This is my perf with FMax/PBO - no PE or anything else except memory at 3733MTs. I'm really impressed and have to experiment more, but this is how things should have worked at launch of Ryzen 3xxx.
Sadly, the memory/sleep bug has returned - I don't know how because I only use hybrid sleep which didn't used to be affected - but it is now.
View attachment 2467235
Nice! I should have tested my system on my 3950x after the flash. I only booted to confirm it worked and to transfer the BIOS Settings TXT File to my Phone. But now I wish I would have taken a few minutes to see if Multi Threaded PBO Actually Worked with the FMax turned on, I had not done any of that testing since we had that ability.
I was too excited to get my 5900x in. I still don't like dropping from 16 Cores with my 3950x to 12 Cores with my 5900x, but, this single core Improvement is HUGE! Right now I have a lot going on in my Windows Build. I normally run a highly optimized version of Windows 10, but because I have been working on a crap ton of projects lately I haven't had the time to clean it back up since installing a bunch of new Software, so its gone to being pretty bloated again, and just before my upgrade I did run a CB R20 & CB R15 Test to get my Single and Multi Core Scores. I am stupid and didn't get a screenshot before, but did document them.
For Single Core on my 3950x running on Auto, I got 510 in CB R20 (Which is one of the highest scores I have received personally for a SC Score, I know most people saw higher, but I never did. And then for MC in CB R20 I got 9939. (I typically run a Per CCX OC, and with that I usually average around 10,300 to 10,400, with 10,469 being my personal best, but I wanted to compare auto to auto.) Only changes in BIOS are Memory manually set to 3600 CL16, with Timings, Secondary, and Tertiary all set manually. Performance enhancer set to Default, no Bias, PBO Set to Auto and on the 5900x FMax turned on, and thats it.
Now with the 5900x, of course my Multi Core score is lower due to losing 4 cores, but I was quite pleased with the Single Core Score, especially considering I am yet to mess with anything yet. I did get screen shots of those:
![]()
![]()
![]()
What really Amazes me are Temps. On my 3950x and 3900x, even running CBR20 with a Room Temp of 25c Would Get me into the low 80s easy, even on Auto. In Fact with my Per CCX OC's My Temps were often lower because I would use much lower Voltage than Auto Would, but at best I was seeing about 79c after a 10 Minute back to back CB R20 Run. I just did the same Run, using the same exact Optimus Block, the only difference being that I now have a 5900x installed, and I switched from using Kryonaut paste to Kingpin Paste, and my Room Temp is actually 2c Higher than in my previous examples, during these tests it was sitting at between 26.6c and 27.5c. And During a 10 Minute back to back CB R20 MC Run I topped out at 68c! That is incredible to me, especially considering it appears to be using voltages close to what I was using Manually with my Per CCX Run. The Amps appear to be about 8amps less or so, which I am sure accounts for a lot of the temp difference, but a 10c Temp Difference is more than I would expect for only a few amps less power draw at about the same voltage.
Anyways, I am just barely getting starting with testing. This is just what I have noticed first.
Anyone else noticing temp reductions from their move from the 3000 series to the 5000 series?
