We Found Problems: AMD Ryzen 5 8600G & R7 8700G APU Benchmarks & Review

but it's okay ASRock because we wouldn't expect you to validate the literal first option on the first page of BIOS benchmarking the apus today has had one massive Challenge and frustration and it's this chart this is the chart that started a conversation at AMD leading to the brand new R7 8700g and r58 600g getting immediate changes in the near future that will improve their performance this huge drop in frequency is from a power limit that kicks in around the 100 second Mark into running this particular workload on this board in fact we delayed our entire review to research this behavior when we used our power measurement device interposer to log the power behavior during cyberpunk gaming testing in this chart you can see that for almost 5 minutes of gameplay the 8700g was holding at about 80 Watts through the EPS 12vt cables then around the 2802 Mark we had a hard wall that dropped power to about 60 watts this also happened with the 8600 G at around the 110 second mark this also happened in balers Gate 3 testing where these three runs are near 80 Watts with the fourth run dropping to 60 watts in some ways it's familiar to Intel's old tow behavior all of this can be quickly summarized with these three numbers these three are the same CPU the only difference is that for the Baseline entry at 15.8 minutes render time for blender we launched from idle and we had staff on so it was stock Behavior with the no staum option we had disabled it the reduction was 8.3% alone from that change and then the pre- staum result was terrible it was 17.4 minutes so from worst to best we have a 17% Improvement the reason it's worse with preloading with some other application is because staum isn't accumulated value so when we exited a different application and immediately went into blender and started rendering our animation this was the result and this pretty well tells the story of why we had some challenging decisions to make for the reviews process which of these do you choose before that this video is brought to you by us and the GN store discount code stap stap for 10% off while the code lasts you can use the code stap for supporting us while getting something in return like the disappointment PC build t-shirt listing all of 2023 worst Hardware launches in a tour date Style on the back we're running low on stock but we still have Heather charcoal with the exploded memory diagram on the front and we've still got some of the cotton in stock it's fitting for the apus to get that memory explosion diagram you can also grab one of our highquality and Ultra heat resistant thick silicone soldering mats with tool holders our large gn15 mod mats with pinout diagrams and more on store. Gamers access.net this is a carryover from devices like the Ally that you see now it's from skin temperature a wear power power management from the mobile and handheld parts from which these apus are derived but skin temperature aware power management or stap them as AMD pronounces it is intended to reduce the shell temperature of a handheld device or laptop and it shows where amd's Focus has been shifted lately it's been in those devices there's no reason to induct this feature to the desktop variant of the same silicon though and since it can be controlled at a firmware level it's not intrinsic to the Silicon despite sharing a solution we brought this up to am right at embargo lift because we needed a sanity check from someone particularly qualified at AMD to speak on the matter we've actually shown them on the channel a few times this is the first am5 part um that I killed and that started a process where AMD moved impressively fast and got the engineers together and came back to us with an answer by end of review embargo day and here it is quote we found out that staum is being incorrectly applied to the desktop Parts it shouldn't be applied to the desktop Parts a future bios update should correct this Behavior if nothing else you caught something that is going to help a lot of customers so better to catch it sooner than later thanks for finding this so AMD plans to ship a bios update in the next Aiza patch it'll take a little bit for that to come out and that will change the skin temperature aware behavior of the desktop apus because the obvious argument against it is why it's in a motherboard in a case you're probably not holding the case and you probably don't have your Tower in your lap so it doesn't quite deserve the same types of protections as a laptop or a handheld M now for end users the difference between sta him being toled on and off it might not be noticeable it really depends on the scenario because since it's an APU it's got CPU and GPU components they get loaded differently so in different scenarios we've seen a mix of impacts from close to zero to actually noticeable like what we showed earlier but from a testing