5900x with peerless assassin 120 temps and clock speed?

[Gohardgrandpa]

I just got a 5900x, was previously using a 5600x. After selling the 5600x I ended up paying $50 for the new to me cpu. I needed more cores/threads for productivity so that was a big win. 

 

The only cooler I have is a wraith prism. In cinebench r23 multi core I’m getting 21104 avg score, 85C on a 10 minute run and clock speed of 4.05-4.1ghz. 
I get similar results when playing games. 
 

I don’t like aio coolers, and from what I can see the peerless assassin 120 seems like the best bang for your buck. If I have to go aio I will but I’d prefer not to.  

Is anyone here using the PA120 and if so what are your temps, clock speed and score running cinebench/gaming with a 5900x? 
 

I’m interested in seeing results with other coolers too. My case is the be quiet dx500 with 6 140mm be quiet silent wings fans. 
 

 

[johnno23]

your temps are fine on a 10 minute run but your 5900x should hit 4.5 on all core. 

I also run a 5900x but run it on an x570 Aorus master. 

I have zero overclocks and xmp dissabled and my cinebench scores are 22841. my cpu hits 82c sometimes 83c

[Gohardgrandpa]

5 minutes ago, johnno23 said:

your temps are fine on a 10 minute run but your 5900x should hit 4.5 on all core. 

I also run a 5900x but run it on an x570 Aorus master. 

I have zero overclocks and xmp dissabled and my cinebench scores are 22841. my cpu hits 82c sometimes 83c

As the temp goes up I see the clock speeds drop til it gets to 4.05ghz. I don’t have any overclock on here, running it stock settings because I was worried the cooler wouldn’t be able to keep up which is clearly the case.
What cooler are you using? 

[RONOTHAN##]

41 minutes ago, johnno23 said:

your temps are fine on a 10 minute run but your 5900x should hit 4.5 on all core. 

I also run a 5900x but run it on an x570 Aorus master. 

I have zero overclocks and xmp dissabled and my cinebench scores are 22841. my cpu hits 82c sometimes 83c

As someone who has that exact same setup, not really. Boost algorithms are very weird with how it does frequency, and it's not uncommon for it to report one frequency but perform like something else. I've seen it do 23000 points while reporting 4.4GHz, and I've seen it do 18000 points while reporting 4.8GHz (both with realtime priority so background tasks don't interfere with the score), all depending on the BIOS, AGESA, etc. The only thing you can judge is the score itself, the reported frequency is kinda irrelevant. 

 

To OP, I haven't used a PA120 on my 5900X, though I did have it on a Scythe Fuma 2 for a week. Temps were fine, mid 80s, and it PBO cinebench scores were in the 22500 range (give or take 500 points, this was something like a year ago). The PA120 is a very similar cooler, performance should be better than the Fuma 2 going by GN's numbers, so I'd expect it to be perfectly fine. It'll be a major improvement over the stock cooler. 

[Gohardgrandpa]

14 minutes ago, RONOTHAN## said:

As someone who has that exact same setup, not really. Boost algorithms are very weird with how it does frequency, and it's not uncommon for it to report one frequency but perform like something else. I've seen it do 23000 points while reporting 4.4GHz, and I've seen it do 18000 points while reporting 4.8GHz (both with realtime priority so background tasks don't interfere with the score), all depending on the BIOS, AGESA, etc. The only thing you can judge is the score itself, the reported frequency is kinda irrelevant. 

 

To OP, I haven't used a PA120 on my 5900X, though I did have it on a Scythe Fuma 2 for a week. Temps were fine, mid 80s, and it PBO cinebench scores were in the 22500 range (give or take 500 points, this was something like a year ago). The PA120 is a very similar cooler, performance should be better than the Fuma 2 going by GN's numbers, so I'd expect it to be perfectly fine. It'll be a major improvement over the stock cooler. 

Thanks, I’m gonna order the pa120 right now so I can get it tomorrow 

[Gohardgrandpa]

@RONOTHAN##

In cinebench r23 what’s the all core clock speed running the multi core test? I got my cooler in today and I’m gonna test it out, curious what I should be expecting to see when using pbo 

[RONOTHAN##]

Just now, Gohardgrandpa said:

@RONOTHAN##

In cinebench r23 what’s the all core clock speed running the multi core test? I got my cooler in today and I’m gonna test it out, curious what I should be expecting to see when using pbo 

Don't pay too much attention to the clock speed during R23, look at the score. I've seen it run at anywhere between 4.4GHz and 4.65GHz at complete stock but score within 100 points of each other because PBO glitches out reported clock speeds. I can get it to report 5GHz in R23 if I mess with the right PBO variables and use the appropriate BIOS revision, it just won't perform anywhere near what you'd expect 5GHz to do. 

