervolt using Kombo strike mode 3 then 2 hours with it at stock. I much preferred stock so have left it like that. Thankfully I dont spend much time mid game looking into my PC case thats under my desk to check the temp.. I do notice when there are stutters and certainly would if my system reset itself mid game!

 

[GuruJockStrap]

Maybe the undervolt you tried with PBO and Kombo Strike wasn't actually stable with your CPU, hence the issue you were experiencing in WarZone 2?

 

[Nexus18]

I read mixed reports on that kompu feature, pretty much everyone said better sticking with pbo 2 tuner

 

[Guest2]

Maybe the undervolt you tried with PBO and Kombo Strike wasn't actually stable with your CPU, hence the issue you were experiencing in WarZone 2?

I tried both PBO tuner2 and Kombo strike (combo strike first) both were fine and stable when stress testing.

Give it a try. Play Warzone 2 with PBOtuner set to -25 all core, then try on stock settings. Like I said, for me I could tell the difference hence leaving it alone. Sure the stress testing is higher temps and lower all core now but im not bothered about stress testing, only for it to be worse when using it for what it was intended for!

 

[Richdog]

I don't know why anyone is even bothered about running Cinebench, Prime and other tools with the 5800X3D and getting high all core boost and/or higher benchmark scores.

If you don't get the context that people are using it to test thermal performance, boost rates and stability then I'm not sure what to tell you; because it's a very well known way to very quickly and painlessly test those things.

 

[Guest2]

If you don't get the context that people are using it to test thermal performance, boost rates and stability then I'm not sure what to tell you; because it's a very well known way to very quickly and painlessly test those things.

However it's surely a flawed test on a gaming chip where the GPU and other components are under heavy load too. Test away, but using productivity tests and pure CPU testing is pointness if you want to use it for gaming.
The best benchmark would probably be real bench which tests the GPU, CPU, RAM and other components together. Best benchmarks being the games themselves. A big park on Planet Coaster is a good one as really hammers the CPU and GPU when settings are maxed. Just grab a preconfigured park and use it in sandbox mode, many like Alton towers recreations are good.

Example.
I bought the 5800X3D, plopped it in and reconfigured it to use my 5800X settings - 1900IF, RAM at 3800Mhz. I tested for a couple hours in Cinebench, Prime, DRAM Calc for Ryzen, Karhu, TestMem5. All worked perfect, just as before.
Then tried a game, CitySample (Matrix demo) within 1 minute the sound started crackling and things were slowing down then the system reset. This Infinity fabric needed lowering to 1800, so running the CPU and RAM tests only was a waste of time.

If people want to only test with CPU stress testing and find that fun then fine, just dont expect CPU stress testing to mean it's going to work when other components are thrown into the mix.

It would be nice to have a stability test app that tested the CPU, IF, RAM, GPU, Storage & sound all at once. I guess that's just a test similar to 3D Mark, but one that's not a 'benchmark' and really does uncover stability issues instead of going for highest epeen score only instead 'most stable'

People wouldnt like a benchmark (and it wouldnt sell as well) if 80% of people overclocked/underclocked only for it to crash their system after a few minutes

I just tried Planet Coaster with Alton towers sandbox map, 30-40 fps navigating around the park on my 5800X3D & 3090 Ti. Still very playable at those frames and looks fantastic. CPU hit 70c quickly

 

[GuruJockStrap]

I don't think people are suggesting purely to use Prime or Cinebench to test stability of the undervolt for the 5800X3D.

I see using those as part of a suite of tools to test stability, thermal performance, integrity, etc. In addition to gaming tests.

The more varied things I can throw at it and not show up any issues then the more I'm comfortable that the undervolt is stable. It all depends on your risk appetite and how much time you want to spend testing.

Yes its a gaming chip but some folks on here will be using it for other tasks as well.

For me, -30 was stable in gaming (I'm only playing MW2 multiplayer at the moment).

I ran Prime small FFT torture test to see have my air cooler would cope in the most extreme conditions. Ambient temp for me at the moment is around 21oC. So they way I saw it is if my air cooler can cope with a Prime test now, then when I'm gaming in summer, I should be fine (gaming temps are way below what Prime small FFT torture test can push it to).

Running that test for me at -30 PBO kept load temps around 84oC but threw up an error after 10 or so minutes. That personally doesn't sit right for me as I see a CPU as performing mathematical tasks and if its erroring in Prime (granted its extreme calculations) then its not stable. So I backed off to -25, retested overnight (economy 7 tariff ftw) and all is well.

