Changed motherboard, CPU runs harder but gets same same CB score as before

[Illyaimi]

I checked last year in around 5-10C+ room temps and I used to get runs like these.

 

Multi and single

 

 

This was with my old x370 Taichi, PBO on with default settings the board was giving me so PPT TDC EDC all on 0 (ryzen master showed stuff like 740 480 650 for the values)

 

Now I have a x570 Dark Hero and I get runs like these (for this mobo PBO on 0 0 0 sets these numbers 395 255 200)
Multi

 

Single

 

Yeah the ram is running at 2133c15 compared to 3200c14 from last year results, but the CPU goes to higher voltages/mhz and temps now even though the result is pretty much the same. Is there anything I can do to get back to around the same max temp levels I did on the first screenshot? The bios options should be identical outside of ram, but the results are different enough tempwise where it bothers me, since it could easily go 80c+ in the same situation if the room temps were higher. and that's at least 10c+ from my old results under the same bios options.

 

I'd appreciate any opinions or suggestions

 

Edit 9/1

 

Decided to put the problem on the backburner for now. I might get 8300/434 in cinebench but my temps are good and until I notice performance issues in either blender or other workloads(or games) I will stick with the current settings as my usual setup has the CPU run on low 40s tempwise, which is way better than the 50-60 I was used to previously. Game + 4 core emulator + another emulator/game chrome with 150+ tabs with a stream running and other random shit.

 

I will make an update when I figured out what was causing it(only thing I haven't tried is disabling ftpm but it's on the list) in case someone else encounters an issue like this and is looking for info.

 

Edit 9/3 

 

I didn't change anything and I get 9000/456. It's pretty weird though that for no reason the performance changed that much.

[Mark Kaine]

what was the reason for this motherboard change?

 

Why aren't you running the RAM at the same speed?

 

 

[Illyaimi]

Previous motherboard had a dying ethernet chip or something got corrupted that I wasn't able to figure out, disabling/refreshing/uninstalling/installing or even just lightly touching the driver for it would crash the entire pc even through multiple clean Windows installs.

 

The ram is running at 2133c15 currently because I'm testing for stability of my ram setup(2x16 3200c14 + 2x16 4000c16.) Gonna do 10 hours of tm5 tonight and if it goes fine I'm going back to 3200c14 xmp.

 

Would the ram running slower really affect the max voltage/freq/temp I get during a workload?

[Mark Kaine]

2 hours ago, Illyaimi said:

Would the ram running slower really affect the max voltage/freq/temp I get during a workload?

I don't think so, but the bios will be slightly different,  so its possible you would actually get more performance with faster ram, with the slow ram it kinda cancels out probably. 

[Illyaimi]

31 minutes ago, Mark Kaine said:

I don't think so, but the bios will be slightly different,  so its possible you would actually get more performance with faster ram, with the slow ram it kinda cancels out probably. 

Makes sense actually. Might try that then and see if I get a higher score in which case I'll try to make adjustments to bring down the temps to around what I was used to.

[Illyaimi]

 

Enabled xmp and got an even lower score and temps didn't go lower at all. Something is not right

 

 

oof

 

Wait why is this lower lmao 

3200c14

 

2133c15

[Illyaimi]

I put asus pbo amd pbo and xfr pbo all on auto now and I got the worst score so far.

 

 

Temps were good as shit but this CPU can 100% do 9k+ on 65-70C so there's still work left to do

[Illyaimi]

Yeah dunno what to do now. I can't even get a similar run to the one where I had 9250 500 even with the same settings now. On default bios options I get 8300 score and single is something dumb like 430. It's bad.

[SamClan]

okay, we start from the beginning. 

 

 

0. have you tried Bios update? 

 

1. do you put enough thermal paste on your cpu? 

2. cleaning? you wouldn't believe where dust can settle. As well as check your fans and the fan settings in bios. 

3. Power plan in windows, not sure if it still influence your system, but it had heavy influence in the beginning.

4. check if your XMP Profile is really dialed in correctly. (I have manual set it as well as higher voltage)

 

 

[Illyaimi]

51 minutes ago, SamClan said:

okay, we start from the beginning. 

Good idea. It's around midnight here so I'll go agane tomorrow.

 

Yeah I updated Bios.

As for the other points, I evenly applied the paste, enough to not have it spill over the CPU

I cleaned everything when when I swapped the motherboard, there was a lot of dust, probably need to clean around once a month ideally.

Power plan is on this. I used this same power plan on last mobo as well

The CB scores in the first post are on stock ram, that's what I don't understand the most. Now I can't get the same score no matter what setting I try. I enabled XMP (3200c14 1.35v) and my score went down to 8600. Now it doesn't matter what speed I run my ram at, I get shit scores either way.

On full stock bios settings I legit get like 8000 in cb20.

 

Going to start over tomorrow with new motivation

[Montana One-Six]

On 8/28/2021 at 3:12 PM, Illyaimi said:

2x16 3200c14 + 2x16 4000c16

Could be part of the issue. Try your benchmarks with only one of those kits.

 

Generally mixing two different RAM kits will often lead to issues.

[Illyaimi]

16 hours ago, Montana16 said:

Could be part of the issue. Try your benchmarks with only one of those kits.

 

Generally mixing two different RAM kits will often lead to issues.

Well, they should be the same kits if part numbers are anything to go by, so there shouldn't be much of an issue if they are running under the same profile. Both of them are dual rank and have this on the chips '04213X8810B' despite being sold as 3200c14 and 4000c16. F4-3200C14D-32GTZR + F4-4000C16D-32GTZR.

 

I had a bsod spree a few reinstalls ago where I thought it was the ram, but it stopped after another reinstall and haven't had an error since both on 2133c15 stock and both with 3200c14 XMP enabled. I tested the 2 kits together on 2133c15 with tm5 for 3 cycles(around 8 hours) and got 0 errors. I have yet to test them on the current xmp profile, but if it was because of that then I don't get why the scores from the first post didn't come back even after reverting to stock which did give me 9300/50.

 

Like this is all I did there and yet the performance stayed even after reverting to stock ram speeds.

 

 

Shame that I didn't save the bios profile with those settings but it just means that I have to keep trying.

[Montana One-Six]

18 hours ago, Illyaimi said:

Well, they should be the same kits if part numbers are anything to go by, so there shouldn't be much of an issue if they are running under the same profile. Both of them are dual rank and have this on the chips '04213X8810B' despite being sold as 3200c14 and 4000c16. F4-3200C14D-32GTZR + F4-4000C16D-32GTZR.

 

Could be but do you know for sure? I have seen kits from the same exact model number having different chips on them, those weren't G.Skills but who says they don't do that?

 

It's your problem and your decisions on what you want to do to figure out the problem. I was just giving you a suggestion and if I were facing those problems I would definitely test those kits individually just to be sure. I mean swapping ram is easy and you can swap and test both kits individually in under 10 min.

[Illyaimi]

3 hours ago, Montana16 said:

Could be but do you know for sure? I have seen kits from the same exact model number having different chips on them, those weren't G.Skills but who says they don't do that?

 

It's your problem and your decisions on what you want to do to figure out the problem. I was just giving you a suggestion and if I were facing those problems I would definitely test those kits individually just to be sure. I mean swapping ram is easy and you can swap and test both kits individually in under 10 min.

I will definitely try that too. Good to eliminate as many possible issues as I can