Gadgetory

and these are r7 3,700 X's MSRP is three hundred thirty dollars and the 3,800 axes is four hundred currently the i7 9700 case it's between them at just under four hundred dollars or three hundred sixty-five dollars on Amazon so this is a tough place for the 3,700 X to be it's following up the r5 3600 which is much cheaper and has many of the same advantages of the 3700 X and the 3900 X has more cores and threads and higher stock clocks the 3600 X 3700 X and 3800 X will all be fighting it for a narrow slot between AMD's affordable and well balanced r5 and their high core account premium chip but today we're reviewing the r7 3700 X to see where it fits in before that this video is brought to you by thermal grizzlies conduct a not liquid metal conductor not as what we've used in all of our liquid metal and delayed thermal tests capable of dropping cpu thermal significantly and replacing the stock thermal interface over cpu thermals don't just allow better overclocks but also lower noise levels because the transfer efficiency is increased the mix of gallium and indium makes for a thermal conductivity of 73 watts per meter Kelvin outclassing traditional pastes significantly learn more at the link in the description below with all of AMD's rise and launches to date there have always been a few CPUs where it just they don't they don't quite fit there are other CPUs that you can overclock to silver performance if not the same performance and often get them for cheaper and that's not to say anything other than this is a good way of doing product segmentation and we like that the lower end chips are still overclocked well you can often meet those more or less pre overclocked CPUs like the 2700 to the 2700 x7z hundred to the 1700 X and so forth and obviously not everyone wants to overclock anyway so there's use cases for those x-series chips but the 3700 X is interesting because we were many years ago in an era where it was not uncommon for people including us to say and i-5 is enough for gaming and now it's it's sort of flipping and i-5 hasn't been enough for gaming compared to some of the alternatives lately you can still game on an i-5 and 9600 K certainly does far better than its predecessors with fewer cores but the point is that these days the frame time variability is getting high enough on high five CPUs that we don't feel as good about recommending them as we might have maybe six to eight years ago so it's quite a bit different in terms of the market the landscape for games has changed and we're seeing the frame time performance dip with those i5s and specific games like there's peculiar bugs in GTA 5 for example some of the other games we test you just see the 0.1% and 1% lows dip I my fives but now an R 5 is in that same position where it's enough for gaming we saw that with our five thirty six hundred review it's a really good CPU and we're very confident and recommending it so that then puts the R 5 as one of the primary competitors to the r7 3700 X for a gaming machine the 3700 X obviously has some very clear advantages in production workloads it's got more threads and so a lot of things like blender do actually make use of that so we'll go through all those numbers today we're going to be looking at a few things one is should you just buy an r5 3600 instead if you're only gaming or well if you're doing very little non-gaming tasks when should you buy the ice of a 9700 K which is a still a big competitor here for the 3700 X and then separately the 3900 acts of flanking the 3700 X when does that come into play should you ever just try and save up and jump to the 3900 X instead so that's what we're looking over for additional notes on testing things we did check our previous reviews the r5 3600 has more of that notation before the charts start just a quick note before we really get into this one so this review was recorded on Sunday on launch day and we obviously went back into the boost testing for the other video after the review was filmed but before published so as an update if you missed it basically what we found was that changing the bios versions for our testing did not affect data in fact when we said the maximum impact was 2.7 percent in GTA 5 we actually went back later and tested that about 15 more times for other testing and found that the difference was reduced to more or less zero percent so it's like 0.9 percent max not even 2.7 after additional retesting with another 15 or so test passes which is enough to average and call it the actual result so functionally zero difference absolutely zero difference to our production testing which is all core for the BIOS change so jisub version 1 0 0 2 vs 1 0 0 3 a + 1 0 0 3 a B on our motherboard with our CPUs our testing does not impact the data and so that's your gonna be looking at the data from the f5c bios which is the one we used for all these CPUs maximum impact to the our 5 3600 is 0% maximum impact to production benchmarks is 0% maximum impact to gaming is about 1.1 percent now after doing all the retest again so that's all you need to know about that and if you want to know more about the ageia and bios differences you can watch our previous video although just know that we went and retest and even more and