Rumors flying about an on-silicon B2 fix from AMD. Generated from a Canard tweet, whatever that is.
RYZEN 3 Spotted!
Are there any 10gbe boards yet or just 5gbe
So, I was just talking with someone on the pgp stream and it seems like my chip might be pretty ok. Up to this point I just cranked multi to 38 and left everything else on auto because ... lazy and shit just worked. Memory was running at 2666MHz, 16, 16, 16, 16, 36 at 1.35V.
I bumped the SOC to 1.1V and the memory is now running fine at 2933 with the same timings. After that I tried lowering vcore. I haven't done lots of testing but so far I was able to finish a blender rendering (the classroom demo file, takes about 12:20 on my rig) at 1.25V at 3.8GHz. 1.2V crashed instantly. So somewhere in that range....
So, what are you at? Tell me about your OC experience.
1 Like
You did basically what I did,,, but then I used bclk to get ram past 3200 and of course break the 3.9 barrier for cpu.
"Both AMD’s Zen 2 and Zen 3 x86 processors will be made in 7nm. "
Bout fell outta my chair when I read that sentance, yea I think I can live with my FX for a lil while longer 
3 days till Ryzen 3
Aug 10th looks .....nevermind @DerKrieger just beat me to the ripper launch date
Put a button on one of the lower corners, some kind of antenna on the top and you have a CRT-TV mockup.
Reaction from Santa Clara, California
insert expletive here
The blue man group collectively just browned their trousers 
1 Like
Looks really nice.
So, I got a little bit more testing in, still just using the classroom blender demo file. I tried 3.9GHz and at 1.25V it crashed instantly. I gave it a bunch more power to play with and after dialing it down step by step it turns out 1.325 crashed half way in while the next step 1.33125 made it through. I feel that is still fairly good and just to give it a bit of headroom I settled now for 1.35V @3.9GHz. 4.0GHz is not gonna happen on this chip. Not even on 1.4V and I am not even gonna try any further.
I would like to know if that is somewhat similar to what you all have experienced so far because to me it looks like there isn't much "lottery" in ZEN silicon. Anyone got a 4.1GHz unicorn? Or a 3.8GHz potato?
Just glancing through the overclocking stuff on other forums one guy got to 4 with 1.408v
Mostly I was investigating BCLK oc'ing on ryzen and it seems at 6% the pcie drops from level 3 to 2. Which is a shame since the turbo has a really nice swing of .7ghz
Something isn't stable yet. Have to do more testing. It might be a balancing act between the memory which is at 2933 on four sticks and the CPU at 3.9GHz. I probably have to dial back at least one of the two. I mean, I'm not on a high-end board and I am on air cooling, so realistically I can't complain.
24 hours from now Ryzen 3 will start shipping and all they hype is for threadripper
Don't know if that is good or bad
4 ghz 1.36v-1.37v (depending on vdroop) with 1800X on air for me.
Ill probably try to get the Crosshair VI extreme for it as I'm going to be building a custom water build soonish so will see what it can do.
I've been testing more precise settings for my R7-1700 lately after finally getting the RAM stable at its rated speed. Focused on finding a decent clock while keeping the power demand/noise low, I ended up settling for 3.8 GHz. Going up to the ca 1.4 V required for stability at 3.9 GHz didn't feel worth it, for a mere 2.6 % boost.
Load testing with Prime 95(In Place FFT for max power draw) I ended up at either a plain 1.30 V core, or a combined setting of 1.25 V togehter with a core LLC at 3. Both settings give a peak recorded core voltage of 1.384 V over the course of a 30 min test. (It was more common to see the voltage linger around 1.30 V.)
Out of curiosity I got 3.7 GHz stable with a plain 1.225 V core. A level of performance/power perhaps fitting a smaller system that doesn't have a cooler like this Noctua NH-D15S that I'm using.
I started out my testing using voltage offset values instead of set values, but that proved to be more temperamental stability-wise and I don't feel like taking it up again. Aside from wishing for a motherboard with P-state support I'm pretty much finished testing this setup.
That sounds familiar to what I am experiencing. What is your memory running at?
Here is a funny thing: My memory controller likes 1V SOC. It does not like less, it does not really care about more. My memory seems to run stable at 2933MHz at 1.0V SOC only. It actually crashes at higher SOC voltage.
For the chip, depending on what the load is it might be perfectly fine at 3.9GHz or not. And it looks like voltage needs to spike massively to even get there. So 3.8 is gonna be my speed as well. That is what I set it to in the first place and I have nothing to complain about, it runs great. I mean, we already are beating a stock 1800X in multicore stuff, so…
The best solution to get my four sticks of 2666 RAM stable proved to be nudging up their voltage from 1.20 to 1.30 V. In the newer firmwares for my Asus Prime X370 the default SoC voltage has gone up from 0.90 to 0.95 and that’s doing fine here now.
Yeah, I just dialed in 1.35V on the memory and that does help a lot. I am also on four sticks.
ryen 3 1300x now on newegg $129
also ryzen 3 1200 $109
should be 4 cores 4 threads. newegg specs show 4 cores 8 threads