Re: Why do 240HZ monitors have more lag than 144HZ AHVA
Posted: 23 Jan 2018, 13:44
Tom's Hardware seems to use Leo Bodnar (60Hz) so drop that result.KKNDT wrote:"Lag competition between 240HZ TN and 165HZ IPS" is a very hot topic around me. However my knowledge is insufficient to asnwer it.That's why I open this thread here. Till now most people I know do trust TFTcentral's result. I believe 240HZ TN is sure to win regarding lag randomness factor explaned here. But who wins on absolute lag? I still can't find a authoritative saying whose data is more reliable?
RTINGS: XL2540 3.7MS VS XB271HU 4.6MS
TFTCENTRAL: PG258Q 4.6MS VS PG279Q 3.25MS, SMTT 2
TOM'S HARDWARE: PG258Q 20MS VS XG2703-GS 31MS
EDIT: Not quite, but something similar via high speed video
RTINGS and TFTCentral is much closer to true absolute lag for 240Hz.
RTINGS currently electronically measure to the GtG1% point (essentially, the beginning of visibility of pixel transition)
TFTCentral currently visually compares a SMTT 2.0 photograph, to find the beginning of GtG.
In TFTCentral's graphs, they add a "band" representing manufacturer GtG (that's, however, the GtG10%->90%). I don't know what Adam Simmons of TFTCentral declares the beginning of GtG, but I think first "faint fade" edge of pixel response -- meaning the lowermost faint appearance of lag digits in SMTT 2.0 photo. That would correspond to the beginning of GtG. For a 1/1000sec photograph on a 1000fps VSYNC OFF test (SMTT 2.0), the error margin is ~1ms (1080p@120Hz 135KHz scanrate, that error margin is 135 scan lines, or about 8% the height of a 120Hz refresh cycle). So I would ignore decimal digits of TFTCentral numbers.
However, TFTCentral numbers are pretty uncannily accurate (compared to Leo Bodnar) considering they are human-eye inspection of SMTT 2.0 photographs -- it is darn impressive how a human-eye inspection of a SMTT 2.0 photograph can yield ~1ms lag-test accuracy (confirmed by me). Just mostly discount the numbers to the right of the decimal point, though. If many runs are done, e.g. 10 runs, you can get better than 1ms accuracy for SMTT 2.0 lag tests. But only to a certain point.
I do trust the numbers to the left of the decimal point for RTINGS numbers and TFTCentral numbers for max-Hz absolute lag measurement to an accuracy margin of approximately 1ms error (for all the unknownn factors, like digital cable latency -- like DisplayPort micropackets, etc).
Cable lag too. There are lag differences in cables.KKNDT wrote:As far as I can conclude, display-only lag is affected by:
1. Signal processing lag (scaling/buffering... etc.)
2. Scanout lag
3. GTG
4. maybe something missed
In lowest-to-highest, is VGA -> (HDMI, DVI) -> DisplayPort.
- VGA is analog and RAMDACs have little lag (virtually lagless).
- HDMI, DVI is digital and requires a little bit of codec lag to convert the digital into a modulated signal over the cable, which is then demodulated on the remote end.
- DispayPort is micropacket based, so there's also packetization lag added on top of digital codec lag.
Even different HDMI version numbers and different DisplayPort version numbers have lag differences (in the hundred-microsecond difference scales). Meaningless for most, but makes decimal digits in most lag benchmarks harder to compare across sites.
It's very tiny (often sub-1ms difference) but when attempting to measure display lag to decimal digits (e.g. 0.1ms) cable lag becomes a massive error margin for decimal digits after the millisecond number.
That's why I would mostly discount decimal points in most lag test numbers (especially if different HDMI version numbers, or only treat them as "relative lag" numbers, as long as the same cable connection is used.
It's still useful to publish the decimal points, since comparing modes, comparing cables, comparing between monitors, on a relative basis (Same outout of the same graphics card), the decimal digit can then become useful (if the error margin of the testing method allows it -- not SMTT 2.0), albiet not for comparing across different websites.