Gadgetory

alongside the r9 3900 X and r7 3,700 acts that we're also reviewing AMD launched its r5 3600 today to the public we got a production sample of one of the r5 3600 cpus through a third party and after seen its performance we wanted to focus first on this one for our initial rise in 3000 review this CPU is supposed to be a $200 part we've been recommending and these are our five CPUs since the first generation as intel's i5 see views have seen struggles lately and frame time consistency and are often close enough to AMD that the versatility same time consistency and close enough came in performance have warranted r5 purchases today we're revisiting with the r5 3606 core 12 thread cpu so look at gaming production workloads like premier blender and v-ray and more power consumption and overclocking before that this video is brought to you by gigabytes X 570 ARS master motherboard built with a true 14 phase vrm and for a high-end rise in 3000 series built the gigabyte X 570 master uses a properly finned heat sink for VR I'm cooling accompanied by RGB LEDs in the i/o cover to makes it looks and performance gigabyte also includes BIOS flash features to update the motherboard BIOS without a CPU or RAM the overbuilt BRM is the major feature on this board though and gigabyte has a brand new BIOS for overclocking the new rise in 3003 CPUs learn more at the link in the description below a couple of quick notes on this this CPU was not sent out by AMD and so this is a production sample but it's from a third party and that means that you probably won't see too many of these reviews just yet but they are going to be sold so these will be $200 to the r5 3600 there's an X Q as well the XQ has a 200 megahertz higher advertised boost frequency whereas this one's at 3200 that would be at 3400 megahertz so those are the main differences that there's price difference as well but just like previous generations where you can buy the non XQ and an overclock it to match the XQ that's basically the only difference it's the frequency and everything else is the same between the XQ 3600 and the non XQ so you could buy this one an overclock it or because we've already overclocked it in our charts you can look at that data and then extrapolate where the X would land if you do prefer that one our 3900 X 3,700 X overclocked with much better voltages although the same frequency then our 3,600 and some of that is potentially because of the higher NCP is requiring higher quality silicon so that's part of it probably will note also that the CPS a and V cent had thermal paste left on them so something to take note of as well but our 3600 although it overclocked reasonably it did require a lot more voltage and then our 3900 X we'll talk about that more in the 3900 X review which is coming up after this one and after our 5700 XT review so those were the main notes other than that all Windows updates up to 1903 have been applied for our new CPU test bench and BIOS mitigations have been applied for Intel CPUs we're on the latest bios for just about everything unless there was a reason not to you for example the AMD BIOS is for a risin 3,000 there are a couple that are questionable right now so for by us we're using fc5 for the gigabyte X 570 master that's the mother would we used for all of our new rise in testing that you'll be seeing this week and we do have other boards as well but we're starting with that one the F clock is another thing I want to mention so the Infinity fabric clock is going to be a big part of overclocking for this generation of Verizon from what we understand we've done some F clock overclocking and it won't be included in this review because without also overclocking the memory it doesn't really do a whole lot and at least vs. auto so we're going to be visiting F clock in a separate piece and we're also doing some xoc liquid-nitrogen overclocking hopefully this week with Rison and X 570 CPUs so we'll have a lot of depth on that coming up as we tinker with it with liquid nitrogen timings for this updated bench have been very heavily controlled so we've improved our test bench and our testing methodology significantly for this round of testing we are now manually controlling more or less every single timing and BIOS before we did the important ones but now we've controlled all of them so our FC for example can have a pretty big impact and it's to the extent that it just it tightens the air margin so it's an extra plus or minus 1% closer every time we do the runs and now sometimes it's 0.1 FPS different run to run which fantastic let's get straight into it we like to illustrate CPU behavior on new architecture launches part of this is to look for boost duration limitations or power fluctuations something we'll cover next the first chart shows frequency over time with a blender workload which is a 3d rendering application hitting all cores nearly equally the average