Install Steam
login
|
language
简体中文 (Simplified Chinese)
繁體中文 (Traditional Chinese)
日本語 (Japanese)
한국어 (Korean)
ไทย (Thai)
Български (Bulgarian)
Čeština (Czech)
Dansk (Danish)
Deutsch (German)
Español - España (Spanish - Spain)
Español - Latinoamérica (Spanish - Latin America)
Ελληνικά (Greek)
Français (French)
Italiano (Italian)
Bahasa Indonesia (Indonesian)
Magyar (Hungarian)
Nederlands (Dutch)
Norsk (Norwegian)
Polski (Polish)
Português (Portuguese - Portugal)
Português - Brasil (Portuguese - Brazil)
Română (Romanian)
Русский (Russian)
Suomi (Finnish)
Svenska (Swedish)
Türkçe (Turkish)
Tiếng Việt (Vietnamese)
Українська (Ukrainian)
Report a translation problem
It's related to the card's power management (TDP limits) usually. That is, if there's not anything else limiting it (Vsync or frame limit in-game or otherwise). The benchmark in BAK will run unlocked FPS, but in-game there is a default 90 FPS cap and that may be what is limiting your FPS and GPU usage/clocks if you haven't manually changed that cap. You can unlock that pretty easily if that's the issue.
If you're sure there is no FPS cap, then you may need to get Wattman and use it to increase the card's power limit. Additionally, if it still seems to be hitting the TDP cap, you can undervolt the card small amounts and check for stability. That will reduce the amount of power it's using and allow it to ramp the clock up higher.
They don't need to. In some cases it might be the issue Ancient mentions, but that's not always going to be the true (assume it really is an issue, and not just laymen guessing).
In general, a Processor is only utilized as much as necessary to meet demands. Think of it like a car's engine. You don't run them at max RPM's all the time. If you did they would quickly breakdown from excessive heat and wear and tear. The engine speed varies depending on what it's currently being asked to do. The same is true of CPU's and GPU's. they vary their voltage levels and clock speeds as necessary.
I think you misunderstand how the architecture works. The GPUs operate in states 0-7. 0 being the minimum frequency/voltage state and not configurable, while states 1-7 can be configured, with 7 being the highest frequency/voltage state.
So you have a curve like this.[www.amd.com]
Essentially, what is wrong with Vegas[forums.anandtech.com] is that AMD sets them at an unnecessarily high voltage for the upper states. It means they are always stable despite the silicon lottery, but are likely pulling much higher draw than is needed for the frequency of each state. This wouldn't matter too much but each GPU has a predetermined TDP limit based on specific temp ranges assigned in it's VBIOS depending on the manufacturer's cooling solution and this limit is usually not high enough for the crazy high voltage most models try to use for states 6 and 7, so they hardly ever maintain those states at all. That cap is usually what limits the GPU's clock speed even when it is not technically overheating.
When the card gets too hot the state is reduced. Likewise when the card hits it's TDP limit in terms of power draw the state is reduced. The two, temp and power draw, are also intrinsically linked in the algorithm that determines which state the card should operate at. So at lower temps, the card may temporarily allow that high power draw, but as it ramps the temp up to 60C+ the actual wattage limit decreases and so it will drop it's state to compensate.
The symptom of this which is common on Vegas is that the GPUs are configured to use too high of a voltage for states 6 and 7 and end up running in state 4 and 5 most of the time even when they aren't really that hot because they keep hitting this sliding TDP/wattage limit. Now, the reason AMD did this is obvious: some of the chips are "bad" silicon lottery-wise and may actually need that much voltage to run stably. But if yours is a moderate or even a good chip and can run it's 1500-1700MHz states just fine with significantly less voltage, you'll never know it until you try undervolting it and testing for stability yourself because that default voltage is going to pull so much draw that the card will virtually always try to limit itself under load.
The fix is to increase the powertarget to +50% as you have and then fine tune the voltage as low as you can get it for each of the problematic states while maintaining stability. This is just like finding the right (lowest stable) vcore for a CPU OC and requires making incremental voltage changes and then putting the GPU under load. When you go too low and it crashes a game/benchmark or bluescreens, then bump it back up just a bit and give it a thorough stress test and if it remains stable you've likely found your GPU's unique sweet spot of stability for that state/frequency.
In Radeon Settings:
Gaming > Global Settings > Global Wattman
Find "Power Limit" and increase the slider. My RX 470 has a power limit boost of 25% and has no trouble hitting its maximum boost clock.
Hope you have a good PSU.