Gadgetory

p'nay sees a quick and easy review of the AMD r7 3800 X CPU with the primary comparison being against andis own r7 3700 X that we already reviewed the two CPUs are identical in every physical way but the only difference is being frequency boosting so the 3800 X goes to 4 point 5 gigahertz single core allegedly and has a base of 3.9 gigahertz while the 3700 X boosts to 4.4 gigahertz single core by the box and as a 3.6 gigahertz base the 3700 X should sell on average for $330 u.s. MSRP with the 3800 X at 400 MSRP this $70 Delta is major while the frequency difference really isn't today we'll be looking into whether it's worth it to spend $70 more on the same part with a higher stock frequency before that this video is brought to you by US and the gamers Nexus toolkit on store documents axis net our brand new toolkit just launched and contains 10 custom made drivers for video card disassembly reap hasting and tear downs the 8 core tools are made of high-quality chromium vanadium alloy steel that's built for long service life and resistance to wear during use the other two tools are carbon steel hex heads that were custom ground down for capacitor clearance on video cards all the tools are easily mounted to a pegboard or stored in the GN made tool bag for easy transport learn more at the link in the description below this will be similar to our 3600 X review or really just focusing on the intra and the comparison 3,800 X versus 3700 X all the other cpus will still be on the chart but if you want the comparison of and these new 8 core cpus versus a and these previous cpus reverses intel cpus check our 3700 x review for a specific discussion on that the charts have everything in them today we're just gonna breeze through them though and focus on the two comparisons for and these eight core 16 thread 3000 series cpus and that's the main focus so should be pretty quick one thing here this is developing to be sort of like the discussion of the 1800 X CPU which was a poor decision in general to purchase have $500 versus at the time of 1700 for 330 which you could then overclock on every single CPU a few minutes to equate 1800 extra ones it's looking sort of like that but a bit less extreme on the price differences and this there's no longer an overclocking requirement so another comparison be the 2700 X versus the 2,700 2,600 X versus 2600 there's that tax for the extra letter you get some of them have a cooler difference to you but we ignore that difference because for the most part we think you should buy a separate cooler so we don't really factor that into the conclusions but overall it's the same vein except now both of them have have an X at the end of it so that's what we're looking at today let's let's go through some numbers we'll go through the benchmarks and talk conclusions at the end a quick frequency check will help us out first so we ran this frequency check before digging into the 3800 x starting with the max frequency line plotted out in Cinebench our 20 we see that the multi-threaded workload shown at the beginning of this chart ends up maxing at about 40 to 25 megahertz on the fastest cores during the all core workload single threaded performance averages closer to 40 for seventy-five megahertz 25 megahertz below the advertised speed and hits 40 500 megahertz about a dozen times during the test blender was another good frequency test logging frequency during a blender workload we measured the all core frequency at 4200 megahertz with some deviations to 41 seventy-five megahertz average across the run overall it seems reasonable to expect about a forty 200 megahertz all core frequency with comparable cool in two hours or about 44 seventy-five megahertz single core with a some spikes to 4500 megahertz and keep in mind that temperatures really effects these frequency numbers with just precision boost to not PBO just the stock settings and that's something we explored in our frequency thermal response chart in a previous content piece next we're going into the game benchmarks and the production benchmarks and for those we have the 3800 x in most of the charts we've got overclocked numbers but not all of them because ultimately an overclocked 3800 acts and 3700 X for the same frequency are functionally the same chip any difference you're seeing is mostly within round around variants all the rise in 3000 CPUs fall fairly close to one another with SMT enabled in the total war campaign test from 155 point 2 FPS average for the stock 3600 to about 160 five FPS average the stock 3,800 X did show a 3.4 percent improvement over the stock 3700 X here but overclocking the 3700 X brings it closer the difference is down to frequency and the 3800 X does boost higher beyond the 4.3 gigahertz all core overclocked managed on the 3700 X and so it makes sense that it places ahead although it's not much this is the biggest improvement that we'll see out of the tests following next up is 1440p the campaign benchmark is less prone than some others to GPU limitation but even so the 3700 X and 3800 X scored almost the same here at one 57.8 FBS average for the 3700 X and 150 9.4 FPS average for the 3,800 X with 1% in point one percent low is actually falling below the 3700 axes results and within reasonable run to run variants an overclocked correct those lows but didn't improve the average FPS of the 3800 acts beyond margin of error in the total war battle benchmark at 1080p the advantage for the 3800 X is smaller still just 1.5 percent over the 3700 X and average FPS stock