all core frequency ends I've had about 40 104 megahertz the CBA was under a to 80 millimeter CLC with 22 degree ambient for this test so boost behavior was not thermally affected the CPU has an advertised boost clock of 42 hundred megahertz which will apply for workloads that don't fully load the core is that 100% there's minimal fluctuation with this frequency plot and it does not appear that any boost duration limits come into play boost eration limits are typically close to the 100 second mark for example for a better view of this well plot each core clock individually and with the y-axis constrained to Jeff's 40 50 to 40 200 megahertz the point of this is to magnify the data as observed here pre-testing starts with cores bouncing off of the 40 200 megahertz a limited core turbo while still an idle once the work begins the cores take turns bursting up to 40 125 megahertz and falling back to 4100 with rare dips down to 40 80 will keep drawing other cores as we go here there doesn't seem to be a preference for which core boosts up to 40 125 but they all do it at some point a pattern emerges where cores 0 and 1 will pass the ball back and forth and then on to cores 4 & 5 & 2 & 3 seems to do similar at the end of the day they all take turns boosting will use Shadow the Tomb Raider at 1080p to give a gaming look at frequency behavior well so skip straight to the zoomed in chart for a better highlight of the boosting in this chart we stay closer to an average of 41 75 mega Hertz boost went under heavier loads with the lighter threaded scenes rendered we see single core boosts up to 40 200 megahertz the maximum stock boost before falling back to 41 52 4175 as for the core number 5 dip below the chart range that spike falls to about 3300 megahertz we use two of total war hammer tears baked in benchmarks starting with battle designs a test performance on the RTS map we also test with the campaign for a look at turn-based vantage points strategy games are a good place to demonstrate performance for CPUs as AI processing is often CPU intensive the higher end Intel CPU is a 1080p approach AGP bottleneck with a maximum average FPS of 174 for the 5.1 gigahertz 9900 K and these are five 2600 and are five 1600 came nowhere close to this limit previously with the overclocked 4.2 gigahertz 2600 maxing out at 145 FPS average the r5 3600 makes a big step towards closing that gap with a stock average of 159 fps a twenty seven point three percent improvement over the stock 1600 and a 10% improvement over the overclocked 2600 4.2 gigahertz core overclocked in the 3600 yielded a minor 3 FPS improvement to 162 FPS average but leaving the frequency stock and disabling SMT bumps the 3600 score up to equal the stock 9600 K placing at 166 FPS average we've observed in the past that total warhammer - and some other games can react negatively to hyper 13 or SMT so this is one way of leveling that out with these six core six thread 9600 K we first broke this information in the first generation rise in reviews and it seems that this persists here we wouldn't recommend actually disabling SMT as the usefulness is a wash between games some like it some don't and ultimately it's better to have the extra threads that is half the reason you're buying the processor anyway this is more of an exercise to demonstrate behavior in Zen to versus the original House for the 9600 K the stock performance that leads the stock 3600 by 4.4 percent they're close enough that other applications may matter more than this Delta we'll look at production workloads later the total war 1440p battle benchmark is a GPU bottleneck so we can't see any CPU differences until we get to the bottom of the chart we'll move on from this one quickly and leave discussion for it in the article below if you want to read a bit more the second Total War benchmark uses the campaign map as seen in the grand strategy portion of the game it's much more CPU dependent than the battle benchmark so we'll start with 1080p at 1080 P there are 5 3600 stock 155 FPS average shows a 7% improvement over the 2600 overclocked to 4.2 gigahertz and a 33% improvement over the stock 1600 this beats the full range of older six core 12 thread a and the parts overclocked or not overclocked in the 3600 itself was ineffectual in this test as it was with the total war battle benchmark but again disabling SMT offered a to performance that made the 36 hundreds performance around the stock in 96 hundred KS both at 161 to 165 FPS average that's stock to stock frequency the 9600 K has much more overclocking Headroom with a 13 percent to FPS improvement from stock the 5.2 gigahertz in this test and a not an impressive 183 FPS average the 1440p results aren't as GP bottlenecks like they were in the battle benchmark but that also means that much of the chart lines up with the 1080p results the 3600 with SMT disabled lost some of its edge here with the 3600 