Question Why is my 5950x a space heater?

May 17, 2021
3
0
10
0
Hey guys o/

I recently upgraded from a 3900x to a 5950x, and I've noticed something peculiar:

When I use Ryzen Master to "auto oc" the 5950x, my voltage rises to almost 1.5, and my H150i radiates so much heat that I don't think I'll ever need to buy a space heater in the winter, that I can just run R23 and warm the entire house.

I have the H150i's fans and pump set to balanced.

The ironic thing is that auto oc decreases performance on tests like Cinenbench and Furmark. My scores average 200 points less on R23 when overclocked.

When not auto oc'd, my voltage averages 1.087 according to RM and 1.02 according to HWiNFO. The three screencaps were taken about a minute into an R23 test under default settings. The auto oc R23 test is the burnt orange. The test on default is highlighted.

My temps are under 60 so I guess I'm thankful I don't have the same cooling issues as others, its just that the processor vomits so much heat its concerning.

I don't want to brag about speed or anything but it will reach 5 ghz on cores 3 and 6 (I've even seen 5.075 on core 3) with default settings.

My question then is - do I have a problem, or is my 5950x dogged with the same issues early Ryzen had that they're just so power hungry they produce inordinate amounts of heat?

Or am I doing something wrong? Is 3.7-3.9 ghz slow for an all-core workload?

For memory I'm using a 32GB (4x8) kit of 3600 CL14 g.skill neo.

I appreciate any insight you guys can give me,

Thanks! - A fluffy pinecone.

 
May 17, 2021
3
0
10
0
Let's just start by saying that ocing, especially the 5000 serie, brings little to no benefits anymore. So i would just leave it at stock. If you do decide to oc because why not (again, pointless but hey, it's not my pc) please do so with pbo2.

View: https://youtu.be/dfkrp25dpQ0
I undervolted by 5.

I now have an all-core average clock of 4.3, and my average temp is upwards of 75 degrees.

My R23 score jumped by 4,000 points.

Holy crap.

Its definitely a space heater now.
 

Mr.Spock

Respectable
Dec 8, 2019
1,144
134
1,890
127
it'll run hotter as you benchmark because you are using 16 vs 12 cores.
it'll seem slower in some tasks as the base clock is 3.4GHz vs 3.8Hz

the processor will also adjust performance as temps rise so even though it's at 60'C it may be reducing voltage and clock as needed to keep that temp

Hey guys o/

I recently upgraded from a 3900x to a 5950x, and I've noticed something peculiar:

When I use Ryzen Master to "auto oc" the 5950x, my voltage rises to almost 1.5, and my H150i radiates so much heat that I don't think I'll ever need to buy a space heater in the winter, that I can just run R23 and warm the entire house.

I have the H150i's fans and pump set to balanced.

The ironic thing is that auto oc decreases performance on tests like Cinenbench and Furmark. My scores average 200 points less on R23 when overclocked.

When not auto oc'd, my voltage averages 1.087 according to RM and 1.02 according to HWiNFO. The three screencaps were taken about a minute into an R23 test under default settings. The auto oc R23 test is the burnt orange. The test on default is highlighted.

My temps are under 60 so I guess I'm thankful I don't have the same cooling issues as others, its just that the processor vomits so much heat its concerning.

I don't want to brag about speed or anything but it will reach 5 ghz on cores 3 and 6 (I've even seen 5.075 on core 3) with default settings.

My question then is - do I have a problem, or is my 5950x dogged with the same issues early Ryzen had that they're just so power hungry they produce inordinate amounts of heat?

Or am I doing something wrong? Is 3.7-3.9 ghz slow for an all-core workload?

For memory I'm using a 32GB (4x8) kit of 3600 CL14 g.skill neo.

I appreciate any insight you guys can give me,

Thanks! - A fluffy pinecone.

 
Reactions: Bassman999

David0ne86

Upstanding
Mar 11, 2021
328
84
290
10
Btw, up to 80 degrees are normal ryzen temps. So i wouldn't consider your 5950 x a space heater lol. My 5800x stock easily reaches 78 degrees with a 280 arctic liquid freezer 2 in prime95 (which is not really something a cpu would normally go through). My cb temps are like yours, around 73 degrees on an all cores 4.6ghz.
 
Reactions: Bassman999
The 5950X is probably not drawing much more power than your prior 3900X, and may even put out less heat depending on the workload. I doubt you would be able to notice a significant difference in the heat output from the back of your case, at the very least, assuming you were running the same stress-testing load on both processors for the same period of time. You can't really compare the heat output of one processor running typical light to moderately-threaded desktop applications and games on one processor against the heat output of another getting pushed near its limits with a 32-thread rendering workload.

And since each core is more efficient, it will get more work done for a given amount of heat produced. So, for example, even if it were drawing a little more power and in turn putting out a little more heat at any given time while performing a rendering workload (like what Cinebench is testing), it would complete that workload faster, resulting in less total heat being generated overall. So, it's not exactly a "space heater", or at least, it wouldn't make for a very good one. : P

it'll seem slower in some tasks as the base clock is 3.4GHz vs 3.8Hz
I doubt one is likely to encounter workloads where the 5950X "seems slower" unless something were configured very wrong. Keep in mind, the 5000-series processors also have improved IPC over the 3000-series, so even if both processors were running at their base clocks, the higher IPC of the 5950X would generally put it ahead. And of course, like most modern desktop processors, one isn't likely to encounter those base clocks under load unless a processor is overheating, as they will tend to boost to at least some degree even with all cores loaded, and the 5950X has generally been shown to boost more aggressively than the 3900X. And of course, for heavily-multithreaded workloads, it also has 33% more cores to work with.
 

Mr.Spock

Respectable
Dec 8, 2019
1,144
134
1,890
127
point is even if each core were 10-15% more efficient, having 33% more cores means a net increase in power used... in a synthetic benchmark

The 5950X is probably not drawing much more power than your prior 3900X, and may even put out less heat depending on the workload. I doubt you would be able to notice a significant difference in the heat output from the back of your case, at the very least, assuming you were running the same stress-testing load on both processors for the same period of time. You can't really compare the heat output of one processor running typical light to moderately-threaded desktop applications and games on one processor against the heat output of another getting pushed near its limits with a 32-thread rendering workload.

And since each core is more efficient, it will get more work done for a given amount of heat produced. So, for example, even if it were drawing a little more power and in turn putting out a little more heat at any given time while performing a rendering workload (like what Cinebench is testing), it would complete that workload faster, resulting in less total heat being generated overall. So, it's not exactly a "space heater", or at least, it wouldn't make for a very good one. : P


I doubt one is likely to encounter workloads where the 5950X "seems slower" unless something were configured very wrong. Keep in mind, the 5000-series processors also have improved IPC over the 3000-series, so even if both processors were running at their base clocks, the higher IPC of the 5950X would generally put it ahead. And of course, like most modern desktop processors, one isn't likely to encounter those base clocks under load unless a processor is overheating, as they will tend to boost to at least some degree even with all cores loaded, and the 5950X has generally been shown to boost more aggressively than the 3900X. And of course, for heavily-multithreaded workloads, it also has 33% more cores to work with.
 

ASK THE COMMUNITY