GTX1080 in 3D Fire Strike

[PopsicleHustler]

Hi

 

I've spent some time trying to figure out balance in my system

 

Right now I'm running

 

R5 1600 3.8Ghz 1.35V

GTX 1080FE

Corsair Vengeance 16GB 3000Mhz in XMP.

 

It seems to me that CPU is bottlenecking me. I play mainly in BF1, and GPU at 1080p and even 1440p often dips. in 1080p it can easely dip to low 70s/high 60s % usage, in 1440p down to 80s. FPS obviously suffers. In BF1 1440p it can dip to low 60s even high 50s with average 80-85, on 1080 it jumps from 130 to 70. I've tried running 3D Mark firestrike to see if GPU performs on par with other 1080s. Heres my score

https://www.3dmark.com/3dm/29647808

 

It seems a little low as others have their 1080s getting 23-24k while mine got 21,639, but could that be down to founders? Or CPU bottleneck? 

 

I'm currently considering investing into Intel CPU, most likely 8700k. Do you think that would give me performance boost?

[Ashiella]

6 minutes ago, PopsicleHustler said:

Hi

 

I've spent some time trying to figure out balance in my system

 

Right now I'm running

 

R5 1600 3.8Ghz 1.35V

GTX 1080FE

Corsair Vengeance 16GB 3000Mhz in XMP.

 

It seems to me that CPU is bottlenecking me. I play mainly in BF1, and GPU at 1080p and even 1440p often dips. in 1080p it can easely dip to low 70s/high 60s % usage, in 1440p down to 80s. FPS obviously suffers. In BF1 1440p it can dip to low 60s even high 50s with average 80-85, on 1080 it jumps from 130 to 70. I've tried running 3D Mark firestrike to see if GPU performs on par with other 1080s. Heres my score

https://www.3dmark.com/3dm/29647808

 

It seems a little low as others have their 1080s getting 23-24k while mine got 21,639, but could that be down to founders? Or CPU bottleneck? 

 

I'm currently considering investing into Intel CPU, most likely 8700k. Do you think that would give me performance boost?

I personally get 21,734 graphics score with my 1700 @4ghz and a Hybrid 1080. It's because the graphics bench doesnt use all 12-16 threads, it bottlenecks it slightly in benchmarks as well as games due to them not using all the threads available to it. The 8700k has way higher single-core performance so; yes, it would give you a performance boost in most games and benchmarks.

[JPStone]

16 minutes ago, Ashiella said:

I personally get 21,734 graphics score with my 1700 @4ghz and a Hybrid 1080. It's because the graphics bench doesn't use all 12-16 threads, it bottlenecks it slightly in benchmarks as well as games due to them not using all the threads available to it. The 8700k has way higher single-core performance so; yes, it would give you a performance boost in most games and benchmarks.

Yea its not the CPU Bottlenecking you. Pull up a system Monitor like CAM or msi Afterburner. No way the CPU is pegging at 100% . R5 should be able to cope with any Firestrike benchmark may be a little sluggish during the CPU test but should not effect graphic tests so much since, as was mentioned previously, the benchmark and most games for that matter will use at most 4 cores. The GPU is the limiting factor. I also run a 1080 and it's a great GPU but it not the 1080ti or Titan V or the 20 series cards. It's great for high end 1080P and maybe 4K but unless you're into LN2 overclocking your not gonna get insane performance over say a 1060 or 1070. It will be a noticeable Delta but not in another league better. For more practical results use game play as your Benchmark and not Synthetics. Benchmarks like in 3DMark or Superposition are designed to push you system to the absolute limit of their capability to see how they respond when pushed that hard. In most gaming applications your 1080 is not gonna be pegged at max boost clocks.

 

[Ashiella]

38 minutes ago, JPStone said:

Yea its not the CPU Bottlenecking you. Pull up a system Monitor like CAM or msi Afterburner. No way the CPU is pegging at 100% . R5 should be able to cope with any Firestrike benchmark may be a little sluggish during the CPU test but should not effect graphic tests so much since, as was mentioned previously, the benchmark and most games for that matter will use at most 4 cores. The GPU is the limiting factor. I also run a 1080 and it's a great GPU but it not the 1080ti or Titan V or the 20 series cards. It's great for high end 1080P and maybe 4K but unless you're into LN2 overclocking your not gonna get insane performance over say a 1060 or 1070. It will be a noticeable Delta but not in another league better. For more practical results use game play as your Benchmark and not Synthetics. Benchmarks like in 3DMark or Superposition are designed to push you system to the absolute limit of their capability to see how they respond when pushed that hard. In most gaming applications your 1080 is not gonna be pegged at max boost clocks.