versus stock overclocking the two CPUs to 4.3 gigahertz put them on about the same level as they should be equal frequency equal core count equal thread count f1 2018 is next the high average FPS results of f1 usually make performance gaps between CP is visible where they wouldn't be in other games but even comparing the raw averages of the 3700 acts at 277 point 8 fps and the 3800 acts at 279 point 3 FPS is barely a difference at all that's far less than 1% improvement for the more expensive CPU with overclocking on both chips yielding similarly minor uplift 1440p is next for the same game the rule of thumb is that differences are even harder to spot at 1440p than at 1080 with CPUs with the averages pushed down and leveled out by the GPU limitations that normally emerge all for average FPS results for the 3700 ecstasy and 3,800 ex stock and OC fall within a range of less than 2 FPS from each other civilization 6 is measured in turn times and is very reliable for CPU performance testing for this one the r7 3800 x stock cpu finishes in 30 3.1 seconds per turn on average which plays to just be the 33 second results of a 4.3 gigahertz overclock and within variance of a thirty three point three second results only 3,700 acts at four point three gigahertz these are all within roughly the same area there within variants run to run the 3700 ax stock CPU at thirty three point seven seconds which is outside of the run to run variance is still not particularly important in the real world Assassin's Creed origins at 1080p positions the 3800 acts at 119 FPS average tied with the 3700 X with some run to run variants showing better 0.1% lows although this is within testing variance there is no meaningful improvement at 1080p between these CPUs 1440p placed the 3800 acts at 112 FPS average tied again with u 3700 X the overclocked 3700 ax is minimally ahead and it's worth in run two run variants once again posting no difference for the 3,800 x @ 1080p GTA 5 positions the r7 3800 x stock cpu at 109 FPS average which has it tied with the 3700 acts at 4.3 gigahertz and functionally tied about 0.6% ahead with the r7 3700 x dock cpu there's no meaningful lap lift here and either CPU would be fine but the 3700 ax would save money that could be put towards something more important like a good GPU the same thing happens at 1440p where the 3800 ax ends up about tied with the 3700 X only 0.2 FPS difference on average with the overclocked 3,700 X not providing any meaningful op lift let's move on and shadow of the Tomb Raider at 1080p that 3800 X doc CPU ended up at 144 FPS average with an overclock with an error at 143 the 3700 X stock CPU performed at 142 FPS average which illustrates an improvement of about 1.2 percent stock the stock hitman 2 is up next for this one tested verse at 1080p the 3,800 x stock CPU operated an average FPS of about 120 which is tied with the overclocked 3,800 x result the 3700 AK stock CB ran at 118 FPS average allowing the 3,800 acts an improvement of 1.2 percent and that's not worth it at all 1440p is satisfying here and that it produces basically the same result as the previous 1080p charts we observed the 3,800 CPU at 1:17 white six FPS average 3700 X doc CBO 117 FPS average at over Fox for both at around 118 and within run to run variants of each other Photoshop is the best demonstration of how the 3800 acts OC and 3800 X scale relative to each other aside from some gaming tests and this benchmark Photoshop has always demonstrated strong favor towards higher frequencies and has minimally favored at higher core account as we always say this is the best illustrated by highlighting the tie between the 9900 K and the 9700 K both at 5.1 gigahertz overclocked where the thread difference has provided nearly zero value the 3,800 x stock CPU scores 1024 points out performing these overclock 3800 X by 20 points like we've seen with other rise in CPUs this is because the single core boosting frequency is higher than an all-court override frequency and so the 3800 X does better when stock then with an all-court overclock at a lower overall value the 3700 X stock CPU is bested marginally here but only marginally it's a difference of about 0.6% 7 zip compression positions the AMD are seven 3800 x stock CPU at 75,000 MIPS or million instructions per second which has it ahead of the r7 3700 X stock CPU by 1.6 percent the sad thing is that this is a rather large gap compared to the other tests but is still generally meaningless the compression in seven zip positions the 3800 acts at ninety seven thousand MIPS to the 3700 axes 95 thousand seven hundred Maps these two are once again functionally the same and so there's no real reason to buy the 3800 x over the 3700 acts in this workload in blender and rendering our in-house monkey heads test workload really heavy workload with a lot of different transparencies and types of materials applied the r7 3800 acts completed the render in eighteen point six minutes putting it roughly tied with the r7 3700 acts at eighteen point eight minutes this improvement is 1% the overclocked 3700 X at 4.3 gigahertz ends up at 18 point two minutes or three point two percent shorter time than the stock 3700 X so there's not much room for improvement with overclocking here overclocking got us barely anything and is almost within error range of the 3700 XOC result or practically is anyway and that's true for a lot of these other benchmarks as well the GN logo render had the r7 3100 X finishing in twenty two point eight minutes which is again only