madVR - high quality video renderer (GPU assisted)

[SamuriHL]

I don't want "madvr pro". I want "madvr rtx". ;)

[*MidnightWatcher*]

Doubtful. That's way too expensive for your average user, IMHO. And it will really motivate the crackers.

No it isn't. A decent video card costs much more than that.

[videoh]

False analogy. Hardware is not software and their cost drivers are completely different.

[*MidnightWatcher*]

False analogy. Hardware is not software and their cost drivers are completely different.

It wasn't an analogy. Someone willing to spend $500 on a video card for an HTPC can buy madVR for $100.

[ryrynz]

I think that basically leaving things more or less as is and probably selling envy hardware at consumer friendly price points would be the way to go.. Like if it's an extra few hundy on top of a standard RTX or something then I'd be keen to just buy the hardware. Can't imagine the software is going to evolve in future anywhere near the levels of previous development.. I expect 1.0 is probably now very close to done in terms of a feature set compared to his initial plans. Thems the breaks I guess.

[brad86]

Shame MadVR updates have pretty much stopped. By the time Envy releases, TV processing would have advanced enough via software/hardware to make a dedicated hardware solution like Envy costing thousands of Pounds to be completely mute. Outside of maybe a few dozen buyers, it's dead-on-arrival, and it boggles my mind that it's still being developed as thas must realise this.

Thankfully, other media players are now doing HDR to SDR conversions, only lacking behind when it comes to tone mapping, which hopefully they can at least be on par with MadVR's pretty soon. I see MPV surpassing it in the near future.

[ryrynz]

Remember updates haven't stopped, maybe there will be a new version THIS year.. but for those of us that already have TVs capable of good HDR tonemapping there's probably not a lot left on the table to provide us software users over what we already have with the current build aside from quality of life improvements.

There's money on the table, the enthusiast and professional AVR crowd reasonably well aware of madVR. Of course as with any business failure to adapt to the market means the possibility of becoming basically null and void as you say.
I expect it's all in at this point to strike while the iron is kind of hot, sales and our own voices will very likely guide the future here along with whatever madshi wants/is able to do, will be interesting to read his next major post.

[NM20]

Sorry if I have missed this, but Is there a guide to setup the Helix 3D Fix for Nvidia with MPC HC and Mad VR to get my 3D .MKV working again?

[jmonier]

I think that basically leaving things more or less as is and probably selling envy hardware at consumer friendly price points would be the way to go.. Like if it's an extra few hundy on top of a standard RTX or something then I'd be keen to just buy the hardware. Can't imagine the software is going to evolve in future anywhere near the levels of previous development.. I expect 1.0 is probably now very close to done in terms of a feature set compared to his initial plans. Thems the breaks I guess.

The one thing about the Envy that's not clear to many people is that for it to work with input devices such as streaming devices, BluRay players, etc., it has to handle HDCP protection from input to output. That requires a special HDMI input card that is not allowed to be sold separately and full HDCP certification for the entire box.

So, you can't just buy the hardware.

[videoh]

It wasn't an analogy. Someone willing to spend $500 on a video card for an HTPC can buy madVR for $100. Wrong again. Sure, they "can" spend $100 bucks but will they? I spent beaucoup bucks on a 2080RTX but wouldn't spend $100 for MadVR. If the author thought he could get people to cough that up he would have done so long ago. And the crackers would break it and start selling it for $50 or give it away free.

[jkauff]

Wrong again. Sure, they "can" spend $100 bucks but will they? I spent beaucoup bucks on a 2080RTX but wouldn't spend $100 for MadVR. If the author thought he could get people to cough that up he would have done so long ago. And the crackers would break it and start selling it for $50 or give it away free.
Agreed, not everyone would be willing to spend $100. However, there are plenty of us here who would like to reward madshi appropriately for all that he's given us for free over the years. Maybe it should be a combination of purchase price and optional donation.

[videoh]

Sure, and I would pay a reasonable amount too.

At $100 people start looking for cracks. The crackers even go after my $15 stuff (which is actually just a few dollars when you consider that one donation gets you all my protected apps). They do it for prestige. Cracking madVR would be very prestigious! My defense is to have
somewhat tough to break code and to release new builds very often. They can't keep up with me and cracks after the first don't carry the same prestige. :) I also play games with build numbers to keep them a little confused.

