Any news update on VRM temps? This board is now available on newegg again and I'm planning purchase it now cause I miss it 3 times. But seeing overclock3d video make me stop and confused. I'm planning use this board with R7 1700, wraith spire stock cooler and normal usage (no overclocked)
Motherboard GIGABYTE AB350N-Gaming WIFI
- Thread starter jeshikat
- Start date
I'll have VRM stress test and idle numbers for you tonight, then.
Regardless, if you're talking "normal usage," you should be totally fine. I'm only getting high temps running full throttle on synthetic benches.
Well I have had the worst night of BSODs I think I've ever had. Hoping that finally getting the AMD chipset drivers to install will fix some of this. Looks like I'm not getting to those benchmarks tonight, though.
I received my motherboard yesterday, and have been trying to overclock my 1600 today without much success. If I change the vCore in BIOS at all my multiplier in Windows changes to 15.5. I put vCore back to auto and I get the multiplier I set. I've tried reflashing the F2 BIOS, rolling back to F1, and going back to F2. No change at all. I can overclock with Ryzen Master, but I'm a BIOS overclocking kind of person. Have any of you experienced this? I'm new to Ryzen itself, but certainly not to overclocking. Maybe I'm missing something that guys could shed some light on.
My AsRock board was running janky as hell untill I downloaded and installed the AMD chipset drivers directly from AMDs site. It installs the Ryzen power plan for windows as well. You may know this already but thought to chime in.
http://support.amd.com/en-us/download/chipset?os=Windows 10 - 64
I actually am reinstalling Windows cuz it was so bad. But updated chipset drivers will definitely be one of the first things going on the new install.
Here are some of my numbers:
1) The max numbers here are after 10 or so minutes of playing Doom.
2) The max numbers here are after 5 minutes of OCCT.
This is in a Nano S with no overclock.
OCCT numbers are definitely bad, but seems like regular use is fine
1) The max numbers here are after 10 or so minutes of playing Doom.
2) The max numbers here are after 5 minutes of OCCT.
This is in a Nano S with no overclock.
OCCT numbers are definitely bad, but seems like regular use is fine
Last edited:
I was having this issue too with my 1600X, was getting locked at x22 multiplier. Try disabling all of the power saving options in BIOS--I'm at work right now so I don't remember exactly what settings I disabled but under Advanced Frequency Settings > Advanced CPU Core Settings, I disabled Core Performance Boost, Cool & Quiet, SVM Mode, and Global C-State Control. left SMT and downcore control on auto.
I forget what exactly solved my problem, let me know if that doesn't work.
By the way, good luck with manually setting voltage. I didn't have any luck... At 3.8GHz no matter what I input in BIOS I couldn't get my vcore to go below 1.3v, unless there was a lot of vdroop at load. Let me know if you have similar issues.
I should have mentioned that I did disable all power saving options. However, I did find other people running into this problem with other boards. X series getting stuck at a 22 multiplier and non X getting stuck at 15.5. Seems like it sporadically happens to people. Some seem to have luck with clearing CMOS, reflashing, and then reinstalling Windows (or some long drawn out process like that). Seems very odd that some people with the same CPU/motherboard combos don't have the problem while others do. I'm going to give that a shot this evening. If you wouldn't mind double checking your settings I would appreciate it.
So did you end up just using offset for your overclock or did you just reduce speed to address the increased vCore?
I'll update if I have any luck. When I got this and saw how sparse the options were I thought it was going to be a nice easy overclock compared to what I've been used with some platforms in the past...I was wrong.
Here we go! Managed to get the Creator update on this thing, and that's solved my issues so far. No BSoD in sight!
Found some interesting results:
My CPU was only boosting to 3.15 on all cores, though I'd previously always seen 3.2
Setting the frequency to 3.15GHz in bios resulted in Windows reading 3.09
Setting the frequency to 3.2GHz in bios resulted in Windows reading 3.19
Take that as you will, but that's why the 1.15 undervolt + manual clock isn't set at 3.15 like the auto clocked tests are.
Single-core boosting up to 3.6 GHz seems to need as much power as overclocking all the cores to 3.6 GHz. At 1.1875V, it would work, but as soon as it tried to single-core boost at 1.15V, the PC gave up the ghost.
Stock speeds and voltage together seem to give the lowest-possible load power consumption. Even at 3.2GHz manual, I couldn't get to post any lower than 1.15V, and the combination of stock speed and voltage brings that down to 1->1.1V.
Stock speeds and voltage together are also the lowest-possible idle power consumption, as long as everything is 100% idle. As soon as anything happens, the processor draws 1.4V! Simply crazy!
Underclocking the CPU frequency doesn't appear to remove the ability to dynamically change its clocks with Windows power plans. Overclocking does. Not that this seems to have any effect on idle power consumption.
