LG 27GK750F-B - 240Hz, 27" 1ms Blur reduction, Freesync

[GammaLyrae]

.

[giubox360]

@a solid lad
Do you prefer the alienware 240 hz freesync more than the new Lg with the Motion blur reduction?

Yes, I do prefer it.

Here's a video I just made, in which I explain my problem with the LG:

phpBB [video]

Can you please try also a "human reaction test" online to test the different input lag of the monitors? And also the other program: LatencyMon from resplendence. Thanks in advance

[A Solid lad]

Aaargh... just made a post and connection error lost it to me

In short, can you please use LatencyMon from resplendence, nowadays afaik its the only valid tool to measure DPC latency.

Also you should try reset-all.exe from CRU since last time you're having "a lot" of different monitors, just to make sure that nothing interferes wrong with Nvidia drivers.
Feeling of worse responsiveness also could be a mind trick, smaller size picture will always feel more rapid. I mean that 24" could produce feeling of better responsiveness, when set next to 27".

DPC latency though is a complex thing and personally I hardly beleive that one-digit-microsecond (1/1,000 of a millisecond) can produce felt by a human difference. But I'm aware that growth of DPC latency can indicate system low level problems, usually caused by drivers.
That's a strange since actually monitors do not use drivers apart from GPU (Nvidia in your and mine case). I'm really sceptic that other monitor can produce higher (or lower) DPC latency.
In my case, DPC latency havent changed when I switched to LG from VG278H.

That's how it looks like on my system (measured at idle)

Again, the dpc latency metrics are not my problem.
Felt responsiveness is, but I can't show any direct evidence of that, as I don't even know where the problem stems form.
Increased dpc latency is just the side effect of the problem that is causing this unresponsiveness... and I don't know what's the problem.
I've reset EDID settings with cru, and tried switching monitors while powering off my pc (also doing a reset before every power off) and tried using all outputs on my video card, as well as all inputs on the monitor, with which I could connect it.
I've already optimised my PC with many tweaks, which I don't have the time to list, and it shows when using my old Benq or the Alienware.
The problem is with the LG monitor, period.

My guess is that they use timings for the "PC" resolutions (120, 144 and 240hz @1080p) which don't work well with my hardware... and I don't have the nerve to fiddle with CRU for days to POTENTIALLY find a custom resolution which fixes this responsiveness issue...and that's assuming it's fixable... I have to send the monitor back tomorrow, otherwise I'm stuck with it.
And maybe my assumptions about the root cause are totally worng... and timings aren't the problem.
That would mean I don't even have a chance of fixing it with different timings... (but I do know that using 60hz "TV" timings mostly fixes the problem...so that's what meakes me believe the timings are at fault)

If you don't believe me, that's fine, but I can tell there's a difference, and it's not placebo, nor caused by 24" vs 27" (I knew somebody would reply with that...) I've been playing CS: Source competetively since 2010, my game and PC is really optimised for it... it runs consistently and I know the game so well, that I can distinguish problems caused by IT or any OTHER source... (pun not intended)
so whenever I get a new piece of hardware/do a new tweak, I just fire up the game, and after playing it for 5 minutes I can tell whether something's up or not.
But alas, I know very few would believe me, just based on that... that's why I've tried to use dpclat, to have some numbers to back my claims up... and what do I know, people blame my test methodolgy...

I understand your disbelief, and you trying to protect your purchase...but I'm not willing/able to sink more time into trying to quantify the problem, just to convince other people it's actually there...

[k2viper]

Yes, LG uses slightly different timings at higher hertz. Stock 100 VBI (VT 1180) at 240hz (more then others 240hz having 1125-1150 VT), and thus lower HT 2040 (others I've seen used 2080).

What GPU model and driver do you have? I have Nvidia 1070 with 385.69 driver (I always stick with some driver version I found stable and responsive, for a long time).

I can beleive that you definitely feel game responsoveness right (I'm also a competitve player), btw did you set 1:1 scaling mode in OSD?
And you may try VT tweaked timings I use:



I just measured with LatencyMon in 60hz and 240hz and looks like yes, at 60hz there is a lower "minimum" measured DPC latency (I also saw one digit numbers at 60hz, two digit numbers at 240hz as a minimum). But an "average" numbers I see on my system both in 60 and 240hz are still between 25-30 microseconds. And same "max" numbers at 60 vs 240, so there's no additional spikes.
What I also note is that my old Asus VG278H at 120hz used the same 2080 HT that LG uses, and it resulted in the same level of measured DPC latency, thus I havent seen any increase after switching to 240hz LG.

It would be interesting when instrumental input lag measurements of this monitor will came out (tftcentral should do these tests).

You just sowed the seed of doubt in my belief that I got best possible 240hz for my game

[A Solid lad]

Yes, LG uses slightly different timings at higher hertz. Stock 100 VBI (VT 1180) at 240hz (more then others 240hz having 1125-1150 VT), and thus lower HT 2040 (others I've seen used 2080).

What GPU model and driver do you have? I have Nvidia 1070 with 385.69 driver (I always stick with some driver version I found stable and responsive, for a long time).