found the differences we're even smaller than presented there that's what we've got so let's get it's like gaming benchmarks we'll look at production stuff as well and then talk when it's worth it total war Warhammer two's campaign benchmark is up first for gaming we'll look at the more GPU constrained battle benchmark next there's more room for CPUs to differentiate themselves in the campaign benchmark but it's still the same game and the issues of the battle benchmark apply here as well the stock 3700 expertly exceeds the performance of the stock rise in 536 hundred with the 3600 at 155 FPS average and the 3700 X at 158 FPS average stock the difference very nearly within error margins disabling that some tea on the 3600 pushes it above the 3700 X again but we'd still advise against disabling SMT for actual use this is just an experiment the differences between stock versus overclocked and 3600 vs. 3700 X are barely there all four of the average results for the two CPUs with SMT on fall within a 5 FPS range which is about 16 or more total test passes they are functionally the same the base clock is the same for the 3600 and the 3700 X and the max single core boost is just 200 megahertz higher on the 3700 X so theoretically any workload that doesn't use the 3700 X's 16 threads could be done almost as well on a 3600 and perhaps even slightly better on a 3600 X the 3700 X does beat its predecessor the 2700 X by 12% but the stock i7 9700 K by Intel beats the 3700 X by 11% in this test not close enough to call it a serious competition overclock in the 9700 kay pushes it to 181 FPS average about 13% ahead of the overclocks 3700 X currently at $365 on Amazon but 9700 K is a strong competitor in the gaming arena it's the same story at 1440p but with an even tighter range between the worst 3600 score and the best 3700 X score there's almost nothing that makes the 3700 X strongly preferable to the risin 5 3,600 a CPU less than two-thirds its price in at least lightly threaded workloads like games total war Warhammer - continues its unique behavior with high core account CPUs we're moving on to the battle benchmark next at 1080p with the 8 core 16 thread 3700 X scoring slightly better than the six core 12 thread 3600 but worse than the 6 core 6 thread 36 hundred thousand T disabled we see some interesting scaling it makes it tricky to compare to the rest of the risin stack since the 12 core 24 thread 3900 X score is worse than the 3 700 X just because of SMT overhead still this is a real game at its exhibiting real-world performance the stock 9700 K is 6.9 percent ahead of the stock at 3200 X but overclocked in the 97 RK did not improve performance this indicates a GPU bottleneck and the potential for a wider gap in an unconstrained workload for example a different game or if he brought down the graphics settings here we try to keep our testing as close to being useful for showing real differences while not dropping the settings to be unreasonably low the campaign benchmark doesn't face this same constraint as you saw a second ago and plots this differences closer to 11 to 12 percent the 9700 K also maintains stronger low performance with one percent frame x and 0.1% frame x improved overall in the 9700 k f1 2018 is next giving us traditional DirectX 11 implementation with the codemasters ego engine all the FPS numbers are big in f1 2018 but that doesn't mean the percent difference between results has changed the 3700 axis stock to 78 FPS average is 4% ahead of the 3600 stocks at 267 FPS average result and the i7 9700 K stock is 8.8% ahead of the 3700 X ranked at 302 FPS average so far the 3700 X looks like more of an equivalent in gaming to the 780 700 K than anything we've tested from the 9000 series which isn't necessarily bad but it doesn't read well on marketing materials the $330 MSRP for the 3700 X is cheaper than the current 365 dollar price of the 9700 K on Amazon but it's the closest Intel competitor to it and the 3800 X which we don't have yet as for generational improvements the 3700 X beats the 2700 X by 13.9% we won't be discussing the 3700 X overclocking much in the games section of this review because the results are barely outside of margin of error the all core 4.3 gigahertz overclock isn't much of an improvement over stock for gaming in general and could be even worse in limited thread workloads but it was the maximum stable and trying to go any higher resulted in instability or shut down the GPU limitations at 1440p mean that there's barely any difference whatsoever between the thirty six hundred and thirty seven hundred X overclocked or not and the 2700 X isn't far behind either civilization six is our next test this one runs with turn times rather than framerate so we're looking at how long it takes the CPU to compute the next AI player action in a turn-based strategy game the 3700 ax Falls where it should in the civics benchmark between the 3600 and the 3900 X results the 97 hard K completed turns in 30 1.1 seconds average calculated across five players with four test passes each and benefited from a 7.8 percent time reduction from the 3700 X stock versus stock the 3700 exa needed 6.4 percent less time than the stock 3600 that puts the 3700 x encouragingly close to intel on