• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD confirms Ryzen 7 5800X3D launches this spring, Zen4 Raphael in 2H 2022

It was always clear there were issues which is why we never got the Vcache 5900X which was the cpu AMD showcased when revealing the tech. Maybe the 4ghz they used in the demo was all it could do so they had to revert to a 5800X3D with 2/3 of the cores and half the L3 cache to actually make it work but even then 4.5 was the best speed they could get.
 
Twinz said:
I remember people saying that the lower clocks were just to keep the cores and cache under the original TDP limit and we would get the clock speed back with a simple overclock.

Now that that isn't happening, people are saying we don't need the clock speed anyway.

Whatever AMD does is great, and if it isn't, some will find a way to rationalize it as great anyway.

If the part performs well for the price, great. I am still running AMD in my newest rigs and the recent price cuts may get them another CPU purchase out of me.

However, I'm not going to spin bad news for them. Even if the vcache ends up being a net gain, the clock speed regression and and a lack of headroom shows that there are still engineering trade-offs with this new tech.
Click to expand...
I remember when Intel fans were hypeing new motherboards for every new skylake generation lol, clock alone isn't important, the only important thing is performance, this is gaming cpu so no one will buy it for productivity.
 
humbug said:
PBO overclocking is a bad idea on a 5800X anyway, i know some reviewers do it but its idiotic as the extra heat actually reduces clocks, anyone who knows anything about Zen 3 knows to use Curve Optimiser tuning for the 5800X, you get the best results doing that.

Removing PBO overclocking is probably a good thing, too many idiotic reviewers use it to try and overclock and then present it as slower.
Click to expand...


Well have to see what happens, it could be that curve optimizer and the voltage and clock control is also disabled
 
Twinz said:
I remember people saying that the lower clocks were just to keep the cores and cache under the original TDP limit and we would get the clock speed back with a simple overclock.

Now that that isn't happening, people are saying we don't need the clock speed anyway.

Whatever AMD does is great, and if it isn't, some will find a way to rationalize it as great anyway.

If the part performs well for the price, great. I am still running AMD in my newest rigs and the recent price cuts may get them another CPU purchase out of me.

However, I'm not going to spin bad news for them. Even if the vcache ends up being a net gain, the clock speed regression and and a lack of headroom shows that there are still engineering trade-offs with this new tech.
Click to expand...

Grim5 said:
Well have to see what happens, it could be that curve optimizer and the voltage and clock control is also disabled
Click to expand...


HlSj3LY.png

Allowing the 5800X to consume more than its 105 Watts makes it run too hot to boost higher. Otherwise known as "PBO overclocking"

The correct way to overclocking is to use a negative curve optimiser which reduces the power consumption and then use that headroom on the boost clock override.
 
I should probably explain it.....

The 5800X has a single CCD at 105 Watts, a 5950X has 2 CCD's at 105 watts, or about 55 watts for each CCD.

The heat density is a lot higher on the 5800X, there is no room for a higher power overclock, hence using a negative curve + boost override, that way you can get to 4.8 - 5.1Ghz, you can use PBO on the 5950X because all you're doing is allowing each CCD to run at 100 Watts + and get to the same 4.8Ghz +, its how you get 32,000 Points in R23, literally double a 5800X. Because its two 105 watt 5800X's.
 
Gondar said:
I remember when Intel fans were hypeing new motherboards for every new skylake generation lol, clock alone isn't important, the only important thing is performance, this is gaming cpu so no one will buy it for productivity.
Click to expand...

You were one of the people who said AMD was just meeting a TDP target and that we could overclock:

Gondar said:
AMD decreased Zen3d clock to put them in 105tdp but you can overclock it, and also all zen3d will be B2 stepping so little better efficiency and overclock potential, enough to compensate that cache.
Click to expand...
 
Joxeon said:
It doesn't mention "Unlocked for Overclocking" on AMDs own product page whereas all the other ryzen CPUs say this.

https://www.amd.com/en/products/cpu/amd-ryzen-7-5800x3d
Click to expand...


Good find, that's probably as good an indicator as any at this point

we can only speculate on the reasons why; I suspect the heat density is right at its thermal limits with current cooling and perhaps amd testing showed if they enable pbo the chip very quickly overheats and gets damaged
 
Could preventing overclocking be damage control against the Zen 4 release? I.e. if this chip does happen to be a significant step up in gaming and Zen 4 only delivers marginal gains above that, it may sour the Zen 4 release.
 
I think they asked for this to ensure there aren't masses of heating failures, thus damaging the 3D stacked marketing image for future cpus.

Someone WILL end up overclocking it to see what happens but OCing the 3d Stuff won't be an option I'd say until the 7th or 8th generation. AMD first need to iron out flaws and refine designs as this is a new type of CPU for mainstream, stopping OCing ability is just a precaution Id think at this stage.
 
That being said, he did say that the 5800x3d can be tweaked in unofficial non AMD software - so while it can't be modified in the Bios, some 3rd party software is for now allowing the user to tweak the chip
 
random_matt said:
A failed experiment to some extent.
Click to expand...

Either need lower power draw and voltages or different forms of cooling; heat rising through a stack of transistors isn't a good idea - that's why Intel hasn't bothered stacking yet, it's working on and waiting for next gen cooling to catch up - with stacking you need to be able to cool between the layers if you want to get the best performance from it, can't just stick a passive IHS on top of the top layer and expect great heat transfer
 
Considering it's AMD's first gen with stacked cache I had a feeling there would be a fair few trade-offs. They should've kept this on the server/professional side of things and iron out the issues before bringing it to the mainstream market. I think they jumped on this too fast in an attempt to stay ahead of Intel.
 
You must log in or register to reply here.
Back
Top Bottom