[el Filou]

your dvd preset for 24 should be never true because even a 23p decoded DVD has a deint FPS of 48.Except in the cases where you manually force film mode/force disable deinterlacing (=weave)/enable IVTC upstream of madVR.

[huhn]

Let's look at an example. The following script selects one of 4 profiles, depending on the source dimensions and the frame rate after deinterlacing. I think the script is pretty much self explaining:

the profile doesn't care if the source is deint or not using deintfps simply means assume it is deinterlace and report the value after deint is applied.

that's why the official "guide" doesn't used FPS check if not needed doesn't start with elseif and doesn't end with elseif.
if (srcWidth <= 1050) and (srcHeight <= 768) and (deintFps < 31) "SD 24fps"
else if (srcWidth <= 1050) and (srcHeight <= 768) "SD 60fps"
else if (deintFps < 31) "HD 24fps"
else "HD 60fps"

and we have this too:
fps, deintFps, bitDepth source frame rate, framerate after deinterlacing, bitdepth

[*MidnightWatcher*]

Wrong again. Sure, they "can" spend $100 bucks but will they? I spent beaucoup bucks on a 2080RTX but wouldn't spend $100 for MadVR. If the author thought he could get people to cough that up he would have done so long ago. And the crackers would break it and start selling it for $50 or give it away free.
No, there is nothing objectively wrong about it. Someone willing to spend $500 on a video card for an HTPC, not to mention all the other required components, can buy madVR for $100. It's not a matter of wallet size, but rather whether they choose to buy or not to buy.

I'd personally pay double that for madVR if a madVR-friendly version of KODI v18 or higher were made available for download. IMHO, madVR is worth it.

[videoh]

Someone willing to spend $500 on a video card for an HTPC ... can buy madVR for $100. It's not a matter of wallet size, but rather whether they choose to buy or not to buy. You're very slow on the uptake, but thank you for agreeing with me.

[el Filou]

the profile doesn't care if the source is deint or not using deintfps simply means assume it is deinterlace and report the value after deint is applied.Sorry but that's incorrect, deintFps represents the actual framerate before presenting, not a supposed framerate if deinterlacing was applied to the source. I use deintFps in my script too, with LFR and HFR profiles, and with interlaced sources it dynamically switches profiles depending on whether deinterlacing is actually active or not.

[*MidnightWatcher*]

You're very slow on the uptake, but thank you for agreeing with me.

You said that $100 was "way too expensive for your average user" who can spend $500 on a video card alone. I'm not agreeing with you at all, but nice try. $100 would not be too expensive.

[huhn]

Sorry but that's incorrect, deintFps represents the actual framerate before presenting, not a supposed framerate if deinterlacing was applied to the source. I use deintFps in my script too, with LFR and HFR profiles, and with interlaced sources it dynamically switches profiles depending on whether deinterlacing is actually active or not.

ok found why it doesn't work correctly for me it's progressive DVD that make it fail. so this is wrong:

your dvd preset for 24 should be never true because even a 23p decoded DVD has a deint FPS of 48.

while it will "never" be true is "correct" but the reason is not because the deintFPS is 48 it's because the reported frame rate is 29 and this only applies to progressive decoded soft telecine dvds.

[videoh]

I'm not agreeing with you at all Send madshi a PM informing him that he is leaving so much revenue untapped. You could be a star!

[*MidnightWatcher*]

He's well aware, and it's not entirely up to him I'm sure.

[videoh]

He's well aware, and it's not entirely up to him I'm sure. Who is it up to?

[SamuriHL]

He has business partners that he has to answer to now.

[Stereodude]

So are these new Ampere based Nvidia RTX cards going to be absolute monsters for MadVR? The new $500 RTX 3070 has ~50% more TFLOPs than even a RTX 2080Ti (20.4 vs. 13.4). The $1500 RTX 3090 has 35.7 TFLOPs! 2.66x as much as the former king (RTX 2080Ti).

[chros]

And just imagine that they haven't started to sell the Envy officially with the old, already depricated (!) GPUs!