one side and discouragingly close to AMD and the cpu tooth years below it on the other side it does reduce turn times by 10.1 percent versus the stock 2700 x Assassin's Creed is the best opportunity to see CPUs differentiate themselves based on core count before we get into workstation benchmarks we found Assassin's Creed origins to have the best balance between core and frequency dependencies in our gaming test so it's good for that even then the 3,700 X only exceeds the our 536 hundreds average FPS marked at 119 to 115 FPS average 53.9% and the 2700 x is stock 107 FPS average is exceeded by 11.5% based on our experience with Assassin's during this round of testing those may be the most quote normal numbers on which to base expectations for the other games we test the 9700 K is 17.1% the head of the 3700 acts here all stock or 18 percent when both are overclocked although note that we are beginning to reach the top and with the 9700 K where GP limitations may begin to exist as usual 1440p results behave like 1080p but with a smaller range of results thanks to limitations outside of the CPU that puts the 3700 acts even closer to the 3600 in performance and the 3700 acts closer to the 9700 K although Intel's chip is still good l haven't presented ahead even with the GPU bottlenecks closing it GTA 5 is next another dx11 title that's still wildly popular as a game that cares more about frequency than threads and all modern CPUs we've tested GTA 5 doesn't break from the trend by once again putting the 3700 X only about 4.8% ahead of the 3600 or 109 FPS average vs. 104 FPS average and the 9700 K is 122 FPS average is 11.8% ahead of the 3700 X the generational improvement over the 2700 acts is about 16% a bit better than in the other tests for what that's worth this is another instance where the 3700 axis results are much more competitive with last gens i7 8700 K than the modern night 700 K and there are 5 3600 is more compelling from a value angle than the 3700 axes GTA 5 at 1440p continues to scale cleanly from 1080p with barely any change in fps numbers between the two tests this indicates that we remain largely cpu-bound with our settings so this is a more perfect environment for comparison so even if playing at higher resolutions you'll still see the CPUs impact framerate you can't fully escape that the GPU isn't always the bottleneck even with those higher limits a high resolution limitations Shadow the Tomb Raider at DirectX 12 title is next and becomes GP limited towards the top bounds of the results the 9700 K is one of those CPUs near the upper bounds it's still about 17% ahead of the 3700 X and average FPS so it doesn't seem to be too limited 3700 X only beats its 6 core 12 thread r5 3600 sibling by 3.3 percent in this test which is pretty lean even compared to the already small performance gap in the other games so far and it's only 7% past the 2700 x1 last game in benchmark to fully and complete we beat this dead horse and hit man 2 using direct x12 the 3,700 access 2.7 percent better than the r5 3600 ranked at 118 FPS average versus 115 FPS average 14 percent better than the 2700 acts and the united 700k is 18.8% better than the 3700 ex in gaming the generational improvements are good the improvements over the lower part of the product stack are bad and the competition with the 9700 K is rough to non-existent game is only half of our CP been sweet so we're not saying the 3700 is a bad CPU just that it doesn't match up to and these chosen competitor and gaming performance quite as well as a MIDI might like and more importantly that the r5 3600 chips away mower at the 3700 X's proposition for gaming than perhaps anything else does and it's a lot cheaper moving into production benchmarks next we can start looking at how the 3700 X compares to the 3600 and the 3900 X and V ray renderer by chaos group we have some other benchmarks to this benchmark is measured in time required to render so lower is better and these are our seven 3700 X finishes the render in 1.0 nine minutes when stock in 1.0 four minutes when overclocked to 4.3 gigahertz a marginal improvement of just four point six percent the ro933 ender in a massive 31 percent faster time in 0.75 minutes instead of 1.09 the r7 3700 X finishes the render 25% faster than the r5 3600 CPU on both our stock posting more noteworthy improvements than a lot of the games would otherwise show where the r5 3600 might be an easy choice where's the 3700 x4 gaming systems that the 3700 X does prove value for this type of work look Adobe Photoshop is next for this one we already know that Photoshop refers frequency over all else which is something that we always demonstrate by highlighting the 5.1 gigahertz 900km 9700 k results at near equivalence despite the double thread count on the 9900 K as for the 3700 X the stock CPU score is 10 18 points in our benchmark basically tied with the all core overclock because some tests in this application are limited thread loads overclocking is typically worse or ineffective versus stock the 3900 X doc CPU scores 1053 points an improvement of just 3.4 percent it's not worth the 3900 X for only Photoshop although the story changes for premiere later for this workload if Photoshop is the only thing you do or