The Point is Proven: CPU is a Waste of Silicon at its Current Pricing

In our production benchmarks, we have proven that this CPU is a waste of silicon at its current pricing. The point is proven when you consider power consumption. Our test showed that the EPS 12 volt cables and using the same motherboard with the 3700 X results in a power draw of 85 watts after five minutes, while the 3800 X and 3,800 XT pulled 118 watts - that's 39 percent more power for almost nothing to show for it. That's just bad.

It's not all doom and gloom, though. If we could make this CPU disappear, we would. Because the 3800 XT is a waste of silicon. It's so bad that we can't even bring ourselves to recommend it. In fact, if we could replace it with something else from AMD's lineup, we would do just that.

Andy is in an interesting spot because it has gotten good enough now where it's competing with itself. It's also become a real enough competitor in this landscape that's been almost entirely Intel for a decade up until the rise of Ryzen and the rest started happening. Now Andy is starting to play by the same rule book that Intel has - just stuff the market with as many products as you can to try and flood it.

Cynical Viewpoint

We have a cynical view of this whole situation, which we've publicly talked about. We think maybe they're trying to flood the stack so that when we compare two products, we're only comparing two Nvidia products because that's just how much stuff they have there. You start to either forget about or not have a direct price comparison for the competition. If AMD is trying to do that, it didn't work here.

But what if we were being too harsh? Maybe AMD is actually trying to raise ASP - average selling price. And if that's the case, then maybe the 3900 X is where it's at. The 3800 XT is a don't buy it. The silicon quality is better, but not in a meaningful way unless you're really serious about overclocking.

Overclocking is Where it's At

But for those who are serious about overclocking - like our reviewer was with the 3600 XT - this chip is awesome. It hits for point whatever depending on the CPU, and there's a good chance you can push it to that for all core. So, if you're really serious about tweaking and tuning stuff, maybe the 3800 XT is worth considering.

However, for overclocking enthusiasts who want the best, we have to say that once again, the 3900 X makes way more sense. Our concern now is that Andy is competing against itself as much as it's competing with Intel. It's also doing so to the extent that it might start getting more negative attention and critical attention as a result of it.

Verifying AMD's Claims

AMD says the 3900 XT is 4 percent better than its competitors, but we'll verify that for you. The best-case scenario is around 3.7 percent improvement. Not bad, not great - but still an improvement. So, if you're looking to upgrade and want a chip that will make a difference in your system's performance, maybe the 3900 XT is worth considering.

But for most people, we have to say skip this one. The 3800 XT is not worth buying, and the 3900 X is where it's at. Thanks for watching, and don't forget to support our content directly by going to our Patreon page or subscribing to our channel for behind-the-scenes videos and patrons-only questions.

"WEBVTTKind: captionsLanguage: enwe get it Andy you can make a 3700 X just like Nvidia can make a 1080 TI three times AMD wants to make sure everyone knows that it can keep making 3700 axes this time the 3700 X is called the 3800 XT which is one of them these three new processors that launched today the 3600 XT reviews already on the channel we have a whole lot to say about that one we've got a really dense section on frequency voltages maximum frequency minimum voltage and the comparative silicon quality changes versus previous 3600 series CPUs but this one we're more laser focused on a few key items primarily the 3800 XT as it compares to the 3900 X another AMD part but which happens to be priced the same now at $400 for each one of that before that this video is brought to you by us and the GNX 570 chipset metro poster the chipset metro poster is an educational look at the i/o capabilities of X 570 and it's company and CPUs laid out with artistic creativity as a subway or a metro map this map accurately depicts the X 570 chipsets PCIe USB SATA and other i/o features each with its own stop along the subway line with the chipset and the CPU represented as major hubs we aren't making many of these and have already sold through 70% of the posters we ordered so if you'd like to get one visit store dock gamers Nexus botnet or click the link in the description below so I've stated in the 3600 xt review this is a business move by AMD Andy is trying to get its average selling price back up its ASP needs to increase and there's not really a ton of change in terms of the silicon you're getting with the extra T at the end that you're paying for but there's a silicon quality change and that can be valuable which again is discussed in great detail in the 3600 xt review you should watch that one for the full story this one's gonna be more focused on just the one part rather than setting the whole stage so the 3800 XT is well the 3800 x we should start with that's one where we said don't waste the money that was our verdict for the 3800 X we thought it was a waste of money and that was especially as compared to the 3900 accent just a little bit more the 3700 X which we were kind of iffy