Input and Processing Latency

So I was super late on getting the input latency section done, my sincere apologies if you've been refreshing endlessly every day as one day turned into an additional three. The delay was no fault of the PA301w, it was entirely just the sheer amount of stuff being worked on behind the scenes during just another busy week at AnandTech. Enough with my excuses though, let's get on to the results.

Processing and input lag are both serious considerations for gamers, but also noticeable after a certain point doing even the most mundane of desktop productivity tasks. It's that nagging - something feels off - sensation that really drains on one after a while. We've been doing our best to characterize input latency so far by doing some real world tests alongside CRTs. First with a 17" Princeton, then a Sony G520 20" CRT. The results so far have actually shown a fair amount of variance in input latency. 

I measured latency on the PA301w the same way we've done it in the past. I run the 3DMark06 wings of fury demo on loop, and take photos of the display in test alongside a mirrored CRT at shutter speeds at or faster than 1/160th of a second. Resolution is set at XGA so we can crank the refresh all the way up on both displays (on the PA301w that's 75 Hz at XGA, 60 Hz at native). I snap a bunch of photos, and average a ton that show the difference in frame number between the two, and through some math we can get a decently accurate feel for what input latency is like. 

You won't find it in the normal OSD, but in the Advanced OSD the PA301w has an interesting option labeled "Response Improve," which I can only assume is essentially overdrive control. I tested with this set to on and off, and found something interesting:

Processing Lag Comparison (By FPS)

Remember that I'm driving both the CRT and PA301w at 75 Hz. At that refresh rate, one frame is 13.33 ms, two frames is 26.66 ms. Given how close those numbers are to the measured times, it looks like turning response improve on adds one more frame of latency. Interestingly enough, the difference between response improve turned on and off is actually visually noticeable. With response improve on, the numbers I read out of super fast shutter images of the PA301w are crisp and readable. With response improve off, almost 50% were lost to two frames being present and making it hard to discern the numbers. This is again the tradeoff you make - better panel response and less ghosting through overdrive, but at the cost of some additional processing latency if you look at it with a camera. I spent some time playing the Crysis 2 multiplayer demo and couldn't detect any latency subjectively with the setting on or off - you're a better gamer than I if you can pick that 10 ms difference out in actual practice.

Custom Refresh Rates

Ganesh (our resident HTPC expert) has also been asking me to add some custom refresh rate testing to our display review suite. Refresh rates like 23.976, 24, 25, 29.97, 50, and 59.94 Hz. I entered the custom refresh rates into the NVIDIA Control Panel and tested them on my GTX 470. All of the above refresh rates were supported at 1920x1080, which is the most likely place to encounter a bunch of those. The OSD appears to round appropriately for all refresh rates but 59.94, which it displays as 59.9 Hz. Curiously enough 23.976 Hz displays as 24 Hz, but displays just fine. 

Brightness Uniformity Power Consumption and Final Thoughts
Comments Locked

92 Comments

View All Comments

  • 63jax - Tuesday, March 1, 2011 - link

    I'm also very curious about Samsung's PLS matrix. i also want to change my monitor with an IPS or PVA panel in 22'' range, but i'm really undecided. Maybe i will wait to see what PLS are capable of. Good review by the way, thanks.
  • softdrinkviking - Tuesday, March 1, 2011 - link

    At 22", you can get a lot of great quality LCDs for reasonable prices. If you can spend 300-400, I would go for IPS, if you want to stick closer to 200, you can get a VA screen from BenQ that is 23" widescreen and looks pretty good to me. The VA screens have between the viewing angles of the IPS and the TN, but without the color shift I noticed on TNs. When a VA goes out of view, it just looks dark, not pink or something weird like a TN.
    But I don't know how "calibration-friendly" VAs are. If photography is a concern for you, and you need the print to match the image in your screen, I think IPS is better.
    This new samsung technology is anybodys guess, but I wouldn't get one until they have been put through the paces a bit and prove reliability.
    NEC and LG have had a long time to work the kinks out of IPS, and it's widespread adoption in critical industries did not occur by chance.
  • eaw999 - Wednesday, March 2, 2011 - link

    va's calibrate fine. many eizo's use va panels. but va panels have off-angle gamma shift, though, and it doesn't take a lot of angle to notice it. for me, anyway. this is most evident in darker shades. i tried an eizo monitor once and i couldn't stand watching movies on it because if i moved my head even slightly, shadow details would 'shimmer'. the only thing i liked about the eizo was that it had phenomenal black levels, as modern va panel usually do.
  • legoman666 - Tuesday, March 1, 2011 - link

    It's amusing to see that Anandtech staff run adblock+ on their own website.
  • quiksilvr - Tuesday, March 1, 2011 - link

    ROFL! Even they admit their adverts are annoying as hell.
  • Stas - Wednesday, March 2, 2011 - link

    lol. Sponsors won't be happy about that ^.^
  • gc_ - Thursday, March 3, 2011 - link

    I expect the advertisers paid for ads alongside articles, not for in-article placement. I expect advertisers also paid to have ads displayed during a certain period (say, a week or a month), not necessarily to have it display in-perpetuity in the archives. I don't know if AnandTech is syndicated and/or translated for different markets, but if it is, then if they included the ads they might have to redo the image for every market.
  • DanNeely - Tuesday, March 1, 2011 - link

    "Interestingly enough, the PA301w has a uniformity setting inside the OSD which defaults to 5 (maximum) and seems to definitely improve overall brightness uniformity and Delta-E consistency. Whatever secret sauce NEC has apparently works. "

    If this works the same way I think it does on the 3090, the internal LUT adjusts different parts of the screen differently to boost uniformity. I'd be curious what happens to it in 10bit mode though, since the 3090 did this by mapping 8 bit input to a 10bit panel/LUT.
  • DanNeely - Tuesday, March 1, 2011 - link

    Should've looked at the specs before asking. The pa301w has a 14bit lookup table to use for onboard calibration/adjusting for screen variation.
  • DanNeely - Tuesday, March 1, 2011 - link

    Is this any better than in the multisync 90 series? The ones on my 2090/3090's were really flimsy and if you tried picking up the monitor the wrong way some of the tabs holding them on would snap off.

Log in

Don't have an account? Sign up now