 

The cpu is bottlenecking in the graphics portion of the Firestrike benchmark. When using a 7700k @5.1Ghz, my GPU score went up to the average. According to multiple programs my 1080 is constant at 1898mhz.

[PopsicleHustler]

45 minutes ago, JPStone said:

Yea its not the CPU Bottlenecking you. Pull up a system Monitor like CAM or msi Afterburner. No way the CPU is pegging at 100% . R5 should be able to cope with any Firestrike benchmark may be a little sluggish during the CPU test but should not effect graphic tests so much since, as was mentioned previously, the benchmark and most games for that matter will use at most 4 cores. The GPU is the limiting factor. I also run a 1080 and it's a great GPU but it not the 1080ti or Titan V or the 20 series cards. It's great for high end 1080P and maybe 4K but unless you're into LN2 overclocking your not gonna get insane performance over say a 1060 or 1070. It will be a noticeable Delta but not in another league better. For more practical results use game play as your Benchmark and not Synthetics. Benchmarks like in 3DMark or Superposition are designed to push you system to the absolute limit of their capability to see how they respond when pushed that hard. In most gaming applications your 1080 is not gonna be pegged at max boost clocks.

 

BF1 is loading 6 threads to around 80-90% constant load. Other 6 stay around 30-40

 

Btw, whats No load Limit, its jumping 1 to 0 with power and temp limit.

 

Those are after a clutch BF1 64 player match, perfect for testing.

[JPStone]

No load limit means your GPU is hitting you max power target which is normal you can see on that graph that the CPU usage was at 74% MAX which is not ideal but on a Benchmark like Firestrike 3D would be expected with a R5. CPU is not Bottlenecked and the GPU is supposed to run at max power and clocks during a synthetic benchmark. That's why just firing up a graphics intensive game like IDK....CSGO of PUBG or Fallout 4 or something like that and playing and seeing what your FRAPS are and what your CPU and GPU are doing is a more practical and realistic benchmark. Like I said the Synthetics are designed to push your gear to the maximum to give you an overall picture of the hypothetical maximum potential of your system. They do not represent performance during actual gameplay in most instances.  

[PopsicleHustler]

5 minutes ago, JPStone said:

No load limit means your GPU is hitting you max power target which is normal you can see on that graph that the CPU usage was at 74% MAX which is not ideal but on a Benchmark like Firestrike 3D would be expected with a R5. CPU is not Bottlenecked and the GPU is supposed to run at max power and clocks during a synthetic benchmark. That's why just firing up a graphics intensive game like IDK....CSGO of PUBG or Fallout 4 or something like that and playing and seeing what your FRAPS are and what your CPU and GPU are doing is a more practical and realistic benchmark. Like I said the Synthetics are designed to push your gear to the maximum to give you an overall picture of the hypothetical maximum potential of your system. They do not represent performance during actual gameplay in most instances.  

These graphs are from 23 minute match of 64 player Battlefield 1.

[JPStone]

There you go it's not a bad thing still that your GPU is pegging out. Your CPU usage is still below 100% at most and was usually around 50% so I don't think you have anything to worry about. Bottleneck is if your GPU still has room to move but your CPU is holding it back and this clearly was not the case.

[PopsicleHustler]

41 minutes ago, JPStone said:

There you go it's not a bad thing still that your GPU is pegging out. Your CPU usage is still below 100% at most and was usually around 50% so I don't think you have anything to worry about. Bottleneck is if your GPU still has room to move but your CPU is holding it back and this clearly was not the case.

Then what 8n the world is holding the GPU back? Usage is low, so is FPS. It's gotta be CPU because 6 cores that are used are 80+%, rest is almost idle because engine is cannot utilize them.

[JPStone]

The usage looks fine to me at least for the CPU. It's not hitting 100%. Any game is not going to use more than 4 cores at a time so you will not see increased activity across all cores evenly. Sometimes it might use the same four cores (say 1-4) or it may bounce the work around between the cores but no more than 4 CPU cores will be actively working on rendering the game. The GPU is hitting the load limit which is fine but that is showing that the GPU and not the CPU is the limiting factor. Try playing around with some Overclocking settings if you feel your GPU is underperforming (at your own risk of course) but really it's almost impossible to fry modern GPUs like a 1080. It wont let you set something to higher than what it can safely handle at least for short period and it will throttle itself even if you manually overclock if temps or voltage get to high. Try setting you power target a little higher. Maybe a +5-10 offset. See if works better. You will see slightly higher temps but it will also increase the boost core clocks and mem clocks