Try this: 1. Download and Install the newest MSI Afterburner (Unless you have it) 2. Now go to settings button and click it 3. A screen like this will pop up - 4. Check Disable ULPS 5. Click the dropdown of "Unoffical overclocking mode" 6. Select Without PowerPlay Support 7. Press OK and you will be asked to restart MSI AB. (Possibly the PC too) 8. Restart it
After that give it all a try again with power limit reset to +0%.
The 380 G1 and SOC look to have this problem constantly. The power saving settings are exceptionally aggressive and cause these massive clock variances.
Yerp. It's only a problem with the Gigabyte 380s though which is weird. Sorry, you lost the draw on GPUs. If you're in warranty you could trade out for another brand maybe?
Unless you can get it working with the afterburner tweaks. (The last few posts on this that I've found got moderate to limited success from those changes.)
With AMD cards, power limit and thermal target are hard-linked and you cannot adjust one without the other. Crank your power limit all the way to max and call it a day. As long as you dont add any voltage you aren't putting yourself at risk.
95% of the time AMD cards will fail to reach advertised clock speeds without manually adjusting your power limit first.
When it comes to using something to overclock, I recommend MSI Afterburner, however in order for it to work properly without constantly resetting clock speeds on you or conflicting with Crimson, you have to reinstall the Crimson Driver, and DO NOT ENABLE Overdrive in Crimson.
Tune CPU the mans way from UEFI, and tune GPU using tools meant for one task: tuning GPUs.
While this is all wonderful advice, you appear to have skipped reading the thread responses as the OP did fix the problem. Power Limit was not the fix.
That is a very broad generalization that in actuality only applies to a small number of cards. Hawaii and Fiji are the only two with intentional throttling either due to a current cap or a thermal cap. Tonga should never under any circumstance be throttling below 90C unless there is a defect with the card or supporting device such as the CPU.
I see this 70c cap on AMD cards all over the place, happens on my Gigabyte 280, my Sapphire 7950, the Sapphire 390x and XFX 390 I have worked with in the last 6 months.
The 280 and 7950 are both in my rig across both a 970 and 990 board from Gigabyte running on 6300/6350, 390x running on ASRock H170 with an i5-6500, and 390 runs on an Asus Z170i board with i7-6700K clocked @ 4.0.
I've also had Overdrive, and Adobe Flash fuck with clock speeds and reset them or lock them to internal profile settings numerous times. (Either don't use flash, or disable Hardware Acceleration. It doesn't play nice with Radeon hardware)
Contrarily, I have NOT seen this 70C cap on ANY cards other than the R9 Nano. Experience with testing eight different 7950s, three different 7970s, two 290X's, the aforementioned R9 Nano and many lower end GCN cards from Oland to Pitcairn.
Power limit caps are common place on any card with a boost state or aggressive ULPS profiles within the VBIOS, but thermal caps [of 70C] are not.
