i5 vs i7 - Worse results with i7??

[PCGuy_5960]

Just now, Enderman said:

Well in this case it's negatively affected by a better CPU.

So tell me again how is it a GPU benchmark...?

200 points lower.... In other words margin of error (1FPS difference)

[Enderman]

Just now, Gameborn said:

gpu load is at 99% in both tests

So?

It was still affected by the CPU, and that affected the result, so it's not really a good GPU test then is it.

[Enderman]

1 minute ago, PCGuy_5960 said:

200 points lower.... In other words margin of error (1FPS difference)

He said he ran it multiple times and still got worse with the i7, that means it is not margin of error.

[PCGuy_5960]

Just now, Enderman said:

He said he ran it multiple times and still got worse with the i7, that means it is not margin of error.

I ran Superposition 4 times and I wasn't able to beat my highest score. And 1FPS is within the margin of error...

[Enderman]

1 minute ago, PCGuy_5960 said:

I ran Superposition 4 times and I wasn't able to beat my highest score. And 1FPS is within the margin of error...

Did you not read the OP?

"I ran it three times, same result."

Margin of error would mean that some of those 3 times the i7 would have done better than the i5.

Please go look up the definition of margin of error.

[PCGuy_5960]

Just now, Enderman said:

Did you not read the OP?

"I ran it three times, same result."

Margin of error would mean that some of those 3 times the i7 would have done better than the i5.

Please go look up the definition of margin of error.

Did he run it 3 times with the i5 as well?

[PCGuy_5960]

@CostcoSamples Any progress?

[Hunter259]

Are you trying to compare an OC'ed i5 and a stock i7?

[JoostinOnline]

My guess is that the 4790K is thermal throttling.  I've got the same one.  If you applied the thermal paste wrong, or used the pre-applied paste on the stock heatsink, it gets really hot.  Using my own thermal paste took like 10 degrees off of it.  It was ridiculous.  You should also install Intel Extreme Tuning Utility and follow these instructions:

 

[CostcoSamples]

31 minutes ago, PCGuy_5960 said:

@CostcoSamples Any progress?

 

35 minutes ago, Enderman said:

Did you not read the OP?

"I ran it three times, same result."

Margin of error would mean that some of those 3 times the i7 would have done better than the i5.

Please go look up the definition of margin of error.

 

30 minutes ago, Hunter259 said:

Are you trying to compare an OC'ed i5 and a stock i7?

 

30 minutes ago, JoostinOnline said:

My guess is that the 4790K is thermal throttling.  I've got the same one.  If you applied the thermal paste wrong, or used the pre-applied paste on the stock heatsink, it gets really hot.  Using my own thermal paste took like 10 degrees off of it.  It was ridiculous.  You should also install Intel Extreme Tuning Utility and follow these instructions:

 

 

OK, I used DDU and reinstalled drivers.  The results are starting to make more sense.  I ran Superposition twice and Timespy twice.  In both cases, results were within 5 points of each other.

 

I have not checked temps, but it should be fine.  I'm using high quality thermal paste and a Noctua cooler.

 

 

[CostcoSamples]

Note that on Timespy, the GPU score is basically identital to the previous score using the i5.  Only the CPU score is higher, as you'd expect.

[Enderman]

I don't get why reinstalling GPU drivers would change your CPU performance, but good you figured it out.

[CostcoSamples]

46 minutes ago, PCGuy_5960 said:

200 points lower.... In other words margin of error (1FPS difference)

I think margin of error is much tighter than 200 points on these types of tests.  In a gaming test, 1 fps is reasonable.

[CostcoSamples]

1 minute ago, Enderman said:

I don't get why reinstalling GPU drivers would change your CPU performance, but good you figured it out.

I agree, it's weird.  Thanks for your help.

[PCGuy_5960]

2 minutes ago, CostcoSamples said:

OK, I used DDU and reinstalled drivers.  The results are starting to make more sense.  I ran Superposition twice and Timespy twice.  In both cases, results were within 5 points of each other.

Told ya  Glad you figured it out

[done12many2]

28 minutes ago, CostcoSamples said:

OK, I used DDU and reinstalled drivers.  The results are starting to make more sense.  I ran Superposition twice and Timespy twice.  In both cases, results were within 5 points of each other.

 

I have not checked temps, but it should be fine.  I'm using high quality thermal paste and a Noctua cooler.

 

 

 

 

If you want to remove CPU performance from being a factor in Superposition,  you'll want to run 1080p Extreme, 4k Optimized, or 8k Optimized only.  1080p Medium and those lower then 1080p Extreme can still be influenced by CPU performance. 

[PCGuy_5960]

Just now, done12many2 said:

If you want to remove CPU performance from being a factor in Superposition,  you'll want to run 1080p Extreme, 4k Optimized, or 8k Optimized only.  1080p Medium and those lower then 1080p Extreme can still factor in some CPU performance. 

1080p Extreme is more intensive than 4k Optimized

[done12many2]

Just now, PCGuy_5960 said:

1080p Extreme is more intensive than 4k Optimized

 

Yeah, Unigine set out to crush GPUs with there latest benchmark and succeeded.  1080p Extreme is definitely hard on a GPU, but that 8k Optimized is brutal too.  

[JoostinOnline]

31 minutes ago, CostcoSamples said:

I have not checked temps, but it should be fine.  I'm using high quality thermal paste and a Noctua cooler.

You should still check them, simply because of weird things that can happen with that CPU. 

[CostcoSamples]

On 5/21/2017 at 2:53 PM, JoostinOnline said:

You should still check them, simply because of weird things that can happen with that CPU. 

Will do!  thanks for the heads-up.