on too for a lot of other reasons but another 3,600 which we really liked so our positioning was you know 3,600 makes a lot of sense 3,900 X makes a lot of sense both of them are incredibly well positioned on the market and then the 3700 X we thought yeah for the most part if you can afford it you should really either go 3900 X or just stick with the 36 so that was our feeling on it overall the 3600 went on to be one of the best sellers of Andy's entire lineup and the 3900 X has gotten a 3900 XT so obviously it was well liked also the 3800 XT is weird it's it's $400 4.7 gigahertz claimed maximum boost frequency that's a limited core boost so that's not all core 3.9 base it's about 200 megahertz max boost over 3800 X but several problems exist here one is that there aren't that many workloads that are that limited in thread count and when they happen intel has other advantages like in those gaming scenarios that add up to give it the gap that it still has for gaming performance so since this isn't an architectural change it's not a huge tremendous overhaul as for the rest of the CPU we'll just quickly throw a few stats out there our 3,800 XT did hit 1900 infinity fabric for the clock and we were able to do 4.6 gigahertz all core the voltages were improved over the 3800 X 3700 X so just like we talked about the 3,600 X TPS it seems like silicon quality has gone up but that's not the only reason you should really buy one of these especially because the 3900 ax is going to be straight better in a lot of scenarios like workstation and production tasks this one's gonna be quick today let's just get through the gaming and the production benchmarks we'll talk about power as well that's significant here because this is horribly inefficient for the performance that you're getting out of the XT and then we'll talk about the conclusions and what this really means for AMD for games we'll start with the Three Kingdoms grand campaign strategy test the 3800 acts wasn't worth it when it first launched but like other isin xcp is it later become more sensible as the price was pulled down towards equivalence was whatever it was supposed to replace in this case the 3700 X this one enters in at the pricing of the original 3800 acts more or less so we're back to that positioning of probably not really being where the 3,700 X does 125 FPS average here with Louis at 87 and 78 for 1% 10.1% lows the 3600 XT is about the same and even though 3300 X is more or less identical as we sat in the 3300 x review & r3 is enough for gaming in the current climate the 3,800 X only jumps to one 25.8 FPS average as compared to the 3700 ax so that's a gain of 0.4% and is clearly within test variance or under on the 3,800 XT gains 1.7 percent over the 3,800 X or a 2.2 percent over the 3700 X in other words that's about $64 per percentage point of improvement not great value this is starting to remind us of some of Intel's launches where it's just sort of boring meanwhile on the Intel front the 10 600 K for reference leads the 3,800 XT stock to stock by 7.7 percent the lows are in favor of the 10th 600 K but not meaningfully different between them overclocking the 10 600 K without a heavier tune gets it to 146 FPS average or 12.8% ahead of the 4.6 gigahertz 3,800 XT as for frame times the only place they're disproportionately low is with the two older 7700 K for core 8 thread entries the rest of the low entries here are proportionate to the averages that we're seeing everyone 2019 is next useful for its extreme CPU scaling at 1080p the 3,800 x stock CPU ran at 238 FPS average growing by 0.3 percent so error as opposed to the 3,800 XT the 3800 XT at 4.6 gigahertz limps to 239 FPS average or 0.5% ahead of the stock 3,800 X it's just under the 7700 can 5.1 gigahertz and the 10 600 K stock leads to 3,800 XT stock by 8.3% OC vs. OC the 10 600 K at 267 FPS average leads the 3,800 XT OC by 12% when we don't do any advanced tuning to either CPU again you can check back for custom tuning in a separate video later the 3700 X 4 reference runs around equal to the 3800 X and the XT it's with an error of the 3,800 X and could trade places based on run - run variance the 3300 acts more impressively is able to keep up with these results the extra cores don't help here at least not much but the single CCX configuration is beneficial hitman two scales with both cores and frequency as the ten 900k and the ten 600k especially its 5.1 gigahertz number demonstrate clearly the 3,700 acts completed this benchmark while holding an average of 118 FPS with a 3,800 X offering no meaningful change and the 3,800 XT gaining at Most Wanted a half percent more performance over the 3700 X lows are the same the 10 600 K stock CPU ran at 128 FPS average leading the 3,800 XT by 7 percent stock to stock or 13% OC to OC as for the 3900 X that one's more interesting here just because it shows an increase in performance where we don't typically see one in gaming scenarios it's pushing 124 FPS average so although that's clearly not good value it does illustrate that extra cores can be beneficial in this title we see that also with the 3800 XT is 2.2 percent lead over the 3600 XT Red Dead Redemption 2 is next with our more limited data set at medium graphics settings first up we haven't fully retested everything here as stated previously so this is the working data set we have for now the 3,800 x stock CPU held 126 FPS average with lows at 92 and 81 that places it above the 3600 xt stock results of 121 FPS average by about