 

I'd be expecting the score to be somewhere between 22,000 and 23,000 depending on how clean your Windows install is. If you tweak your PBO settings, possibly mess with the curve optimizer you might be able to get a little over 23,000. 

[Gohardgrandpa]

12 minutes ago, RONOTHAN## said:

Don't pay too much attention to the clock speed during R23, look at the score. I've seen it run at anywhere between 4.4GHz and 4.65GHz at complete stock but score within 100 points of each other because PBO glitches out reported clock speeds. I can get it to report 5GHz in R23 if I mess with the right PBO variables and use the appropriate BIOS revision, it just won't perform anywhere near what you'd expect 5GHz to do. 

 

I'd be expecting the score to be somewhere between 22,000 and 23,000 depending on how clean your Windows install is. If you tweak your PBO settings, possibly mess with the curve optimizer you might be able to get a little over 23,000. 

Thank you for replying. I’ve got my old results with the wraith prism above so I’ve got some kind of baseline for what I’m getting by swapping coolers in terms of temps and r23 score. 

I’m gonna test it stock than with pbo and with a -.1 voltage offset and see what I get. I’ll update this tonight when I get to test it in about 2 hours. 

[RONOTHAN##]

5 minutes ago, Gohardgrandpa said:

m gonna test it stock than with pbo and with a -.1 voltage offset and see what I get.

So the thing with Ryzen chips is that they clock stretch (perform worse while the clocks are the same), with the best way to get them to clock stretch being to set a negative voltage offset. If you want to undervolt, you need to use the curve optimizer and not try to set negative offsets. 

[Gohardgrandpa]

2 hours ago, RONOTHAN## said:

So the thing with Ryzen chips is that they clock stretch (perform worse while the clocks are the same), with the best way to get them to clock stretch being to set a negative voltage offset. If you want to undervolt, you need to use the curve optimizer and not try to set negative offsets. 

Ah ok, I didn’t know that. Here’s what I got so far

previous results. 
wraith prism pbo on, limit set to motherboard 
10 minute R23 run 4.05ghz 85c 

MC score of 21104

 

with pa120 voltage on auto, pbo off

10 minute R23 run 4.1ghz 62C

MC score of 20805 

 

with pa120 voltage on auto, pbo on +200 pbo limit set to motherboard

10 minute R23 run 4.3ghz 74c

MC score of 21695 

 

with pa120 voltage on auto, offset -.1v, pbo on +200 pbo limit set to motherboard

10 minute R23 run 4.35ghz 69C

MC score of  22076

 

 

 

[RONOTHAN##]

1 minute ago, Gohardgrandpa said:

Ah ok, I didn’t know that. Here’s what I got so far

previous results. 
wraith prism pbo on, limit set to motherboard 
10 minute R23 run 4.05ghz 85c 

MC score of 21104

 

with pa120 voltage on auto, pbo off

10 minute R23 run 4.1ghz 62C

MC score of 20805 

 

with pa120 voltage on auto, pbo on +200 pbo limit set to motherboard

10 minute R23 run 4.3ghz 74c

MC score of 21695 

 

with pa120 voltage on auto, offset -.1v, pbo on +200 pbo limit set to motherboard

10 minute R23 run 4.35ghz 69C

MC score of  21957 

 

 

 

Didn't realize you were doing 10 minute runs, those will be a bit lower than the scores I'm referring to. I generally only do 10 minute runs when I'm doing case thermal tests, for actual scores a single pass is usually fine. I also tend to run with realtime priority set, that can get anywhere from about 100 to 1000 points higher score depending on how bloated your OS install is, so keep that in mind. 

[Gohardgrandpa]

6 minutes ago, RONOTHAN## said:

Didn't realize you were doing 10 minute runs, those will be a bit lower than the scores I'm referring to. I generally only do 10 minute runs when I'm doing case thermal tests, for actual scores a single pass is usually fine. I also tend to run with realtime priority set, that can get anywhere from about 100 to 1000 points higher score depending on how bloated your OS install is, so keep that in mind. 

I did the 10 minute runs to see how the temps changed between the different settings. 
The -.1v offset scored the highest out of all the runs and kept the speed the highest 

[RONOTHAN##]

Just now, Gohardgrandpa said:

I did the 10 minute runs to see how the temps changed between the different settings. 
The -.1v offset scored the highest out of all the runs and kept the speed the highest 

There might be an AGESA version where it's not completely broken. Most of the ones I've used -.1V would dramatically lower performance in certain workloads. 

[Gohardgrandpa]

I just checked the version I’m running 

Bios version 1.B0   08/10/2022

Agesa version CombiSM4v2PI 1.2.0.7

 

Looking on the Msi site, there’s a bios that came out on 3/14/23 that updates the agesa to 1.2.0.8 for my board. I wonder if that’s the one your talking about