Hard to Justify - Intel Core i7-10700K CPU Review & Benchmarks vs. 3900X, 3700X, 10600K
**Article: In-Depth Review of Intel i7 10-700K CPU**
---
### Introduction to the Intel i7 10-700K
Patrick and I purchased the highly anticipated Intel i7 10-700K on launch day, setting off a round-the-clock effort to complete this review under 24 hours. The journey began with receiving the CPU through the mail, marking the start of an intense testing phase.
### Pricing and Market Positioning
The Intel i7 10-700K was purchased for $410, situating it between the more affordable AMD Ryzen 3600 ($290) and the premium Intel i9-10900K ($500+). On Amazon, competitors like the Ryzen 3900X were priced at $420, establishing a competitive landscape.
### Thermal Performance and Frequency Behavior
Intel's die-sanding approach has shown improvements in thermal performance. The i7 10-700K exhibited a maximum quarter-core delta of 6°C under full load (134W), with peak temperatures reaching 9°C across all cores. This marks an enhancement over previous generations, though uncertainties remain regarding solder variants.
### Gaming Performance
#### Total War: Three Kingdoms
The i7 10-700K demonstrated impressive gaming performance, achieving 166 FPS at stock settings. Overclocking to 5.1 GHz yielded 147 FPS, closely matching the i9-10900K's 148 FPS.
#### GTA V
In Grand Theft Auto V, the CPU showed consistent frame rates, with minor gains from overclocking. The Ryzen 3900X led at 162 FPS, while the i7 10-700K trailed slightly at 150 FPS.
#### Tomb Raider: DX12
The i7 10-700K struggled marginally in this title, achieving 140 FPS stock and 153 with overclocking. This highlighted limitations under heavy GPU constraints.
### Production Applications
#### Blender Render Times
The i7 10-700K completed a GN MonkeyHead render in 14 minutes, matching the Ryzen 9 2700X. Overclocking to 5.1 GHz improved times by up to 10%, but the Ryzen 3900X remained superior at 9.6 minutes.
#### Adobe Premiere Pro
Rendering a 4K project took 11 minutes stock and 10.5 with overclocking, trailing the Ryzen 3900X's 8.9 minutes.
### Power Consumption
Under full load, the i7 10-700K drew 268W, less than the i9-10900K's 280W despite higher voltage. Idle power consumption was manageable at 134W, with efficient performance under light loads.
### Compression and Decompression
The CPU achieved impressive results in Handbrake, finishing transcoding tasks in 14.6 minutes stock and improving to 13.5 minutes when overclocked. This solidified its position among top CPUs for media processing.
### Conclusion and Recommendations
The i7 10-700K excels in gaming but faces steeper competition in production workloads. For gamers seeking high FPS without premium costs, the Ryzen 3600 offers excellent value. The i7 10-700K finds its niche among enthusiasts willing to invest for marginal performance gains over AMD alternatives.
In conclusion, while the i7 10-700K is a formidable processor, its role is best suited as a mid-tier option in a competitive market. For detailed insights and further discussion, check out our YouTube channel or support us on Patreon for exclusive content.
"WEBVTTKind: captionsLanguage: enwe bought the elusive Intel i7 10 700k on launch day and then overnighted it then Patrick and I worked literally around the clock to get the review turned around in under 24 hours from the time we received the cpu through the mail so the price we pay it was four hundred ten dollars for the ten 700k that puts it between expectedly the ten 600k at about 260 280 range and theory and the ten 900k which will be over five hundred dollars the pricing on Amazon for the nearest competition would be and these are our nine 3900 X currently 420 dollars and it's similar elsewhere on the web that's supposed to be the new permanent price the 3,700 X might be another apt comparison if you wanted to save some money on something that competes potentially with a ten 700k that one is about two hundred ninety dollars these days so we're gonna be looking at the tensile mark a obviously four quarter core Delta's for thermals we looking at the frequency behavior the expected frequency behavior and stock specification from Intel we're looking at 3d rendering video rendering compression decompression applications code compile gaming obviously overclocking and more all in under 24 hours this was done before that this video is brought to you by Squarespace Squarespace is what we've been using for years to manage our own gamers Nexus store and we've been incredibly happy with the choice Squarespace makes ecommerce easy for those interested in starting stores but it also has powerful tools to build all types of websites photo galleries for photographers resume and portfolio sites and small business sites are all easily done through Squarespace having built a lot of client websites the old way before running GN full-time we can easily recommend Squarespace has a powerful fast solution go to squarespace.com slash gamers nexus to get 10% off your first purchase with Squarespace the 10700 kind of reminds us of the 90s have a hard K where it's sort of in a weird position in an Intel's pricing structure but the 9700 K although it felt terrible to buy because it didn't have hyper threading enabled on an i7 at least had the advantage of having a couple extra