four point four percent and it ties with the 2017 7700 K at five point one gigahertz the XT CPU offered no meaningful benefit as a reminder even Andy's own reviewer guide often shows differences that are plus or minus 1 FPS which means they probably just ran reran the tests over and over until they could pick a number to show a slight improvement the 10 600 K leads by a massive margin here at about 14% ahead with its 143 FPS average stock result like we said about the 3700 acts and the 3,800 accent launch it seems like the 3600 or maybe now the 3600 XT or the 3300 acts for that matter would be better buys unless those extra cores are needed explicitly for something like workstation applications and the 3600 acts he would really be worth it if you like overclocking which is something we talked about in the review for the 3600 xt GTA actually shows direct scaling for once and isn't entirely relegated to error that's because it's more thread limited and so this proves a valuable benchmark for establishing where the differences actually lie for these CPUs you basically need to play a game that's like this where it's not really leveraging all those threads in order to see a benefit from higher boosting frequencies and these types of games are becoming increasingly rare all core frequencies aren't as significantly different as the single or dual core load frequencies are as we showed previously the 3,800 XT places at 1:12 FPS average stock so it's 1.8 percent ahead of the 3800 X or 2.7 percent ahead of the 3700 X the 3800 XT OC at 1 12.8 FPS average almost meets the stock 10 600 K which then jumps ahead with its own basic overclocked Assassin's Creed has always been interesting for its more balanced approach to thread and frequency dependencies depending on which part of the game you're analyzing in this test the 3700 ax runs at 119 FPS average which is within error of the 3600 XT and it's higher single core boosting clocks the 3800 X runs at 123 FPS average stock that positions the 3800 acts at a riveting earth-shattering 3.9% ahead of the 3700 X which is the biggest increase we've seen thus far from this otherwise boring set of chips the 3,800 XT gains on that by about 0.2 FPS average which is error there is no provided benefit in our benchmark here the over clogs variance is also under one FPS average ahead and for reference the 10 600 K stock runs at 128 FPS average while the 10900 K establishes our ceiling at 141 FPS average one stop production benchmarks are up now which stressed the CPUs and often have reduced focus on frequency that's especially true in tile-based renderers like blender or Chaos groups v-ray which we use in-house for designs like our wireframe mouse mat design where everything was 3d modeled and rendered in blender and the cycles renderer we're using that same renderer for our benchmarks here although with different images the monkeyhead render completes in 13.9 minutes 3,700 acts proving that the 3,800 acts at thirteen point seven minutes remains a horrible value at the original pricing that's a mind-blowing extremely exciting time reduction of 1.4 percent the 3,800 XT goes down to thirteen point four minutes a reduction against the 3,700 X of three point six percent and against the 3,800 acts of 2.2% the overclock gets us to twelve point seven minutes a 5.2 percent improvement over the 3,800 XT stock results at the price the 3,800 XT is a complete and utter waste of money for production applications at least the 3600 XT had some may be sort of sometimes kind of uplifting qualities but the 3,800 XT is embarrassingly bad for the price and the modern pricing structures of the 3900 acts make that especially so the 3900 acts it now cost about $400 that's the 3800 XT pricing the 3900 acts is also for more cores and finishes this render in 28% less time than the 3,800 XT but it costs the same the 3,800 XT is therefore a joke and it shouldn't have been launched it just stands to make AMD look bad against its own products the GN logo render moves a couple things around in the hierarchy and blender but it's otherwise similar the 3,800 XT ends up at sixteen point two minutes to complete the render versus sixteen point five for the 3,800 X and sixteen point nine for the 3700 X overclocking gets us to fifteen point two minutes just ahead of the ten core twenty thread and ten 900k the 3900 acts more importantly does all of this work in eleven minutes a reduction of 29% in time required against the 3,800 XT again that's just embarrassing for the 3,800 XT and for AMD Adobe Premiere rendering is next using real gamers access YouTube videos for render time benchmarks the 1080p video is simple and short film that a convention center as a news report the 3700 X finishes this render in 3.8 minutes adjacent to the 9900 K and the 3,800 X there's no tangible non error uplift here with the 3,800 acts instead the 3,800 XT not shockingly is also the same the 10700 k is also the same as this although the 10600 k dragged somewhat distantly behind at five minutes the 3,800 xt OC runs 3.6 minutes again not exciting finally the 3900 X actually posts a meaningful uplift at three point four minutes as compared to something like the 3700 X or the 3,800 XT stock value that means the 3900 acts out matches the stock 3,800 XT by 11% and costs the same remember that this scales against video length if we were doing a more complex a roll shot that's half an hour long like most of our videos but at 1080 P instead 10 to 11% might be something like 5 minutes