Which has better silicon? 5800X or 5900X?

[YoungBlade]

I've been looking into why 5900X CPUs tend to run cooler than 5800X CPUs despite having 4 extra cores. The two main hypotheses I've found are:

1. The 5800X has only one CCD with 2 extra cores active, so the heat is more concentrated and harder to dissipate.
2. The 5800X is lower binned silicon, so it requires more power and thus runs hotter.

My understanding is that the 5800X uses a lower binned 5950X CCD and that the 5900X uses 2 CCDs with 1-2 failed cores each, but that those CCDs are better binned than the ones used for the 5600X.

 

To me, that implies that the 5900X actually has worse silicon than the 5800X, because it is made of chiplets with failed cores. But the 5900X is binned over the 5600X, so maybe the surviving cores are better.

[Somerandomtechyboi]

Is this just a curious question or do you wanna look at a cpu for ocing?

 

Tbh ryzen doesnt gain much from ocing unless you really push your clockspeeds and volts

 

 

5800x from what i know does run hotter cause of the concentrated heat due to only having 1 ccd

 

Though not sure on the 5900x being better binned than the 5600x

[Mateyyy]

7 minutes ago, YoungBlade said:

My understanding is that the 5800X uses a lower binned 5950X CCD and that the 5900X uses 2 CCDs with 1-2 failed cores each, but that those CCDs are better binned than the ones used for the 5600X.

 

To me, that implies that the 5900X actually has worse silicon than the 5800X, because it is made of chiplets with failed cores. But the 5900X is binned over the 5600X, so maybe the surviving cores are better.

I think you're looking a little too far into it. 

 

The 5800X has its 8 cores packed much more close together than the 5900X's 12 cores, making its thermal density higher and thus making the heat transfer between the CPU and the CPU cooler less efficient - so basically, your first hypothesis.

 

I doubt there's much in the way of 'binning' being done with these chips, both because they're just regular, mainstream parts, and because it wouldn't really be practical considering the ongoing semiconductor shortage.

[YoungBlade]

15 minutes ago, Somerandomtechyboi said:

Is this just a curious question or do you wanna look at a cpu for ocing?

 

Tbh ryzen doesnt gain much from ocing unless you really push your clockspeeds and volts

 

 

5800x from what i know does run hotter cause of the concentrated heat due to only having 1 ccd

 

Though not sure on the 5900x being better binned than the 5600x

I'm just curious. I already purchased a 5900X and I've noticed that my temps are lower than those looking at temps for the 5800X.

10 minutes ago, Mateyyy said:

I think you're looking a little too far into it. 

 

The 5800X has its 8 cores packed much more close together than the 5900X's 12 cores, making its thermal density higher and thus making the heat transfer between the CPU and the CPU cooler less efficient - so basically, your first hypothesis.

 

I doubt there's much in the way of 'binning' being done with these chips, both because they're just regular, mainstream parts, and because it wouldn't really be practical considering the ongoing semiconductor shortage.

Those aren't my hypotheses. Those are the two primary reasons I've seen given on sites like here and reddit. I agree that the first one is more likely, but perhaps there's something I don't know.

 

There's certainly a degree of binning happening, otherwise the 5900X and 5950X wouldn't have higher boost speeds than the 5600X and 5800X. If all the silicon was identical, they would all be able to boost to the same level.

 

I'm pretty sure all AMD silicon is binned, because there's no difference between the silicon for servers and the silicon for consumers. They all use the same chiplets, so you can't bin for only the server-grade ones, because the only difference is the binning.

[Guest]

Doesn't matter if the chips are really binned. 

 

At most the binning is for useable core count and yields are high enough the frequency between say a 5800x and a 5900x are extremely similar.

 

So the yields of production are really good. 

 

Binning outside the manufacturer is a different story. In some cases a few elite competitive overclockers will be given (or purchase) an entire tray (or more) and will test each cpu. Sometimes this is done with extreme cooling.... and bam world record just from trial and error.

 

Fabrication process is much better today then 10 years ago. Most everyone's chips are basically the same, will reach about the same speed at the same voltage with little deviation.

 

Then looking at the nature of Ryzen chips, they hit a pretty hard wall without going sub zero. So binning for most people to gain 50mhz more then the next guy is wasted time.

[Mason Peterson]

23 minutes ago, YoungBlade said:

I'm just curious. I already purchased a 5900X and I've noticed that my temps are lower than those looking at temps for the 5800X.

Those aren't my hypotheses. Those are the two primary reasons I've seen given on sites like here and reddit. I agree that the first one is more likely, but perhaps there's something I don't know.

 

There's certainly a degree of binning happening, otherwise the 5900X and 5950X wouldn't have higher boost speeds than the 5600X and 5800X. If all the silicon was identical, they would all be able to boost to the same level.

