Gadgetory

welcome back to harbor unbox today we are exploring the impact disabling hyper-threading has on Intel processors and obviously we'll be doing this on models that actually support hyper threading now I've done this in the past and I've done it because it's an interesting test and that's partly why I'm remaking this video however another motivator is the fact that Intel's latest security vulnerability actually impacts hyper-threading Tim's already talked about zombie load and similar micro architectural data sampling vulnerabilities that affect Intel processors in his news corner segment so I'm not going to go over it in too much detail basically there are four MDS attacks that affect Intel processors are the most serious of which has been named a zombie load and is quickly for those of you wondering AMD processors are not impacted at all so none of the four variants are a threat to those using AMD CPUs so yeah that is a relief for AMD I would imagine and especially those using their processors as for Intel users the only way to mitigate or at least minimize these vulnerabilities short term is to disable simultaneous multi-threading or as Intel brands at hyper threading as it stands Microsoft is pushing out OS level updates to address the for MDS vulnerabilities and you'll get those with this month's 19:03 update however this doesn't mitigate the problem entirely for that we need new motherboard BIOS updates and apparently Intel has released the updated microcode to motherboard partners but so far no new biased revisions have been released to the public or at least at the time of making this video so at this point I'm I can't test the real impact these updates will have for Intel processors however I believe we can test a worst case scenario by disabling hyper threading and for older platforms that may not get updated this might end up being the only solution so I've grabbed the core i7 87 okay along with the 77 or okay and I've tested them with a battery of gaming applications with and without hyper threading enabled I didn't bother testing the core I $9.99 hundred K given that it is an eight core part so the missing threads won't really be an issue here particularly for gaming performance though there will still be a negative impact on application performance and that really should fall in line with what we said from the 8700 K and 7700 K if I was afforded a bit more I'm about to fly out for Computex and the testing for this video already took up a few days but given more time I would have liked to have test some dual-core hyper threading enabled CPUs as well as I suspected the impact there will be rather brutal but I do have some old data that we can fall back on and have a look at later in the video anyway getting on with it all testing has been conducted this week using the Windows 10 build 1903 the Intel processors and we tested with ddr4 3200 memory and the GeForce r-tx 28 ETI has been used to reduce GPU bottlenecks though the gaming benchmarks do take place at both Teddy P and 1440p but before we get to the game benchmarks let's start with some application testing okay so first we have some Cinebench our twenty results and looking at the core i7 eighty seven ok we see a 24% reduction in performance with hyper-threading disabled naturally that kind of performance drop won't go unnoticed furthermore in terms of performance we're essentially turning the 87 okay into a 77 okay so that's a fair old downgrade for those that would have bought an 87 hundred K meanwhile the 7700 K becomes 26% slot with hyper-threading disabled and now we have a plain old quad core or Core i5 of that generation so naturally for applications that heavily leveraged all cores disabling SMT or hyper threading has a big impact on performance moving on to some testing with WinRAR and here we see a massive 36% reduction and throughput for the 87 or okay clearly hyper threading works really well for this type of workload likewise we see a massive drop-off for the 77 RK as well a 39 percent reduction in this case so hyper threading is very advantageous for this kind of processing the second last application benchmark I'm going to look at is Corona and here the 87 okay so a 31% performance decrease with hyper threading disabled while the 77 our case or a similar 33 percent drop off again both were obviously significant performance decreases so depending on how much these updates impact hyper threading performance we could see some pretty big moments drop-offs in rendering and encoding workloads then finally the last application I'm going to look at is blender and like cinema jar 20 we're seeing around a 25% drop off for the 87 or okay with hyper threading disabled I say only 25% because we have seen up to a 36% decrease but a 25% D Risa's certainly nothing to take lightly then we have the 7700 K with its fewer cores and it does suffer a little more and here we see a 29 percent performance reduction before we move on a gaming I just want to quickly note some total system power consumption numbers and granted we're not looking at the individual process of consumption here so it is hard to comment on efficiency but you can see that the 87 are ok with hyper-threading disable doesn't really save us that much power in this test we're seeing just a 5 percent reduction in total system consumption the 77 or okay was better leverage for hyper-threading enabled and here we saw an 11 percent reduction which is probably more in line with the performance drop-off that we saw again keep in mind we are looking at total system power consumption anyway time to move on to some gaming benchmarks and first that we have the ternary peer results for Assassin's Creed Odyssey here the 87 ok only sorry 13% reduction for the average framerate and no change the 1% low the 77 okay on the other hand with its fuel costs or a massive reduction for both