standpoint it triggered a lot of tests and retests because we were trying to understand why the performance was slowly dropping as we spent longer in the game running those back-to-back test passes uh and specifically we encountered this in games where we don't load a save game file to do the next pass so ones where we basically stay in the game uh and just walk back to the start of the test pass that's where it popped up for us the most oh and also asrock's PBO menu item the literal first thing in the Bios doesn't work we found that and many other bios bugs while working on this review for the most part staum doesn't affect the results which is good for the review but there are a few key areas where it does and we'll talk about again how we're kind of averaging over that uh sometimes even when the power drops though it's simply not enough to move the frame rate meaningfully one way or the other but in either case it'll be a little bit different and whenever the next bios update ships for users who have the 8600 and 8700g AP that we're reviewing today it really just depends on the game and how heavily it loads things but first let's cut over to yesterday Steve and learn about the basics of the apus how much they cost and what they're supposed to do today's review is a little different and simple by our standards reviewing the AMD 8600 G and 8700g apus these have disproportionately powerful integrated Graphics components in them they're using the 760m for the 8600 G and 780m for the 8700g uh and the pricing makes for an interesting comparison at the ultra budget end it's $230 for the 8600 G and then 330 for the 87 for apus again one of the most important aspects is simply the system memory that's because for a video card you have the GPU core that's located centrally on this one it is surrounded in this case by memory that is dedicated to the video card to the GPU and that's going to operate at a bandwidth uh latency and actually a physical distance that is advantageous as compared to something like this single solution going out to system memory and physical distance is one aspect of that but there's a lot more again like just the bandwidth and the latency uh being key elements so because the igps today don't have any vram of any meaningful capacity or any at all on the package they just have the normal cash uh there's really no way to make up for that other than try and go for faster system memory and it will disproportionately bump the performance as opposed to if you were just using a normal ryzen desktop CPU still benefits yes but not nearly as much as in an igp constraint scenario where you need that speed another factor to consider is that apus it is not architecturally identical to the 7,000 series CPUs that they have right now so pcie generation as an example goes down a step but to get you the basics here's a quick run through of the specs for the Parts in review today the 8700g is an 8 core 16 thread CPU has 8 megabytes of L2 and 16 of L3 for the GPU it's using a Radeon 78m with 12 cus and clocked at 2900 MHz the r 58600 g has a six core 12 thread part with 6 megab of L2 and 16 of L3 the CU count on the 760m for the 8600 G is just 8 but that's still a significant improvement from the 7700 X's cus at a significantly slower 2200 MHz and actually the AMD Z1 extreme processor found in something like the r Ali is an 8 core 16 thread part with 12 cus as well it's very familiar the biggest difference between this and the desktop variant other than socket ability is the extreme power constraints on the Ally and we're back to modern day Steve for my next trick I'll show asrock's broken ass bios bug it's it's pretty good trick so when testing the azrock board we noticed some peculiar interplay between sta them and PBO uh PBO Precision boost overdrive enabling it in asrock's main Bios page the literal first page unfortunately didn't work properly and uh it gives the appearance of a PL but this chart shows it doesn't the stap them limit climbs and hits 65 Watts after a brief period of 88 Watts package power it should be at least sustained 88 Watts but likely higher with Unbound PBO but to prove that this isn't supposed to happen we next went to the amdoc menu in BIOS and we did the literal same thing we enabl PBO but this time we didn't use the ASRock redundant main page menu item that's right front and center in front of everyone and it worked in this chart you can see that staum continued to climb and that package power remained elevated there were multiple other bugs that we found in this launch but that's the one that's really the most important to talk about cuz it's the one that's most likely to effects you a lot of people use pobo it's a quick way to get a boost of performance if you're not worried about thermal or power uh it's okay though because fortunately it's unlikely anyone would ever encounter this option but it's okay ASRock because we wouldn't expect