It downclocks and sleeps with changed multiplier ok, still would be livid if wouldn't see any 4.6 boost at all in your case. Running mine manual 4375 atm cause it is cooler (35-47 light use,54 gaming), as you learning about your cpu voltage etc, defo could tune it a bit, but thinking about: absolute max performance would be to check how far pbo would take it and leave on boosting over 4.4(should be over 4.6 with 3900x), would be hotter no doubt.
-
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.
Zen 2 Owners Thread!
- Thread starter SiDeards73
- Start date
I did not touch PBO, just set a offset of -0.1 and I'm hitting 4.6 in some cores, 4.45/4.5 in others. I've seen once hitting 4.65 when cpu settings were completely stock. 3900x taichi x370 here.
Regarding IF I can't say much as I'm using x370, so I'm pretty much locked at FCLK 1500. I can go up to FCLK 1500 and MCLK 3733/2 in settings, but then I got way higher latency. However I got better scores in cb15/20 and not so much better ram read/write numbers comparing with 1500/3000, that was about 1000mb/s higher.
I'm gonna wait for some new bioses, if it does not happen I'm gonna get a x570 board next month after checking reviews properly.
FIY
1500/1500
1500/1800
Caporegime
How does one get the free 3 months game pass?
What are you using to test / monitor max core clock? HW monitor seems to mess with core frequency's and watching in ryzen master is hit and miss. Will experiment with CPU settings more tonight.
On my x570 IF seems to be happy running at 1800 so ram speeds and latency are good:
Starting to do some tweaking now.
X570 Auros Master
3900X
Standard LLC
PBO Enabled (Advanced with +200Mhz override on Auto OC)
CPU Vcore Auto
XMP Profile for 3600CL16
For Cinebench R20 seeing an all core boost of 4125Mhz, load temp of 70 Deg C and around 1.37 to 1.4 Vcore. Score of 7360 which is inline with other peoples runs.
I however have a VERY poor single thread boost clock of only 3950Mhz despite having PBO and Auto OC enabled. How on earth is my single core boost LESS than the All Core boost?!
Edit: Messing around further I get my best scores/clocks by simply leaving everything at Default/Auto.....bonkers behaviour.
X570 Auros Master
3900X
Standard LLC
PBO Enabled (Advanced with +200Mhz override on Auto OC)
CPU Vcore Auto
XMP Profile for 3600CL16
For Cinebench R20 seeing an all core boost of 4125Mhz, load temp of 70 Deg C and around 1.37 to 1.4 Vcore. Score of 7360 which is inline with other peoples runs.
I however have a VERY poor single thread boost clock of only 3950Mhz despite having PBO and Auto OC enabled. How on earth is my single core boost LESS than the All Core boost?!
Edit: Messing around further I get my best scores/clocks by simply leaving everything at Default/Auto.....bonkers behaviour.
Last edited:
Associate
My 3700x arrived this morning,
Ordered on friday came today thanks ocuk!
Ordered on friday came today thanks ocuk!
More testing and still confused how PBO and XFR are working.
Test 1
AMD CBS > XFR Enhancements PBO set to Enabled
AMD Overclocking > PBO Set to Auto
Any combination of either of those settings together or individually results in the similar sort of outcome. ~4.1Ghz All Core boost and ~4.2ghz single core boost (Still miles away from expected single thread boost clocks though)
Test 2
AMD CBS > XFR Enhancements PBO set to Enabled
AMD Overclocking > PBO Set to PBO Advanced and +200Mhz Core Clock
Multi core is largely unchanged and in the right sort of ballpark with ~4.1Ghz All Core. But single core absolutely nose dives. Am seeing 3.9Ghz single core and significantly degraded Single Core benchmark results. This is the strangest outcome of testing so far.
Test 3
CPU Voltage Auto vs CPU Voltage Normal
Auto voltage results in slightly degraded clocks and benchmark scores for both Multi and Single threaded. Normal seems to have a very slight uplift.
Test 4
CPU Voltage Normal +/- 0mv offset vs CPU Voltage Normal - 0.05mv offset
Setting a - offset reduces core clocks and both Multi and Single threaded benchmarks scores. I envisage pushing an even great - offset will continue to degrade the observed results.
So far the testing is saying leave PBO/XFR Enhancements on Auto/Enabled with a Normal CPU Voltage. I'm still unsure why I am seeing such low Single thread boost clocks.
Test 1
AMD CBS > XFR Enhancements PBO set to Enabled
AMD Overclocking > PBO Set to Auto
Any combination of either of those settings together or individually results in the similar sort of outcome. ~4.1Ghz All Core boost and ~4.2ghz single core boost (Still miles away from expected single thread boost clocks though)
Test 2
AMD CBS > XFR Enhancements PBO set to Enabled
AMD Overclocking > PBO Set to PBO Advanced and +200Mhz Core Clock
Multi core is largely unchanged and in the right sort of ballpark with ~4.1Ghz All Core. But single core absolutely nose dives. Am seeing 3.9Ghz single core and significantly degraded Single Core benchmark results. This is the strangest outcome of testing so far.
