But we have just received very unofficial, very secretive source modified bios for this card that gives us on limited power.

You guys, I got new merge.

It's available now.

Probably dot com slash tasty sense.

Guys slip up hoodies.

We got tri blend.

We got a new logo.

I digress.

Since 2012 it's a rest.

A little.

He has been asking for that anyway.

Way kinds of stuff.

Zip hoodies, Beanies, polos don't take my word for it is obviously I can't do this ad.

So just looking description below When you guys find the link.

Thanks way tested this card on air.

We set the fan curve to max 100% because normally, if you set the fan curve to Max, it uses a few watts.

And if you're on the edge of stability, five watch or 10 watts to the fans could actually reduce your overall stability.

We actually saw on air this thing pull 420 watts, and it's 4 20 watts guys for 20 watts.

It's Max draw while over clocked telling us that we could theoretically And that was that, like 65% of its power dolls, what it said.

So you do the math on that?

How far this allow us to go?

So the question is, can a software mud like this allow us to actually forgo any sort of shut mining or any sort of saw during or any of that stuff?

It's a bit of a Band Aid.

I mean, there's still gonna be things in place, but the software at least, isn't gonna be getting involved.

And what I really looted, too.

When I said Spoiler alert, I think I said it in the stake.

Remember, we were able to pretty much Max what we were getting with her ex see ultra, with the modified bios on chilled water on air on the further 13 card.

Now I want to go straight to the King pin card and or the Titan have talked about the fact that the Titan cards I want to use in the gold build and that's still happening.

I'm just being lazy because I've got to take the seed out of my work station, which it's my workstation.

So until I'm ready to finish that I can't have that system down.

Order of events, obviously.

But I thought, well, I still would love love to beat Steve score without using titans.

And the next logical jump up is gonna be the for the Win three card, which has a very overbuilt V R M.

Delivery or power delivery.

Very, very stable.

And like I said, we saw 2100 and 60 megahertz on the core on this guy on air, something we couldn't even come close to achieving on the exit ultras, even when we had the air conditioner plumbing up to it.

So now I'm gonna take my home group once again my little home brew stuff.

That's why quick disconnects really come in handy.

We're gonna hook this up.

You're gonna chill it as well as our CPU, which we're not going to.

Ellen Ellen doing the CPU today.

Now we're doing just CPI test.

We know how far the CPU will get.

We're doing GPS.

We know far the CPI will get 5.9 gigahertz and I want to see now how much more headroom we have on water with this Probably won't be a lot boat again.

Every bit counts in this competition.

So let's go ahead and get this thing ready for water.

That is one beef cooler right there.

If you don't ask me, it's about the cooler.

Well, this one.

Check that out.

Every time we do this, I feel like we improved the design slightly.

Maybe not so much aesthetically.

But at least this time we've got the conversation sort of contained in here, wrapped up with, uh, electrical tape and lots of paper towels that we can avoid drips because the fittings, anything metal that touches the air is where all the conversation's gonna form.

It is a form of the plastic.

Yeah, some does form around the block on the bottom, But believe it or not, that's so close to the heat source that that conversation isn't really a problem there with the chilled water Cindy on the backside, we never really see any sort of conversation form on the back side.

So, um, scratch the back.

All right?

It's gonna get this over to the test bench and see if we can go any higher than we did on air.

If we can, then we'll get ourselves another one of these cards.

hopefully, if they're available, and then we will go for it again in another video.

So another proof of concept video?

Yeah.

This is how this works, you proof of concept, like we did with just the CP.

We see how far we can go with that.

Theoretically.

Say, yeah, we should able to win and then lose and then go over here to our Jeep.

You would be like, Well, theoretically, if this is higher independently with a higher CPU, then theoretically we should win the problems.

I know Steve's probably working on Ellen to or dry ice right now with that 1980 xy with his Titans, which means the moment we win, he's gonna win again, making all this work for nothing.

Way go.

Yeah.

So here's what I do.

Then I'm gonna go ahead and load our 5.9 profile, but that I'm gonna pull everything back from there.

So let's do 50 I mean, the nice thing about using an icy ex card those we can at least see what the different temperatures are on the board.

Ironically, RVR ems came down like 10 degrees versus when the cooler was on there with fan running, and it's over clocked right now, but also loaded our previous settings plus 1 75 on the core, plus 1 11 80 on the memory.

So I guess we're going to see if it passes that test.

Okay, so our graphic scores a 78 82 these are the same settings, though upstairs that it was crashing on air.

No, we don't get that on camera, but we were crashing.

If we went above 1000 megahertz on the Ram were at 11.

80 on here.

If we went past 1 70 on the core, we were at 1 75 Right now, now five Maker heard increments doesn't happen.

It's like 12 megahertz or 15 megahertz increments on this.

But we're gonna go ahead now and start pushing it and seeing if this card on water with unlimited power limit will allow us to actually get a higher score.

Okay, so 79 16.

We are getting higher scores, and I have really not started pushing the clocks too much, So I think I'm gonna push memory before I pushed core a little bit because it's one of those things where depending on the test.

