Hello, I'm a long time lurker in this website and forum, but finally decided to register in.
I have been interested about input lag in gaming for some time. I read a thread from link beneath and got interested in msi mode. I turned all my devices to msi mode one by one starting from nvidia card and got my isr/dpc latencies to drop drastically. However, I'm not so sure if the actual input lag got higher or lower in games after the switch.
http://forums.guru3d.com/threads/window ... ts.378044/
I'm asking would blurbusters team be interested in doing an input lag test comparing msi mode vs legacy? You have the skills and equipment to do this kind of stuff reliably.
Input lag test: msi mode vs legacy
Re: Input lag test: msi mode vs legacy
If you're talking about the end to end latency test setups, that's a bit of a difficult question, because DPC latency optimization is on the order of tens to hundreds of µs, and the uncertainty in the timing of inputs vs frame intervals(~1ms at 1000fps, ~17ms at 60fps) means you need hundreds to many thousands of samples in order to confidently say overall latency is better or worse by that small a quantity, or what exactly caused the difference.
Honestly, I think I'd need to take a course in statistics before attempting this measurement.
Honestly, I think I'd need to take a course in statistics before attempting this measurement.
Re: Input lag test: msi mode vs legacy
I highly doubt it makes any difference at all. Unless you're getting higher fps from it.
Re: Input lag test: msi mode vs legacy
I got about 5% fps increase atleast in one game. Hard to say. It just feels little bit different now, but I'm not sure if its different in terms of input lag or not.sharknice wrote:I highly doubt it makes any difference at all. Unless you're getting higher fps from it.
- lexlazootin
- Posts: 1251
- Joined: 16 Dec 2014, 02:57
Re: Input lag test: msi mode vs legacy
Typical gaming latency (mouse to screen) should be anywhere between 9 and 15 ms if your using a fast monitor, mouse and whatnot. I've seen hundreds of tweaks online that can "improve" your latency. So that means that users are feeling SUB ms differences in latency, or if we apply all of these latency tricks at the same time we will get -350ms latency
OR it's all bullshit and we are just fooled really easily.
OR it's all bullshit and we are just fooled really easily.
- Chief Blur Buster
- Site Admin
- Posts: 11653
- Joined: 05 Dec 2013, 15:44
- Location: Toronto / Hamilton, Ontario, Canada
- Contact:
Re: Input lag test: msi mode vs legacy
It's hard to say but there are lots of snowball effects from certain tweaks -- things that makes things better and things that make things worse.
I've heard of tweaks that is supposed to improve latency but actually makes it worse. I've also heard of tweaks that improves latency supposedly by only microseconds but actually improved framerates by a benchmark-noticeable 3fps or 4fps. While many others are so insignificant that it's placebo.
Over the long term, we'll make latency tests easier allowing more obscure scenarios to be tested. See what's crapshoot and what's real -- measurable in real lag tests. But many priorities first!
I've heard of tweaks that is supposed to improve latency but actually makes it worse. I've also heard of tweaks that improves latency supposedly by only microseconds but actually improved framerates by a benchmark-noticeable 3fps or 4fps. While many others are so insignificant that it's placebo.
Over the long term, we'll make latency tests easier allowing more obscure scenarios to be tested. See what's crapshoot and what's real -- measurable in real lag tests. But many priorities first!
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter
Forum Rules wrote: 1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
2. Please report rule violations If you see a post that violates forum rules, then report the post.
3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!
Re: Input lag test: msi mode vs legacy
I was also little bit sceptical about msi mode affecting input lag any significant way. When I see several people report that it helps them of course I'm still not fully convinced. Placebo can affect a lot. The tweaks might work for some PCs but not others. The thing is, more and more people start to fiddle with these settings because they read the same reports. If there was some input lag test to clearly prove that it's "snake oil", then people would not waste their time anymore in this matter.