[SamuriHL]

I'm banking on the 3080 being an absolute monster for madvr, yes. I was considering the 3090 but I don't think the price/performance curve is quite there given how fast the 3080 is and the price. 699 50% over the 2080ti is insane. And I'm on board with that.

[Stereodude]

I think moving to a RTX 3070 from a GTX 1080 will be a nice upgrade for me in my HTPC. More than 2x as many TFLOPs.

Though getting another 50% TFLOPs performance boost from the 3070 for another $200 is tempting too.

[NoTechi]

Since 3D is important to me without having to deal with some outdated drivers and I am currently very satisfied with my RX 5700 XT, my next upgrade, if any, will be with AMD Big Navi, as my current configuration is already more than enough for my use case (madvr for 1080p, 4k movies and 1080p 50Hz live TV upscaled to 4k).

NoTechi

[chros]

Yes, AMD cards will be also interesting to wait for, hopefully they'll drop the nvidia prices :D

[Siso]

Yes, AMD cards will be also interesting to wait for, hopefully they'll drop the nvidia prices :D

I wouldn't count on that :D

[SamuriHL]

I wouldn't, either. nVidia planned this out VERY carefully. If AMD manages to whack the 3080, nVidia will immediately release a 3080ti with slightly higher clock speed and 20gb of ram for 999. AMD is screwed this time around I'm afraid.

[huhn]

you don't know what the AMD card will be able to do.

consoles have about 12 TF @ 52 CU a 5700XT has 10-13 TFlops @ 40 CU. so without massive IPS gains that's it.

i complained hard about the 2000 series pricing and that you get similar performance for the price as with the old generation which was terrible... that doesn't seem to be the case this time not at all so we are final back to competition without competition.

madshi "fixed" 3D for new driver many month ago he simply never released the build.

[SamuriHL]

You're right, I don't know what AMD will be able to do. But I have SERIOUS doubts they're going to beat the 3090. And if they don't, if they only beat the 3080, then nVidia is going to come back with a 3080ti with 20 gigs of ram for a grand and call it good. So, we'll see if AMD can pull a rabbit out of its, uh, hat but I have my doubts. Historically, it doesn't bode well for them.

[Asmodian]

I would not have believed the 3090 as announced if I heard it as a rumor. It is close to the rumors but doubling up the fp32 is pretty surprising. It also looks great for madVR, but honestly my 2080 Ti is already sufficient for basically anything I want to do. I suppose the 3090 will be useful for HQ deblocking of 4K YouTube streams. I wonder what clocks can be sustained after a shunt mod + water cooling. ;)

I would need a lot more info about AMD's next big GPU to get hyped but a monster GPU from AMD is definitely possible. However, AMD releasing something crazier than a 3090 within the next year would be really surprising.

[SamuriHL]

Oh I'm not suggesting it's impossible, but, let's face facts. AMD has been playing catch up for quite some time and nVidia went balls out on this one. If this was a little speed bump over 2080ti I definitely agree they could have caught and surpassed them. But you see what that 3090 looks like. I just don't see any realistic way they can beat that. The 3080? Maybe. I'm not counting them out on that one. But nVidia knocked this one out of the park.

[huhn]

they don't have to beat the 3090.
they only need to release a card that works (GG at this point) and is cheaper then the nvidia cards by a "huge" margin.

you don't need the biggest best card you need the best performance for the buck the market share of the 2080 ti is tiny compared to the 2060.

and that's the problem AMD is only even with a huge processing advantage to the 2000 series. now they need to beat the 3000 series with there old 7 nm process that wasn't able to "beat" 2000. that really unlikely. they need either huge IPC gains or huge dies so silicon price. is TSMC cheaper then Samsung. it's complicated. what ever it's worth Samsung 8 nm seems to be pretty "bad" in terms of clock rate the 3000 looks great even with these "low" clock rates so yeah whatever.

beware our current knowable about the 3000 series and it's FP32 performance is based on the fact that one core can do 2 FP32 operation instead of 1 like turing. proper testing is need to see if this works as advertised and doesn't behave like "hyperthreading" or more like bullldozer.