you to validate the literal first option on the first page of BIOS no that that would I'll say this I'm glad azrock has reminded us why we don't use their motherboards for reviews if we can avoid it it's always pre-launch problems with them but if you want to use PBO do it through ryzen master or through the amdoc menu they both worked fine for us okay so here's some more real world data showing how these CPUs behave we'll look at all core frequency this also helps us better understand the performance and gaming scenarios between single and all core workloads this chart shows the 8700g first initially running a high frequency of 4600 to 4900 MHz on average for all cores that expired relatively quickly and gave way to the system sustained clocks of about 4200 MHz it's not shown on this graph but sapam was at the 65 watt limiter here so this should change with the BIOS update whenever that ships plotting the 8600 G now we observed a similar boost to base dropping Behavior which aligns with the earlier power Behavior moving from 4900 MHz to 4550 MHz during the course of testing it's less severe it's normal for clocks to be lower in all core workloads but for perspective here's the 7700x it held 5200 MHz during the same test when we first reviewed it what this means is that staffen will have a greater impact in our testing scenarios where the CPU cores are most heavily loaded and the power consumption of the CPU package power itself is at the highest point but there's one other thing which is again that this is a cumulative thing that means depending on the order of the testing and what came before the current test and how long it ran and how intensive it was for example if we run blender before a game that'll change the accumulated staff and value and potentially the resulting frequency now some quick validation of the single core boosting to ensure it alignance with the advertising here our 8700g was typically around 5100 to 5150 MHz which is equal to or slightly above the advertised claims and that's a good thing the 8600 G held 5050 MHz solid also about 50 MHz over advertized so AMD has done well to stick with its promise from about a generation or so ago now when it said that it would try to under promise and overd deliver for frequency for reference the son 700 x had a single core Max of 5500 to 5550 MHz that'll be the biggest contributor to its advantage in many CPU bound tasks especially when that limit kicks in in blender and after waiting for the CPU to hit steady state power consumption we found the 8600 G and 8700g to both pull about 61 to 62 Watts at EPS 12 volt rails this has the two CPUs about equal with an R5 5600 for power consumption which itself pulled 65 watts and the I3 13100 f is also close by at 66 and functionally equal compared to the PRI G Series Parts the 5600 G was at 72 Watts so AMD has definitely improved for efficiency and the 5700 G was at 95 watts in this test one thing we did notice was higher initial power consumption than longterm or steady state draw this simple chart shows the initial spiked power consumption versus the long-term and steady state power consumption the initial draw had these CPUs closer to 80 to 86 Watts but it dropped down rapidly as they began hitting their power targets now because the impact of power consumption here is related to our earlier staum discuss we'll stop discussing it pretty I thought that was a good one looking now at Power efficiency and using a blender all core workload we see the 8700g up toward the top of the chart at 16.1 wat hours to complete the fixed unit of work it's behind only the 79 adx and is just ahead of the 14700 K when locked to 91 Watts the 8600 G took longer to complete the render so despite similar power consumption its efficiency is pred predictably lower time to get into more testing so now we're going to look at the CPU only results without the igp being utilized that'll help us establish the ceiling for maximum performance out of the CPU with the current bios configuration uh if you were to upgrade to a dgpu later and stop using that igp so that's the purpose of these tests but as we get into it some methodological notes this launch is a mess the fact that we've been testing an APU in what is currently it's default configuration but will soon change but we don't know when it's challenging that means we have some considerations for the test in process and we have to make some decisions on how we want to do the testing but in addition to all of this there's one other aspect to consider and that's that Andy both claimed and didn't claim at various times that 2400 MHz should be the effect of default configuration for Infinity fabric for the reviews process uh but like Schrodinger's fabric ultimately it kind of it was unclear what it was supposed to be and the company ended up emailing this they said quote I wanted to follow up and clarify on the note about f- clock from