5900x vs 5950x Question

LGabrielPhoto

2[H]4U
Joined
Jan 5, 2006
Messages
3,103
is the 5950x hotter than the 5900x , being its a binned cpu , I had 9900KS and it was way cooler than my regulars 9900k binning helped alot with
I think most of the benchmarks I've seen had the 5900x running better on average boosts. My 5950x gets a R20 single score of ~615, and boosts between 4.8 - 4.9 during the benchmark, but regularly sits at 4.9 - 5.05 during gaming/normal usage. I'm also running ambient temps about 10F higher than normal right now because I live in Texas, so there's a chance it could be a bit better.
That is boosting low for single. Maybe temperature related add you mentioned. I have mine tweaked to get me a good balance of single and multi.
 

Attachments

  • CTR_CB644.JPG
    CTR_CB644.JPG
    63.9 KB · Views: 0
Last edited:

schizo

[H]ard|Gawd
Joined
Nov 6, 2004
Messages
1,654
4.9-5.0 is low? That's as high as Zen3 ever gets without sub-ambient cooling.

I'd say 4.6/4.7 is low, 4.8 is normal without tweaking, and 4.9/5.0 you did good. There isn't a ton of headroom in these chips, they boost pretty close to their limits at stock.
 

bobzdar

[H]ard|Gawd
Joined
Jun 6, 2003
Messages
1,822
is the 5950x hotter than the 5900x , being its a binned cpu , I had 9900KS and it was way cooler than my regulars 9900k binning helped alot with temps

It's actually not hot at all - at stock at least. I was only hitting a little over 60C at stock with 240AIO, where 5800X's seem to hit 15C higher at stock during all core stress testing. That's the product of spreading 142W over 16 cores vs. 125W over 8. 5900X will be similar, you're spreading 142W over 12 cores instead of 16, so it'll run a little hotter. Cranked up it can get hot, sure, but I'm at board limits (190A EDC) hitting around 215W and still topping out at 85C. Performance is nuts, though. I can crank through photogrammetry builds in 30-40% less time than the 3950X, a function of PBO doing almost nothing on that chip (so I ran it stock), the 5950X being 15-20% faster at stock, and then getting another 10-15% running PBO+CO. For games it just doesn't matter, any of the r9's are overkill and most games stop scaling at 6 cores, so even a 5800X won't be running flat out and PBO+CO seems to let it get similar single core speeds.

A 5900X with pbo+CO will be a decent amount faster than a 3950X with PBO, which is a pretty amazing generational jump. The 5950 is another level altogether. Zen 3 threadripper is going to be fucking nuts.
 

sonofson

n00b
Joined
Feb 24, 2021
Messages
3
Guys, I'll say right away I'm new to this, so don't kick me if I say something wrong. Why are you only comparing 5950x and 5900x? After all, you can also compare it with the Core i9-10900K. The parameters in the benchmark are about the same, but the i9 is half the price.
 

Digital Viper-X-

[H]F Junkie
Joined
Dec 9, 2000
Messages
14,269
Guys, I'll say right away I'm new to this, so don't kick me if I say something wrong. Why are you only comparing 5950x and 5900x? After all, you can also compare it with the Core i9-10900K. The parameters in the benchmark are about the same, but the i9 is half the price.

the 10900k is not half the price of a 5950x, though it is cheaper, and really they don't compare, that is why people are not comparing it. 10 core intel chip vs a 16 core AMD chip where AMD is already faster core for core. 5900x Already beats out the 10900k.
 

bobzdar

[H]ard|Gawd
Joined
Jun 6, 2003
Messages
1,822
Guys, I'll say right away I'm new to this, so don't kick me if I say something wrong. Why are you only comparing 5950x and 5900x? After all, you can also compare it with the Core i9-10900K. The parameters in the benchmark are about the same, but the i9 is half the price.

A lot of the debate is around boost clocks in the chiplet architecture, which is completely different to the monolithic design of intel. 10900k is also overkill for gaming.
 

SpeedMan88

n00b
Joined
Apr 27, 2019
Messages
3
I just got a 5950x in my machine. After lots of tweaking the PBO + CO settings, LLC, and voltage offsets, I've got it boosting to 5025mhz consistently in single threaded benchmarks and about 4500mhz +/- 25 in all core benchmarks. CB20 scores are 640 single and 11400 multi. The multi core benchmark maxes out at about 73C on my 240 AIO cooler. I have seen a boost clock recorded at 5200mhz in hwinfo64 a few times, but it never sustains this.