marginally faster than the r7 3700 X stock CPU there's functionally no difference here although blender likes frequency a bit it prefers threads first and foremost you'd be better off just running a big cooler on the 3700 X to get similar performance uplift Adobe Premiere is next starting first with the 1080p 60 h.264 render we observed the 3800 X completing the render in 3.8 minutes when stock tying the 3700 X exactly the 3800 X overclocked completed the render in 3.7 minutes placing it within error of the test passes for reference the 3700 X overclocked also exhibited this behavior placing with an error of the stock 3,700 acts CPU this is an instance where the boosting behavior is minimally equal in how beneficial it is to an overclocked especially because premiere doesn't remain fully pin to 100% load per core and fluctuates in its load level / core allowing cores to boost higher for brief periods in this test there is definitely no advantage to the 3800 X or at least not one that we can find with our 4k 6 th t-64 render using similar settings to our youtube uploads the 3800 acts have finished the render in 11.1 minutes and completed the render 6 seconds faster than 3700 X in other words they're basically the same again overclocking is again minimally beneficial and not worth it especially if running at voltages above 24/7 safe voltages like we are v-ray like blender is a threat intensive renderer on CPUs in this benchmark the r7 3800 X finishes in functionally the same time as the 3700 X posting only a 0.6 second improvement between the 3800 acts of the 3700 X this is within test variance anyway so there's no gains here and there not only in substantial they're imperceptible overclocking gets us at 1.2 seconds faster completion time on the 3800 X at 4.3 gigahertz which again isn't worth it at all so that's a these are seven 3,800 x over all the differences are definitely not worth it you should save the $70 if you want one of these two CPUs it should be the 3700 X and this is no longer even a situation of overclock to equate the performance it's it's run out of the box and you're basically there there's max maybe 3.5% something like that difference in one of the games and that game in particular does have some run to run variance that's wider than the rest of the games and then we're seeing one to two percent differences in a lot of the other titles and production workloads it doesn't seem to overclock any difference it differently we still hit a wall at 4.3 gigahertz and so at the end of the day you should just buy the 3700 X between the two we don't necessarily recommend the 3700 X it's not a bad CPU it does fine but it is much more embattled than the flanks so the 3600 is pretty easy to recommend it's it's one of the it is we think the best in class at its price right now for a modern generation CPU now gets a bit different when you start talking about rising 2000 series chips freaking of the 20s over hard for about the same price for cheaper you have two more cores which might matter in some workloads not others but you have 3,600 modern generation pretty easy recommend 3900 x4 what it does is pretty easy to recommend doesn't always win but the 900k still kind of wins the game in scenarios they're just clock for clock it is still in the lead and that matters but those 2 CPUs from name the 3900 X and it's targeted work workloads and 3600 and pretty much everything easy to recommend 3700 X it still falls in that we're not quite so sure middle ground we don't typically like to recommend middle-of-the-road components because we think that the argument is much stronger on either end of it where if you can't afford the more expensive one it might be worth going for the cheaper one and then just spending that difference on a better video card instead which will likely matter more it has used cases of course if you do need the extra 4 threads on the 37 or 38 series and you do stuff like blender and you can't quite afford a 3900 X then yeah it's it's way easier to defend that purchase but you can find all that discussion with a 3700 X review for this one the conclusion don't buy the 3800 exit to waste of money he should just get the 3,700 X instead save the money and this is this is just part of Andy's market strategy where they saturate every single price point they can think of use the same dyes across all of them which is that part's brilliant by the way and end up with roughly the same cost to assemble the chips but can get $70 more out of it and have a way higher margin on that 3800 X so without a guarantee to hire bin which is relevant here because they all overclocked about the same without meaningfully higher performance we just can't recommend it and it's as simple as that it's again it is not a bad CPU but in this instance it is an overpriced CPU and it's not even because of Intel it's because of AMD's home products that they launch the same day so that's it for this one pretty clear decision on this thank you for watching and subscribe for more Oh quick note the overclocking numbers we do so we just we just pump the voltage and get it stable at whatever we can all core which is 4.3 here and you'll want to run lower voltages so you might not even be able to hit 4.3 all core because we just do a preview and out for daily use but anyway subscribe for more we go to store down here in Texas net sports directly by buying things like the shirts the mod mat or the toolkit or you go to patreon.com/scishow numbers and axis thank you for watching I'll see you all next time