yesterday we do believe most most parts will be stable and provide the best performance at 2400 F clock that being said the default Behavior will remain at 2,000 F clock without manual intervention thus results with 2400 would require specific and direct user intervention to accurately represent the default behavior when Expo is loaded please keep F clock at 2000 then they apologized for it uh and said we value accurately conveying performance and behaviors to your readers as much as you do if you have any questions let us know and this all relates back so the higher F clock is an isoc option but just like the staff and behavior we have a strong policy here of representing the product as close to out of the box what you actually get as possible we do enable Expo and XMP uh but that means we don't review a company's hopes and dreams it's the same as with RTX when RTX launched for the 20 series and there were literally zero Games available we didn't review it based on envidia promise of RTX we reviewed the product they shipped same thing here so we're not reviewing what AMD uh hopes that people run it at or or what their staff and behavior might be those are admirable and and good objectives we'll look at them when they ship so methodologically we're leaving staum alone uh and that means there going to be some scenarios where in the run around variants is higher than we would typically like but that is an actual behavior of the product and so we view it as a valid Behavior Uh to test and represent so because of that this means that it sometimes exits some of the testing standards what we're doing is just averaging the passes so you get a couple before STA and a couple during uh the limit applied when it applies doesn't always and it doesn't always affect the results so it really has to be something where the CPU package power is boosted High which means a higher CPU workload uh and that's not all scenarios it won't it won't be there for everything but let's get into the original edited benchmarks and review and talk about how these CPUs behave now we're going to look at CPU only performance this is in production work workloads where we're assuming that you're basically only doing work on the CPU and we've coupled a 4090 with it the purpose of this is to see what is the maximum performance of the CPU component only so if you were to buy the Apu and then add a GPU later we get to see what your limitations are and where you would compare versus having gone with a dgpu and a normal CPU in blender Cycles rendering with one tile per thread the R7 8700g completed the GN logo render in 15.8 minutes which has it similar to the R5 7600 X CPU with the same parts so the 8 core Apu is about the same as the 6 core CPU the 7700x 8 core 16 thread CPU reduces the total render Time by 20% against the 8700g amd's Apu with eight cores isn't as powerful as its CPU with eight cores but that's not a revelation the 8600 G 6 core part and its 20-minute result has it closest to an R7 3700x 8 core CPU from a few Generations back the most Sim similar six core desktop Parts by performance would be the 5600x which allows the 8600 g a reduction in render time of 16% or the 7600 X which reduces the render time from the 8600 G by 15% itself it's basically in the middle of these two comparing the two new apus to each other the CPU performance with the 8700g gives it a 20% render time reduction as compared to the 8600 G still CPU only compression performance with seven zip and measured in millions of instruction per second is next this one has the 8700g at 105,000 mips or behind the 7800 x3d and ahead of the R5 7600 X the improvement over the 7600 X is about 10% with improvement over the 8 core 5800 X about 12% the 8700g is behind the modern 8 core 7700x which leads it by 16% as for the 8600 G the new apu's 84,000 mips result has it behind the 12600 K from Intel and ahead of the the 3700x from AMD compared to six core AMD Parts the 7600 a leads by 14% And The 8600 G leads the R5 5600x by 133% in decompression and CPU only the 8700 G's result was 125,000 mips with the 8600 G at 97,000 mips in this one that gives the 8700g a lead of about 30% which means its performance is disproportionately high and decompression compared to compression the 8700g sits between a 7800 x3d and a 5800 X for performance with the 7700x leading by 15% over the 8700g the 8600 G is between the 3700x and the 5600x leading the 5600x by about 8% now we get into the game testing so as we were saying earlier the unique nature of the product and the timing means that this is a simpler review for us and so we're really only looking at a couple parts on each chart as opposed to the usual like 208 or whatever it was for the gpus uh so the key focus is going to be Memory uh that means when you're reading the charts make sure to read it carefully and identify which memory kit is used with with each line item we consider the ddr5 6400 kit to