the average framerate and 1% low result here we saw a 23% drop up for the average framerate and 21% drop up for the 1% low so those with quad-core processors will be impacted significantly more by a reduction in hyper-threading performance now if you're primarily GPU bound the 80 simmer ok does just fine without hyper threading enabled and we see that here at 1440p even with an RT X xx 82 I that said the 77 error case still suffers an 18% reduction in the average framerate performance so again those with quad cores and God forbid a dual-core any reduction hyper threading efficiency is really going to sting battlefield 5 is a very CPU intensive title though for this one we only had time to test the single-player portion of the game and even then we needed to origin accounts just to finish our testing because of that delightful Hardware lockout anyway keeping on the subject we don't really see much of our performance decline here the biggest drop-off was up to 12% we're looking at the 1% low margin for the 77 her ok increasing the resolution or 1440p didn't really help with the 1% low performance for the 77 ok and here we see in both instances disabling hyper-threading does reduce performance again it's not major but I doubt any gamers will welcome a 5 to 12 percent reduction in performance oh wow this is pretty brutal look at the performance drop off 77 okay with hyper-threading disabled in the division - that's similar to the reduction we saw in testing with winrar so here the average frame rate is reduced by 37% and the 1% low result by 38% again pretty brutal stuff I'm sure there's a few core i5 quad core owners watching this thinking hey my PC plays the division - quite well and well I bet it does we're still seeing over 60 FPS at all times but we're also seeing a massive reduction of performance with hyper-threading disabled the performance impact for the six karate 700 K isn't nearly as extreme but even so a 13% dip in 1% low performance won't be appreciated by most gamers moving a 1440p and now with hyper-threading disabled 87 or okay isn't the performance limiting component rather that would be the RT x 28 DTI however we're still seeing a 32% reduction for the 77 okay we're looking at 1% low performance okay so if you're looking at these results a little cross-eyed then let me explain farcry new dawn is a title that plays much better on the 9700 k than it does the 87 okay and at the 97 okay allows for around 120 FPS on average at 1080p I bring this up because when we disable hyper threading on the 87 okay it matches the 9700 K in this test despite having two less cause so basically six cores and six threads is more efficient this total and six cores with 12 threads and we do see this quite a bit in games when testing SMT that said the 7700 K with its fewer cause doesn't suffer the same issue with hyper threading enabled though it was still a smidgen faster with it disabled so at least for this game running without hyper threading is pretty much a non-issue and in fact it's likely going to be beneficial though probably not if you have a dual-core moving to 1440p and we see that the core count isn't an issue here it's actually hyper threading that's slowing down the 7700 okay and 87 huh this is similar to what we saw at 1080p though the effect seems to be more amplified here moving on we have hitman - and here we say that disabling hyper threading has no real impact on performance for the 87 or okay however once again we see the impact for the 77 okay is quite devastating the average frame rate dropped by 18% but far worse than that was the massive almost 30% reduction in 1% low performance granted we're still seeing over 60 FPS at all times but for those chasing big frame rates this kind of performance here is brutal even at 1440p the 7700 K is hit hard with hyper-threading disabled as we're still seeing a 25% reduction in 1% low performance next up we have a rage too and here the 87 okay so almost no declining performance with hyper-threading disabled the 77 okay on the other hand here we see that although the average frame rate is virtually unchanged the 1% low performance drops about 20% so that's obviously a significant reduction however once we increase the resolution to 1440p that is enough to remove the cpu as the performance limiting component at least when looking at the hyper threading law 77 or okay so here disabling hyper threading has no impact on performance we found in the past that shadow of the Tomb Raider is a very demanding title and we're getting a pretty clear reminder of that here the 8700 K saw a 10 to 12 percent performance drop off with hyper-threading disabled while the 7700 K saw a 24 percent drop off though the one percent low margins were similar to what we witnessed with the 87 or okay in either case disabling SMT does have a big impact on performance in this title even at 1440p the effect is quite significant at least for the 77 RK the 87 era case still saw a small performance to climb but it was nothing like the 20% drop off the 7700 case suffered last up we have some World War Z results using the low-level Vulkan API and here we see the game runs just fine with four cores so neither CPU suffers with hyper-threading disabled that being the case we naturally see a similar thing at 1440p both CPUs are able to extract the maximum performance from the RT X xx atti okay so we've now got a pretty good idea of how intel's 4 & 6 core cpus perform with hyper-threading enabled and then disabled but to quickly summarize for those of you who just skipped over the results in court heavy applications the performance reduction is typically anywhere from 25 to 35 percent the impact on gaming performance does vary quite a bit and this will depend on things such as the game used that's probably important but there are other factors such as the resolution for quality settings and