faster that starts to add up for the 4k render the timescale stretches and positions the 3,800 acts XT and X EOC all at about the same relative positioning we've really proven the point by now but to drive at home the 3900 ax stock result finishes in 18% less time than the 3,800 XT stock result the 3700 X being equal to the 3800 acts and XT just further shows what a waste the XT is for this type of application there's zero benefit to it our next test is 7-zip decompression measured in millions of instructions per second or MIPS higher is better for decompression the 3,700 X and 3800 acts were with an error of each other the 3,800 XT scored all of 1% higher and the 3,800 XT OC gained about 6% over its stock result that said the 3900 X puts both to absolute shame it's at 145 thousand bits versus the 100,000 mips range of the 3700 X 3800 acts and 3,800 XT at 145 km it's the 3900 acts at $400 achieves 44% more performance than the r7 3,800 XT but at the same price it really doesn't get much worse than that if this were an Intel part no doubt that the AMD fan base would be dragging its bloody corpse through the streets for months on end the 3,800 XT is embarrassed thoroughly by the 3900 X chromium code compile is our next test this one uses clean CL and ninja to compile the chromium code base and its measured in time to complete the compile the 700x finishes the compiled in 91 minutes with the 3800 x stock at about 88 minutes that's an improvement of about three point four percent less time to compile and the 3800 xt was about 2.4 percent less time than the 3800 X that's the riveting edge-of-your-seat style uplift you can expect out of this CPU the 3900 X meanwhile completes the compile in 25% less time than the same price 3800 XT stock result at 65 minutes 286 minutes for reference to Intel the 10 700 K is probably the closest competitor nearby in price but at five point one gigahertz it can't keep up with the 3900 X if you're considering the 3,800 XT for a mix of workstation tasks and a focus on gaming you might as well get an Intel CPU instead otherwise get the 3900 X and break transcoding is next this one will have to rework eventually in our next methodology overhaul but it still scales well enough at these thread counts going higher becomes a bit of a challenge the 3900 ax completes the trance code in twelve point eight minutes while the 10 700 K stock requires fourteen point six minutes that out does the 3,800 XT stock results up fifteen point one minutes that positions the 3900 X has about 15% faster than the 3,800 XT stock here and the 3,800 XT stock is about the same as the 3700 X and 3800 X both of which are at fifteen point four minutes we're going to have to stop there for production benchmarks because the point is proven this CPU is a waste of silicon at its current pricing power consumption really starts to tell the story test it out the EPS 12 volt cables and using the same motherboard the 3700 X and blender after 5 minutes so after tau for Intel polled 85 watts at the cables the 3800 X and 3,800 XT pulled 118 watts for this that's 39 percent more power for almost nothing to show for it that's just bad all of that said then if we could just and make it disappear we would because the 3800 xt is a waste of silicon it's so again that could be something else instead something that was from AMD's on line up so Andy's in an interesting spot because it's gotten good enough now where it's competing with itself it's also become a real enough competitor in this landscape that's been almost entirely Intel for a decade up until the rise and stuff started happening it's become a real enough competitor here where Andy's starting to play by the same rule book that Intel has which is to just stuff the market with as many products as you can to try and flood it there's a few things here if we're being cynical of stuff then the same kind of viewpoint we have of one Nvidia does this might apply to AMD when Nvidia launches functionally the same product over and over and over with a different name and sometimes a slightly different price our cynical look that we've publicly talked about is that maybe they're just trying to flood the stack so that when we compare two products we're only comparing two Nvidia products because that's just how much stuff they have there you start to either forget about or not have a direct price comparison for the competition and if AMD is trying to do that well it didn't work here but we could see maybe an argument for that where it's just filling the stack with things for just just more AMD's more get in people's faces more listings on product websites whatever but realistically what we think they're actually doing outside of the sort of weird reviewer conspiracy type stuff is probably trying to raise ASP and for that we'd say by the 3900 X instead so yeah 3800 XT is a don't buy it it's a waste of money the silicon quality is better but not in a meaningful way unless you're really serious about overclocking so if you're really serious tried overclocking like you said about the 3600 xt 3600 xt is an awesome chip at least ours was and this should apply mostly universally awesome chip for overclocking enthusiasts people want to tweak and tune stuff maybe you're competitive with it whatever it's it's great for that because it's been you know that it hits for point whatever depending on the CPU but 4.6 4.7 on at least a couple of the cores there's a good chance you can push it to that for all core so that's cool but