you do it professionally it's probably best to stick to higher frequency intel parts for the current generation even the 9600 K at 5.1 gigahertz highlights the lack of usefulness for threads in this workload and that's another point to show the frequency requirement by Photoshop Kanu compiler collection is next this is a compile workload that is functionally a cache benchmark and Rises victim cache combined with high cache quantities upwards of 32 megabytes out three on the 3600 alone means that it finishes the compiles significantly faster than other parts as stated before we can't represent all types of code compiled in one charge so this one will primarily or only apply to those using this type of GCC with cygwin or m62 and mingw environments so r7 3,700 x does just as well as the 3900 X and results are within error margins both have the same cache and so we learned the limitations of this particular benchmark are related to l3 cache for cash in general and not core count or as the 9900 k illustrates clock speeds 7-zip is next for this one we're looking at seven zip compression results first then we'll look at decompression the test is to see how many millions of instructions per second each CPU can manage while dealing with compressing or decompressing files the r7 3,700 X scores 74,000 MIPS 1 stock with an overclock under 1% better not really worth it generationally the r7 3,700 X stock Seaview outperforms the previous gen our 727 and X by a notable 26% and the r7 1700 a 3.9 gigahertz basically a 1700 X overclocking them got them to about the same frequency by a massive 48% improvement versus the first gen rising processors compared to end how the closest to the 3700 axes the 9900 ka 5.2 gigahertz which is more expensive and does require an OC although to be fair into overclocking is much more effective than AMD overclocking and this test and basically all others it's just that Andy's got a strong lead anyway for decompression and these are seven 3,700 X stock CPU runs at 96 thousand MIPS roughly with an overclock giving us an improvement of two I understand it's better than compression but still not worth the loss of higher boosts on one core the 3900 X has a massive improvement of almost 50% over the 3700 X stock to stock which is thanks to the higher core account if working heavily and daily with compression and decompression even though it's almost $200 more expense the r9 3900 ax is worth serious consideration a 50% increase in performance is something that will affect daily use meaningfully for everyone else the 3700 ax is a fine middle ground part this is more easily justified against the 3600 than in gaming although the 3600 is the best in its price class and we strongly recommend it the 3700 X does have meaningful uplift here blender gives us another popular usage environment with 3d modelers and animators and one to 2.79 with our custom-made monkey heads render the 3700 X ends up a bit faster than the 9900 K stock CPU completing the render in 7% last time although overclocked in the 9900 K does put it back in the lead the 3900 X leads the 3700 significantly halt thanks to thread dependencies in this workload finishing in 32% last time at twelve point eight minutes versus the 2700 X doc CPU the generational improvement to the 3700 X is a reduction from 23 minutes to 19 or about 20% for reference the 3600 stocks he finishes the render in 25 minutes so again there's a large difference between these two and thread bound workloads particularly popular in the non-gaming segment the GN logo rendered positions the 3700 x stock cpu at 23 minutes about 5 minutes faster than the stock 27 an X or 18 percent and 6 minutes faster than the overclocked r7 1700 x 3.9 gigahertz where Andy started losing ground to the 1900 K when it first launched shown were the 9900 a stock performance outperforms even the overclocks 2700 X although AMD always had a cost argument now with the 3700 X Andy has both the cost and the performance arguments in favor for a blender adobe premier is another Adobe application that often likes frequency but unlike Photoshop it does actually make use of the threats that's why the 3175 ex showed meaningful uplift in our review of the 28 core CPU for this one-hour 1080p60 Eng and convention report renders in 3.8 minutes with the 3,700 X operated in conjunction with CUDA as all these results do for a Youtube upload it's it's good I mean it's faster which ties it with the 9900 K overall overclocking does not improve the results for the 3700 acts at all overclocking the 9900 K gets it about 5% to uplift ahead of the 3700 X with the 3900 acts leading both by a meaningful margin it's easy to compare the 9900 K to the 3700 X here just because they're close to each other on the chart but reality is the 3,700 acts is a lot cheaper at $330 versus the raw 485 to $500 price range of the 9900 K at overclocked 297 30 K is probably the closest price comparison that still meaningfully competes with a 3700 X at that point you might as well just buy the 2500 X from last gen with our 4k render the 3700 X doc CPU ends up at eleven point two minutes ahead of the 1900 KS stock CPU and I 700k overclocked CPUs and behind the RO 9:30 to 900 X the 3900 acts is really the one that surprised us