be our Baseline for this testing then we also have 6, 5600 at the low end just to set the floor for you as a note getting into this too any of the older data or data that is with desktop CPUs unless otherwise specified as an igp is using our standard Hardware that includes a 490 starting with 1080p medium the 8700 G's Peak Performance with the 490 is 87 FPS average with ddr5 at 6,400 Ram or 83.4 with DDR 56000 the memory is a limiting factor even at this performance level for the CPUs but not because of the igp it's just ryzen Behavior here frequency becomes the next limiter for the CPU the 8600 G with a 490 shows that the cpu's gating performance of the 490 with a 6% lead from the 8700g over the 8600 G when looking at ddr5 6000 only results used as you'd expect though the 8 700g with DDR 56400 and the igp runs at 29 FPS average 14% better than the 8600 G at 25fps with DDR 56400 and functionally infinitely better than the 7700x with its igp which was completely unplayable turning our attention now to memory the 8700g with DDR 56400 ran 7.5% ahead of the DDR 56000 results and an impressive 12% ahead of the ddr5 5600 kit at 25.6 FPS average remember timings come into play here too memory definitely matters with an igp this isn't news and using PBO and igp of 200 MHz on the actual GPU component via bios gets the 8700g to improve to 31 FPS average that's an uplift of 10% from Baseline so not bad you can definitely get more out of this with memory tuning as well as for the discreet options the I3 1200f and 6500 XT ran significantly ahead of the Apu at 55 FPS average which is a lead of 120% over the cost equivalent 8600 G the 1200f and RX 6600 are CPU bound otherwise the 6600 would provide much more uplift here but this is cool data because we can see where the ceiling of usefulness is for a CPU in a given budget an i5 or equivalent modern R5 would do much better here we have some data from October will show that shows a higher FPS ceiling and it' be somewhere around 5 to 10% higher still if retested with the newest patch but this is only here to demonstrate that the 1200 f is limiting the 6600 only for purposes of showing the ceiling of this test we also added the 7800 x2d and again note that it'd be higher with patches and needs retesting but it serves its purpose all CPUs below it are limiting the 490 pairings as a quick note too we're not going to show a chart for 1080p low but we did do some testing with it running around the city with the 8700g just ad hoc to see how it did in a holistic way and for that we saw typically about 40 FPS average in the densest most NPC heavy areas of the city and it stretched up beyond that as you got out into the Wilderness finally for balers gate at 1440p medium the heaviest GPU load of this game that we're testing right now the r 78700 g memory scaling lessons the 8700g with a 490 versus the 8600 G with a 490 produces a 4.7% gap whereas the 8700g with igp and ddr5 6400 memory versus the 8600 G and the same is a 16% Gap from 18 to 21 FPS that shows us that although the CPU does does contribute to limitations at the top end the GPU or the igp has a much bigger impact Rainbow Six Siege is up now the data here has changed and isn't comparable to our October data set so all that's been removed the 7600 X on the 490 set the ceiling at 640 FPS average there'd be room with the 490 to scale for say a 7800 x3d so we're still CPU Bound in these entries the 8700g with a 490 achieves 467 FPS average illustrating the limits of the 8700g CPU component the core frequency ends up the limiting factor here with DDR 55600 there's an 18% gap between the 6,400 results 82 FPS average and it's 69 FPS average memory really matters for the igp and we're seeing that repeatedly here going to ddr5 6000 from DDR 55600 is a 7% Improvement and from there to 6,400 it's 10% Like for like the 8700g leads the 8600 G by 16% when both use ddr5 6400 memory and igps that's a 7% Gap when both use RTX 490s as for the RX 6600 that leads the 8700g similar price part by 162% with the 6500 XT and 1200f leading the 8600 G by 86% imposing a heavier GPU constraint pushes the 8700g and 8600 G entries with a 4090 closer together so either would scale similarly if adding the same DGP later for this game the 5600 result sets the floor at 42 FPS average giving away way to the 6,000 result at 7.4% uplift like for like the 8700g leads the 8600 G by 15% the price equivalent options still lead significantly but it may not be long before AMD gets to Apu levels that equal parts like the 6500 x t there there's a t in there and in either case the apus are actually in playable territory especially at the lower 1080P or with a settings drop to medium here Starfield is up next tested at 1080p low for this one with these settings the 800g using DDR 56400 memory landed at 30 FPS average with the DDR 56000 kit at 28.6 the DDR 56400 result is about 6.3% higher than the 6,000 result and like for like