overclocked outperforming it slightly and the stock to 9600 K just beyond that so 1440p sometimes will matter for the CPU choice and sometimes it won't like with a battle benchmark where we saw they all more or less equal if you have 1440p there with any of the top end or even mid-range CPUs you're going to be limited by the GPU choice f1 2018 is up next a DirectX 11 standard implementation of the game on the ego engine everyone 2018 runs out higher frame rates than any human could require in our 1080p testing but it still shows good scaling between CPUs the stock 3600 to 67 FPS average is again well ahead of the previous best score for the six core 12 thread AMD parts surpassing the overclocked 2600 to 35 FPS average by 14% and the old stock and the RFI of 1600 by 34 percent overclocking at 3,600 again had very little effect and he's done a good enough job at boosting per core frequency that is difficult to improve the performance with an all core overclock at least in lightly threaded applications like games everyone is another title where disabling SMT raises performance although not quite to the level of the 9600 K this time we climb to 273 FPS average with SMT off and stock clocks it's possible that disabling hyper threading would give a slight boost to many of the CPUs on our charge but it's rarely worth going into BIOS and cutting a CPUs thread count in half for a minor performance increase in some video work and some gaming workloads the only reason we're treating the CPUs differently is because it's a new architecture and we want to see if SMT overhead has changed in the past two years here's a quick frame time chart with have 120 18 at 1080p remember that lower is better but more consistent is better than just being lower overall the r5 3600 remains close to 4.3 to 5.0 millisecond frame to frame intervals and typically we don't notice stutters unless there's an excursion equal to or greater than eight to twelve milliseconds frame-to-frame something that only happens once in this benchmark for both tests and it happens on both the 9600 K and the 3600 the same spot and the same test overall these two processors have similar frame time pacing and consistency with at least this game with the 9600 K faster by 0.3 milliseconds on average frames of frame 1440p f1 is next the CB is at 240 FPS average and above are almost entirely GP limited but all of our 3600 results fall just short of that range the narrower range of results means that all three stock overclocked and SMT disable to the 3600 averaged almost exactly the same the 9600 case still went out by 4.5 percent stock versus stock technically speaking although the value of the 3600 is a tough match for it particularly in later tests and with the lower price the top-end of the results hit the GPU limit so note that all differences are within error margins that's why the 9900 KOC isn't at the top it's because it bouncing off the limits so any Delta here is just run to run variance we do multiple passes of the new gathering storm AI benchmark for Civilization six each of which takes an average completion time for 5 turns and then averages those numbers together it's the only game test we do that isn't too measured by frame times and the results are extremely consistent we took a look at turn time instead as this is entirely CPU dependent and heavily impacts how enjoyable the experience is unfortunately despite being an AI benchmark it shows a strong preference for frequency over thread count as can be seen by the r5 1600 at 3.2 gigahertz base outperforming the r7 1700 at 3 gigahertz base the 9600 K stock took 34 seconds to complete each turn reducing the average turn time by 5% versus the 3600 stock and the overclocked to 9600 K is far beyond the overclocked 3600 and performance at 30 point 3 seconds versus 35 point 3 multiplied across all five AI turns that means it take an extra 25 seconds per turn for the human player to have input again and that be more exaggerated as the game grows complicated Assassin's Creed origins has revealed itself to be one of the most balanced titles we test in terms of benefiting from frequency and core count at the same time disabling SMT you had an appropriately negative results for once so it's the stock six core 12 thread 3600 results that slightly edges out to the 96 hundred K by one 1 FPS average more or less within error margins at that point so they are functionally tied interestingly the overclocked results for the two CPUs are fairly close as well the 12 threads of the 4.3 gigahertz 3600 bring it closer to the 5.1 gigahertz 9600 K than we've seen in the previous tests with only a 7.4 percent advantage for the Intel chip the 3600 is also the first of AMD's X 600 series CPUs to bring 60 fps and a 0.1% lows which it does even at stock frequencies at 1440p the 3,600 3,600 OC and the stock 9600 k are all within error margins of each other the results are pushed