[SamuriHL]

Yes, it'll be nice to see the reviews. And I didn't suggest AMD NEEDED to beat 3090. I'm simply saying that if they don't, nVidia has more levers they can pull to beat them down. Including price points. They're already better than last gen, but, I suspect that if AMD came in with something comparable at 100 bucks cheaper, nVidia would respond one way or another. Who wins in that scenario? Come on, I know you get this. :D I PRAY AMD is successful! ;)

[ryrynz]

A lot of people will be able to pick up some great 20 and 10 series cards for excellent madVR performance with the 30 series release.. A big win for everyone here really.

that doesn't seem to be the case this time not at all so we are final back to competition without competition.


Pioneering raytracing and tensor tech the obvious reasons, now it's just evolutions of what's already been established now and all helped by 7nm shrink too.

Yes, AMD cards will be also interesting to wait for, hopefully they'll drop the nvidia prices :D

With the performance increase offered this time round the entire stack has to drop prices considerably, I doubt AMD can offer anything really to decrease prices further at any price point.

If AMD manages to whack the 3080

LOL, not even next year.. They're a full gen and a half behind.

[huhn]

Pioneering raytracing and tensor tech the obvious reasons, now it's just evolutions of what's already been established now and all helped by 7nm shrink too.
which is barely used... with pathtracing results of sub 60 FPS no wonder.

BTW. they use a nearly for sure inferior 10nm samsung optimised for nvidia named 8nm not the TSMC 7nm.
but if this is what keeps prices down then please keep doing that.

the cores have been massive changed again there is no word about int 32 so it's still rumoured that the INT 32 cores can be re purpose as FP32 cores as you pleases.

tensor cores where already "old" with turing.

[RXP]

I'm always interested in reducing power consumption. Is it right to think that the new range of cards will be able to run identical settings on at lower power consumption?

[huhn]

with current information it's a yes.

nvidia talks about 1.9 times(context as "always" with nvidia is not really given) it looks more like.

a 220 watt GPU is faster then a 250 watt so 1.2-1.5.

[chros]

beware our current knowable about the 3000 series and it's FP32 performance is based on the fact that one core can do 2 FP32 operation instead of 1 like turing. proper testing is need to see if this works as advertised and doesn't behave like "hyperthreading" or more like bullldozer.
Ohhh, that's why the cuda cores in the leak the other day are exactly the half that nvidia states! Thank for this!
Am I the only one who feel bullshit in the air? :) It's like amd would say that a 6 cores SMT CPU is 12 cores. And if that's the case then why nvidia felt the need to lie like this, instead of put out good marketing for "hyperthreading"? :)
I guess we will have to wait for benchmarks...

[huhn]

the more you look at it the clearer it get's it 3090 has 2 set of ~5k cuda cores in turing one set was INT32 and the other normal CUDA FP32 in ampare they seem to be able to switch between INT32 and FP32 i have no clue if all core can do that or only half.

we will know when official data is available with the INT32 performance.

[nevcairiel]

The "CUDA cores" were always a bit of theoretical thing, its not like you can access them individually. But if it can handle twice as many operations per tick, does that really conflict with saying it has twice as many cores?

[huhn]

if you get the information of FP32x2 and 10K cores do you now have to expect the performance of 20k turing cuda cores?

beware these are not on the same spec sheet!

[SirMaster]

The CUDA core definition has not changed from Turing to Ampere.

But the scenarios where you can utilize 100% of CUDA cores are sometimes limited as the streaming multiprocessor has been reworked to do FP+FP, FP+INT, etc., rather than FP *or* INT.

[NoTechi]

madshi "fixed" 3D for new driver many month ago he simply never released the build.

This is interesting because I assumed that it would not work without nvidia 3D driver support. Then I look forward to a new madvr 3D with nvidia re-enabled version :D

NoTechi

[chros]

The "CUDA cores" were always a bit of theoretical thing, its not like you can access them individually. But if it can handle twice as many operations per tick, does that really conflict with saying it has twice as many cores?
I think it does:
It's like amd would say that a 6 cores SMT CPU is 12 cores.

[nevcairiel]

I think it does:

No, thats hardly the same thing, because SMT doesn't actually mean it can run twice the instructions. Its an apples to oranges comparison entirely.
The CUDA core layout was reportedly changed so that it actually has hardware to run two FP32 computations, instead of just one.

[chros]

Hm, cheers.