I can beleive that you definitely feel game responsoveness right (I'm also a competitve player), btw did you set 1:1 scaling mode in OSD?
And you may try VT tweaked timings I use:



I just measured with LatencyMon in 60hz and 240hz and looks like yes, at 60hz there is a lower "minimum" measured DPC latency (I also saw one digit numbers at 60hz, two digit numbers at 240hz as a minimum). But an "average" numbers I see on my system both in 60 and 240hz are still between 25-30 microseconds. And same "max" numbers at 60 vs 240, so there's no additional spikes.
What I also note is that my old Asus VG278H at 120hz used the same 2080 HT that LG uses, and it resulted in the same level of measured DPC latency, thus I havent seen any increase after switching to 240hz LG.

It would be interesting when instrumental input lag measurements of this monitor will came out (tftcentral should do these tests).

You just sowed the seed of doubt in my belief that I got best possible 240hz for my game

I've 1060 with 391.01 drivers. As you've witnessed similar results with a different driver, I think we can rule that one out.
Yes I've set 1:1 as suggested earlier, and also tried the other two scaling modes...no difference imo.

Thanks for testing the dpc latency yourself, I was just about to ask...
in this case, just like with many other things, averages don't tell the whole story...

If I wouldn't have had my old XL2411Z for comparison, I wouldn't have been able to tell that there's any additional delay... but alas I had.
And I now have the Alienware on my table as well, and that feels even more responsive... even at 144hz, which is strange, since my pc has a little bit higher average dpclat, when connected to the alienware, as opposed to the Benq, as you could see in the vid.
My guess is that the better overdrive (and maybe faster signal processing?) makes up for the slightly higher dpclat spikes on the alienware.

[k2viper]

I highly doubt that overdrive nor signal processing of the monitor have something to effect on system kernel level (drivers dpc latency), I'm not totally sure but I always thought that monitor has no feedback to the gpu. At least such internal things like overdrive.

So, I think the case is this exact timing set, paired with some logics in Nvidia drivers are playing that game for us.
I'll experiment with custom 240hz later, will try to make 240hz with 2200 Horizontal total just like 2200 HT at 60hz (thanks LG has a headroom of 640mhz dotclock).

Sadly I dont have any fast gaming monitor to compare my responsiveness feeling, because I dont see any dpc spikes and higher averages of dpc's.

Maybe Chief can tell us, may theoretically different res timings set affect gpu+monitor input lag in some way, or not. E.g. I remember you stick to GPU scaling and maybe exact timings set affects how fast GPU scaling is processed (I remember Chief explained, that GPU scaling nowadays is line-by-line scanout)

[A Solid lad]

Interesting ideas at the end of your post.
I wrote about signal processing, because I still was not a 100% sure, that the more sluggish feeling was caused by higher dpc lantency... but thinking about it now, it pobably is.

[giubox360]

I highly doubt that overdrive nor signal processing of the monitor have something to effect on system kernel level (drivers dpc latency), I'm not totally sure but I always thought that monitor has no feedback to the gpu. At least such internal things like overdrive.

So, I think the case is this exact timing set, paired with some logics in Nvidia drivers are playing that game for us.
I'll experiment with custom 240hz later, will try to make 240hz with 2200 Horizontal total just like 2200 HT at 60hz (thanks LG has a headroom of 640mhz dotclock).

Sadly I dont have any fast gaming monitor to compare my responsiveness feeling, because I dont see any dpc spikes and higher averages of dpc's.

Maybe Chief can tell us, may theoretically different res timings set affect gpu+monitor input lag in some way, or not. E.g. I remember you stick to GPU scaling and maybe exact timings set affects how fast GPU scaling is processed (I remember Chief explained, that GPU scaling nowadays is line-by-line scanout)

When you finish the test, please tell us more about the real "input lag" on the LG @240 hz

[k2viper]

When you finish the test, please tell us more about the real "input lag" on the LG @240 hz

I dont have other monitors to compare it side by side with LG, and dont have an instruments to measure input lag.
For me LG's 240hz is absolutely fine, rapidly fast, good "connected feel".
System DPC latency measured at 60hz, has definitely lower minimums then at 120, 144, 240hz. I looked at the timings sets of these resolutions and see, that 60hz has 2200 HT, 120-144hz has 2080 HT and then 240hz has 2040 HT.

I only can suppose that Nvidia driver may be working "better" with higher HT like 2200 and struggling to work with HT 2040.
Most others 240hz monitors resolutions that I saw, used HT 2080 at 240hz.

But still, even if I can measure system DPC latentcy differences at 60hz vs 240hz, that difference is not so prior.
On my system, I'm getting like 25-30 microsecond "average" and about 70-80 max at idle.
If I set 60hz, I can sometimes see a one digit "minimum" numbers like 4-9 microseconds. At 120-144hz minimums were about 18-19. At 240hz they are about 22-23.

In A Solid lad's video there is the same thing, minimums go lower but averages are still about the same when he choose 60hz.

It's just 0,01-0,02 ms DPC delay difference with no additional high spikes.

My opinion is that the slugginess A Solid lad mentioned, is not an additional DPC latency related (since an additional 0,01ms is really not much), but likely Nvidia GPU scaling related. Because if GPU scaling on Nvidia Pascal GPUs are used, it should be line-by-line unbuffered scanout processing, AFAIR Chief explained that. And thus, different resolution timings set can possibly affect how that scaling is processed.

So, I think it's possible that there will be no actual difference if user will not use GPU scaling. Anyway, we need an instrumental measurements to clear this out.

[GammaLyrae]

At least a tftcentral review is coming soon. Maybe it'll be out around the time the monitor is more readily available at or near MSRP through Amazon directly, instead of exclusively through third parties.