Everyone will have a different definition of what they define as a stable system based on their risk appetite and how much time they have to invest. I personally drew the line at running something like Core Cycler, which is basically running Prime on each core to potentially run different PBO offsets per core. Don't have the time for that

 

[Richdog]

However it's surely a flawed test on a gaming chip where the GPU and other components are under heavy load too. Test away, but using productivity tests and pure CPU testing is pointness if you want to use it for gaming.
The best benchmark would probably be real bench which tests the GPU, CPU, RAM and other components together. Best benchmarks being the games themselves. A big park on Planet Coaster is a good one as really hammers the CPU and GPU when settings are maxed. Just grab a preconfigured park and use it in sandbox mode, many like Alton towers recreations are good.

Example.
I bought the 5800X3D, plopped it in and reconfigured it to use my 5800X settings - 1900IF, RAM at 3800Mhz. I tested for a couple hours in Cinebench, Prime, DRAM Calc for Ryzen, Karhu, TestMem5. All worked perfect, just as before.
Then tried a game, CitySample (Matrix demo) within 1 minute the sound started crackling and things were slowing down then the system reset. This Infinity fabric needed lowering to 1800, so running the CPU and RAM tests only was a waste of time.

If people want to only test with CPU stress testing and find that fun then fine, just dont expect CPU stress testing to mean it's going to work when other components are thrown into the mix.

It would be nice to have a stability test app that tested the CPU, IF, RAM, GPU, Storage & sound all at once. I guess that's just a test similar to 3D Mark, but one that's not a 'benchmark' and really does uncover stability issues instead of going for highest epeen score only instead 'most stable'

People wouldnt like a benchmark (and it wouldnt sell as well) if 80% of people overclocked/underclocked only for it to crash their system after a few minutes

I just tried Planet Coaster with Alton towers sandbox map, 30-40 fps navigating around the park on my 5800X3D & 3090 Ti. Still very playable at those frames and looks fantastic. CPU hit 70c quickly

That's a lot of words that say very little. Many of us have been into pc hardware for a long time so we know the ways to quickly stress test CPU clocks and thermals... and Cinebench is one of those ways. You are reading far too much into it.

 

[Guest2]

That's a lot of words that say very little. Many of us have been into pc hardware for a long time so we know the ways to quickly stress test CPU clocks and thermals... and Cinebench is one of those ways. You are reading far too much into it.

I was providing more detail. Less words - Stress testing individual components is pointless

 

[Richdog]

I was providing more detail. Less words - Stress testing individual components is pointless

You are wrong, it's not pointless for the reasons I clearly stated that I do it for. Anyway I'm fine to agree to disagree than get into a back and forth about common sense things.

 

[Skynet5]

I only run prime95 for about 10mins. Then I run 200% kharu memory tester.

After that I just use the machine for what I use it for.

If I have an issue I make a change and repeat.

Is my machine stable, in all scenerios? Maybe not.

But actually since moving to the 5800x3d I dont think there is any point doing much. I've just left everything stock...

Although I am investigating why my mobo actually powers down and restarts 3 times on boot when I tighten my primaries. It's not a soft reset, it's hard...audible clicking from somewhere psu, mobo..

 

[opethdisciple]

With no fiddling mine goes up to 4.4GHz.

Only tested by opening like 20 browser tabs and continuously refreshing them however.

But load hits 100% and the cpu turbos to a little over 4.4GHz.

 

[Woodsta888]

I only run prime95 for about 10mins. Then I run 200% kharu memory tester.

After that I just use the machine for what I use it for.

If I have an issue I make a change and repeat.

Is my machine stable, in all scenerios? Maybe not.

But actually since moving to the 5800x3d I dont think there is any point doing much. I've just left everything stock...

Although I am investigating why my mobo actually powers down and restarts 3 times on boot when I tighten my primaries. It's not a soft reset, it's hard...audible clicking from somewhere psu, mobo..

That just not part of memory learning?

 

[Skynet5]

That just not part of memory learning?

In my experience with this mobo, no. When training it would restart, but this feels like it's powercycling due to the audible clicking of some component (which I think is the PSU)

I've parked it for the moment. Blown away with the perf over my 3900x anyway, so might look in the future if I feel I'm needing a little boost and I'll just leave it all at XMP.