against the 8600 G we're seeing a significant uplift moving to the 8700g as for discret Solutions the RX 6600 outperforms the apus by about more than double once again the 6500 XT has been removed from this data because it was was functionally unplayable we noticed during actual gameplay that as we played longer the frame pacing became just completely untenable and the results were all over the place F1 2023 is up now this is another good testing title because it shows a type of game that's playable with relatively high graphics quality on the apus the 8700g overclock makes return here at 57.4 FPS average with our PBO application GP that puts it ahead of the Baseline 6400 result at 54 FPS by 6% the Baseline result leads the 6,000 result by 5 1.2% which itself leads the 5600 result by 7% the 8600 G is just below that entry with the like for like comparison producing a 20% lead for the 8700g the RX 6600 configuration has a significant Advantage largely attributable as you'd expect to the GPU the 7600 a isn't hitting a GPU limit yet either we could go hundreds of FPS Beyond this and so that shows there's still scaling room in the test if the other parts keep up the 8700g with a 490 depicts the 7600 X as able to unleash an additional 22% performance from better utilizing that GPU F1 2023 at 1440p gives us a more GPU intensive look at performance here the 7600 X is at about the same level as before showing us that we're CPU bound at the very top of the chart the 8700g with an overclock led the Baseline result by 4.7% at 37.7 to 36 FPS average the Baseline DDR 56400 configuration outperformed the 6000 config by 5% with the DDR 56000 option now doing the 5600 option by 7.2% the 8600 G lands below that at about 31 FPS average and giving the 8700g with ddr5 6400 a lead of 18% here two discret parts still lead here at 330 bucks but again there are tradeoffs that may not be compatible with a build of that style there's also more of a CPU limit if upgrading the GPU with that I3 in the future so something like a 5600 might start to make sense but but apus potentially end up in form factors that wouldn't fit a GPU anyway so it depends what you're doing Final Fantasy is up next this one validated at the same performance levels as when we tested it in October so old data can be shown and is still intact but of course everything for the apus is brand new for scaling the 8700g ran at 45 FPS with the 6,400 kit or 41.6 with the 6,000 kit it's about a 9% Gap the 8600 G is next in line at 41 FPS average so there's a 10% swing between this and the 8700g in this particular Tas case which is relatively low compared to some of the others likely it's the memory binding us up and to help further solidify that hypothesis we can look at the 2490 entries this is why they're here the 8600 G and the 8700g are within error of each other when using 409s both of them are at about 150 FPS average which indeed confirms that we are more Bound by a non-cpu component and it's not the GPU so the only component left really is the memory because the 490 is definitely not limiting us the 7600 X and the 78 x3d help illustrate that the 8700g at 5600 also further amplifies the memory limitation side of the story producing a large 15% swing as compared to the 6400 8700g entry on this chart the new apus are at about 3700x levels for CPU performance at 1440p the 8700g Baseline performance was 26.6 FPS average yielding to the 6500 XT and the 6600 in a significant way again as an APU though compared to past launches this is impressive to come from one chip the memory scale meanwhile is 7.3% over the 6,000 entry and 11% over the 5600 entry at the top again we're limited on the apus the recently tested 7600 X helps establish the performance of a possible desktop part instead so this product is a weird one it's always been the case with the apus it's been similar conclusion for like a decade now although there have been times where they've broken out especially for the early htpc days uh but it it just first thing out of the way is that it will not make more sense than a cheap CPU and a cheap GPU especially if you're willing to consider used but we didn't even have to and we still blew it out of the water with performance uh if all you want is the cheapest possible gaming computer if you're looking for uh basically the maximum amount of money that I have to spend on the CPU and the GPU together is under $400 you actually can put something together and it's really not bad at least not bad if you're willing to play mostly 1080p maybe some 1440 in there depending on the game the load not really any RT to speak of but you that is a a massive gaming segment what I just described uh and you can do it for under $400 the 65 XT really really do not like that GPU to just be super clear but if you had to buy new you're not willing to do used uh and there's absolutely no room in the budget then that plus the I3 1200f just as an