Placeholder Image

字幕列表 影片播放

  • so clearly there's a lot of construction going on outside of our old set.

  • We're building our new one, so we're kind of in the shooting position right now where it's noisy, no matter what we do.

  • So I apologize for that.

  • But until our set is done and they're done, we get to deal with the noise and being semi homeless.

  • But that's okay.

  • So here we're in here.

  • We're gonna talk today about the new vulnerabilities that are going on with Intel processors, but we're not gonna talk so much about the problem.

  • We're talking about the potential effects of the problem on your day to day.

  • And then we're gonna kind of test whether or not disabling hyper threading Onley affects our performance like 7% which is what Intel is trying to make you believe.

  • Theme newbridge PCU cables from cable mon fuse to separate PC I'II cables together with a single close comb.

  • For an incredibly clean look, you can configure years however you like using their easy to use configuration er and from now until May 27th you could take 15% off your order using the coupon code.

  • May 15 This coupon code is good for absolutely everything on the cable mod global store.

  • So if you've been looking at pimp you rigged with cable mont gear, now is the time to learn more.

  • Click the link in the description below.

  • Okay, So, like I said, we're not going to talk a whole lot about the problem itself.

  • If you guys want to learn more about this new vulnerability, it's called Zombie Load, and it's specifically affecting like predictive computing.

  • And so that is going to specifically be dealing with intel.

  • See pews and hyper threading, and you can drill down into that Mauritz.

  • It's called MDS, but you can drill down into that more if you go and do some searches.

  • But what I can tell you the problem regarding this and zombie load is the fact that it could be affecting clouds.

  • How much information have you sent to the cloud?

  • How much data have you ever just connecting to the Internet?

  • Your information is in the cloud packets are sent packets or received, and so we have server farms and data farms and all that that could be potentially affected by this, So that's a whole different level enterprise solution that's gonna have to be figured out in terms of patching and EMC, you are micro code updates.

  • So I looked at this way.

  • I'm like, what is the impact?

  • And before you, my average viewer is going to be general use computing, maybe some video rendering or live streaming and gaming.

  • So today we're gonna do some testing on both hyper threading, enabled and disabled on our 8700 K.

  • Not our highest NCP is that we have because we want to kind of try and keep it sort of realistic to the real world.

  • We have 2080 on here, not a 20 t I.

  • So this is more like a 10 80 attended t I and I guess, technically not really lower than that.

  • But whatever.

  • What we want to see here is what the impact would be like if you wanted to take the extreme measure of protecting yourself by disabling hyper threading.

  • Now, like I stated earlier, it's a really convoluted, complex problem that I don't know what the solution is gonna be a long term on that don't think Intel does either.

  • But they made a statement that it's okay you can disable hyper threading and you'll lose like 7% performance.

  • Well, it's kind of funny, cause Apple said the same thing.

  • Only they quoted 40% of a performance loss.

  • So I'm kind of curious is what that number really is.

  • We're gonna test that today and see what the real impact is gonna be, too.

  • My average here.

  • All right.

  • So, as you can see, they're 8700 k.

  • It's still running a five gigahertz.

  • That's only like a 300 megahertz over clock on all course.

  • So this isn't out of the box performance.

  • But like I said, what we're testing here is what the bottom line?

  • What did we lose performance wise on my rig by disabling hyper threading.

  • So, obviously, your results may vary depending on your motherboard.

  • If you have a hyper threading CPU, you more than likely even on a like a aby Siri's motherboard would have the ability to disable hyper threading so you could test this on your own.

  • And it's only one switch yet.

  • The flip inside the bios, of course, each bios is gonna be different.

  • So let's do this.

  • I'm gonna go ahead and ah, let me do a three d mark run.

  • First there is why I like three.

  • D.

  • Mark is the fact that it gives us both a physics test, which is CPU, and it gives us a GP test, which is obviously high frame rates, which seep.

  • You can hold that back.

  • We see it all the time, especially when we're doing like our our high, eh?

  • No.

  • Ellen to cooling benchmark runs that the sea few does indeed affect the graphic score.

  • So that is what we're gonna test for sort of our baseline there.

  • Um, probably a Senate bench will do a Senate bench run as well, and we're going to do time spikes.

  • Dream will do both.

  • Screw it.

  • We'll do times by extreme.

  • And we'll do regular times by We'll just kind of do some tests and then we'll just disable hyper threading and do the same death.

  • So here's our baseline.

  • The very first test that we did everything is on hyper threading five gigahertz and all that.

  • No over clock being applied to our graphics card.

  • That's important.

  • There's no afterburner running any of that.

  • So we've got a graphic score 50 to 23 a CPU score of 36 45 for a combined score of 49.

  • 04 Now, we obviously know when we turn on hyper threading it's gonna go down.

  • But anyway, there's our base Mark our baseline 49 04 with two individual scores.

  • So let's see what Cinemark gives us or cinnamon Cinemark bench, whatever.

  • Imagine if they just combine the two tests.

  • Three benches.

  • OK, so far, baseline tests.

  • Right now, I'm not gonna go through the scores.

  • Really.

  • I'm just gonna show you which ones we did, because what we're going to do here is we're gonna put them in a graft format.

  • We care more about how much performance we lost, rather than what the actual score is that we started with because when we end up with this kind of the point here.

  • So we did Senate bench ar 15 with hyper threading on all 12 cores.

  • Six course, 12 threats doing their thing at five gigahertz.

  • We did times by extreme and regular times by because they both respond very differently to workload because it's this is technically a four K test versus a 10 Ap Test or 14.

  • 40 tests.

  • We did it in a teepee.

  • Metro critics in again no other clocks to the GPU.

