Gadgetory

there are a lot of theories about how the lower-end Intel chipsets would impact the eighth generation CPU performance for the current line of Intel processors and there's decent reason for that because Intel did a few strange things they launched the case queue Intel CPUs 8700 K and so forth alongside non-case Q locked CPUs like the i5 8400 with only zseries motherboards this is something we've talked about in the past but basically came down to Intel pulling in the launch cycle pushing out the higher end stuff as fast as possible and then you end up with AI 5 non K parts that have to be coupled with expensive motherboards that frankly make no sense being coupled with an i5 8400 so today we're looking at the final launch of the rest of the eighth generation motherboards before that this video is brought to you by Thermaltake and the view 71 enclosure the view 71 is a full tower case that's capable of fitting three video cards and most configurations it's also one of the better cooling cases in our recent case testing bench lineup the view 71 has hinged a tempered glass doors on either side that make it easy to open and show off and it comes with at least one rain fan though you can get the RGB version if you prefer learn more at the link in the description below let's bring it way back to fourth quarter of last year so if you remember we tested the i5 8400 with two different memory speeds 2666 megahertz and 3200 megahertz the point of that was because we knew that these boards later 8:37 8b 360 and B 360 over here would be limited in their memory support they don't go to 32 so we tested that 2666 as well we collected some data very quick recap some of the games had a 5 to 7 percent difference I think one of the higher ones was 8 percent different with the slower memory of course being behind and for synthetics there wasn't a huge difference at all because it really depends on what you're doing what kind of application workload it is so we already tested 2666 Bridge 3200 I will save you some time if you're wondering how it performs in games the scaling is the same now as it was then doesn't matter if you use a Z series board B Series board whatever have thirty-two hundred megahertz on the Z 370 board versus 2666 on the Z 370 it's the same scaling in terms of percent gains and loss as 2666 on this or this or this versus 3,200 on Z 370 so hopefully if you're wondering about that and about how gaming performance scales we already answered it months ago it hasn't changed i double-checked the performance scaling is functionally identical so we got that out of the way next thing to look at them we did a big content piece talking about multi-core enhancement on motherboards some boards have things like MCE which will basically pre overclock the CPU so we did that content piece talked about it about a week ago if you missed it go watch it but we're revisiting some of those topics with the same testing math it's very simple using Cinebench to establish a baseline score and then using Cinebench to establish a baseline frequency on our multi-threaded and single threaded workloads what we're doing is comparing does the h3 70 or in this case the B 360 motherboard actually impact performance versus Z 370 if all other factors are identical that's what we're looking at right now and then we'll get into some of the blender performance for real-world production workloads and things like that start with something really simple this is Cinebench scores the MSI B 360 mortar motherboard had no MCE options to speak of so performance testing was simplified we tested with XMP and with stock settings and that was it no other BIOS changes necessary this chart compiles a lot of the Cinebench results from our previous MCE testing which you can find already uploaded on the channel any of these numbers confuse you go check that video before you ask why they are what they are the MSI B 360 mortarboard posted a score of 1416 with XMP on or 1407 with complete stock settings again these are all using the same kit of memory single threaded performance or was 196 on both settings and compared to other boards that we previously tested we're about 10 to 20 points lower than multi-threaded P performance with single-threaded performance about the same as for why that happens it's nothing to do with the chipset here's where we recap some other old cons n the difference we're seeing here 10 to 20 points a bit more a bit less sometimes between the B 360 and Z 370 boards it's not the chipset what it comes down to is memory timings and sub timings of motherboards take a lot of liberties here the single biggest differentiator in motherboard performance because motherboards do actually impact performance it does matter outside of thins even like vrm temperatures and fear I'm overheating overclocking outside of all that stuff which is pretty obvious that impacts it the next biggest change is memory profiling some boards have pretty bad memory profiles out of the box some of them support one kit of memory better than another some might not support a particular kind of memory at all and you end up with just loose timings and generally poor performance what's happening here is MSI's got a brand new board that at this point of filming has not yet been released it presently has a pre-release BIOS on it and they have weak memory support it's fine it's doing ok but it's 10 to 20 ish plus or minus 5 points behind the other center bench scores and so the reason for that is if you dig through the memory time it's quite obvious basically comes down to things like TR FC or TR efi settings like that the lower level stuff we're increasing by a couple hundred for our FC or something for example will change your performance somewhat significantly so what it comes down to is this over time if you've ever scrolled through a motherboard page the support page for BIOS downloads and we can maybe show one on screen if you ever looked at one of those almost every note section for every BIOS release will say improves memory support that's what it comes so this will eventually have all these updates that say improve memory support and it will equalize and score so basically what I'm trying to get across here is that when you see a performance difference in something like Cinebench which is a relatively very simple Oh review application for one type of workload when you see performance differences like we're seeing here a couple double digits of points it's coming down to the memory