example sometimes looking at like two exra performance or very nearly in most cases going up a step to the 6600 which is a much better video card and lately has been around $200 uh that plus an I3 12100 f is a phenomenal pairing and actually you could do even better still and you could go get an R5 5600 and the 6600 you'd be at about $330 so it's the same price as 8700g uh and really good performance so if all you care about is gaming performance and nothing else is a consideration uh other than price then that's the direction to go that does not invalidate the product though so uh another aspect that's worth considering is a build without any space for a video card at all I think that's probably what a lot of the people who want to use this in the audience are thinking about is one of those htpc setups where it's a very powerful and capable single solution and and this is a competent performer both of them but the 87 in particular but a capable solution that doesn't even have two slots or one slot for a GPU is just a small box of some kind uh and for some reason you don't want it to be a soldered solution from a a provider that uses an existing uh BGA option or something instead maybe you want socket ability for future upgrades whatever the case may be but for those who have a use case for it like an extremely small system again uh it does well the thermal and the power envelope are impressive here AMD maintains basically among the best efficiency uh and out does Intel for efficiency still and like for like considerations Intel's igp we actually did test it and we didn't bother putting it in any of the charts because it was it was so far down sometimes it was completely unplayable uh it's irrelevant so it is not really a direct comparison like the 14900 K UHD 770 or whatever it is not a linear comparison to these uh it is a significant underperformer so you might as well compare the 79 50x to the 14900 k at that point with igps only that makes more sense than this does different class memory scaling absolutely matters uh if you are going to buy one of these apus and especially if you're going to be playing games with it that's important we would strongly recommend that you buy as close to or at 6,400 as possible 6,000 was okay but actually in some of the games the gap between 6,000 6,400 with the two kits we have remember timings matter to we didn't really get into those today but the gap between those sometimes is like 5 six% so commonly 3% but five or 6% is kind of a lot uh and the Gap from 5600 to 6,400 was sometimes closer to like 19% so you really want to assign some budget for the memory we tried 1 7,400 kit in the time we had we just we couldn't get good performance out of it it was actually worse than the 6400 kit and that was because the timings were just so loose to hit that uh and Frankly Speaking we don't I I personally don't have a ton of memory tuning experience done it we've done some live streams with it but not particularly skilled in heavily tuning memory uh so that's my disclosure to you if you're more into memory OC or something go for it you can get a ton out of these and maybe we'll try that in the future uh but just 6400 if you're trying to use something out of the box is it seems like a good Baseline for these so that's kind of it for the review the short version of the conclusion is if you just want cheap gaming and you have absolutely no other considerations like form fact or whatever future upgradeability options then don't buy this go buy a cheap GPU and CPU and actually you really seriously look at eBay or something for a cheap GPU because that would get you even better than what we were testing here today uh if you go used and you're okay with it and then for those of you who know you want this you have a build idea you got a project uh or you want something really small form factor it seems fine it's not offensive in any particular way the thermal power envelope is good so uh if you want to use it for your project it it there aren't a ton of major considerations other than the memory that's it for this one thanks for watching if you want to support our testing as always go to store. Gamers access.net and please grab one of these shirts these have the memory module exploded on the front which is particularly suitable for the apus given the memory Reliance so go grab one on the store it helps fund our testing efforts directly uh and that's it for this one we'll have another review here in a day or two and we'll see you then

Channel: Gamers Nexus

Views: 365,404

Keywords: gamersnexus, gamers nexus, computer hardware, amd apus, amd 8000g, amd r7 8700g review, amd r5 8600g review, amd ryzen 7 8700g benchmarks, amd ryzen 5 8600g benchmarks, amd r5 8600g worth it, amd r7 8700g worth it, amd r7 8700g vs rx 6600, amd rx 6500 xt vs 8600g, amd apu vs gpu and cpu

Id: MFzegmwHxPM

Length: 35min 24sec (2124 seconds)

Published: Tue Jan 30 2024