in this test we weren't expecting to see the 9900 KD throned and premiere by a comparatively priced part in a few ways this review and similarly to the r5 3600 of you AMD is not really a match directly fourth chose an Intel competitor the 9700 K that's the one that AMD shows itself against in most cases and some games we're looking at 18% leads for the 9700 K and that varies a bit depending on the GPU bottleneck but even we're bottlenecked it's still in the lead and so even if you get an extra 25 megahertz back with some different update to BIOS although we were more or less at the boost tree it seems to vary chip to chip and I guess we can explore that later if we need to but we knew about this issue for a while for a couple days now the content was mostly content complete though and the BIOS updates didn't really seem to have a whole lot of impact on how our 3600 was was boosting but anyway the point is the 90s 500k has a pretty reasonable lead in gaming only applications and that's not going to change a whole lot so if all conditions are the same if you overclock memory on rise and sure it'll improve but it will on Intel as well so all things equal if you're only gaming the 9700 still in a good position but you've got the the additional performance headroom in things like production tasks blender we see improvements in premiere now that we weren't expecting with AMD chips the new ones and if you're doing anything like that in addition to gaming as we've said with some of the other CPUs the 3700 acts become something worth considering but it's not like the r5 3600 with the 3600 we more or less give it a clean sweep when for the purchase in its price category of $200 you could get a 9600 K it's definitely performs better and gaming workloads by a bit but the jumps aren't that big and at that price category you shave a bit off the price of what a 9600 K costs you dodge the variability of 1% 0.1% Louis or frame times that are mixed results by using the 3600 so it's that was more or less a clean sweep with a 3900 X we came to the conclusion of if you do in production this thing's pretty great and it's a good solution for most production applications now there's some that stand out like Photoshop which is frequency bound but for the most part the 3900 acts especially things like compression decompression blender for premiere not Photoshop a premiere surprisingly did really well in all those and so for those applications we can make a pretty good recommendation for a lower-cost h EDT part because that's basically the gap that it's filling in the market for gaming at that price point the 9900 case still better for Photoshop it's still better so that review was more of an it depend scenario the 3700 X review ends with basically if you're spending the amount of money would on a 3700 X you've got on one end the 3600 and then on the other end the 900 K or the 3900 X and so the 3700 falls in this weird middle ground if you're spending more than you would for an r5 3600 at $200 for a gaming focus build it probably best to go for a 9700 K if you're stepping down in price we'd say go for the r5 3600 the r7 3700 X is tough to justifying gaming workloads and doesn't quite fit in as cleanly as the 3600 and the 3600 X might even outperform it in some because of the boosting behavior where it can boost higher with more limited thread count we need to look at that when it comes out but it becomes easier to justify the 3700 acts as a cheaper solution for production tasks for pure gaming we continue to recommend at this point the Intel CPUs in this price bracket the 97 K has come down a bit recently and for any additional tasks like video production that can make use of more cores and these are r7 is the better choice and like the r5 3600 the r7 3700 X shows a respectable improvement over its 2000 series predecessors in gaming that makes it even more attractive for or maybe a streamer compared to the rest of the rise in 3000 stack though we'd probably only recommend the 3,700 acts in a choice against the 3,800 X but we can't be certain without benchmarking it we can however predict based on our seven three hundred versus 1800 X and 2700 versus 2700 decks testing that that would be true otherwise the 3600 has the edge in price and the 3900 X has the edge in performance so 3200 X it's not like it's a bad CPU it just doesn't quite have as strong of a recommendation as a 3600 it's got real competition from Intel and it's in a weird middle ground and we don't typically like recommending weird middle ground products because there's really things that are very good at what they're targeted at on either side and so the 3700 X is it's just a bit it's the odd man out for this one so that's it for this review I think we've I think we've covered most of the products that have been available to us we're working on more coming up we're hoping to do a liquid nitrogen stream at some point maybe this week hopefully so check back for that stay tuned on Twitter to get updates on that at gamers Nexus you got a store that cameras access net to support us directly by picking up one of our tool kits for example which was one under here somewhere or you can pick up one of the mod mats thanks for watching I'll see you all next time