F5F Stay Refreshed Power Users Overclocking Tips for optimizing your GPU overclock.

Tips for optimizing your GPU overclock.

Tips for optimizing your GPU overclock.

Pages (2): 1 2 Next
E
Ender_kat22
Member
171
10-03-2016, 10:45 AM
#1
I've increased my EVGA GXT 1070 SC to 2100mhz and set the memory clock offset to +500, without changing the voltage. The issue is that when I run benchmarks with Heaven or Valley, the app starts skipping and then crashes. It's not my system, and I've tried Overwatch too, which also causes skipping. Any suggestions on how to fix this would be greatly appreciated.
E
Ender_kat22
10-03-2016, 10:45 AM #1

I've increased my EVGA GXT 1070 SC to 2100mhz and set the memory clock offset to +500, without changing the voltage. The issue is that when I run benchmarks with Heaven or Valley, the app starts skipping and then crashes. It's not my system, and I've tried Overwatch too, which also causes skipping. Any suggestions on how to fix this would be greatly appreciated.

F
Frigiel21
Member
187
10-04-2016, 04:17 PM
#2
From numerous overclocked cards, my guidance is that lowering voltage can boost stability, but it may also cause instability in the core. The key is to adapt to working with the Voltage/Clock Curve. Adjust your maximum boost to 1093mV, test, then reduce by one notch, repeat testing.

For memory: around +500 appears to be the limit for most nVidia cards you own. You can fine-tune it to get more overclocking, but it requires a lot of effort and only pays off if you enjoy the process, as the real gains are minimal.

Temperature-wise: Pascal has three throttling points—around 45°C, 61°C, and about 82°C (with a 2°C buffer). What this means is...
F
Frigiel21
10-04-2016, 04:17 PM #2

From numerous overclocked cards, my guidance is that lowering voltage can boost stability, but it may also cause instability in the core. The key is to adapt to working with the Voltage/Clock Curve. Adjust your maximum boost to 1093mV, test, then reduce by one notch, repeat testing.

For memory: around +500 appears to be the limit for most nVidia cards you own. You can fine-tune it to get more overclocking, but it requires a lot of effort and only pays off if you enjoy the process, as the real gains are minimal.

Temperature-wise: Pascal has three throttling points—around 45°C, 61°C, and about 82°C (with a 2°C buffer). What this means is...

A
awesomehabib
Member
86
10-04-2016, 04:46 PM
#3
Watch the tutorial provided.
Perform the overclock in small increments, continuing until you achieve the maximum.
A
awesomehabib
10-04-2016, 04:46 PM #3

Watch the tutorial provided.
Perform the overclock in small increments, continuing until you achieve the maximum.

I
ImUnLucky1337
Junior Member
10
10-06-2016, 02:10 AM
#4
I've increased the overclock to 2100mhz and adjusted the memory clock offset to +500 without changing the voltage. The issue occurs during benchmarking with Heaven or Valley, where the app starts skipping and crashes. Not my system, the problem lies with the app. I also tried Overwatch, but it still skips. Any suggestions would be greatly appreciated.
I
ImUnLucky1337
10-06-2016, 02:10 AM #4

I've increased the overclock to 2100mhz and adjusted the memory clock offset to +500 without changing the voltage. The issue occurs during benchmarking with Heaven or Valley, where the app starts skipping and crashes. Not my system, the problem lies with the app. I also tried Overwatch, but it still skips. Any suggestions would be greatly appreciated.

S
218
10-10-2016, 01:28 AM
#5
From numerous cards that have been overclocked, my guidance is that lowering the voltage can boost stability, while raising it tends to make the system more unstable. The key is to adapt to managing the Voltage/Clock Curve. Aim for a maximum boost of 1093mV, test it, then reduce by one notch, repeat testing and adjusting.

For memory, around +500 seems sufficient for all the NVidia cards I’ve tested. There are methods to fine-tune performance, but they require significant effort and only provide noticeable gains if you enjoy the process; otherwise, the improvements are minimal.

Temperature-wise, Pascal has three throttling points: first around 45°C, second at about 61°C, and a third near 82°C (with a 2°C buffer). It means you shouldn’t expect stable clock speeds when lowering temperature from 70°C to 64°C—while the noise might increase, the actual average clock value will remain almost unchanged.

Again: based on personal experience, I’ve overclocked several Pascal cards, though these are not official results.
S
shacklebolt323
10-10-2016, 01:28 AM #5

From numerous cards that have been overclocked, my guidance is that lowering the voltage can boost stability, while raising it tends to make the system more unstable. The key is to adapt to managing the Voltage/Clock Curve. Aim for a maximum boost of 1093mV, test it, then reduce by one notch, repeat testing and adjusting.

For memory, around +500 seems sufficient for all the NVidia cards I’ve tested. There are methods to fine-tune performance, but they require significant effort and only provide noticeable gains if you enjoy the process; otherwise, the improvements are minimal.