then of course your hardware configuration so your CPU and GPU depending on what CPU use and what GPU use you could be bottlenecked more so by one or the other but I would say for today's games the six core all parts there right on the edge for the most part get away without hyper-threading in today's games but we did see instances where the 1% performance did suffer at times so yeah if you're chasing really high frame rates or a higher refresh rate gamer then the performance drop from losing hyper-threading on a 6 core part will certainly be noticed then for those of you with an eight-course 16 thread part like the 9900 K the impact on gaming performance will be virtually non-existent I haven't tested that in this video but we have looked at that sort of thing in the past and pretty much all games will run really well on an 8 core processor I've seen that when comparing the 9700 K and the 9900 K you know games previously that said application performance we'll see quite a noticeable drop even with the 99er ok that is assuming the application uses all at cause quite heavily and you should expect to see around a 25 to 35 percent performance drop with SMT support and disabled but again that will depend on the application and then of course all of this is well it's it's much more impactful when you have a lower end CPU so we showed how hyper-threading performs with the 7700 ok when it's enabled and disabled and this will be even more significant for dual core parts with hyper-threading which are sort of right on the edge now they're getting away they're there they're reasonable make-do solutions but without hyper threading they're pretty unusable for now we don't know exactly how much of an impact we for MDS mitigations will have a performance at least for these windows-based applications and games but we know there will be a performance hit and we know it will be felt most where hyper threading has the biggest impact for onyx has tested the mitigations on linux and the performance impact ranges from negligible to massive again we don't yet know what the impact will be for Windows users but we're confident there will be a reduction performance of some sort particularly in the scenarios just showing that saw substantial performance downgrades with hyper threading disabled products also found that Intel systems are now around 16% slower out of the box than they were before they Spector meltdown for Shadow and zombie load mitigations meanwhile AMD has only seen a 3 percent performance drop they also went on to say that the mitigation impact is enough to draw the core i7 87 came much closer to the Rison 720 700 X and the core I 9 79 80 XE - the threat Ripper 2990 WX now I'm sure there'll be some Intel fans that shrug their shoulders and say who cares the 98 is still faster than 2,700 X Man well yeah I guess that's fair enough but you'd have to be happy with paying almost twice as much for a CPU that seems to continually become slower over time Korra 999 okay owners aside this is almost certainly going to be disastrous news for those using a dual-core or quad-core Intel CPU that supports hyper threading this includes all core i3 processors from Clarkdale 2 kb lake core i7s up 2 kb lake as well as kb lake and coffee-like pentium processors three years ago now I looked at how the core i3 6100 performed with and without hyper threading and the gaming performance results without hyper threading were quite horrendous in some titles the CPU was choked so badly the performance dropped off by more than 50% the crazy thing is three years ago using a modest GPU of the time the performance hit in almost all tested games was significant even gamers using a gtx 1050 try with ultra quality settings at 1080p we're going to notice big performance drop-offs so any impact on hyper threading is effectiveness for these CPUs is going to be bad news for anyone using them I guess for those of you using older hardware the best performance option is just to never update your BIOS again given the impact it's likely going to have on performance it's probably worth the security risk as I said at the beginning of the video the BIOS updates that enable the mitigations on Windows 10 aren't yet available so we can't test the true performance impact but we can at least show you a worst case scenario so presumably it will end up somewhere in the middle there this does put us in a bit of a pickle though as I was just about to start updating all my CPU performance numbers I know other media outlets have started doing that in anticipation of Zen twos release so yeah that's made things a little bit difficult because if I invest days upon days of updating all the numbers for the Intel processors and then these BIOS updates do come out during or after Computex those numbers are pretty well invalid and I'd have to go do all the testing again which I don't want to do so I've just put that on pause for now I'll see what the deal is after Computex and then hopefully by then we will have am the new bias revisions with the updated micro code from Intel anyway I thought it was interesting to see where hyper threading makes the biggest difference on these processes so hopefully you've found the results interesting and yeah hopefully in the not-too-distant future will actually be able to see what the impact of the MDS vulnerabilities is for Intel processors on Windows 10 anyway that is going to do it for this one if you did enjoy the video please hit the like button that'd be much appreciated I said you can subscribe for more content you guys know how that works but hit the alarm bell because just hitting subscribe is kind of pointless these days and you can support us on patreon you get access to our monthly live streams our discord chat and all that good stuff anyway thank you very much for watching I'm your host Steve and I'll see you again next time you