Test 3
CPU Voltage Auto vs CPU Voltage Normal
Auto voltage results in slightly degraded clocks and benchmark scores for both Multi and Single threaded. Normal seems to have a very slight uplift.
Test 4
CPU Voltage Normal +/- 0mv offset vs CPU Voltage Normal - 0.05mv offset
Setting a - offset reduces core clocks and both Multi and Single threaded benchmarks scores. I envisage pushing an even great - offset will continue to degrade the observed results.
So far the testing is saying leave PBO/XFR Enhancements on Auto/Enabled with a Normal CPU Voltage. I'm still unsure why I am seeing such low Single thread boost clocks.
Associate
Ordered the Ryzen 5 3600 earlier today (for now) along with an AM4 bracket for my Alpenfohn Matterhorn cooler. I have an Asrock X470 Taichi board so I'll be skipping X570. Asrock may even release a BIOS update to enable PCIe 4.0 for the primary PCIe slot and one of the M.2 slots (which Asus have apparently done for some of their previous gen boards).
I'll need to get some new RAM at some point... I currently have 16GB of Hyperx Fury DDR4 2800 (4x4GB from my X99 build). I'm also looking to upgrade my GPU to either the RX 5700 or GTX 2060 Super. I'm currently using an MSI Radeon RX 580 Gaming X.
I'll need to get some new RAM at some point... I currently have 16GB of Hyperx Fury DDR4 2800 (4x4GB from my X99 build). I'm also looking to upgrade my GPU to either the RX 5700 or GTX 2060 Super. I'm currently using an MSI Radeon RX 580 Gaming X.
Soldato
Hmm, hate illogical results like this. I recently did the same. Switched from full auto to setting +200 Mhz core clock.
I would have expected zero change to be honest. Because setting +200Mhz is only saying this is the max you can boost. But PBO will boost up to what ever threshold holds it back. So shouldn't make a difference.
Soldato
I've been testing out the 3900X doing some low volts testing today, and on X470 with a stable(ish) BIOS I found that dropping the voltage made the CPU run much better and cooler. Need to spend more time with it later in the week, but try setting your vcore to between 1.05-1.15v you should get a higher all core boost.
You need to open thread in your CS section.
I've been testing out the 3900X doing some low volts testing today, and on X470 with a stable(ish) BIOS I found that dropping the voltage made the CPU run much better and cooler. Need to spend more time with it later in the week, but try setting your vcore to between 1.05-1.15v you should get a higher all core boost.
No matter what BIOS setting configurations I use I cannot get a core boost above 4125mhz. Be it synthetic multi/single threaded benchmarks or gaming. That seems far too low *shrug*
I'm not thermally limited, my chip runs at 44deg C in single threaded benchmarks with a VCORE of 1.2v and being WELL within the extended PBO limits with it enabled. Even with All Cores loaded I don't see in excess of 70 Deg C which seems well within the limits of thermal throttling?
At a loss tbh, lost the silicone lottery?
Soldato
It seems somewhat hit and miss presently. I do think that due to the huge amount of hardware configurations and and chipset variations that can be used it is causing way more issues that expected, annoyingly.
For me with offset -0.1V, maximum speed is reduced by about 25-50MHZ(in gaming all cores have 4250-4275). But maybe it is actually worth it (10C+ reduced temps), considering that single thread boost is only visible for short moments (CB single thread test had only quick spikes to 4550MHz with offset V).
I think I'll leave it like this at the moment and see how stable it is.
My PBO is probably off as I've not played with it, based on the reports it doesn't work.
I think I'll leave it like this at the moment and see how stable it is.
My PBO is probably off as I've not played with it, based on the reports it doesn't work.
Temporary housing. Old nzxt 340 elite I had spare.
Much like everyone else. Temps idling at 35-45.
Just had a quick few games of overwatch, CPU was usage nothing at about 7% and temps Hardly changed. Cinabench R20 went up to 79deg. (7174 score)
everything on default apart from XMP profile 1.
This is the asrock taichi with 3900x. 16gb of 3200 8pack ram.
Can understand why a lot of high end builders went for this board, not only it got "all the gear", it looks stunning aswell
.The mystery continues!
With a -0.15v offset on the VCORE I get higher boost frequencies but significantly less on benchmark scores. Go figure!
Can confirm, my 3600 was stable with a -0.2v offset but CB score dropped from 3600+ to 3000. Bumped it up to -0.07v which shaves 10deg off load (prime95) temps while restoring my CB score to just over 3600