The voltage does not dynamically change when set to Auto if the Frequency is not also set to Auto.
If you're curious about the temps, they were all leveling off when I snapped the results, all within ~.5C of max temp they eventually reached. I was more concerned with getting consistent timings than waiting for the temps to take their first little dip below max.
My recommendation, @sodhc718 depends what you're trying to do. Unless you're maxing 16 threads on that thing all the time, I'd recommend undervolting, keeping stock frequencies, and calling it a day. That prevents the CPU from drawing ridiculous voltage whenever it decides to boost one core to 3.6 GHz. If you don't mind losing single-core boost, I'd recommend manually setting the frequency to 3.2 GHz and getting the lowest VCore you can on that. 1.15 runs well within fine temps on CPU and VRM, even in my wicked-hot testing environment.
If you're looking, as I am, to just get the best performance while not risking temps, go for that 3.6 GHz OC and 1.1875V. It produces pretty much the same heat and power consumption as the stock frequency, while boosting ALL the cores instead of just one of them. I only recommend stock frequency and voltage if you are looking for good (not as good) performance and really really care about a few Watts/degrees.
Last edited:
Well, nothing I've tried has fixed the issue. I've cleared CMOS, reflashed BIOS several times, reloaded W10 (tried Home and Pro 1703, did not try previous builds though), left the CMOS battery unplugged all night, but I still have the same issue. If anyone else runs into this same issue problem and finds a workaround let me know.
For now I'll just try Ryzen Master since it seems to at least work. All other settings on the motherboard work as expected, and my Flare X 3200MHz kit booted right up without issue on the XMP profile. Maybe I'll get lucky and the kit will overclock pretty well on top of that. Noctua sent me the wrong mounting kit for my cooler so I'm using the stock cooler for now. The connector layout worked pretty well for me for the most part in the Fractal Core 500. I do wish the 8 pin was a bit higher up on the board.
For now I'll just try Ryzen Master since it seems to at least work. All other settings on the motherboard work as expected, and my Flare X 3200MHz kit booted right up without issue on the XMP profile. Maybe I'll get lucky and the kit will overclock pretty well on top of that. Noctua sent me the wrong mounting kit for my cooler so I'm using the stock cooler for now. The connector layout worked pretty well for me for the most part in the Fractal Core 500. I do wish the 8 pin was a bit higher up on the board.
Me too on 1600x, the volt on auto when it boost goes up to 1.5 volts, and if i try to put the vcore on manual instantly locks a x22 multiplier. have tried everything without luck. can you post your setting to get 3.8 and what vcore volts setting? pls.
Scan.co.uk just dispatched mine. So I assume the UK is finally got their first batch of this board.
At noon I asked a support guy if they had any info and said they didnt have any information at this time....sigh.
4.5 hours later. Dispatch notification.
At noon I asked a support guy if they had any info and said they didnt have any information at this time....sigh.
4.5 hours later. Dispatch notification.
I asked them too and I got told they would be in stock the 19th. Colour me happily suprised that my order suddenly got processed a little later.
with bios f1, and lower memory clock i was able to set a fixed volt at 36 core multiplier with coolnquiet and boost off, will wait and use the computer like that for a while before trying anything new
I'm planning purchase this board when it's available in Germany. I'm a complete noob and I'm concerned about the VRM temperatures (I saw the review on YT). So is it really that high? And compared to other ITX boards? I'm using the R5 1600 with stock cooler.
In a cramped case with a small CPU fan, yes the VRMs may get very warm. Otherwise, smart cooler and chassis solution should avoid it. The VRMs are rated to 110c btw.
You underclocked your RAM? Previously when I reported I tried F1, I did this with the RAM at it's default 2400 speed (no XMP profile applied for 3200MHz), all power saving features turned off, but no dice for me.
On another note I saw an update from an ASUS rep on a thread in their forums from people having this issue. They had stated before they knew what the problem was and that it could take some time for a fix to be pushed out so some people asked them to elaborate on the time frame. The ASUS rep said "No ETA for a fix other than it could take a long time." I believe this was the same ASUS thread where they said it was an issue on AMD's side, and they'd have to wait for a fix from them. I also saw another user having the issue say that he RMA'd his CPU, and got a new one back that did not have the issue.
My Gigabyte ticket is still sitting in new status as well. It's been almost a week. I find this unacceptable, but maybe it's just because I'm an IT service provider. At my company if we left a ticket in new status without even reaching out to communicate to the client they would be pretty upset. I think I'm going to order an ASRock board today, and also get a replacement for the 1600.
Similar threads
Power Supply
Will the HDPlex 400W DC-DC be enough?
