F5F Stay Refreshed Power Users Overclocking GTX 780 TI Overclocking Tips - Problems with Unigine Inconsistencies.

GTX 780 TI Overclocking Tips - Problems with Unigine Inconsistencies.

GTX 780 TI Overclocking Tips - Problems with Unigine Inconsistencies.

S
snipsnap27
Member
123
11-07-2016, 03:47 PM
#1
Just installed a GTX 780 TI in my system. Upgraded from a GTX 580. I am following the Safe GPU Overclocking Guide here: I followed the same guide to OC my GTX 580. And every time I would turn the clock up in 5 Mhz increments I would see a slight increase in my Average FPS and in my Score. Using the same method with the GTX 780 TI, some scores would actually be lower than my previous 2 scores even though I've upped the core clock. Even doing 2 runs at stock I seem to get scores that are different. In a possibly related issue, I am running a 3 monitor setup. 1 DVI to 1 monitor, then HDMi to my gaming projector (where I run the benchmarks). Then 1 DVI to another monitor running off the Intel GPU. I do this to try to spread the work load and not affect my GPU too much. What I am noticing is that on the monitors connected to the GPU, if I am running my GPU at load on one monitor (my projector) I will use the 2nd monitor with something like notepad and when I type there is a delay in it showing on screen as if my GPU is not handling it. Can someone figure out what the problem is here?
S
snipsnap27
11-07-2016, 03:47 PM #1

Just installed a GTX 780 TI in my system. Upgraded from a GTX 580. I am following the Safe GPU Overclocking Guide here: I followed the same guide to OC my GTX 580. And every time I would turn the clock up in 5 Mhz increments I would see a slight increase in my Average FPS and in my Score. Using the same method with the GTX 780 TI, some scores would actually be lower than my previous 2 scores even though I've upped the core clock. Even doing 2 runs at stock I seem to get scores that are different. In a possibly related issue, I am running a 3 monitor setup. 1 DVI to 1 monitor, then HDMi to my gaming projector (where I run the benchmarks). Then 1 DVI to another monitor running off the Intel GPU. I do this to try to spread the work load and not affect my GPU too much. What I am noticing is that on the monitors connected to the GPU, if I am running my GPU at load on one monitor (my projector) I will use the 2nd monitor with something like notepad and when I type there is a delay in it showing on screen as if my GPU is not handling it. Can someone figure out what the problem is here?

T
Tommy6774
Member
161
11-07-2016, 03:54 PM
#2
Your overclock might benefit from increased voltage. It's often near the edge of performance when running at high speeds with insufficient power. I commonly notice this issue where higher clock speeds cause lower FPS when VRAM speeds approach instability for the required voltage to drive the main clocks, because it's difficult to allocate power separately to memory and core. This could be due to your power delivery (check the dual molex to 8 pin plug) or your GPU itself.

GPU Boost might also contribute if your 2600 score was recorded during the hottest part of the afternoon and the 2900 during the coldest evening hours. Did you push the thermal and/or power limits to their maximum?
T
Tommy6774
11-07-2016, 03:54 PM #2

Your overclock might benefit from increased voltage. It's often near the edge of performance when running at high speeds with insufficient power. I commonly notice this issue where higher clock speeds cause lower FPS when VRAM speeds approach instability for the required voltage to drive the main clocks, because it's difficult to allocate power separately to memory and core. This could be due to your power delivery (check the dual molex to 8 pin plug) or your GPU itself.

GPU Boost might also contribute if your 2600 score was recorded during the hottest part of the afternoon and the 2900 during the coldest evening hours. Did you push the thermal and/or power limits to their maximum?

A
ApexNinja02
Member
166
11-11-2016, 07:38 AM
#3
While you're doing this, I'd try just running to one screen with your GPU and monitoring temps to make sure it's not throttling. That's my first thought to make sure those aren't a problem.
A
ApexNinja02
11-11-2016, 07:38 AM #3

While you're doing this, I'd try just running to one screen with your GPU and monitoring temps to make sure it's not throttling. That's my first thought to make sure those aren't a problem.

J
Justin9401
Member
211
11-28-2016, 11:45 AM
#4
Just started playing DOOM after a long session of Overwatch. The FPS in DOOM felt much worse than what I got from a GTX 780i I tested on my setup last night. Could an underpowered GPU be the reason for the lower performance? Is my power supply unit to blame? Are there methods to check this?
J
Justin9401
11-28-2016, 11:45 AM #4

Just started playing DOOM after a long session of Overwatch. The FPS in DOOM felt much worse than what I got from a GTX 780i I tested on my setup last night. Could an underpowered GPU be the reason for the lower performance? Is my power supply unit to blame? Are there methods to check this?

N
NerfMe
Member
93
12-02-2016, 11:32 PM
#5
Full system specs?
N
NerfMe
12-02-2016, 11:32 PM #5

Full system specs?

F
Felipe_Lima
Junior Member
26
12-03-2016, 01:01 PM
#6
The build includes an Intel Core i7-4790 Haswell Quad-Core at 3.6 GHz, a Gigabyte GA-H81M-S2H motherboard, a GeForce GTX 780 TI Direct CU 2 with 2 OC, an Antec EA-650 W, a 8GB PC3-12800 DDR3-1600 2Rx8 RAM, a Thermaltake Versa H22 Mid-Tower case, a 240 GB SSD, and a WD 1 TB Blue HDD. There are a few updates: I recently added the 1 TB HDD and the Direct CU 2. Previously, I was using an EVGA Reference 780 TI without the 1 TB HDD, which I think consumes less power than the Direct CU 2. I’m also connecting the Direct CU 2 to an 8-pin PCIE plus 2 Molex to an 8-pin PCIE slot.