together closer by the GPU constraints for this one the 9600 k OC still has an advantage but again everything is clumped together a bit more GTA 5 is next the oldest game on our bench still has some life left in it thanks to some settings tweaks to further load on the CPU and it's also still one of the most played games on Steam it's another title we're just a way now since he allowed for a performance increase of a few FPS but not past the stock 9600 K this time frequency is important in this title and a lack of major improvement with a 3600 overclock all core settings indicates how close the stock boost frequencies already are to the maximum achievable all core overclock we'll be interested to see whether the 3600 2x can justify its $50 higher price with what seems like very little room for improvement and these generational improvements are as strong as ever here with a 20.7% uplift over the stock 2600 and a 37.4% improvement over the stock 1600 pretty good scaling at 1440p between 1080p and 1440p is almost perfect for GTA 5 with only the nearly tied overclocked 2600 ax and Sox 7600 K trading blazes FPS numbers themselves are also barely changed meaning we're a healthy distance from a GPU bottleneck or the observe the FPS cap of about 180 7.5 FPS for this engine you would have to go to a high resolution or even higher settings to limit the GPU down to a point where the CPU is bottleneck shadow of the Tomb Raider is another one of the minority games we're leaving SMT enabled gives us a better result the stock 3600 showed an 18% improvement over the average FPS and the stock r5 2600 but with barely any further improvement from overclocking the generational improvements are big but it's the same story as it has been manufacturers squeeze performance out of parts more efficiently these days so stock performance goes up and walking Headroom can go down as a result the 9600 K is one part that still has some room left and the 5.1 gigahertz I see put at 13% ahead of the overclocks 3600 due to popular demand we've switched to DirectX 12 for our hitman to testing DX 12 support was passed into the game relatively recently and any previously published benchmarks from us in the CPU side were done using dx11 we use both 11 and 12 for GPU reviews know that dx11 has better frame time pacing and performance than DX 12 but so many people seem to prefer the different graphics API that we decided to just move to it for CP testing even though the frame time performance is technically worse hitman 2 showed a reassuringly strong preference for SMT on rather than off with a 7.9 percent improvement from the extra thread slightly better than even the 646 thread stock 9600 K our testing with the 8 core 16 thread 9900 K showed much better results with hyper threading disabled across multiple tests and retests and 16 threads doesn't exceeding some threshold because the 8 core 16 thread 3.9 Giga at 1700 our performance the 6 core 12 thread 3.9 gigahertz 1600 so it must be down to hitman 2 treats thread count the 4.3 gigahertz our overclock on the 3600 was as disappointing as it was in the other gaming tests or most of them leaving hitman two is universally bad your apartment low is unaffected for this CPU you're really getting so much out of the stock performance that a higher power consuming OC is getting tough to argue in the games the r5 3600 out-of-box performance is highly competitive with Intel similarly priced CPUs already included in games and a big leap over the previous two generations by AMD the 1440p results stack up the same way as the 1080p results did 9600 KOC is better than the 3600 OC but by a smaller than usual margin 3,600 stock just barely better than the AI 600 K stock and the 3600 1000 T disabled Tralee and well behind the normal six core while thread result we can get into the production workloads next including Adobe Photoshop Premiere blender v-ray GCC canoe compiler collection and some additional testing so our gue new compiler collection benchmark is basically a cache benchmark something that's Illustrated clearly by this chart this demonstrates how quickly the CPU completes our code compiled in the benchmarking it is not however benchmarking the compilers against each other nor is it testing the compiled quality the faster and higher core count CPUs and especially Intel CPUs would do better in other code compile environments potentially but something compiling GCC with GCC is as Wendell of level one text told us quote cache hits all the way down in this respect we can functionally use this test as an illustration of the impact of having so much l3 cache by all counts the overclocked 2700 X should clearly win against an r5 3600 and pretty much every other test if you happen to work in a similar environment to this basically cygwin or mingw compiling on Windows the higher cache will help we may have accidentally discovered Andy's new favorite benchmark