The Unigine Valley application experienced a crash after a short sequence of screens while performing GPU overclocking.
The Unigine Valley application experienced a crash after a short sequence of screens while performing GPU overclocking.
Hello fellow forum users,
I've been attempting to overclock my watercooled MSI GeForce GTX 1070 with an 8GB RAM setup for the first time. When I set the voltage to maximum and try to push the core clock to +100, the benchmark runs smoothly. However, when I gradually increase the memory boost to around 200, the benchmark crashes after a few screens. The temperatures, voltage, and FPS are all normal, but it just stops responding. It’s not overheating, yet it still gives me random crashes.
My system specifications are:
CPU: Intel i5 6600K, 1.25V, clocked at 4.5 GHz
MOBO: MSI Z170A Gaming M5
RAM: Kingston HyperX DDR4 2666MHz
GPU: MSI GeForce GTX 1070 Gaming x 8GB
PSU: Corsair HX750i
I have a custom waterloop over the CPU and GPU. The pump is strong, and I also use a NZXT Hue+ with four LED strips and a fan controller running six fans, plus one fan connected to the CPU PWM port on my motherboard. This made me wonder if my 750PS PSU isn’t sufficient for this setup, or if my GPU consumes too much power during overclocking. It seems possible that either not overclocking is the only solution, or I need a new PSU. If you have any further details, feel free to ask and reply.
Thanks.
I don't believe energy is the main issue. That power supply seems quite capable for what you have. Boosting the VGA's memory clock usually doesn't offer a good balance between cost and benefit. The instability might appear quickly, while performance gains are minimal. As long as you can maintain a stable GPU clock and keep temperatures low, focus on that. Also, ensure the dissipator makes full contact with the memory chips.
If you still wish to increase the memory clock, consider using a lower GPU and see if it improves things. Identify the highest stable memory speed and then try raising the GPU. But remember, a higher GPU can yield better results, though very high memory and GPU clocks may not work well together.
I don't believe energy is the main issue. That power supply seems quite capable for what you have. Boosting the VGA's memory clock usually doesn't offer a good balance between cost and benefit. The instability might appear quickly, while performance gains are minimal. As long as you can maintain a stable GPU clock and keep temperatures low, focus on that. Also, ensure the dissipator makes full contact with the memory chips.
If you still wish to increase the memory clock, consider using a lower GPU and see if it improves things. Identify the highest stable memory speed and then try raising the GPU. But remember, a higher GPU can yield better results, though very high memory and GPU clocks may not work well together.
Thanks for the update. I'll experiment with the numbers again once I'm back.
I think the most obvious answer is that you're overclocking your memory too far, past what it is capable of. So it's unstable and you get crashes. Not much you can do about it.
As an aside, I'm pretty sure increasing voltage in an overclocking utility just increases core voltage, so it doesn't aid in memory overclocking.
TJ Hooker :
I think the most obvious answer is that you're overclocking your memory too far, past what it is capable of. So it's unstable and you get crashes. Not much you can do about it.
As an aside, I'm pretty sure increasing voltage in an overclocking utility just increases core voltage, so it doesn't aid in memory overclocking.
I have these settings right now
http://imgur.com/a/MF5JC
set memory back to 0, benchmark runs just fine, it even completes it. but after a minute or 10 MSI Activex service crashes while temps and everything looks just fine. is 170 core clok to high already?
I'm currently at work, though I recall 170 being quite high and difficult to maintain even with the highest voltage (without any mods, of course). I'll look into my 1070 at home later and update you.
I'm currently at work, though I recall 170 being quite high and difficult to stabilize even with maximum voltage (without any mods, of course). I'll look into my 1070 later at home and update you.
Back home, my G1 Gaming is at +130 MHz, but I suspect this exceeds the built-in overclock that comes with the VGA. Above this level, I notice artifacts start appearing. The voltage is at full capacity, yet temperatures stay below 50 ºC.
ClowReed :
ClowReed :
I'm at work right now, but as far as I remember 170 is pretty high and hard to stabilize even at maximum voltage (without any sort of mod, of course). I'll check my 1070 back at home later and let you know.
I checked back home and my G1 Gaming is at +130 mhz, but I bet this is on top of the OC that already comes with the vga. And as far as I remember, anything above this starts to give me artifacts. The voltage is also at +100%, but the temps rarely rises above 50 ºC.
Thanks for looking it up. Im at +150 mhz now and +0 memory with voltage fully open, gaming hasnt crashed yet. I'll benchmark soon and post how it went.
Assuming the power limit is at +100%, your voltage offset should be in mV, not percentages. Could you clarify the exact value you're referring to?