free of box stock usage you really shouldn't be buying the 3800 XT or the 3600 XT and frankly for overclocking usage you shouldn't really be buying the 3800 XT anyway because once again the 3900 X just makes way more sense so our concern now is that although andis competing against itself as we've been talking out for over a year at this point it's also doing so to the extent that it might start kind of getting more negative attention more critical attention as a result of it whereas before it was like okay great well whatever its it beats the other AMD parts and it's meaningful and it's not a ripoff so yeah that's fine whatever sure not super exciting necessarily but at least it's a meaningful and grooving now we're talking about sometimes less than 1% like 3 or 4 at absolute best that's AMD by the way AMD says 4 percent better it's and that's we will verify that it's like 3.7 in the best-case scenario so if that's their advertising it's not gonna be better than that so at least not not commonly so yeah skip this one the 3900 XT we'll talk about that separately but so far overclocking is where it's at for these so thanks for watching you can support this type of content directly because AMD sure as hell won't want to by going to support our cameras excess net you can pick up things like mod mats our wireframe mouse mat is on backorder and coming back in very quickly at this point or you can pick up the x5 70 chips at metro poster that we have just recently restocked and we're shipping now alternatively patreon.com/scishow and resist for behind the scenes videos and patrons ask Gaea and subscribe for more thanks for watching we'll see you all next timewe get it Andy you can make a 3700 X just like Nvidia can make a 1080 TI three times AMD wants to make sure everyone knows that it can keep making 3700 axes this time the 3700 X is called the 3800 XT which is one of them these three new processors that launched today the 3600 XT reviews already on the channel we have a whole lot to say about that one we've got a really dense section on frequency voltages maximum frequency minimum voltage and the comparative silicon quality changes versus previous 3600 series CPUs but this one we're more laser focused on a few key items primarily the 3800 XT as it compares to the 3900 X another AMD part but which happens to be priced the same now at $400 for each one of that before that this video is brought to you by us and the GNX 570 chipset metro poster the chipset metro poster is an educational look at the i/o capabilities of X 570 and it's company and CPUs laid out with artistic creativity as a subway or a metro map this map accurately depicts the X 570 chipsets PCIe USB SATA and other i/o features each with its own stop along the subway line with the chipset and the CPU represented as major hubs we aren't making many of these and have already sold through 70% of the posters we ordered so if you'd like to get one visit store dock gamers Nexus botnet or click the link in the description below so I've stated in the 3600 xt review this is a business move by AMD Andy is trying to get its average selling price back up its ASP needs to increase and there's not really a ton of change in terms of the silicon you're getting with the extra T at the end that you're paying for but there's a silicon quality change and that can be valuable which again is discussed in great detail in the 3600 xt review you should watch that one for the full story this one's gonna be more focused on just the one part rather than setting the whole stage so the 3800 XT is well the 3800 x we should start with that's one where we said don't waste the money that was our verdict for the 3800 X we thought it was a waste of money and that was especially as compared to the 3900 accent just a little bit more the 3700 X which we were kind of iffy on too for a lot of other reasons but another 3,600 which we really liked so our positioning was you know 3,600 makes a lot of sense 3,900 X makes a lot of sense both of them are incredibly well positioned on the market and then the 3700 X we thought yeah for the most part if you can afford it you should really either go 3900 X or just stick with the 36 so that was our feeling on it overall the 3600 went on to be one of the best sellers of Andy's entire lineup and the 3900 X has gotten a 3900 XT so obviously it was well liked also the 3800 XT is weird it's it's $400 4.7 gigahertz claimed maximum boost frequency that's a limited core boost so that's not all core 3.9 base it's about 200 megahertz max boost over 3800 X but several problems exist here one is that there aren't that many workloads that are that limited in thread count and when they happen intel has other advantages like in those gaming scenarios that add up to give it the gap that it still has for gaming performance so since this isn't an architectural change it's not a huge tremendous overhaul as for the rest of the CPU we'll just quickly throw a few stats out there our 3,800 XT did hit 1900 infinity fabric for the clock and we were able to do 4.6 gigahertz all core the voltages were improved over the 3800 X 3700 X so just like we talked about the 3,600 X TPS it seems like silicon quality has gone up but that's not the only reason you should really buy one of these especially because the 3900 ax is going to be straight better in a lot of scenarios like workstation and production tasks this one's gonna be quick today let's just get through the gaming and the production benchmarks we'll talk