字幕表 動画を再生する 英語字幕をプリント What's up, guys? Just two cents here. And um, yeah, we may or may not be seeing new graphics cards in the near or not near future, and that's why this test bench exists. I built this test bench years ago, and I have seen a lot of revisions and changes. But the only thing that really gets changed on here often is the graphics card. Obviously, now we were running the 8700 k on here pretty much ever since the eighth Jen launched. The reason why I've been doing that is because I've always heard and even preached myself that the more cores you add, the more threads you add, the less performance you actually get in games somewhat negligible, depending on the graphics card in the game to a lot of different variables. So it's admittedly, why have left the 8700 count here for so long, even though I used to run x 2 99 and X 99 on here forever? Actually, no, I never ran extra night. That's what we do today. I am preparing for the new launches by putting on my ex to 99 dark from E V. g et an amazing over clocking motherboard. Extremely overbuilt V. R m. And the reason for that is I'm finally taking my 79 80 xy out of the box, which I've never used for any project because this was such an unpractical CPU. So I figured minds will test that theory. Are we going to lose performance by going from 8700 K to the 79 80 exit? That's why we're using this board because of e. R. M. Can actually take it without exploding. But I just realized I probably have two small the power supply now. But that's OK m s. I brings ultrathin gaming to life with their new GS 65 stealth, the 4.9 millimeter basil 144 hertz 10 80 p panel in video graphics and eighth Gen Intel CP use Make the new GS 65 self a true powerhouse with a small footprint. Find out more and had a pre order by following the link in the description below, you know? So I'm gonna be running to benchmarks here to give me sort of a baseline, because what tends to happen from one platform to the next and, ah, special when you're dealing with Intel Intel. Like Intel, the AMG is obviously a whole different story. But we tend to see whatever happens in a couple of benchmarks kind of follow suit all the way through with the few variables and exceptions. But it's also very important to point out a couple of things. One we're not just switching to. More cores were also switching tomb or memory channels. The mainstream si pues on Intel on Lee used dual channel single single or dual, whereas the ex platform of the Extreme Platform can use anywhere between 123 or four channels. So we are gonna be setting this up in Quad Channel. Gimme a s'more memory bandwidth. We'll be setting up obviously more memory. Because of that, this is gonna be having 32 gigs versus 16 gigs over here. I don't think that's gonna have a huge impact on things, but we could always pull out to six of Rams to compare that, And that's another video coming up. We're testing the 2018 version of Does Memories memory capacity matter with games. I did that back in like 2013 but things have obviously changed incident. We're gonna redo our baseline test right here. This is Metro Last Light. I still really like this because it's definitely a good seat or a GPU test and we're looking for here Is bottle necking with the CPR? I hate that word. It's not really bottling. We're just trying to get numbers to compare settings wise, though we are running 1920 by 10 80. Quality is very high texture filtering that 60 next motion blur normal test elation very high. The reason we're not running four K right now is we want to see if at the higher frame rates that we're gonna get with 10. 80 obviously offloads a lot of that duty to the CPU to handle all those frames. We're going to see whether or not we get a big difference there. We've tested into the ground what happens with four K where CPU is less important with four K than GPU. That's why we won't be testing it today. If you want to see the four k results and just add four k to your search bar and you'll find something so yes. Okay, so here's our result. We ran this test what, four times four times, and we're pretty much within plus or minus two or three. F. P. S highest we saw was a 1 93 Lowest was a 1 89.53 average frame rate. A couple things I wanna talk about. Hero Quick 1 87 100 k is running with single cores on the A Seuss bios, which means it's not adding more frequency, but it is making all the course run at the same frequency. It's a setting. I always leave on number two. I thought this seemed a little bit low, and I went and looked at my original 10 80 t I review none of this exact card, the Masai card, and you could see in 10 80 with the same settings we were getting. 220 f ps. And then I looked at this video and realized I was running X 99 at that time. But then I'm senior, looking at trying to figure out what was running to maybe make me be losing approximately was about 10% 15% close to 15%. Whatever the number is, Phil, come annotate with the number was, but it dawned on me we weren't dealing with the specters and meltdown issue back then. And we have significantly seen a performance reduction in in total max F. P s at really high frame rates like this after Specter and meltdown was patched. Perfect example. Skunk Works is still running Windows eight and a 69 50 x 10 core 20 thread, and it is faster than my D frame running a 79 100 exits over clocked on water with to tighten ex peas. Little peas, which are faster Jeep use than my regular tight necks past kill cards on skunk works. Skunk works pizza in all F. P s testing. It beats it in three d mark and it beats it in PC mark specifically because of the specter and meltdown patching. So that just shows you how much power or how much F. P s. You actually lost with that. So these are our new bass lines right now, at least for this test 1 89.53 and we're gonna go ahead and run now. Far cry five used. It's synthetic and see if that goes the only thing we did the GP, though we didn't over clock it. We just allowed power limit to go higher so it wouldn't do any sort of throttling with power limitation. And then we up the fan curve to make it a little bit more aggressive because we want consistent results. So yeah, apparently, Now we have to patch far cry five, This is hold. Spend a week away and everything has to be updated. Far cry five in ultra settings. 