Sounds great! I'm eagerly waiting for all kinds of tests released about input lag in gaming. Keep up the good work!Chief Blur Buster wrote:Over the long term, we'll make latency tests easier allowing more obscure scenarios to be tested. See what's crapshoot and what's real -- measurable in real lag tests. But many priorities first!
Re: Input lag test: msi mode vs legacy
If it's actually getting you higher FPS it probably is getting lower input lag. Directly related to the FPS. For example if you just capped at the same FPS it would be the same input lag.
Re: Input lag test: msi mode vs legacy
This does not reduce input lag.
This affects cpu overhead wich can result in reduced cpu usage, so affects performance of the system.
Many devices nowadays are already working in MSI mode. The hardware and driver must support it.
In legacy mode, each IRQ can be used for several devices and this can create problems for (integrated devices) ethernet and sound, sometimes due to buggy drivers. I know cases like these.
Sound: Popping and crackling noises. DPC is high. MSI mode lowers DPC and solves those issues.
Ethernet: Reduced performance (by alot). Using a dedicated card solves it. No idea if MSI mode does too.
Video: Freezes? Stutters? No idea.
This affects cpu overhead wich can result in reduced cpu usage, so affects performance of the system.
Many devices nowadays are already working in MSI mode. The hardware and driver must support it.
In legacy mode, each IRQ can be used for several devices and this can create problems for (integrated devices) ethernet and sound, sometimes due to buggy drivers. I know cases like these.
Sound: Popping and crackling noises. DPC is high. MSI mode lowers DPC and solves those issues.
Ethernet: Reduced performance (by alot). Using a dedicated card solves it. No idea if MSI mode does too.
Video: Freezes? Stutters? No idea.
- Chief Blur Buster
- Site Admin
- Posts: 11653
- Joined: 05 Dec 2013, 15:44
- Location: Toronto / Hamilton, Ontario, Canada
- Contact:
Re: Input lag test: msi mode vs legacy
So, in other words, it can have a Rube Goldberg effect in reducing input lag.MatrixQW wrote:This does not reduce input lag.
This affects cpu overhead wich can result in reduced cpu usage, so affects performance of the system.
A 3 frames per second improvement in a game, from 100fps to 103fps is equal to a frametime latency improvement of
((1 / 100) - (1 / 103)).
Any framerate increase ... is often something that has enough performance freed up ... to allow it to focus on something else that ... happens to reduce lag.
Keep doing enough good, proper, correct tweaks of different kinds -- and you've increased frame rate about 10%. The great thing is many games has framerate counters that will tell you if the game is running at a higher frame rate after a tweak. Whenever a frame rate number consistently reliably goes upwards with a specific tweak -- then it can be tweaking well done!
There can be side effects of some tweaks. A "higher-but-more-erratic/stuttery" framerate can be worse than a lower consistent frame rate. (A great example is badly-framepaced SLI or CrossFire).
Yet other tweaks will raise those nasty 0.1% worst frametimes. Undiagnosed problems or performance bottlenecks that causes more unexpected freezes & stalls -- are everything that adds new erratic freezes/halts that causes big stutters. Like that weird freeze for a fraction of a second that might happen every exactly 15 seconds. (Was it that virus scanner in the background, or was it a hardware conflict caused by that new device I installed? So much work trying to track causes) Fixing those are often quite desirable.
The problem is the large number of wild goose chases and red herrings we have to keep digging through if you decide to be a hard-core tweaker -- digging through the placebos to finally hit those important tweaks that gives much needed relief to a balky computer system!
Tweaking continues today, with today's games, but sometimes more complicated today thanks to all those weird differences of Win7/8/8.1/10/etc ... FreeSync/GSYNC/VSYNC OFF/etc ... FXAA/MSAA/ACME-AA/KitchenSinkAA ...
The days of simpler good old fashioned DOOM and Quake timedemos of the past, and refining a system until those numbers are maxed out. (Those were simpler days in a way!)
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter
Forum Rules wrote: 1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
2. Please report rule violations If you see a post that violates forum rules, then report the post.
3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!