support it's not because this be 360 chipset is implicitly worse at doing Cinebench than Z 317 now there are very clear disadvantages to the lower end Intel chipsets it's not out of the box performance though it's things like overclocking support which is functionally absent on the lower end chipsets things like the amount of high-speed IO lanes or HS io lanes that you get HS io lanes are something that the motherboard vendors have control over what they assign them to higher end chipsets in the past from Intel have gotten more of those for the Z series than the H and B series for example so that kind of influences how many different attached IO devices you can have SSDs for example or nvme SSDs or whatever those are the primary differences so yeah the Cinebench performance difference we saw is actually not that large particularly considering Cinebench is rather large margin for error and variance in its own execution and it comes down to you memory sub timings so then next thing frequencies we can look at those and just double check that it's running at the frequency it should be and before we get into the frequency chart I'll also note the bus speed or base clock was about ninety nine point six for all tests with this board and about ninety nine point five to a hundred on the other boards as well as for frequencies they're more or less the same which is why we can point at memory timings as the primary differentiator in performance the B 360 mortar averages 42 94 megahertz core frequency with XMP on which is within reason for the usual $42.99 megahertz single thread frequency was 44 64 to 4500 which is also within reason and variants there's no meaningful frequency deficit here the differences again derived from other settings in the board things that will probably be improved with time moving on to some real testing our blender scene rendering results show how little difference there really is this is also going back to frequency more than anything else with version 2.7 9 and using the GN monkey had test the 8400 CPU moving off the 8700 we were just testing Li ke skew the 8400 with 32 hundred megahertz memory tested on the ultra gaming Z 370 completed its render and forty two point eight minutes dropping to 26 66 megahertz memory on the B 360 platform rendered in forty-two point four minutes dropping the 26 66 megahertz memory on the z 370 platform rendered in forty three point two minutes these differences are from change in frequency not from board change the splash screen tests rendered next in thirty five point five minutes on thirty-two hundred megahertz memory with the 8400 on a Z 370 board 36 point 2 minutes with the B 360 board at twenty sixty six for the memory and 36.1 minutes well within margin of error with 2666 on z 370 in other words the motherboard chipset when leveling the fields of stock settings has no meaningful impact on performance and to bring it all back to the beginning if you're curious about relative gaming performance of 2666 versus 3200 there's a bit there but your money is far better spent on things like a better cpu so no it's not always a lot of money to go from 26 to 32 hundred megahertz but it depends a lot on the region you're in some of you have sent us screenshots of memory prices in your countries and I've seen the price difference in 3200 to 2600 in different regions sometimes it's completely outrageous you absolutely should not be paying a lot of money to make that jump because we're talking about a couple percentage points and you're also talking about 15 plus dollars more you might as well buy a better other component GPU or CPU and go with twenty six hundred megahertz memory cuz you're gonna get more there than you will from the memory for gain for these CPUs though anyway the i5 8400 as we've said from the start the sort of the stance we've had on it remains the same the original stance was it's not a bad CPU but also you shouldn't buy it with Z 370 it's just not how it was meant to be it's unfortunate Intel did not ship these boards sooner but that's the way it happened so now that we're in the future with these boards again we would recommend the i-5 8400 as still a pretty good CPU but that it be paired with B and H Series motherboards unless you have a very specific reason and you would know better than I would to buy Z 370 maybe one of those reasons is you want to buy it now with a cheap CPU and upgrade later not something I'd recommend but there are completely valid reasons to do it so that would make sense but generally speaking we'd recommend the non K CPUs with these motherboards they're fine the performance difference the impact from performance from the chipset change is functionally zero outside of HS i/o and things that would actually make sense from changing the chipset as noted earlier so if you're worried about if the motherboard if is e 370 is better and air quotes then B 360 HD 70 strictly on grounds of being a different chipset and for no other reason the answer is no it is not inherently better the better is derived from things like a higher supported memory frequency on the XIII sounding boards this caps 2626 66 and from things like again HS i/o and generally just kind of higher quality boards and the overclocking department because these don't do that either we're not in any meaningful way so that's what it comes down to we'd recommend the non K CPUs with these boards basically is what I'm saying because there's functionally no performance difference what performance difference exists comes from the limited memory tuning for boards that just came out pretty standard and if you run all the Cinebench tests again in a couple months more updates on this board they'll probably do better and blender just didn't care at all so once you get out of synthetics you get into real applications it's all kind of a wash anyway you'll run into other bottlenecks first most likely so that's it that's it for this one very simple very straightforward stuff as always subscribe for more of this we have a couple of cool content pieces coming up especially on well the CPU that this came from and otherwise I'll PAX East this weekend at a panel a pcmr panel I think bit wit Kyle will be there with me along with Corsair rep and I think an AMD rep will also be on the panel so it'll just kind of find a thing and we'll film it for you if you're not gonna be there otherwise stop by go to store it on cameras nexus net to pick up a mod mat or one of our 2-tone hoodies I'll see you all next time