threads whereas the 9600 K without hyper threading a lot of the time it was dead in the water despite being objectively better in a lot of games than am these closest competitor the 3600 the 9600 K was occasionally abysmal and it just depended on what game you were looking at now dames that exceeds six active threads to the extent that it tanks frame times aren't too common normally the CPU can deal with it a little bit better than that in the form of just lower performance but not game ruining performance that wasn't always the case though and those games that did achieve ruinous levels of performance made it hard to justify the 9600 K over just losing 13% of your performance for a 3,600 instead it's different now though the 10 700 K has brought hyper-threading back it's a 9900 except it's about $100 cheaper and the frequencies are changed around a little bit so that's where we are for the 10-7 it's a 99 heart a drop-in replacement for cheaper which is how the markets supposed to work as Thane's age but Intel still getting back into it after getting pretty thoroughly clobbered by AMD for the last couple of years so let's just get straight into the numbers we've got a lot to go through today we need to start with some thermals so the court scored deltas if you want to learn about what cooling solution we're using you can check our in-depth CPU testing methodology for more information on the coolers we use and why we chose them for this it's an ex 60 to 80 mil CLC at 100% fan and palm speeds we illustrated with both the 10900 k and 10600 k that intel's approached to die sanding has had actual meaningful improvement on core to core Delta's at least it has so far as we continue to build our sample size will better understand how much of that is solder variants versus improvement from the 300 micron reduction in the thigh height and the subsequent increase in the IHS thickness in blender all core workloads the 10 700 kg maximum quarter core Delta was approximately 6 degrees Celsius at full stock for 134 watts this always gets more exaggerated as the power that goes up but we still max out at 9 degrees Celsius core to core this is much better than we've seen on previous Intel CPUs for core to core Delta's so it appears that following the 10 600 K nine hundred K and 700 K results we've now collected Intel's die sanding approach has legitimately improved the thermal performance as we've done for the previous two 10 series CPUs we need to start next by defining the expected stock behavior of the CPU not doing this means that you could end up running the CPU way out of spec which some motherboards enable by default asus gets a lot of credit for following Intel guidance rather than cheating to try and beat the other motherboard makers the 10 700 K has a pl one of 125 watts PL 2 of 229 and tau of 56 it should never exceed 5.1 gigahertz on three active cores with the rest set to 48 48 and then 47 down the stack so our CPU like in the previous reviews was running at the proper Intel guidance if you don't know what that means check our other video on it and also note that Intel guidance is going to be conservatively low as compared to the motherboard makers who just blast the frequencies without any regard to spec to better understand how the CPU behaves with regard to boosting in frequency here's a benchmark of cinnamon charge 20 single threaded runt this shows the maximum clock at a given interval we hit 50 100 megahertz twice during the run checked across all cores with the average at 4900 to 5000 mega Hertz for a given interval we don't want to make this video even longer so we won't break it out per core but the numbers match the previous image that we showed before getting into the gaming charts one more look at a frequency plot this one shows all core frequency under a load that hits tau and then expires the behavior is as expected for all cores loaded it starts closer to 40 900 megahertz but immediately drops to 40 700 megahertz then hits the tau expiration period and drops to about 4300 243 75 mega Hertz all core average Total War three kingdoms campaign will start us off for games this one is interesting because the very top end is actually memory bound rather than GPU about something that we rarely see in gaming applications given the prevalence of heavy graphics loads first and foremost we discovered this in our 10 600 K tuning benchmark which you can find on the channel separately if you want to learn more the 10 700 K stock CPU doesn't get too close to that 152 FPS average memory bandwidth barrier and so it's scoring has it between the 9900 case stock and 9700 k 5.1 gigahertz results the overclock gets it to 147 FPS average about tied with the stock 10 900 K and these closest CPU is the 1350 a stock CPU down at 127 FPS average which allows the 10 700 K a lead of about 13% if you wanted to look at it another way because stat math is a great way of portraying whatever scenario you want the 3300 ax 4.4 gigahertz result at 124 FPS average is able to achieve a five percent of the performance shown by the ten 700k 5.1 gigahertz results the are three 3300 acts really is enough for gaming for most users and if all you're doing is building a gaming machine without much regard for the absolute highest frame rate performance or care for other applications the 