as one of our patreon subscribers noted and it's functionally a look at the cache so a bit of a unique test but not one that necessarily scales to the way most people would compile the code we need to look into expanding this testing going forward moving on to the next one our next benchmarks are for compression decompression with 7-zip with compression the r5 3600 pushes 55,000 that million instructions per second or 55,000 MIPS ranking it between the 1,700 at 3.9 gigahertz and 9700 K a 5.1 gigahertz the bigger story here is that a 3,600 at 4.3 gigahertz all core performs about where a 5.1 9700 K does and not distant from the 2,700 X dot CPU generationally we see an improvement of 27 percent over the twenty six hundred and forty six percent over the r5 1600 cpus decompression is next in this test the r5 3600 pushed of 72,000 MIPS roughly tying it with a 9700 K a 5.1 gigahertz and holding a strong lead over the price comparative 9600 K which isn't even in consideration at this point Adobe Photoshop is next like premiere Photoshop refers frequency first performing transforms warps applying filters color changes and resizes we see the 9900 K illustrate what Photoshop likes in a processor the 9700 Kay's 5.1 gigahertz result being so close to the nine nine hundred KS 5.1 karakurt's result is useful for demonstrating that frequency matters first for photoshop the r5 3600 ends up nearly tied with the i5 9600 k stock cpu leading it by about two percent overclocking the r5 3600 gets it to 979 points an improvement of just 2.3 percent the 9600 K with an overclock despite 13% from 942 points to 1065 points and these are five can hold its own here but frequency dependence and Photoshop does shift the recommendation to be less strongly towards the r5 3600 blender 2.79 is next this is a real application for 3d modeling and animation and is the very one we used for our own GN intro animation these videos and for designing a lot of the GM products on stored I Kara's excess net the GN monkeyhead render gives CPUs a mixed about heavy workload to crunch for this one the r5 3600 stock cpu finishes in twenty four point eight minutes ranking it as faster than the 8700 K and just under a minute slower than the five point one gigahertz 9700 K blender has an organic use for the hike or count on a.m. these mainstream CPUs and doesn't lean as heavily on frequency though it obviously still matters generationally the 3600 finishes the render faster than the 2631 minute result by 20% with the 1635 minute results reduced by 30% on the 3600 the 3600 completes the render significantly faster than intel similarly positioned to 9600 case stock and overclocked results overclocked in the 3600 ties it with the r7 1700 at 3.9 gigahertz illustrating that Andy has brought $330 performance from 2017 to the $200 price class in 2019 but also highlighting that a used processor in the r7 class may be a good consideration if you can get it for cheaper like maybe the 2700 the GM logo is a heavier workload not much changes but the 9700 km 5.1 guardsman was up the ranks more than earlier meeting the 2,700 X the r5 3600 is close to both of these one stock and an overclock gets at only a time reduction of 4.2% not much from the limited OC Headroom in this one our Adobe premier benchmarks are next using a 1080p show report project that we can hopefully show on the screen with a roll and b-roll followed by a 4k project that's heavily comprised of b-roll shots we're rendering without Vig P in the case of Intel so there would be some potential performance uplift if I GPUs is acceptable in your organization and workflow the 1080p show report renders in 4.8 minutes on the AMD r5 3600 as discussed in the past Premiere and Photoshop are still heavily frequency-dependent but the 3600 does well to reduce render time versus the stock are 526 hundreds 5.9 minute results and 18% decrease in render time stocks and generationally is a major lift where AMD needed it and he has been weak in Adobe Photoshop and Premiere previously so the IPC and clock increase help here for reference in overclocks 2600 rendered the file in five point five minutes with the 1600 stock CPU from 2017 rendering it in six point seven minutes making the 3,600 about twenty eight percent last time intensive compared to the i5 9600 K a processor with comparatively fewer threads and these are five thirty six hundred finally begins to ball ahead in one of Andy's weakest realistic production workloads Intel stock 9600 K and it's five point six minute results it closer to the r5 2600 at four point eight gigahertz overclocking the Intel CPU to five point one it ends up about tied with the 3600 the 3600 finishes in about the same amount of time as the 3.9 gigahertz 700 from a few years ago basically a 1700 x4 reference there are more or less the same when overclocked and not far behind the Sutton 9700 K the 4k renders I have your workload for this