about power as well that's significant here because this is horribly inefficient for the performance that you're getting out of the XT and then we'll talk about the conclusions and what this really means for AMD for games we'll start with the Three Kingdoms grand campaign strategy test the 3800 acts wasn't worth it when it first launched but like other isin xcp is it later become more sensible as the price was pulled down towards equivalence was whatever it was supposed to replace in this case the 3700 X this one enters in at the pricing of the original 3800 acts more or less so we're back to that positioning of probably not really being where the 3,700 X does 125 FPS average here with Louis at 87 and 78 for 1% 10.1% lows the 3600 XT is about the same and even though 3300 X is more or less identical as we sat in the 3300 x review & r3 is enough for gaming in the current climate the 3,800 X only jumps to one 25.8 FPS average as compared to the 3700 ax so that's a gain of 0.4% and is clearly within test variance or under on the 3,800 XT gains 1.7 percent over the 3,800 X or a 2.2 percent over the 3700 X in other words that's about $64 per percentage point of improvement not great value this is starting to remind us of some of Intel's launches where it's just sort of boring meanwhile on the Intel front the 10 600 K for reference leads the 3,800 XT stock to stock by 7.7 percent the lows are in favor of the 10th 600 K but not meaningfully different between them overclocking the 10 600 K without a heavier tune gets it to 146 FPS average or 12.8% ahead of the 4.6 gigahertz 3,800 XT as for frame times the only place they're disproportionately low is with the two older 7700 K for core 8 thread entries the rest of the low entries here are proportionate to the averages that we're seeing everyone 2019 is next useful for its extreme CPU scaling at 1080p the 3,800 x stock CPU ran at 238 FPS average growing by 0.3 percent so error as opposed to the 3,800 XT the 3800 XT at 4.6 gigahertz limps to 239 FPS average or 0.5% ahead of the stock 3,800 X it's just under the 7700 can 5.1 gigahertz and the 10 600 K stock leads to 3,800 XT stock by 8.3% OC vs. OC the 10 600 K at 267 FPS average leads the 3,800 XT OC by 12% when we don't do any advanced tuning to either CPU again you can check back for custom tuning in a separate video later the 3700 X 4 reference runs around equal to the 3800 X and the XT it's with an error of the 3,800 X and could trade places based on run - run variance the 3300 acts more impressively is able to keep up with these results the extra cores don't help here at least not much but the single CCX configuration is beneficial hitman two scales with both cores and frequency as the ten 900k and the ten 600k especially its 5.1 gigahertz number demonstrate clearly the 3,700 acts completed this benchmark while holding an average of 118 FPS with a 3,800 X offering no meaningful change and the 3,800 XT gaining at Most Wanted a half percent more performance over the 3700 X lows are the same the 10 600 K stock CPU ran at 128 FPS average leading the 3,800 XT by 7 percent stock to stock or 13% OC to OC as for the 3900 X that one's more interesting here just because it shows an increase in performance where we don't typically see one in gaming scenarios it's pushing 124 FPS average so although that's clearly not good value it does illustrate that extra cores can be beneficial in this title we see that also with the 3800 XT is 2.2 percent lead over the 3600 XT Red Dead Redemption 2 is next with our more limited data set at medium graphics settings first up we haven't fully retested everything here as stated previously so this is the working data set we have for now the 3,800 x stock CPU held 126 FPS average with lows at 92 and 81 that places it above the 3600 xt stock results of 121 FPS average by about four point four percent and it ties with the 2017 7700 K at five point one gigahertz the XT CPU offered no meaningful benefit as a reminder even Andy's own reviewer guide often shows differences that are plus or minus 1 FPS which means they probably just ran reran the tests over and over until they could pick a number to show a slight improvement the 10 600 K leads by a massive margin here at about 14% ahead with its 143 FPS average stock result like we said about the 3700 acts and the 3,800 accent launch it seems like the 3600 or maybe now the 3600 XT or the 3300 acts for that matter would be better buys unless those extra cores are needed explicitly for something like workstation applications and the 3600 acts he would really be worth it if you like overclocking which is something we talked about in the review for the 3600 xt GTA actually shows direct scaling for once and isn't entirely relegated to error that's because it's more thread limited and so this proves a valuable benchmark for establishing where the differences actually lie for these CPUs you basically need to play a game that's like this where it's not really leveraging all those threads in order to see a benefit from higher boosting frequencies and these types of games are becoming increasingly rare all core frequencies aren't as significantly different as the single or dual core load frequencies are as we showed previously the 3,800 XT places at 1:12 FPS average stock so it's 1.8 percent ahead of the 3800 X or 2.7 percent ahead of the 3700 X the 3800 XT OC at 1 12.8 FPS average almost meets the stock 10 600 K which then jumps ahead with its own basic overclocked Assassin's Creed has