144Hz vs 240Hz

[majkel]

Hi,

obviously, there is a slight boost in 3D when increasing refresh rate from typical 60 or 75 to 144Hz and I have experienced that myself.

But currently I plan to choose a new display and I am wondering what to choose (performance wise).

I suppose there is also a difference between 144 and 165 and even further to 240Hz.

Can u guys advise on what can be expected beyond 165 and to 240Hz? Is the difference between 144Hz and 240Hz much lower that from typical 60 to 144hz?

 

Also, do you see any plug dependant differences, e.g. DP vs HDMI differences?

[suzuki]

Where do you see this differences ?

In what benchmarks ? (Numbers not placebo)

[majkel]

http://forum.hwbot.org/showthread.php?t=166712&highlight=144hz&page=2

 

here is 178k @ 60Hz and 185k @ 144Hz using GTX 1070 and 3dm03...

 

I seem not to have a placebo with my recent GT1030 scores -

3dm03 70k with 60Hz and 73k with 144Hz (same settings)

heaven Xtreme ~ 1070X on 60Hz and 1160X on 144hz (same settings).

Seems like noticable difference.

[yosarianilives]

Yep, it's well known that higher refresh rate monitors provide slightly higher score. What also works well is to just pull monitor during run if you don't believe it do a quick test in aquamark or 3dmark03, you'll see your score go way high. Although it will make your subtests look off for 3dm03 and will get taken down if you're doing it for cheapaz. As far as 144 vs 240 hz? I couldn't tell you but there probably is a benefit but it may not be huge. I don't know where the performance scaling stops. It also helps to set lowest desktop res, not just for ice storm.

[Guest]

Not sure about this, but it depend what the benchmark count for a valid score.

If it is the effective frames calculated by the gpu or the frames displayed on the screen.

 

The gpu have a number of frames that come out from the buffers, but only the frames allowed by the screen are effectively displayed.

The effective frametime outputted by the gpu and the one displayed by the screen are not the same thing. You could have a lower frametime on the gpu compared to the screen one.

 

A 60Hz monitor could display a frame each 16.6ms, a 144Hz each 6.9ms, a 240Hz each 4.1ms. As said if the gpu is handling the bench quite easy and could display 144fps it would match the monitor refresh rate. If the monitor would be only 60Hz but the gpu could display 144fps, roughly only 1 frame of 2 would be effectively displayed.

 

So again it all depend what the bench consider, if it consider the numbers of frames outputted from the buffers or if it count the frames displayed.

If it count the frames displayed, it could be that you would get better scores with a high refresh rate monitor over a low refresh rate one since effectively there are more frames displayed compared to a low refresh one.

Edited January 16, 2018 by Guest

[suzuki]

There is no difference in my runs of 2003 between 60 hz and 120 hz. Maybe we do different tweaks which can be affected by this.

Here is my run with 980 Ti,used to be on 19th place :

http://hwbot.org/submission/3462125_suzuki_3dmark03_geforce_gtx_980_ti_274070_marks

 

I do not recomend benching with no image on the monitor,you will never know if it was a bugged run.

I have even 284k score but i know it is bugged,the image was black for 2 seconds on gt2 hence the higher score.

[Guest]

Well the guy who beat all the R9 290 pushed the card to 1400Mhz, hence he posted the video about, you clearly see that the image went off multiples times during the bench.

I don't like it and always bench with a image, with a lot of artifact but still an image.

Still it is not against the rules i suppose.

 

Also i suppose the connection to the screen would have some effect on overall latency, lower is latency of the protocol used better should be the result. But i don't really know the differences between DVI/HDMI/DP protocols latency, usually the newer it is the lower the connection latency it is, i suppose.

Edited January 17, 2018 by Guest

[Splave]

Hey if u can make sure you can have a dvi connection. I bought a 1440p 165hz and it's only hdmi and dp. Tried dp to dvi and had no signal. Another thing to consider.

[buildzoid]

Well the guy who beat all the R9 290 pushed the card to 1400Mhz, hence he posted the video about, you clearly see that the image went off multiples times during the bench.

I don't like it and always bench with a image, with a lot of artifact but still an image.

Still it is not against the rules i suppose.

 

Also i suppose the connection to the screen would have some effect on overall latency, lower is latency of the protocol used better should be the result. But i don't really know the differences between DVI/HDMI/DP protocols latency, usually the newer it is the lower the connection latency it is, i suppose.

 

The thing is Hawaii black screening could just mean the display drive circuit is dropping out while the GPU is still running all the calculations. Lots of AMD GPUs will do this if the Vcore is high enough.

[Guest]

The thing is Hawaii black screening could just mean the display drive circuit is dropping out while the GPU is still running all the calculations. Lots of AMD GPUs will do this if the Vcore is high enough.

 

I have no experience with the card having this behavior so i couldn't say if it is good or bad, nor have opinion to this regard.

Nevertheless i always try to bench with an image, surely because in my case i did not pushed the card so far.

My 290 vrm are still under air but the gpu is under water, so i could not crank up the voltage, the card stop being stable passed 1300Mhz or so with +250mv and i don't want let the vrm hit more than 90° with that current flowing through.

My gpu just crash the bench.

[Tame]

I have no experience with the card having this behavior so i couldn't say if it is good or bad, nor have opinion to this regard.

Nevertheless i always try to bench with an image, surely because in my case i did not pushed the card so far.

My 290 vrm are still under air but the gpu is under water, so i could not crank up the voltage, the card stop being stable passed 1300Mhz or so with +250mv and i don't want let the vrm hit more than 90° with that current flowing through.

My gpu just crash the bench.

+1 What Buildzoid said... At +380 mV and lower than normal Vdroop setting in Bios the display output circuit signal gets corrupted on Hawaii without modding the 0.95 V rail which I have not done. That's why I now use DVI and 1080p (or lower) @ low refresh rate for less video bandwidth which helps with this. The scores still scale as you would expect with GPU Mhz regardless of the display drop out. Also I didn't use that run for submitted scores, since I got better runs with different conf and actually pretty much continuous video signal. I have EK full cover blocks with upgraded thermal pads on the VRM which prevent them burning out (so far).

http://www.overclock.net/t/1638551/testing-results-r9-290-gpu-vrm-vcore-power-measured-with-multimeter-at-different-vid-and-load-line-settings/0_20

[majkel]

Thanks for all the info above, it's very useful. Still, let's do not forget about main question: Have you guys saw any difference between 144 and 240Hz results?

[FireKillerGR]

@dancop has tested it and might be able to share any differences. I was considering getting one but got 144hz 1440p instead