Temperature-wise, Pascal has three throttling points: first around 45°C, second at about 61°C, and a third near 82°C (with a 2°C buffer). It means you shouldn’t expect stable clock speeds when lowering temperature from 70°C to 64°C—while the noise might increase, the actual average clock value will remain almost unchanged.

Again: based on personal experience, I’ve overclocked several Pascal cards, though these are not official results.

Z
Zsniper1274
Member
169
10-10-2016, 01:57 AM
#6
The cause of instability with higher voltages is the increased heat generation. Pascal prefers to stay cool... Boost 3.0 modifies clock and frequency across all operating temperatures. The specific adjustments vary depending on the GPU model.
Z
Zsniper1274
10-10-2016, 01:57 AM #6

The cause of instability with higher voltages is the increased heat generation. Pascal prefers to stay cool... Boost 3.0 modifies clock and frequency across all operating temperatures. The specific adjustments vary depending on the GPU model.

E
EletrikFlamez
Junior Member
4
10-19-2016, 09:39 AM
#7
The reason higher voltages cause instability is because they generate more heat. Pascal prefers keeping things cool... Boost 3.0 will modify clock and frequency across all operating temperatures, though the adjustments vary slightly depending on the GPU. Not always, but it's definitely a factor. I've seen overclocked watercooled cards where voltage changes had minimal effect on temperature—especially with aggressive water cooling—and different voltages produced varying stability results. In one case I struggled to reach stable 2.1GHz at 1043mV or higher, but lowering it to 1031mV achieved the optimal balance, while 1025mV proved unstable again.
E
EletrikFlamez
10-19-2016, 09:39 AM #7

The reason higher voltages cause instability is because they generate more heat. Pascal prefers keeping things cool... Boost 3.0 will modify clock and frequency across all operating temperatures, though the adjustments vary slightly depending on the GPU. Not always, but it's definitely a factor. I've seen overclocked watercooled cards where voltage changes had minimal effect on temperature—especially with aggressive water cooling—and different voltages produced varying stability results. In one case I struggled to reach stable 2.1GHz at 1043mV or higher, but lowering it to 1031mV achieved the optimal balance, while 1025mV proved unstable again.

T
TheMiiki14
Junior Member
11
10-21-2016, 07:51 AM
#8
Temperatures remain consistent throughout the entire temperature range. From -7c down to the lowest tested, it indeed matters constantly. The voltage being excessive can lead to issues even if temperatures are stable, so it's important not to overlook this aspect.
T
TheMiiki14
10-21-2016, 07:51 AM #8

Temperatures remain consistent throughout the entire temperature range. From -7c down to the lowest tested, it indeed matters constantly. The voltage being excessive can lead to issues even if temperatures are stable, so it's important not to overlook this aspect.

C
Chiller9592
Senior Member
670
10-21-2016, 09:56 PM
#9
Interesting facts:
Temperatures vary throughout the entire temperature range. From -7c down to the lowest tested point. Indeed, it plays a role consistently.
Hmm... what's going on here? Didn't anyone mention this before?
C
Chiller9592
10-21-2016, 09:56 PM #9

Interesting facts:
Temperatures vary throughout the entire temperature range. From -7c down to the lowest tested point. Indeed, it plays a role consistently.
Hmm... what's going on here? Didn't anyone mention this before?

G
GmoneyKeef
Junior Member
3
10-23-2016, 06:58 AM
#10
TehPenguin :
Vellinious :
Temperatures vary throughout the temperature range. From -7c all the way down to the lowest I measured. Indeed, it really does matter consistently.
Uh... what? Didn't anyone mention that, or am I overlooking something ^^
It seems so.
Overclocking pascal focuses mainly on core temperatures. Keeping the core cool is essential for everyone. Adding voltage can assist slightly, but only if a custom fan curve is used to handle the higher voltage.
I emphasize the "temperature" aspect as often as needed, since it remains the most critical factor during this overclocking process... and will likely become even more significant as the die continues to shrink.
The reason I referenced the temperature tests I've conducted is because you mentioned "throttling thresholds" and their current status. I stepped in there, since those thresholds don't disappear below specific temperatures. Boost 3.0 makes constant adjustments.
G
GmoneyKeef
10-23-2016, 06:58 AM #10

TehPenguin :
Vellinious :
Temperatures vary throughout the temperature range. From -7c all the way down to the lowest I measured. Indeed, it really does matter consistently.
Uh... what? Didn't anyone mention that, or am I overlooking something ^^
It seems so.
Overclocking pascal focuses mainly on core temperatures. Keeping the core cool is essential for everyone. Adding voltage can assist slightly, but only if a custom fan curve is used to handle the higher voltage.
I emphasize the "temperature" aspect as often as needed, since it remains the most critical factor during this overclocking process... and will likely become even more significant as the die continues to shrink.
The reason I referenced the temperature tests I've conducted is because you mentioned "throttling thresholds" and their current status. I stepped in there, since those thresholds don't disappear below specific temperatures. Boost 3.0 makes constant adjustments.

Pages (2): 1 2 Next