Sometimes memory really does give you a bigger increase in GPU.

We've talked about that, and I know it's Stevens talked about that gamers nexus before where?

Memory clock.

I can give you a bigger increase sometimes in GP course.

So if you pull the core back but clock speed up on the memory, you see bigger gains.

So we have to sort of whac a mole this and kind of figure out one before we go to the other and back and forth.

Same thing with CPU testing when it comes to the C P over clocks, we saw much.

We got our Max CPU scores by getting memory and memory timings tighter, and there's still a lot of times on there.

We have to optimize.

There's some pretty bad matches there, but we saw bigger gains from doing that than course speed.

Remember, we really did l into the first time.

We got all the way up to, what, 5.7 before we even matched our 5.4 on dry ice when the memory was left at like its stock settings.

So it all has to work the system 79 42 very minimal gain but stable nonetheless.

At 12 25 of the memory clocks 50.

And if we look at our temperature, the hottest Arcor got with 16 c.

But we know the first, like state of throttling starts at, like 28 c, so we've got a lot of headroom.

He'll push this card 439.

What's so?

Which is, like 65% of what the power limit will allow us to pull.

Yeah, the pump tied that could have been wide, that it crashed because the jeep you got and in the sea if you shut down, which, as we're just sitting here like would it was happening Shut down.

Probably thermal shut down, but the pump is turning for sure, but it's making that whining sound because it can't move any water.

J.

I think I found the problem.

Yeah, Phil found it.

This wasn't clicked in all the way.

No, no.

Let's turn it on.

And that should come shooting down.

Thank God is before empty the loop.

This is why we have Phil's fills.

Our important, um, we know that if we could match our dry ice scores on water on the sea few that Ellen T will bring our scores up, remember?

Wait to make up 100 points for Steven.

We have from Stephen.

We have to do it before he goes with Ellen to or dry ice.

Our only hope at that point is that he pulls a new mistake and blows his shit up like I did, but that probably won't happen.

So my second hope is that his hair accidentally dips in the dry ice.

It breaks off.

All right.

So 79.

32 again on the graphic score.

We're not improving anymore with extra memory, so I'm gonna go back to 12.

50.

Now, we're gonna push the core up a little bit.

Our limited power of the GPU is pushing the cloud gate father than you can handle.

So you have to our set it with zero slider.

If the slider goes up in both sliders, go up.

The math is all wrong.

I can't do that.

Look at this.

It is passing right now.

8000 on the dot.

By increasing the CPU to five point twos.

We're just gonna see how high we can get the CPU now.

Just for the hell of it Because again, remember, the whole point of this is testing one card to theoretically see if they can match and or beat the other card at much higher seep you if we can match it with lower CPU that we know what we're really high in the CPU.

We win.

Theoretically, we also Theoretically, I thought we were gonna win last time.

Shut up.

Felt shut up, Phil Ashes.

One more test I can do right now.

So we determined that obviously increasing the CPU speed right now isn't helping us.

For some reason, we're actually seeing a diminishing return right there, which we know isn't the case when we use Ellen to.

So something else going on right there.

I'm not gonna worry about that too much, but I test something we know 2010 or 2 10 crashed with the memories super high.

I'm gonna drop the memory back down to, like, not 12,000 1200 said 1300 or 12.

75?

Actually, no, we're looking for their 11.

50 and I want to see if the core will survive with the memory reduced.

And if that works and then score goes up that everything I said about leapfrogging, it's wrong, you know, I think you guys have figured out by now in this competition, we're just guessing.

In fact, I think one of the extreme over Clockers is even like, Yeah, J did pretty good for not knowing anything about what he's doing.

I'll take that as a compliment.

I really do.

Can take that as an insult.

He's absolutely right.

Well, that didn't work.

And 20 to 35 we can't seem to get stable at this current temperature.

So the next step is gonna be for us, too.

Drop the temperatures even further.

We're going to somehow get in Ln to pot hooked up to this thing, and we are going Thio go forward.

I think we're gonna for 2300 at least not today.

That's another video later on.

I'll give you guys some time to sort of simmer down a bit over this competition.

You guys think we're milking this, but you do understand we are learning.

We're learning through failure and we were educating through learnings of our failures.

Hey, guys, time to go.

Thanks watching today's video, but we did learn Indeed, that won the further 13 card with the modified bios that removes the power limit entirely.

The air cooler never saw any harder than 54.

See when the fans were 100% when we over clocked it to 2100 and 45 is what we were seeing on air.

And ah, that's a hell of an air cooler.

And then once you add colder temps, we were able to push that all the way up to 20 to 20 which now tells us we go even colder.

This even get it around zero, see under load we could potentially hit 2050 2075.

Who knows?

We're experimenting.

We're taking guys along for the ride, which is the point.

The video's thanks for watching.

And I got to clean this mess up because now we have a dedicated area for this that we we came in on an off day and actually cleaned sort of Justin, turn the camera around.

We can actually chip the ship.

I just painted 120 barrels yesterday.

That's the whole point of these videos were experimenting.

Can you guys alone for the right, so be it.