I ran the Unigine benchmark at stock settings today and achieved a score of 2600, which improved to 2900 the previous night. This inconsistency in scores is puzzling, suggesting the GPU might be underpowered. I’m wondering if GPU Boost settings could vary based on power delivery, even though I read that insufficient PSU power doesn’t hurt performance. If the system would crash or fail to boot with a weak PSU, that would be concerning.
F
Felipe_Lima
12-03-2016, 01:01 PM #6

The build includes an Intel Core i7-4790 Haswell Quad-Core at 3.6 GHz, a Gigabyte GA-H81M-S2H motherboard, a GeForce GTX 780 TI Direct CU 2 with 2 OC, an Antec EA-650 W, a 8GB PC3-12800 DDR3-1600 2Rx8 RAM, a Thermaltake Versa H22 Mid-Tower case, a 240 GB SSD, and a WD 1 TB Blue HDD. There are a few updates: I recently added the 1 TB HDD and the Direct CU 2. Previously, I was using an EVGA Reference 780 TI without the 1 TB HDD, which I think consumes less power than the Direct CU 2. I’m also connecting the Direct CU 2 to an 8-pin PCIE plus 2 Molex to an 8-pin PCIE slot.

I ran the Unigine benchmark at stock settings today and achieved a score of 2600, which improved to 2900 the previous night. This inconsistency in scores is puzzling, suggesting the GPU might be underpowered. I’m wondering if GPU Boost settings could vary based on power delivery, even though I read that insufficient PSU power doesn’t hurt performance. If the system would crash or fail to boot with a weak PSU, that would be concerning.

G
GetUSom
Member
194
12-03-2016, 08:11 PM
#7
I discovered that gpuz will reduce performance and adjust the GPU to the clock I configured during benchmarking. I set it using the benchmark tool, then used this method to force the GPU to throttle before running my game test.
G
GetUSom
12-03-2016, 08:11 PM #7

I discovered that gpuz will reduce performance and adjust the GPU to the clock I configured during benchmarking. I set it using the benchmark tool, then used this method to force the GPU to throttle before running my game test.

M
Marc9225
Member
204
12-04-2016, 12:20 AM
#8
Your overclock might benefit from increased voltage. It's often near the edge of performance when running at high speeds with insufficient power. I commonly notice this issue where higher clock speeds cause lower FPS when VRAM speeds approach instability for the required voltage to maintain core clocks, because power can't be allocated separately to memory and core. This could be due to your power delivery (check the dual molex to 8 pin plug) or your GPU itself.

GPU Boost might also contribute if your 2600 score was recorded during the hottest part of the afternoon and the 2900 during the coldest evening hours. Did you push the thermal and/or power limits to their maximum?
M
Marc9225
12-04-2016, 12:20 AM #8

Your overclock might benefit from increased voltage. It's often near the edge of performance when running at high speeds with insufficient power. I commonly notice this issue where higher clock speeds cause lower FPS when VRAM speeds approach instability for the required voltage to maintain core clocks, because power can't be allocated separately to memory and core. This could be due to your power delivery (check the dual molex to 8 pin plug) or your GPU itself.

GPU Boost might also contribute if your 2600 score was recorded during the hottest part of the afternoon and the 2900 during the coldest evening hours. Did you push the thermal and/or power limits to their maximum?

K
Kxte
Junior Member
12
12-04-2016, 03:22 AM
#9
I noticed I begin with a high overclock offset for my core and keep the stable offsets active. For power, I use a temp priority with an 88 power target and a 76 c target, achieving solid overclock results at around 73°C and reaching 1110 oc with a 110 gpu clock offset. I’ve played Tomb Raider and Playstation without any issues. I plan to gradually adjust the core clock incrementally until I find the best setting for my power goal. I prefer avoiding crashes by playing daily until I reach optimal performance. It seems the system runs more smoothly in the 70°C range, though success drops below 110°C. The ASIC 80 value helps reduce power usage, minimizing thermal concerns and allowing focus on core clock stability. I make small adjustments to the core clock and monitor temperatures closely, aiming for temps under 76°C until my card reaches its maximum performance. The setup includes two fans—one at the inlet and another at the rear of the card.
K
Kxte
12-04-2016, 03:22 AM #9

I noticed I begin with a high overclock offset for my core and keep the stable offsets active. For power, I use a temp priority with an 88 power target and a 76 c target, achieving solid overclock results at around 73°C and reaching 1110 oc with a 110 gpu clock offset. I’ve played Tomb Raider and Playstation without any issues. I plan to gradually adjust the core clock incrementally until I find the best setting for my power goal. I prefer avoiding crashes by playing daily until I reach optimal performance. It seems the system runs more smoothly in the 70°C range, though success drops below 110°C. The ASIC 80 value helps reduce power usage, minimizing thermal concerns and allowing focus on core clock stability. I make small adjustments to the core clock and monitor temperatures closely, aiming for temps under 76°C until my card reaches its maximum performance. The setup includes two fans—one at the inlet and another at the rear of the card.