  • So we want to see what happens to our average frame rate because this is a high frame rate game, 208.9 frames for seconds.

  • We want to see what happens to that.

  • Once we cut the that workload, potentially in by 1/3 I'm hyper threading.

  • People think like doubles your performance.

  • It doesn't, um, same thing in 14 40.

  • And we also did shadow of the Tomb Raider in both 10 80 p and 14 40 mostly because this is a very CPU central game as well.

  • So we also did a premiere of Our Kingpin video.

  • So we did a premier render with the 9900 K.

  • We're doing this on Phil's test bench, and then we're gonna re render it or edit rate whatever.

  • Yeah, it's a test bench today, and then Phyllis gonna also re render it with the exact same settings and same export settings with hyper threading disabled.

  • So those are our test, so let's go ahead and disable hyper threading now in your particular motherboard, it's going to vary.

  • We haven't a snooze motherboard here This is a strict Z 3 70 R Z to 70.

  • Hey, they started working again.

  • Apparently that union break is over.

  • So anyway, I digress.

  • Um, let's go ahead and go into the BIOS here.

  • No, I'm just mashing Delete some of my B f ate whatever.

  • Look at your motherboard manual.

  • I can't tell you how to get into bios on your motherboard, but I can tell you in a suit you can mash, delete, and then what?

  • You're going to get as many that looks like this.

  • It's an RG board.

  • It's It's red.

  • It might be blue, depending on which motherboard you have.

  • Um, so you just go over here to advanced CPU configuration, Hyper threading, disabled F 10 save.

  • So now we just turned off hyper threading entirely.

  • So now we have a six corps six thread CPU.

  • Do you think we're gonna lose more than 7 to 9%?

  • All right, so we ran all the tests and stuff.

  • So the question is, now, how does it all stack up?

  • Well, if you take a look at these charts that were kind of showing you right now, you'll see that hyper threading for gaming really doesn't make that much of a difference.

  • Believe it or not, it's a subject I haven't really did.

  • Devon did David Dive, Dive Dove in Dover.

  • I have not done a lot of testing with how hyper threading directly relates to gaming performance.

  • But we do know is si, pues and stuff become or utilized in games and multi threaded.

  • It could indeed have an impact now.

  • What you gonna notice with these percentages is that this is just the one task.

  • It's the one game running and the comparison of hyper threading on hyper threading off, you know, half the logical course, but we saw as low as 0% loss like 14 40 Metro, all the way up to about what 15.6 air 16.8% loss in Premier Fun.

  • Fact about premier, though, is if you're running a CPU, that's got an I g.

  • Peeling it.

  • We showed in the past on how you could enable hardware acceleration where the Eye GP built into the CPU.

  • I can handle the H dr six foreign coating on the hardware level, so it doesn't really matter at that point because want to enable that we got the exact same score or a time pretty much about 11 minutes with hyper threading on and off.

  • But that's because it's offloading it to a whole separate die anyway.

  • An interesting thing, though, is with times by We saw the exact same percentage drop with hyper threading on and off in both extreme and non extreme.

  • But what's interesting is that we did say, see a bigger reduction in normal versus extreme because of the lower resolution, higher frame rates, which, of course, the CPU could start to hold it back.

  • But we saw in Time Spy.

  • We actually saw the graphic score increase ever so slightly, like 60 points or something like that, probably still within the margin of error.

  • But we did, of course, see the CP drop on both of the physics test, cause that is 100%.

  • See, few seems to be roughly around where Intel was saying the performance loss would be, you know, 7% to 9% nowhere near the 40% that Apple is saying the performance lost could be.

  • But we don't know what those testing perimeters were.

  • We don't have any details about what Rick they use with CPU they used.

  • We think Apple could be wanting to make until look a little bit worse than maybe it really is, because who knows what their future holds in terms of proprietary hardware?

  • They could also be trying to cause a little bit of pressure for Intel to do something more quickly.

  • We don't know.

  • So at the end of the day, I'm much more comfortable now telling people like, Look, if you're building a system just for gaming, it makes sense to save some money and go with, like a 96 100 care in 9700 K versus a 99 100 K.

  • But again, this is a six core CPU where a four core would obviously show us more of a gap.

  • But yeah, so this kind of really changed my way of thinking on the way.

  • I might build some of my future systems when we start thinking budget and even high end, because gaming systems just for the example of gaming, do not get hit that hard with hyper threading, even on the multi threaded stuff like far cry and shatter the Tomb Raider and all that sort of stuff.

  • But again.

  • It really depends on what you're doing to computer.

  • You have tons of programs running a background, tests going if you got V EMS, all that sort of stuff that obviously you want as much of the horse powers you can possibly get.

  • All right, guys.

  • Thanks for watching.

  • I just wanted to kind of put this out there.

  • If you were curious about what performance you lose if you turned off your hyper threading on your intel CPU.

  • Not as much as we initially thought, which really, truly surprised me.

  • So our guys, if you think this is something we should investigate further in terms of hyper threading and games and how they correlate, let's know what titles you think we should test.

  • There's so many titles out there, it's impossible for us to really determine what the most universal titles are.

  • So this is where you guys help us.

  • Try and determine that.

  • All right, we're gonna go now.

  • Thanks for watching.

  • We'll see you in the next one.

  • And I got a set to build because I'm done with this crap.

  • So we're gonna do Senate bench now.

  • Things time to start.

so clearly there's a lot of construction going on outside of our old set.

字幕與單字

單字即點即查 點擊單字可以查詢單字解釋

B1 中級

超線程是否有必要?ZombieLoad影響測試 (Is Hyper-Threading Even Necessary? ZombieLoad Impact Testing)

  • 2 0
    林宜悉 發佈於 2021 年 01 月 14 日
影片單字