Question Potentially low multi-core score for 5900x in Cinebench, is this normal?

[Fireflasher361]

Specs: Ryzen 9 5900x
Noctua NH-U12S Cooler
16 GB RAM 3200Mhz
Corsair RM750 PSU
NVIDIA RTX 3080 FE
ASUS TUF GAMING x570 PLUS Wifi

Peak speed on multicore test never exceeds 4000-4050 MHz and PPT TDC & EDC are all maxed out

 

[Mr.Spock]

you didn't list your m/b, kinda key?

 

[CountMike]

Which MB ? Key is in those red circles, it's redlining all power from MB:

 

[Fireflasher361]

you didn't list your m/b, kinda key?

Which MB ? Key is in those red circles, it's redlining all power from MB:

ASUS TUF Gaming x570 PLUS Wifi, editted the post to reflect that

 

[drivinfast247]

107 watts out of a 105w TDP CPU.

Enable PBO and see what happens.

 

[Fireflasher361]

107 watts out of a 105w TDP CPU.

Enable PBO and see what happens.

Is there anything I should do here or elsewhere besides setting PBO to enabled?

 

[drivinfast247]

Is there anything I should do here or elsewhere besides setting PBO to enabled?

You can adjust the scaler settings. See which one gives the best performance vs. heat.

 

[Karadjgne]

There's 4? settings for PBO, default, levels I-III (OC) . There's also upto a 200MHz bump. If you have the cooling for it and voltages are low enough you should see power used get higher than default settings. While default is a 105 TDP cpu (which in AMD terms is somewhat funky) it should be able to stretch upto the 142w limit without issue. The VRM's are the same as the higher grade ROG boards.

But you'll have to take that with a grain of salt. It's not Intel and the scores don't scale well with temps, it's possible to see higher scores with slower clock speeds if the temps are better. I get @ 5100 in CB20 on a 3700x at 4.28GHz and 62°C. If I push the clocks and volts to 4.40GHz it hits 82°C and I'm pulling @ 4950 instead.

 

[Fireflasher361]

You can adjust the scaler settings. See which one gives the best performance vs. heat.

I tested out PBO and it improved the results somewhat. Multi-core score in r23 was around 20700 and speeds were about 4150 to 4250 MHz. I don’t think I’ll keep this on as it decreased my single core performance too much. But either way, I’m mostly concerned because I’m seeing countless people report scores of ~21000-21500 at stock with PBO disabled and no tweaks. I’ve seen footage of speeds in the 4249-4300 MHz range while running Cinebench. All at stock. Meanwhile I’m struggling to hit that even with PBO enabled. Which is why I’m wondering if there’s something wrong. Is this a case of bad silicon? I’ve heard ram speed influences ryzen’s performance and I’m only running 3200 MHz, could that have any effect? Just feeling kinda exhausted as I’ve been searching all over the web and come up with nothing up substance.

 

[drivinfast247]

I tested out PBO and it improved the results somewhat. Multi-core score in r23 was around 20700 and speeds were about 4150 to 4250 MHz. I don’t think I’ll keep this on as it decreased my single core performance too much. But either way, I’m mostly concerned because I’m seeing countless people report scores of ~21000-21500 at stock with PBO disabled and no tweaks. I’ve seen footage of speeds in the 4249-4300 MHz range while running Cinebench. All at stock. Meanwhile I’m struggling to hit that even with PBO enabled. Which is why I’m wondering if there’s something wrong. Is this a case of bad silicon? I’ve heard ram speed influences ryzen’s performance and I’m only running 3200 MHz, could that have any effect? Just feeling kinda exhausted as I’ve been searching all over the web and come up with nothing up substance.

What BIOS version? You install the latest chipset driver for your motherboard?

 

[Fireflasher361]

What BIOS version? You install the latest chipset driver for your motherboard?

Already installed the latest chipset and BIOS prior to testing. BIOS is version 4204 and chipset driver is version 4.03.03.431

 

[Karadjgne]

One thing to remember with Ryzens is what you discovered already. They are dynamic performance based cpus, not static. They run the best performance they can according to temps, loads and voltages. There's going to be differences in cpu silicon, some ppl inevitably got Bronze cpus, some ppl got Golden ones etc. What you don't see is their cooling solutions, they could be running full custom loops in a 20°C ambient and the cpu isn't getting beyond 60° under heavy loads, at startup temps before the loop has had time to acclimate. What you aren't seeing is if they've used Dram calculator and/or CTR2 to fine tune their cpu, and/or are pushing multi-core scores at the cost of really bad single threads.

There's lots of possibilities as to why others got higher scores, but you are in the ballpark, which is good enough.

 

[Mr.Spock]

definitely if you're running @stock 3200MHz; even running 3600 w/ proportion CL may help

I tested out PBO and it improved the results somewhat. Multi-core score in r23 was around 20700 and speeds were about 4150 to 4250 MHz. I don’t think I’ll keep this on as it decreased my single core performance too much. But either way, I’m mostly concerned because I’m seeing countless people report scores of ~21000-21500 at stock with PBO disabled and no tweaks. I’ve seen footage of speeds in the 4249-4300 MHz range while running Cinebench. All at stock. Meanwhile I’m struggling to hit that even with PBO enabled. Which is why I’m wondering if there’s something wrong. Is this a case of bad silicon? I’ve heard ram speed influences ryzen’s performance and I’m only running 3200 MHz, could that have any effect? Just feeling kinda exhausted as I’ve been searching all over the web and come up with nothing up substance.