Wow,
I am hoping that the Ryzen mobile release of Zen 2 does not go this bad. I am holding off on buying a laptop until Zen 2 due to the performance delta between Zen+ and Zen 2, but man.
Zen+ was actually a much better release on mobile than Zen 1.
I have the newest Bios and I am getting 4.2 all core in gaming workloads.
Under linux my 3900X on B450 tomahawk with PBO off, Cool’n’Quiet enabled and 64GB of RAM at 3466 CL 16 is boosting as follows. Currently on the 7C02v1B4 (beta) BIOS.
I have listed a range because it’s not 100% consistent between runs, just that it’ll be in that range somewhere. Perhaps related to thermals… running stock wraith prism in “h” mode.
- single core: 4350 -> 4525. temp is 50-58C **
- all core: 4100 -> 4200, temp is 70-80C
measurements using /proc/cpuinfo every 1/3rd second and recording max speed.
** i think here i might need to adjust my CPU fan profile because @4350 it’s sitting on 50C but won’t boost any further and fan not maxed out. When it does hit @4500 then the temps are > 50 and it hits the next level on the fan profile (maxes out). The result is still higher temps (56-58) but it can maintain the higher boost.
My 3800X is a champ! all cores have reported advertised boost. 5 of them even hit the hardware max 4.550 at some point - under lightly threaded loads. This is why I was talking with you about my 3900X in the other thread - I was struggling in my mind - thought for sure something was just not right with it because the 3800X had no issues. Even so, hitting advertised boosts or not, that 3900X is still putting the numbers up - so does it really matter 4.5 vs 4.6??? I still got what I paid for.
2 Likes
oh yeah should point out i’m super happy with the performance and don’t really care it’s not hitting the 1T advertised speed. OTOH it’s exceeding the all-core boost and that’s more important!
This is a workstation for my work which is mainly development and running lots of VMs for testing. I have been doing testing with the old and new system and the new system is more than twice as fast on both light or heavy workloads like doing a full rebuild which maxes out the system. My old system was no slouch either with 2x E5-2650v2 CPUs (16c/32t). So 4 less cores and this think is just killing it. All with lower power usage too.
I also gave Cinebench Singlecore a run yesterday. Was a bit weird. It didn’t seem to give me 100% Load on any core. Max was 80% and load was jumping between cores.
I also tried pinning Cinebench to Core#0, which led to 100% load on that core.
In either case: 3600 was not able to hit 4Ghz. It got close with 3990, but that’s still 200Mhz off.
Board is a B450m Bazooka Plus, Bios is latest non Beta 7B90v18. I’ve not tried the beta 7B90v192 yet, as my system is stable and i couldn’t care less for 200 Mhz in the real world. I also didn’t try older Bios Versions.
32G of 3200 CL16 RAM (Balistix Sport).
And just because it was mentioned as a potential “Problem”: I’m running stock cooler and Windows 10 with latest Updates.
The dotted line is whenever the kernel is doing things, so 80% load and that line at 20 means 100% load. Welcome to windows wierdness.
But was it performing like 4.2GHz? As in, was the compute work done at 4GHz the same as 4.2GHz?
The 3600 my friend has is 1530 to 1560 points, in line with what results are online. It hits 4.2GHz (4180MHz something something, 3000MHz RAM, stock cooler (gets 
rather quickly) ).
What if...?
What if AMD managed to address the high idle voltage and “relatively” high powerdraw my doing some clock magic without reducing compute done, what is the fuss about?
No idea since i never ran it on older Bios Versions or ever saw 4.2Ghz. I can rerun the SC Test later and post the scores though.
1 Like
A quick update on progress with my 3700x (ASUS Prime Pro X570 board which I quite like).
I’ve overcome my newb auto-clock mistakes and reset everything back to defaults (a couple of times now!) with no issues.
Playing around with memory timings I got my best single core cinebench score (499) with PBO off, Kingston memory overclocked from 3333 to 3600 (slightly tighter timings thanks to the DDR calculator tool) and a bump in memory and SoC voltage (1.39 and 1.15 respectively).
Booted up today and system was unstable with random windows shenanigans so down locked the memory and SoC to 3400 and 1.125. Things are better now.
Max single core was 4.367 (on a few different cores) temps around 50 to 60 degrees and idling around 40 which isn’t too bad for air I think.
Nothing I seem to do can get the advertised 4.4 right now but I guess 4.36 is close and I really can’t complain about performance - just a weird OCD thing that it’d be nice to see 4.4 
If anyone has any suggestions to test let me know otherwise I might go back to defaults and wait for the bios update!
1 Like
did you test everything on default to see if it reaches 4.4 on its own?
I am commenting on someone’s comment on youtube. Someone suggested to do what wendell suggested in his recent video. I discussed this with wendell when I was having over heating issues. It was reaching the max temp of 95c and shutting down my computer. I tried three different times to reseat my heatsink and each time I reapplyed my thermal paste. So then I contacted wendell who gave me suggestions. Now I see my temps as low as 72c on full load with handbrake. Anyways. This guy I am commenting on has a point, if you use too much paste it will leak out the sides. I did that with my first build ever. The guy is saying if you follow wendell’s method of applying a small dot on each chiplet you will ruin your cpu. He also states he is running his 3900x @ 5.0ghz with extreme cooling. Anyways, I am gonna quit typing right now. This paragraph is enough.
I re organized the paragraph and fixed some information.
4 Likes
How that?
1 Like
Stupid people say stupid things
And it’ll drive you right up the wall
Sometimes I feel the same way too
But it don’t mean nothing at all
4 Likes
He said it will seep onto the motherboard and into the socket. Which isn’t untrue, if you use an obscene amount of thermal paste.
And sometimes us stupid people say stupid things just for the entertainment of frustratin smart people 
Be on guard.
Wasn’t directed at @Argone. He described someone … clearly stupid.
Without the proper context, please take my post simply as a song recommendation. ¯\_(ツ)_/¯
2 Likes
I know it wasn’t directed at me. I just am trying a new approach. Not flaming here about someone on youtube.
Posted about my new leaf. Not being offended, not be petty, being zen and not starting flame wars.
1 Like
Since you are working on being Zen now, has this brightened your hor-ryzen? I will see myself out. Thanks
4 Likes
So Wendell three dot things is harmful Freaking silly.
I was wondering how it would translate to my Bequiet Shadowrock TF ?
1 Like
As promise, some tests on my system:
3600 on B450m Board, Stock cooler, stock thermal paste, no Overclock, Bios default except selecting XMP Profile 1.
Closed all Background Apps.
Running SC Test without changes: 4.074 Ghz max.
Pinning Cinebench to Core#0: 4.076 Ghz max
Multicore test: 3.899 Ghz All Core boost (The 4.074 was at idle right after the test)
Now that it’s colder outside, Temps seem OK with 85C for an all core load. So, scores seem a tad on the low end, but fine. Make from that what ever you want. Repaste and a better cooler might improve those, but since Temps and Powerdraw aren’t close to maxed, i’m not sure why it shouldn’t perform at it’s best atm.