always been interesting for its more balanced approach to thread and frequency dependencies depending on which part of the game you're analyzing in this test the 3700 ax runs at 119 FPS average which is within error of the 3600 XT and it's higher single core boosting clocks the 3800 X runs at 123 FPS average stock that positions the 3800 acts at a riveting earth-shattering 3.9% ahead of the 3700 X which is the biggest increase we've seen thus far from this otherwise boring set of chips the 3,800 XT gains on that by about 0.2 FPS average which is error there is no provided benefit in our benchmark here the over clogs variance is also under one FPS average ahead and for reference the 10 600 K stock runs at 128 FPS average while the 10900 K establishes our ceiling at 141 FPS average one stop production benchmarks are up now which stressed the CPUs and often have reduced focus on frequency that's especially true in tile-based renderers like blender or Chaos groups v-ray which we use in-house for designs like our wireframe mouse mat design where everything was 3d modeled and rendered in blender and the cycles renderer we're using that same renderer for our benchmarks here although with different images the monkeyhead render completes in 13.9 minutes 3,700 acts proving that the 3,800 acts at thirteen point seven minutes remains a horrible value at the original pricing that's a mind-blowing extremely exciting time reduction of 1.4 percent the 3,800 XT goes down to thirteen point four minutes a reduction against the 3,700 X of three point six percent and against the 3,800 acts of 2.2% the overclock gets us to twelve point seven minutes a 5.2 percent improvement over the 3,800 XT stock results at the price the 3,800 XT is a complete and utter waste of money for production applications at least the 3600 XT had some may be sort of sometimes kind of uplifting qualities but the 3,800 XT is embarrassingly bad for the price and the modern pricing structures of the 3900 acts make that especially so the 3900 acts it now cost about $400 that's the 3800 XT pricing the 3900 acts is also for more cores and finishes this render in 28% less time than the 3,800 XT but it costs the same the 3,800 XT is therefore a joke and it shouldn't have been launched it just stands to make AMD look bad against its own products the GN logo render moves a couple things around in the hierarchy and blender but it's otherwise similar the 3,800 XT ends up at sixteen point two minutes to complete the render versus sixteen point five for the 3,800 X and sixteen point nine for the 3700 X overclocking gets us to fifteen point two minutes just ahead of the ten core twenty thread and ten 900k the 3900 acts more importantly does all of this work in eleven minutes a reduction of 29% in time required against the 3,800 XT again that's just embarrassing for the 3,800 XT and for AMD Adobe Premiere rendering is next using real gamers access YouTube videos for render time benchmarks the 1080p video is simple and short film that a convention center as a news report the 3700 X finishes this render in 3.8 minutes adjacent to the 9900 K and the 3,800 X there's no tangible non error uplift here with the 3,800 acts instead the 3,800 XT not shockingly is also the same the 10700 k is also the same as this although the 10600 k dragged somewhat distantly behind at five minutes the 3,800 xt OC runs 3.6 minutes again not exciting finally the 3900 X actually posts a meaningful uplift at three point four minutes as compared to something like the 3700 X or the 3,800 XT stock value that means the 3900 acts out matches the stock 3,800 XT by 11% and costs the same remember that this scales against video length if we were doing a more complex a roll shot that's half an hour long like most of our videos but at 1080 P instead 10 to 11% might be something like 5 minutes faster that starts to add up for the 4k render the timescale stretches and positions the 3,800 acts XT and X EOC all at about the same relative positioning we've really proven the point by now but to drive at home the 3900 ax stock result finishes in 18% less time than the 3,800 XT stock result the 3700 X being equal to the 3800 acts and XT just further shows what a waste the XT is for this type of application there's zero benefit to it our next test is 7-zip decompression measured in millions of instructions per second or MIPS higher is better for decompression the 3,700 X and 3800 acts were with an error of each other the 3,800 XT scored all of 1% higher and the 3,800 XT OC gained about 6% over its stock result that said the 3900 X puts both to absolute shame it's at 145 thousand bits versus the 100,000 mips range of the 3700 X 3800 acts and 3,800 XT at 145 km it's the 3900 acts at $400 achieves 44% more performance than the r7 3,800 XT but at the same price it really doesn't get much worse than that if this were an Intel part no doubt that the AMD fan base would be dragging its bloody corpse through the streets for months on end the 3,800 XT is embarrassed thoroughly by the 3900 X chromium code compile is our next test this one uses clean CL and ninja to compile the chromium code base and its measured in time to complete the compile the 700x finishes the compiled in 91 minutes with the 3800 x stock at about 88 minutes that's an improvement of about three point four percent less time to compile and the 3800 xt was about 2.4 percent less time than the 3800 X that's the riveting edge-of-your-seat style uplift you can expect out of this CPU the 3900 X