Caporegime
- Joined
- 8 Jan 2004
- Posts
- 32,614
- Location
- Rutland
Yeah I dont see the problem?
-
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.
3800x thread.
- Thread starter beany_bot
- Start date
Yeah I dont see the problem?
The 3700X is such good value i went for this over the 3800X. I have a 3900X in my second system.
Soldato
- Joined
- 24 Sep 2013
- Posts
- 2,890
- Location
- Exmouth, Devon
You not work for them anymore? So is there not gonna be any BIOS/chipset enhancements that will leave a gulf of performance difference between the 3700X and the 3800X?
Still can't get my 3800X to draw any more than 60W but it does 4.5+ on 5/8 cores and gaming (PUBG) runs 4.35 on all cores
I assume a 3 core CCX will need to send data over the IF more often than a 4 core CCX?
Caporegime
- Joined
- 8 Jul 2003
- Posts
- 30,063
- Location
- In a house
Yes i still work for AMD, no freebies unfortunately this time.You not work for them anymore? So is there not gonna be any BIOS/chipset enhancements that will leave a gulf of performance difference between the 3700X and the 3800X?
Still can't get my 3800X to draw any more than 60W but it does 4.5+ on 5/8 cores and gaming (PUBG) runs 4.35 on all cores
I am using the stock cooler (Wraith Prism) on both my systems and just let PBO do its thing. I have both systems memory 32GB running at 3600Mhz CL16.
Yes it does, they are ALL Zen2 8c16t chiplets. All Zen2 based Ryzen 3000 chips use the exact same core (ignoring the APU which is still Zen1+). They are binned down, AMD would hope that all cores can meet 3950x spec (or beyond), but in the real world that doesn't happen. That's why we get everything else below that, and is also why i assume the R5 3500 6c6t part has been "made" probably because some cores cant handle SMT either so thats basically the last place for Zen2 chips to go before they go in the rubbish bin. Its why the 3900x and 3950x are out of stock or not released yet, they are at the absolute limit of current 7nm yields so have to stock pile the best of the best for them.
A zen2 chiplet that can do 3900x speeds will be put into that bin, and have two cores disabled, its not two 3600's because a 3600 cant do the ghz of a 3900x.
Associate
- Joined
- 9 Jun 2004
- Posts
- 1,502
Got my RAM and IF at 3800 1:1 without any real trouble. Seems rock steady. Think I'll leave it at that.
Not all 3800x can do that, one of mine didn´t that is why I returned it
Associate
- Joined
- 9 Jun 2004
- Posts
- 1,502
Yea, I've not had any trouble with boosting or anything else really so it seems I've got a good one.
Really, what reason did you give them? lol
Matisse SSE Frequency AVX2 Frequency Voltage % Capable
3700X Not Tested 4.05GHz 1.237V 100%
3700X Not Tested 4.10GHz 1.250V Top 74%
3700X Not Tested 4.15GHz 1.262V Top 21%
3800X Not Tested 4.20GHz 1.275V 100%
3800X Not Tested 4.25GHz 1.287V Top 53%
3800X Not Tested 4.30GHz 1.300V Top 20%
3900X Not Tested 4.00GHz 1.200V 100%
3900X Not Tested 4.05GHz 1.212V Top 87%
3900X Not Tested 4.10GHz 1.225V Top 68%
3900X Not Tested 4.15GHz 1.237V Top 35%
3900X Not Tested 4.20GHz 1.250V Top 6%
3800X looks to be a much better binned 3700x. If you are overclocking then 3800x minimum.
3700X Not Tested 4.05GHz 1.237V 100%
3700X Not Tested 4.10GHz 1.250V Top 74%
3700X Not Tested 4.15GHz 1.262V Top 21%
3800X Not Tested 4.20GHz 1.275V 100%
3800X Not Tested 4.25GHz 1.287V Top 53%
3800X Not Tested 4.30GHz 1.300V Top 20%
3900X Not Tested 4.00GHz 1.200V 100%
3900X Not Tested 4.05GHz 1.212V Top 87%
3900X Not Tested 4.10GHz 1.225V Top 68%
3900X Not Tested 4.15GHz 1.237V Top 35%
3900X Not Tested 4.20GHz 1.250V Top 6%
3800X looks to be a much better binned 3700x. If you are overclocking then 3800x minimum.
Ive seen that before. Why does it say not tested though?
Soldato
- Joined
- 12 Feb 2014
- Posts
- 3,013
- Location
- Somewhere Only We Know
My 3800X does 4.325ghz @ 1.325v does that put it in the top 20% of all CPU's or of all 3800X's then, thats 100% stable, tested, tested and tested some more for hours upon end, IBT AVX, Prime95, Realbench, Memtest, AIDA64, ive also topped the Fabric out at 1900mhz, with RAM at 3800mhz.