One thing I haven't figured out yet that is bugging me is how the Infinity Fabric plays into things. Currently I'm running 1900 IF with my ram at 3800 CL16. My chip is also stable at 1967 IF with the exact same ram timings at 3933 CL16. Yet, my CB20 multi score is over 1000 points lower when I do this (and CB23 multi is 3000 points lower). Anyone have any thoughts on why I'd be seeing lower benchmark scores with a faster IF given everything else is kept the same? AIDA64 reports better memory speeds (61GB/s vs 59 GB/s) and latency (56.8ns vs 57.8ns) with the higher IF, so I know it's at least improving the memory performance. Anyone else with these chips notice something similar?
 

GotNoRice

[H]F Junkie
Joined
Jul 11, 2001
Messages
10,038
5800X's seem to hit 15C higher at stock during all core stress testing. That's the product of spreading 142W over 16 cores vs. 125W over 8. 5900X will be similar, you're spreading 142W over 12 cores instead of 16, so it'll run a little hotter.

Your assumption about the 5900X is totally false. The 5800X runs hotter because it's a single CCD, which means less surface area to transfer heat from the 8 cores to the heat spreader. Both the 5900X and the 5950X use two CCDs with the exact same total surface area connected to the heat spreader - the 5900X simply has two cores disabled on each CCD, but those inactive cores still help to conduct heat to the heat spreader. That actually gives the advantage to the 5900X in terms of heat, as the 5950X will often tend to fully load one CCD (8 cores) before switching to the other. With the 5900X you will only ever have 6 cores active per CCD, so the heat will often be spread over a larger surface area when more than 6 cores are used.
 

evhvis

n00b
Joined
Feb 12, 2021
Messages
39
Your assumption about the 5900X is totally false. The 5800X runs hotter because it's a single CCD, which means less surface area to transfer heat from the 8 cores to the heat spreader. Both the 5900X and the 5950X use two CCDs with the exact same total surface area connected to the heat spreader - the 5900X simply has two cores disabled on each CCD, but those inactive cores still help to conduct heat to the heat spreader. That actually gives the advantage to the 5900X in terms of heat, as the 5950X will often tend to fully load one CCD (8 cores) before switching to the other. With the 5900X you will only ever have 6 cores active per CCD, so the heat will often be spread over a larger surface area when more than 6 cores are used.
When loading all cores the 5950x will have the heat advantage, but in most other cases the 5900x will run cooler. Fully loading 4 cores with prime95 small FFT gives slightly higher temps than 6 cores, even though 6 cores is running 142w and 4 cores runs approx. 130w on the 5900x. The heat in the CCD relates both to highest watt for a core and total watts for CCD. You would get less spot heat on the 5950x when fully loading 24 threads due to having less watts per core. The difference would most likely be minor though (within a few degrees is my guess).
 
Last edited:

bobzdar

[H]ard|Gawd
Joined
Jun 6, 2003
Messages
1,822
Your assumption about the 5900X is totally false. The 5800X runs hotter because it's a single CCD, which means less surface area to transfer heat from the 8 cores to the heat spreader. Both the 5900X and the 5950X use two CCDs with the exact same total surface area connected to the heat spreader - the 5900X simply has two cores disabled on each CCD, but those inactive cores still help to conduct heat to the heat spreader. That actually gives the advantage to the 5900X in terms of heat, as the 5950X will often tend to fully load one CCD (8 cores) before switching to the other. With the 5900X you will only ever have 6 cores active per CCD, so the heat will often be spread over a larger surface area when more than 6 cores are used.
True, there's the same area spreading heat but each core gets more power in the 5900X so the heat will be more concentrated when generated. It won't be as drastic as the 5800X, probably more like 5C hotter vs 15C. But remember, all of the cpus rotate the load through different cores so the less cores, the more localized heat generation as there are less cores to spread the load over. They have the same area to dissipate the heat, but the heat is generated in less area so will be a little hotter.
 

LGabrielPhoto

2[H]4U
Joined
Jan 5, 2006
Messages
3,103
Latest CTR seems pretty stable, much better than all previous version at least on my system
1614557584827.png

Passes BLENDER, AIDA, Realbench etc
 
Top