- Feb 27, 2017
- 4
- 0
- 520
Hey Everyone! Sorry for the long post!
I'm struggling to determine why my CPU frequency, as shown in AMD OverDrive and CPU-Z, fluctuates when I run stress tests. Here is my setup:
AMD FX-6300 CPU
Cooler Master Hyper Evo 212 CPU Cooler
ASUS M5A78L-M/USB3 Motherboard
8GB DDR3 1600MHz
EVGA NVIDIA GTX 960 4GB VRAM
500W Corsair CX500 PSU
Samsung 850 EVO 250 GB SSD
I'm using Prime95 Blend to stress test, CPU-Z to confirm Core Speed, Multiplier, and Voltage as well as Benchmark each profile, and finally AMD OverDrive to monitor/log temps. I began overclocking from stock Core Speed 3.5Ghz and Core Voltage 1.25V by going into BIOS and increasing the core multiplier until I failed to passed Prime95 for 30 minutes. Then I increased voltage a bit, then more multiplier, and so on. I got it stable all the way up to 4.6GHz at 1.45V. At that point, I was reaching a minimum of 15.0C for Thermal Margin (20.0C average) read from AMD OverDrive.
I read somewhere that a Thermal Margin of 0C is 70C Core temp when using AMD OverDrive, and that the max Core Temp I should aim for is 60C for the 6300 CPU, thus I'm trying to aim for 15C Thermal Margin, so that I'm sitting at a comfortable 55C, which is why I stopped when I got to 4.6GHz. I did keep track of temps, Prime95 errors, and CPU-Z benchmark values for each 0.1GHz step up on the way.
In any case, I'm preparing for a 24-hour Prime95 test, but I was wondering if I could get this figured out first. My Core Speed, Multiplier, and Core Voltage all go up and down when I start running stress tests instead of showing a consistent value.They change about every 3 seconds. However, they stay very constant when at idle.
Currently, I have AMD Turbo Core off and Cool N Quiet off with 22x Multiplier and Core Voltage at 1.4V, Reference Clock at 200 MHz, so 4.4GHz, with a minimum Thermal Margin of 18C (20C average) after doing a 2-hour Prime95 Blend stress test, 0 errors. I have done tests with Cool N Quiet off or on, and it doesn't seem to make a difference at idle or under stress (maybe due to my Windows High Performance Power Setting minimum processor speed 100%). I also tested with LoadLineCalibration on Auto, 0%, and 100%. I got black screens and reboots at 0% and 100%, so I put it back to Auto. I did try messing with the FSB/Bus Speed/Reference Clock (up to 205MHz), but I decided to keep it at stock 200MHz for now. It didn't affect the fluctuations.
The fluctuations act like this... At idle, there are no fluctuations, Core Speed is a stable 4.4GHz. Once I start a stress test, within seconds, even before temperatures ramp up in AMD OverDrive, my multiplier switches from 22x to 15x and then back and forth every 3 seconds or so. The Core Speed and Voltage follow suit.
On the one hand, since it fluctuates even on default settings (3.5 GHz, with and without turbo boost) during stress tests, I don't think this is a temperature throttling problem (also because my thermal margin always stays above 15C [30C Thermal Margin at stock speeds], so I should always be at 55C Core Temp or below, unless I'm wrong about the 70C Thermal Margin thing...) On the other hand, when I ramp it up to 4.5 or 4.6GHz, I get worse fluctuations that do seem to follow higher temperatures. When the thermal margin gets to 15 or 14C, the multiplier plummets to 7x for a good 5 seconds and the temperature quickly goes down. Then it goes back to 15x and 22x for a few minutes, then it plummets again, and so on.
So I basically have three questions:
1) Why is my core speed, as shown in CPU-Z and AMD OverDrive, fluctuating under stress tests?
2) Why does my CPU seem to throttle at 4.5-4.6 GHz when it seems I still have plenty of thermal margin left (15C) to be considered safe? Could it be a motherboard throttle because it's a cheap mobo without sufficient heat sinks? I haven't monitored any other temperatures except the Thermal Margin given from AMD OverDrive. Instead of temperature throttle, could it be that my motherboard can't supply enough current to the CPU because of the 95W Max TDP?
3) What can I try to hopefully alleviate the fluctuations and apparent throttling?
Thanks in advance for all the help! I'm glad I finally joined the site! Let me know what other information you need.
UPDATE 2/27/17: I did a bit more research on my motherboard, and it seems like the board isn't the best for overclocking, but the TDP rating is 140W which does leave some room for overclocking the 95W FX-6300. The consensus seems like I should be able to get to about 4.4GHz before I see the motherboard temperature-throttling. SO that answers question 2 above. When I go to 4.5 or 4.6GHz my motherboard is overheating somewhere and throttling down, even though my CPU temps seem okay. I'm fine with that. I know I'm working with cheap equipment. I'll probably keep it at 4.3 or 4.4GHz, but I'd still like to fix the fluctuations that occur at any load. I will run a test different than Prime95 later tonight and see if maybe it only fluctuates under Prime95 Blend for some reason. Is there software I can use to monitor Mobo temperatures during stress testing? What specifically should I be looking at, and what sorts of temps are okay?