Why don't they test the 3700x at the same voltage as the 3800x? Are they trying to make the difference look bigger than it actually is to drive sales of the more expensive chips?
The maximum voltage I would guess is vcore 1.325. https://www.youtube.com/watch?v=-6VlrlS0f4Q see #1 So that would imply that very few 3700x's overclock past 4.15GHz for that voltage in their sample set. I am sure that there are CPU's at higher clocks but they are likely rare.
The best way to overclock looks to be phase change. That way 4.8Ghz all core would be possible. https://www.youtube.com/watch?v=JdYKVLrQL4s
#1
1.325V as long as you do NOT mess with the load line default. That 1.325V number assumes that high-current voltage droop will take it down to 1.245V which is the actual maximum safe voltage for high-current loads. If you mess with the load line, though, it won't droop as much and you will be well on your way to destroying your CPU. At low-current loads (i.e. single-core boost), the maximum safe voltage is much higher at 1.45V (which is why stock systems can boost single cores to 4.6GHz). Don't expect to be able to boost all cores the same as single-core, its just impossible... its a high-current situation that significantly lowers the maximum safe VCORE verses single-core. That's why it is impossible with normal (ambient) cooling solutions. The other problem you are facing is temperatures. Every 20C hotter reduces your OC headroom by 100MHz and increases power consumption noticeably (as per DerBaur's videos). So your 80C stress test is reducing your OC headroom by 150MHz at a minimum and probably requires another 25W+ in the CPU socket. In otherwords, its a positive feedback loop that puts the processor into a thermal run-away. That's why controlling the OC is so difficult. You really need to keep temperature under control if you want a good overclock. Lower temperatures also reduce power consumption on this node. If you can run the system at 65C or lower then you have to reduce your overclock until it drops to 65C. The lower you can go, the better your overclock will be. Ultimately, though, the issue is that above 4.1GHz power consumption goes exponential. The only way to work around this exponential power increase is to go sub-ambient (i.e. exotic cooling). So no matter how good the motherboard is, no matter how good your ambient air or water cooler is, going above a stable 4.2 GHz overclock is going to be difficult. And the crazy thing? The crazy thing is that 4.1 GHz is trivial and 4.2 GHz is reachable on virtually any motherboard, even a B450 (and 4.1 can be reached on a B350). But 4.3 or 4.4... even with a good motherboard, those are hard to get stable and you wind up burning a ton of money on the cooling solution, PSU, case design, etc to get there. All for just 5% more performance. In my book... it isn't worth it for just 5% more performance. -Matt
The best way to overclock looks to be phase change. That way 4.8Ghz all core would be possible. https://www.youtube.com/watch?v=JdYKVLrQL4s
#1
1.325V as long as you do NOT mess with the load line default. That 1.325V number assumes that high-current voltage droop will take it down to 1.245V which is the actual maximum safe voltage for high-current loads. If you mess with the load line, though, it won't droop as much and you will be well on your way to destroying your CPU. At low-current loads (i.e. single-core boost), the maximum safe voltage is much higher at 1.45V (which is why stock systems can boost single cores to 4.6GHz). Don't expect to be able to boost all cores the same as single-core, its just impossible... its a high-current situation that significantly lowers the maximum safe VCORE verses single-core. That's why it is impossible with normal (ambient) cooling solutions. The other problem you are facing is temperatures. Every 20C hotter reduces your OC headroom by 100MHz and increases power consumption noticeably (as per DerBaur's videos). So your 80C stress test is reducing your OC headroom by 150MHz at a minimum and probably requires another 25W+ in the CPU socket. In otherwords, its a positive feedback loop that puts the processor into a thermal run-away. That's why controlling the OC is so difficult. You really need to keep temperature under control if you want a good overclock. Lower temperatures also reduce power consumption on this node. If you can run the system at 65C or lower then you have to reduce your overclock until it drops to 65C. The lower you can go, the better your overclock will be. Ultimately, though, the issue is that above 4.1GHz power consumption goes exponential. The only way to work around this exponential power increase is to go sub-ambient (i.e. exotic cooling). So no matter how good the motherboard is, no matter how good your ambient air or water cooler is, going above a stable 4.2 GHz overclock is going to be difficult. And the crazy thing? The crazy thing is that 4.1 GHz is trivial and 4.2 GHz is reachable on virtually any motherboard, even a B450 (and 4.1 can be reached on a B350). But 4.3 or 4.4... even with a good motherboard, those are hard to get stable and you wind up burning a ton of money on the cooling solution, PSU, case design, etc to get there. All for just 5% more performance. In my book... it isn't worth it for just 5% more performance. -Matt
Last edited:
That would seem to be a good overclock going by that table. Is Load Line at default? What vcore do you get at high load all cores?
Last edited:
It was with the Rainforest one and I said not reaching advertised clock speeds