meanwhile completes the compile in 25% less time than the same price 3800 XT stock result at 65 minutes 286 minutes for reference to Intel the 10 700 K is probably the closest competitor nearby in price but at five point one gigahertz it can't keep up with the 3900 X if you're considering the 3,800 XT for a mix of workstation tasks and a focus on gaming you might as well get an Intel CPU instead otherwise get the 3900 X and break transcoding is next this one will have to rework eventually in our next methodology overhaul but it still scales well enough at these thread counts going higher becomes a bit of a challenge the 3900 ax completes the trance code in twelve point eight minutes while the 10 700 K stock requires fourteen point six minutes that out does the 3,800 XT stock results up fifteen point one minutes that positions the 3900 X has about 15% faster than the 3,800 XT stock here and the 3,800 XT stock is about the same as the 3700 X and 3800 X both of which are at fifteen point four minutes we're going to have to stop there for production benchmarks because the point is proven this CPU is a waste of silicon at its current pricing power consumption really starts to tell the story test it out the EPS 12 volt cables and using the same motherboard the 3700 X and blender after 5 minutes so after tau for Intel polled 85 watts at the cables the 3800 X and 3,800 XT pulled 118 watts for this that's 39 percent more power for almost nothing to show for it that's just bad all of that said then if we could just and make it disappear we would because the 3800 xt is a waste of silicon it's so again that could be something else instead something that was from AMD's on line up so Andy's in an interesting spot because it's gotten good enough now where it's competing with itself it's also become a real enough competitor in this landscape that's been almost entirely Intel for a decade up until the rise and stuff started happening it's become a real enough competitor here where Andy's starting to play by the same rule book that Intel has which is to just stuff the market with as many products as you can to try and flood it there's a few things here if we're being cynical of stuff then the same kind of viewpoint we have of one Nvidia does this might apply to AMD when Nvidia launches functionally the same product over and over and over with a different name and sometimes a slightly different price our cynical look that we've publicly talked about is that maybe they're just trying to flood the stack so that when we compare two products we're only comparing two Nvidia products because that's just how much stuff they have there you start to either forget about or not have a direct price comparison for the competition and if AMD is trying to do that well it didn't work here but we could see maybe an argument for that where it's just filling the stack with things for just just more AMD's more get in people's faces more listings on product websites whatever but realistically what we think they're actually doing outside of the sort of weird reviewer conspiracy type stuff is probably trying to raise ASP and for that we'd say by the 3900 X instead so yeah 3800 XT is a don't buy it it's a waste of money the silicon quality is better but not in a meaningful way unless you're really serious about overclocking so if you're really serious tried overclocking like you said about the 3600 xt 3600 xt is an awesome chip at least ours was and this should apply mostly universally awesome chip for overclocking enthusiasts people want to tweak and tune stuff maybe you're competitive with it whatever it's it's great for that because it's been you know that it hits for point whatever depending on the CPU but 4.6 4.7 on at least a couple of the cores there's a good chance you can push it to that for all core so that's cool but free of box stock usage you really shouldn't be buying the 3800 XT or the 3600 XT and frankly for overclocking usage you shouldn't really be buying the 3800 XT anyway because once again the 3900 X just makes way more sense so our concern now is that although andis competing against itself as we've been talking out for over a year at this point it's also doing so to the extent that it might start kind of getting more negative attention more critical attention as a result of it whereas before it was like okay great well whatever its it beats the other AMD parts and it's meaningful and it's not a ripoff so yeah that's fine whatever sure not super exciting necessarily but at least it's a meaningful and grooving now we're talking about sometimes less than 1% like 3 or 4 at absolute best that's AMD by the way AMD says 4 percent better it's and that's we will verify that it's like 3.7 in the best-case scenario so if that's their advertising it's not gonna be better than that so at least not not commonly so yeah skip this one the 3900 XT we'll talk about that separately but so far overclocking is where it's at for these so thanks for watching you can support this type of content directly because AMD sure as hell won't want to by going to support our cameras excess net you can pick up things like mod mats our wireframe mouse mat is on backorder and coming back in very quickly at this point or you can pick up the x5 70 chips at metro poster that we have just recently restocked and we're shipping now alternatively patreon.com/scishow and resist for behind the scenes videos and patrons ask Gaea and subscribe for more thanks for watching we'll see you all next time\n"