I'm struggling to determine why my CPU frequency, as shown in AMD OverDrive and CPU-Z, fluctuates when I run stress tests. Here is my setup:
AMD FX-6300 CPU
Cooler Master Hyper Evo 212 CPU Cooler
ASUS M5A78L-M/USB3 Motherboard
8GB DDR3 1600MHz
EVGA NVIDIA GTX 960 4GB VRAM
500W Corsair CX500 PSU
Samsung 850 EVO 250 GB SSD
I'm using Prime95 Blend to stress test, CPU-Z to confirm Core Speed, Multiplier, and Voltage as well as Benchmark each profile, and finally AMD OverDrive to monitor/log temps. I began overclocking from stock Core Speed 3.5Ghz and Core Voltage 1.25V by going into BIOS and increasing the core multiplier until I failed to passed Prime95 for 30 minutes. Then I increased voltage a bit, then more multiplier, and so on. I got it stable all the way up to 4.6GHz at 1.45V. At that point, I was reaching a minimum of 15.0C for Thermal Margin (20.0C average) read from AMD OverDrive.
I read somewhere that a Thermal Margin of 0C is 70C Core temp when using AMD OverDrive, and that the max Core Temp I should aim for is 60C for the 6300 CPU, thus I'm trying to aim for 15C Thermal Margin, so that I'm sitting at a comfortable 55C, which is why I stopped when I got to 4.6GHz. I did keep track of temps, Prime95 errors, and CPU-Z benchmark values for each 0.1GHz step up on the way.
In any case, I'm preparing for a 24-hour Prime95 test, but I was wondering if I could get this figured out first. My Core Speed, Multiplier, and Core Voltage all go up and down when I start running stress tests instead of showing a consistent value.They change about every 3 seconds. However, they stay very constant when at idle.
Currently, I have AMD Turbo Core off and Cool N Quiet off with 22x Multiplier and Core Voltage at 1.4V, Reference Clock at 200 MHz, so 4.4GHz, with a minimum Thermal Margin of 18C (20C average) after doing a 2-hour Prime95 Blend stress test, 0 errors. I have done tests with Cool N Quiet off or on, and it doesn't seem to make a difference at idle or under stress (maybe due to my Windows High Performance Power Setting minimum processor speed 100%). I also tested with LoadLineCalibration on Auto, 0%, and 100%. I got black screens and reboots at 0% and 100%, so I put it back to Auto. I did try messing with the FSB/Bus Speed/Reference Clock (up to 205MHz), but I decided to keep it at stock 200MHz for now. It didn't affect the fluctuations.
The fluctuations act like this... At idle, there are no fluctuations, Core Speed is a stable 4.4GHz. Once I start a stress test, within seconds, even before temperatures ramp up in AMD OverDrive, my multiplier switches from 22x to 15x and then back and forth every 3 seconds or so. The Core Speed and Voltage follow suit.
On the one hand, since it fluctuates even on default settings (3.5 GHz, with and without turbo boost) during stress tests, I don't think this is a temperature throttling problem (also because my thermal margin always stays above 15C [30C Thermal Margin at stock speeds], so I should always be at 55C Core Temp or below, unless I'm wrong about the 70C Thermal Margin thing...) On the other hand, when I ramp it up to 4.5 or 4.6GHz, I get worse fluctuations that do seem to follow higher temperatures. When the thermal margin gets to 15 or 14C, the multiplier plummets to 7x for a good 5 seconds and the temperature quickly goes down. Then it goes back to 15x and 22x for a few minutes, then it plummets again, and so on.
So I basically have three questions:
1) Why is my core speed, as shown in CPU-Z and AMD OverDrive, fluctuating under stress tests?
2) Why does my CPU seem to throttle at 4.5-4.6 GHz when it seems I still have plenty of thermal margin left (15C) to be considered safe? Could it be a motherboard throttle because it's a cheap mobo without sufficient heat sinks? I haven't monitored any other temperatures except the Thermal Margin given from AMD OverDrive. Instead of temperature throttle, could it be that my motherboard can't supply enough current to the CPU because of the 95W Max TDP?
3) What can I try to hopefully alleviate the fluctuations and apparent throttling?
Thanks in advance for all the help! I'm glad I finally joined the site! Let me know what other information you need.
UPDATE 2/27/17: I did a bit more research on my motherboard, and it seems like the board isn't the best for overclocking, but the TDP rating is 140W which does leave some room for overclocking the 95W FX-6300. The consensus seems like I should be able to get to about 4.4GHz before I see the motherboard temperature-throttling. SO that answers question 2 above. When I go to 4.5 or 4.6GHz my motherboard is overheating somewhere and throttling down, even though my CPU temps seem okay. I'm fine with that. I know I'm working with cheap equipment. I'll probably keep it at 4.3 or 4.4GHz, but I'd still like to fix the fluctuations that occur at any load. I will run a test different than Prime95 later tonight and see if maybe it only fluctuates under Prime95 Blend for some reason. Is there software I can use to monitor Mobo temperatures during stress testing? What specifically should I be looking at, and what sorts of temps are okay?
Facebook
Twitter
Instagram