I support this. Somehow this 12-core processor is such a beast, that it doesn't need tweaks in order to want some more from it. Not yet, of course, after 3-4 years, maybe........
-
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 Zen 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***
- Thread starter Jamin280672
- Start date
I support this. Somehow this 12-core processor is such a beast, that it doesn't need tweaks in order to want some more from it. Not yet, of course, after 3-4 years, maybe........
Ok, thanks. Btw, what do you mean by 'IF'?
The North Bridge
Ahh, thanks. I'll have to look into that. Currently, I've just set my memory to XMP. That's about all I've done.
Waiting on a bracket so I can use my Noctua from the last build.
Soldato
Infinity Fabric
Yeah. Every time tried to tweak the 3900X resulted to lower speeds than normal. After all 50-100mhz don't make difference to spend time.
What matters is tweaking the subsystems and memory.
Soldato
I've never used it.
Also controversially says that VDDG should be 1.1v which is contrary to what the likes of The Stilt and 1usmus say on the subject.
Soldato
Thanks.
FYI that doesn't mean cannot be used as for manual overclock is the best tool out there for setting the speed per CCX.
But it won't work for PBO on 3900X and consequently 3950X.
There is a good guide for manual overclocking per CCX.
Soldato
Right so this video explains why even with a stable ram overclock I was still freezing. (400% HCI stable)
https://vid.wxzm.sx/watch?v=xSAFs-t0DCg (16:40mins)
It's because these ram testing utilities only stress the ram subsystem, they are not putting any pressure on the IF.
Which is why it makes sense that the missing ingredient to my ram troubles is the VDDG voltage. As that is the one that helps with IF overclocks.
Doesn't explain why I need it at 1800MHz and some of you guys at faster speeds don't.
But maybe it's just individual to the chips themselves.
https://vid.wxzm.sx/watch?v=xSAFs-t0DCg (16:40mins)
It's because these ram testing utilities only stress the ram subsystem, they are not putting any pressure on the IF.
Which is why it makes sense that the missing ingredient to my ram troubles is the VDDG voltage. As that is the one that helps with IF overclocks.
Doesn't explain why I need it at 1800MHz and some of you guys at faster speeds don't.
But maybe it's just individual to the chips themselves.
Last edited:
Soldato
Yeah I read some chips will vary so your on the right track there.
Speaking generally stuff like this usually comes down to individual chips - over the course of production you get quite a range in the actual physical quality of the semi-conductor which will affect stuff like this.
Soldato
If you were to use Ramtest and then ticked the "Stress FPU" box in the advanced section, you would absolutely be putting both IF and the CCX under stress.
I had 1866 MHz (3733 MT/s) working and always passing memory stress tests and even OCCT but crashing in games. Since then I dialled it back to 1800 MHz (3600 MT/s) with the same timings and it is rock solid. The latency isn't as low as I'd like so I'll probably try 1833 MHz (3666 MT/s) at some point but I think 1866 MHz (3733 MT/s) is just a no-go for me, at least with the primary timings I want. I already tried bumping and lowering every related voltage and nothing ever helped.
Soldato
Wonder if we are going back to the quality of VRMs and amount of phases on later boards. Mine wont do 1900 IF on X570 albeit bottom end of these boards. Though I haven't tried on the last 2 BIOS's. Performance wise you'd not notice anything in gaming. I'd run 3600Mhz stable until BIOS's have matured and try for higher IF then.
Can't be watching any video with someone that says 'like' or 'kinda' every sentence. Man, that really makes my teeth itch. (My problem I know).
Soldato
Yes. At the moment, I'm not claiming I am stable. It's hardly been a week yet since I set VDDG to 1.1v.
But I'm giving it ago as it's the only thing I haven't fiddled with voltage wise.
1.1v SOC and 1.1v VDDG as per Ryzen Master documentation.
But I'm giving it ago as it's the only thing I haven't fiddled with voltage wise.
1.1v SOC and 1.1v VDDG as per Ryzen Master documentation.
Soldato
I've just changed mine too now. In BIOS's is this not the voltage that says may approach but not exceed SOC? I set VDDG in RM to 1.1, post reboot shows 1.0979v
I've had a couple of rare and random crashes so I'll give VDDG a go also
Soldato
Yes that one.
I'm not sure why we have this warning in the BIOS about it can "approach but not exceed SOC voltage". And yet AMD's own documentation says to put VDDG to 1.1v.
But any way 1.075v and it goes yellow (at least on my board) so I cant believe VDDG at 1.1v is that high and harmful.
Soldato
Hmm Intel actual getting competitive. Roll on TR3.