one the I 999 hundred K predictably chart topside eleven point nine minutes stock more relevant to our conversation today the 3600 finishes the render in fourteen point two minutes allowing the more expensive and I do 900 K a time requirement reduction of 16 percent when both our stock the r5 3600 our performance of stock 9600 K with an 18% time reduction pretty massive and further manages to push a 7% lead or less time required than the overclocked at 9600 K generationally the r5 3600 stock CPU our performs the stock 2600 CPS 18 minute result by about 20 percent or about 28 percent shorter time than the r5 1619 point eight minute result finally an overclock on the r5 3600 allows us to finish in about three percent last time let me stock 3600 dra is the last production test for us it's by Kaos group and one that workstation users have requested this one measures in render time by minutes so lower is again better the r5 3600 CPU finishes the v-ray benchmark render in about 1.45 minutes landing it near an i7 8700 K stock were under time and ahead of the 1.5 for a minute r5 2600 4.2 gigahertz render time generationally the 3600 stock CPU completes the rendered in 16% less time than the 2600 stock CPUs 1.7 30-minute render or about 26 percent faster than the r5 1600 stock CPU the r5 3600 finishes the rendered about 23% less time and the in tile 9600 K stating that v-ray does actually utilize the threads overclocking the 9600 K closes the gap ranking it at one point six minutes but it's not enough the 3600 with an overclock is near the 1700 I want 3.9 gigahertz and overclocking the r7 2700 would get you to about the r7 2700 X levels of performance or 1.25 minutes this again shows you that there may be even better value in buying an older 2700 CPU on clearance and overclocking it at least there would be for some workloads like this one power consumption testing is measured at the EPS 12-volt rails before vrm efficiency losses but after the wall this is a much more accurate measurement than the wall and gives us a fairly direct read on CPU power consumption without all the variability and noise of the rest of the system the a.m. the r5 3600 measured at 79 watts on the stock gigabyte motherboard with no overclock supplied our 3600 silicon is much worse than our 3900 X 3700 X silicon so the overclocked requires 1.43 volts to hold 4.3 gigahertz all core this lands us at 90 watts down the EPS 12-volt rails with all power limits manually disabled this is the only chart that will contain the r7 + r9 data for our 3600 review but briefly the r7 3700 X stock CPU consumes about 87 watts when stock and 103 watts when overclocked 4.3 gigahertz at one point 3 5 volts note that our voltage required here goes down one full step from the 3600 but core count has gone up this is a silicon quality advantage and that carries to the 3900 X our 3900 acts could impressively hold 4.3 gigahertz all core across 12 cores at just one point three four volts this is really good this and that's a 170 watt power consumption but more importantly we're able to run both the 3700 X and 3900 x with a lower power consumption and lightly threaded workloads that in stock or auto with the motherboard the motherboard pushes higher voltage than the CPUs need and so we can clock higher with lower voltage and power requirements and therefore heat requirements when working manually will talk about this more nor 3700 ax + 3900 X reviews hitman 2 quickly gives us another look at power for this one the 3600 stock CPU stays within Andes defined spec of 65 watt TDP noting that TDP on a processor box is slightly different than actual power through the socket but not by much we were outside of spec and blender but within spec for a more lightly threaded gaming work the 3,600 measure depth of 52 watts overclocking pushed us 256 and the 3900 ex ran at 76 watts when stock is 72 and overclocked that's not variance that's not margin of error that's about testing error that's because we were able to pull lower voltage than the stock motherboard BIOS assigns by manually tuning it but still pull higher frequency conclusions then this is actually pretty fun processor Patrick did a lot of the testing on this and wrote half of the script so we I wrote the production section he wrote the gaming section and we've both come to the conclusion that we feel good about the r5 3600 this is a processor that we can confidently recommend it is more or less superior in all ways to competing Intel i5 parts now technically the i5 9600 K does perform a bit better on average in gaming scenarios you can also overclock it further so if you count overclock to overclock performance yeah it's in the lead but the problem we have with the i-5 now is the same one we've had with the i-5 for the last two years which is that they are much harder to defend at a purchase than say an i9 or an i7 high-end CPU a couple like they did some hundred K a couple