 

I'm pretty sure all AMD silicon is binned, because there's no difference between the silicon for servers and the silicon for consumers. They all use the same chiplets, so you can't bin for only the server-grade ones, because the only difference is the binning.

Most 5800X chips at lower loads / idle will be themselves hotter than a 5900X, but once you put a full load, 5900X will suck more power. The problem comes, as hinted before, 5800X is so thermally dense with that full CCD that it's common to have really high temperatures on the cores since it's hard to remove heat like that.

This could play into binning--but I don't really care--but there is a wide variance (-/+ 8C) in what various 5800X users report with around the same cooling equipment. Aris from Hardware Busters have a stock 5800X doing 91C on a D15S, others can throw a Frost Spirit 140 or Peerless Assassin 120 into a NR200 with the side panel off and get around 73C. Mind you, this is obviously some different sound going on, but it's not like 30% vs 100% PWM. 

[freeagent]

26 minutes ago, YoungBlade said:

I'm just curious. I already purchased a 5900X and I've noticed that my temps are lower than those looking at temps for the 5800X.

I wouldn't really look at it that way. Your cooling setup might be better. From what I noticed with my 5900X is that it boosts higher and runs cooler than most guys using AIO's. Does that make my air cooler better than an AIO? No. Because my air cooler has direct access to vast amounts of fresh, forced outside air. Whereas a lot of people running just a heatsink have quiet fans that barely do anything. Like you could stick your head in the case and blow on the parts yourself and do a better job. Then when they do switch to an AIO all of a sudden their world is better because the cooler is actually getting fresh air. I take most reviews with a grain of salt.. but I still use them when deciding on purchases.

[Guest]

5 minutes ago, freeagent said:

I wouldn't really look at it that way. Your cooling setup might be better. From what I noticed with my 5900X is that it boosts higher and runs cooler than most guys using AIO's. Does that make my air cooler better than an AIO? No. Because my air cooler has direct access to vast amounts of fresh, forced outside air. Whereas a lot of people running just a heatsink have quiet fans that barely do anything. Like you could stick your head in the case and blow on the parts yourself and do a better job. Then when they do switch to an AIO all of a sudden their world is better because the cooler is actually getting fresh air. I take most reviews with a grain of salt.. but I still use them when deciding on purchases.

End result of you guys ambient coolers is the ambient room temp.

 

Obviously the guy with lower ambient temps can get an air cooler to outperform a Liquid cooler.

 

And the end result maybe 50mhx PBO difference. 

 

Self regulated overclocking makes the user OC obsolete.

[Freakwise]

39 minutes ago, YoungBlade said:

I already purchased a 5900X and I've noticed that my temps are lower than those looking at temps for the 5800X.

well theres a lot of factor you have to count as well

like the case, how much fan, what fan is it, what is the ambient temps, what rpm the fans runs at, the heatsink, where the pc is place and so on....

 

just saying

[YoungBlade]

1 minute ago, Freakwise said:

well theres a lot of factor you have to count as well

like the case, how much fan, what fan is it, what is the ambient temps, what rpm the fans runs at, the heatsink, where the pc is place and so on....

 

just saying

Yes, but from what I've found online, it seems that the 5800X tends to report higher temperatures when the cooling configuration is comparable.

 

I started looking around because of this thread on the forums. The OP has an arguably superior config to mine. At worst, we're equal. Yet even when my ambient temp was a bit higher, I was getting 10-15C lower temps. Then another user reported having a custom loop with a 5800X and also getting worse temps than me.

 

I'm quite happy with my cooling solution; I think the Fuma 2 + P300A + extra fans is a good combo, but I have a hard time believing that it's superior to a custom loop with 720mm worth of rads.

[Guest]

8 minutes ago, YoungBlade said:

 

I'm quite happy with my cooling solution; I think the Fuma 2 + P300A + extra fans is a good combo, but I have a hard time believing that it's superior to a custom loop with 720mm worth of rads.

My custom loop is geothermal. Average yearly water delta of 8c.

 

This lower temp does not make chips perform at much higher frequency under a sustained load, however the idle and mild load frequencies are sustained much higher.

 

Also, the max clocks are not increased even with a sub 50c load temp. These chips just wall. 5ghz is a stretch, even on a good day.

 

-40c to -50c, you can jump that wall, perhaps with LN2 or Dry Ice.

 

Binning would only matter to the cooling used. A binned chip on air may not OC well on LN2 for example, these chips then should be tested across different temps and cooling scenarios.

 

But where would that performance measurement come in? Would binning a cpu frequency reflect the Infinity Fabric capability from one chip to another? I think not.

[freeagent]

1 hour ago, ShrimpBrime said:

Obviously the guy with lower ambient temps can get an air cooler to outperform a Liquid cooler.

I'm not sure if it is that obvious.. lots of guys don't think it can be done. Lots of guys know it can..