123 average F p s twice in a row. So I'm going with that number 123. So now we've gotta swap over stuff and do it again. Way ram out because I'm shallow. I wanted to use the burnt Ram. I just can't do it. I can't do it. And I've got this beautiful set of white and black Corsair dominator which will just look so good on this test Bench blue. This this guy has got good moment of truth exception not handle what 79 80 xy and all of its 18 cores and 36 threads of glory. Obviously, before we do our benchmarks of the GPU, I want to see what happens with our seep. You make sure things running fine. And just see all of these cores. Just look at that. Oh, shit. I 84 instantly. 85. 86. 4318. Sentiment, score. Now, you know I can't help myself. And to keep things fair, uh, that was running 4.7 single cores on the 8700 k, which is as much as I wanted to push it with water cooling and attempts we were seeing. So I've been playing around now for about an hour with my over clock to figure out where I want to dial it in. And this is where I decided 4.5 gigahertz. It's actually doing it at 1.2 volts. Yeah, Golden chip probably been. So now that I'm convinced that I am good and I'm stable now, we're going to go ahead and run the same benchmarks again. That's what all this was for The benchmarks. Did I now skew results? Moving forward? This may seem an awful lot like the does court count affect your games video? I did not that long ago, but that was just an 8700 K. We had a six core 12 thread and I were disabled cores and hyper threading with the same chip. Now we're going from six cores to 18 cores with 30 to 3 36 threads, which is a significant gap there. And I forgot to turn on Emma's afterburner. Yes, surprisingly, here's the thing. Our average frame rate came up to 200.71 Um, here's what's interesting about that. We ended up putting, raising and reason why Emma's afterburners open right here is we raised our power limit to 120 percents. We're allowing it to draw 20% Maur power TVP than it's set to from invidious base. And we also set a higher fan curve. This is the exact same settings that I ran on the 8700 k. What did not happen over there while running our metro? Last light was we never saw the power limiter. Come on. Remember how jumps up and down says power, power power that tells you when you're hitting power limit. We saw that during this test we saw we were hitting power limit, which tells us that we're it looks like we were potentially I don't see a bottle necking bottom like is not the right term. But we were limiting the amount of F ps 8700 could handle while we sent a shit ton of Effie s like this to it. So I'm gonna run this test one more time because you can't just run a test one time and get you know, your result, But I want to see if it does it again. If we start seeing power kick on, if if it does, then we'll know that the CPU is actually allowing the graphics card to just run balls to the wall as hard as it possibly can. Power limit did not come on as much this time. There was once once more. At least we saw it a couple times. 200.4. Another 200. So Okay, Metro, last light. At least that title benefits from more course now going from 6 to 18. I mean, we can't what one f p s for not even an FBS prick. Or so our last test here in far cry five with ultra settings. We had ah consistent back to back runs of 123 f p s average. So average MPs here was 1 29 and increase of six. I don't know what percentage that is. Filkin again could put that up on the screen. But the reason we're using far cry five is even the synthetic built in benchmark is still using a lot more seep you than other synthetic benchmarks. So when we were running metro, the CPU just said it like 39 c 40 seeds. It wasn't really doing anything, but you'll notice now the CPU temp has gone up significantly in the GP, usage drops down to more like 80 85. Well, sometimes it the nineties and come back down doesn't get us hot. Which tells us the CPU is doing so much more. But because we're doing far correct, I feel like I should be talking like this. We're gonna do a test one more time. We're gonna make sure the results were the same. We don't want to mislead folk down Rome pass and you know, definitively the CPI was actually gave us our improvement here, but I die grass. It's besides the point. Now I'm expecting probably a pretty consistent results. But if you look at the temperature right here is certainly doing something with CPU. All right, so that's time we got 1 27 versus the 1 29 All right? So that time we got a 125 average. So we've gone down to every time. 1 29 27 25 If you look, the Max from MPs is the same number of frames renders a little bit less on the second test, but our minimum MPs dropped from 99 to 95. What we think is actually happening here is that the GP is getting warmer and we're starting to see the Geep. You see, states or P states actually called PCs and GPU starting to come down a little bit. So this one's probably not one that we can actually draw a conclusion from because yeah, So this one really didn't show much of a difference. Yeah, 1 23 is still lower than 1 25 No one's gonna notice this f PS anyway. So anyway, the whole point is video. Like I said, new test bench. I was curious of whether or not we cause myself a problem. We test the new graphics cards are inevitably coming. I think I'm just gonna roll with it. I'm fine. I mean, there's plenty people doing reviews on all sorts of platforms. And this is just another one to the add to the mix. So, guys, thanks for watching today's video. If you have any ideas for videos we should do in the future makes you hit us up on Twitter or does comment down below. That has always. Guys, we will see you in the next one, you know. Uh huh. Room. Look at these. Uh huh.