years ago and the reason for that is because threads are artificially turned off on these on the 9600 K and it has now gotten to a point that was some of the benchmarks we've done in the past and even recently frame time consistency is just not as reliable on those Intel six core parts as it used to be so interestingly as a separate conversation for those of you who've been in the industry long enough and it's only been maybe eight years since this was relevant you might remember when people pretty much everywhere said an i5 is not for gaming and I said that too many years ago so we're now in a scenario where an i5 isn't enough for gaming for Intel but and our v is enough for gaming from an vo is something we'll talk about more in our our 737 hundred extra view so r5 3600 is genuinely good it is highly competitive we can strongly recommend it and versus the 9600 K in general we would prefer the 3600 and that's because frame time consistency is reliable we know it's almost always going to be good and if it's not there's probably some other problem with the software we know that it is versatile and not everyone needs the versatility so this is something we recommend as a strong point because 3900 ex versatility is irrelevant if you aren't going to actually use it so if you're only gaming that's a different story but for the r5 3600 having strong enough gaming performance plus the versatility it just becomes an add-on feature so that's something that's good the value is good it's $200 so that makes it cheaper than the 9600 K making it easier to purchase or defend the purchase of and we would recommend getting a cooler with it as we do with most CPUs but technically you get one in the box so that is most of the 36-hundred conclusion a couple of interesting notes here AMD has become AMD's biggest competition now with these CPUs so for a video maker with a stricter budget the r5 3600 is superior to its immediately price matched competition from Intel you may be better served though by purchasing an R 7 2700 on steep sale and overclocking it for perspective if you did our benchmarking that land you had our overclocks 2700 X results of four point three minutes for 1080p premiere and that would cost about $200 but that inventory will stop being made at some point if not already even of the 200 to 250 dollar range there's no point in buying a 9600 k if premiere will be part of your regular activities or any rendering software that can make use of more than six cores like blender will be doing streaming benchmarks that later as part of our ongoing rise in 3000 coverage but for now we can at least say that the 3600 is the better choice for those who plan to edit and render footage if AMD is its own biggest competition then they've done a great job on the gaming side of differentiating the 3600 from the 2600 the 1600 excuse or otherwise there are significant generational improvements over the other six core twelve thread parts with clocks being pushed closer to the max out of box there's still freedom to overclock but there's a less and less point to pushing in an all core OC on AMD parts at room temperature we're hoping for better results from PBR precision boost overdrive so stay tuned for that the i5 9600 K outperforms the 3600 and most of our game benchmarks as games have been slow to adapt to CPUs with more than eight threads and the 5 gigahertz overclocking potential the 9600 K makes it an even clearer winner for exclusively gaming but the r5 3600 is again more versatile cheaper the big question is whether the $250 or five of 3600 ex that Andy is also coming out with can justify its $50 price point because the 3600 you can just overclock it on average to about the same point then that's probably the better solution but they'll be close enough so this is a scenario now where for the most part we're recommending picking between a 3600 and it all or 2700 for about the same price and overclock either one of them or don't there the 3600 pretty good out of box the 2700 definitely deserves an overclock view by the non ex though so yeah those are your considerations go look at the old AMD stock first it's good and a lot of it is anyway not all of it but pretty good and competitive and price and with Andy's new stuff and if you're on less of a budget or you just want the newest thing the 3600 is something we can recommend without any concern so that'll be it for this one check back for our r9 3900 ex review that'll be a heated battle versus the I $9.99 hundred K and you can check back for a 3700 ax review after that we also have a teardown of the 5700 XT coming up today and a review of the 5700 XT coming up today so that'll be it for this one thanks for watching gonna store documents XS net to support us directly by like by buying one of our tool kits or our mod mats and you can go to patreon.com/scishow and access as well we'll see you all in the next video today