3300 acts would still be a major cost saver that allows GPU usage pretty much all the way up the stack and you can see our bottleneck testing for that but not everyone wants something lower-end and might have value for some of these mid-range and high-end chips and for that we've got the numbers on the screen for you hitman two has shown expressive CPU scaling and our benchmarks lately the ten 700k stock CPU is within error of the 9900 K at five point one gigahertz it's just ahead of the ten 600 K at five point one and it's just behind the ten 600 K with arraigned overclock the 10900 case stock CPU outdoes the ten 700 K stock CPU by about three point three percent here so nothing exciting for an ante comparison the chart leader for AMD section is the 3900 X doc CPU at 124 FPS average this positioning allows the 10700 stock CPU a lead of about 10% F 1 20 19 gives us a relatively unconstrained look at CPU performance without regard for GPU limitations the top of the chart is occupied by the 10 600 K with a heavy tune and the 10900 K at 5 points each gigahertz showing scallion from core and memory overclocking the 10-7 hard case stock CPU is about equal to the 9700 case stock CPU so the extra threads don't mean a whole lot here we saw this same behavior with the 9700 9900 K both at five point one as well the 10900 k stock CPU leads the 10 700 K stock by about two and a half percent so the Delta is meaningless in a practical sense and the lows are also within error the 10 700 K a five point one scales directly responding with a 283 FPS average from a 5% uplift against stock which ranks it as between the 9900 K at five point one and the 10900 K at five point two gigahertz and these CPUs are mostly led by the 3300 acts with a memory tune and infinity fabric tune at 242 FPS average peak but as for comparisons using our standard test memory the 3700 ax is closest in stock performance at 239 FPS average that's about a 13 percent lead again for the 10 700 K versus the 3700 ex stock now force 40p all the results become truncated by a GPU limitation we include these charts to show realistically what's expected for higher resolution gaming where GPU limitations are real consideration but the framerate is still very high f1 is great for that and these results we still see a direct advantage for the higher-end in two parts with higher frequencies but that advantage becomes more constrained the 10700 K for example is now about equal with the 3700 X and the 10 600 K stock CPUs overclocking still pushes it toward the top but they're all about the same at that point the division 2 is another great example of a realistic GPU bind this is at 1080p medium and that's the limit of what we're willing to run for a benchmark although we could force more scaling by further driving the settings into the ground at some point we have to draw the line for what's reasonable or useful to the end user sure we could show more scaling here by dropping the resolution comically low but ultimately outside of speculation on what that might mean for the future which is always a dangerous game to play there's not a whole lot of practical information that we can derive from that everything between the 10 700 K and 99 ook at 5.1 gigahertz is roughly with an error or close enough since it's all limited by the GPU no meaningful differences emerge until we get to a.m. these parts statistically speaking and even those differences aren't necessarily appreciable to most users but they'd be important for people who are frame snobs and we don't mean that in a critical way from the standpoint of a strict CPU review these numbers are invalid for measurement of differences because the 10s days would peak higher and so you'd have a bigger gap however they're valid for illustrating that real limitations outside the CPU are often encountered first shuttle the Tomb Raider with dx12 is next the 10 700 K stock CPU ran at 166 FPS average stock with frame x pastes consistently behind as compared to other CPUs on the charts while the overclocked posts no meaningful gain note that this is a title where we saw a good scaling out of cash ratio overclocking on our 10 600 K in that content so that would be the next move to improve 10 700 K performance the 10 700 K ends up functionally the same for performance as the 10 6 in gaming it has a tactical advantage as in a measurable and repeatable one but not an appreciable advantage the ten nine stock CPU runs about seven percent better here but also runs into a GP limit at the top illustrated by the tied results of the ten nine hundred K stock and the OC benchmarks that gap is wider in some of the earlier games we showed if you wanted to see true scaling the ten six hundred K is also competitive and although this is an arms race where you could also OC the cash of the ten seven the ten six can be made competitive at least with these results as for Rison that establishes its top end at about 150 FPS average in this game so the 3700 X is led by the ten 700 K by about twelve percent stock to stock this is the same Delta that we saw in some of the earlier games and is about the max range that you can expect to see in less constrained scenarios that don't have as much of a GP limitation as the previous game the tens of 100k also runs technically better frame time consistency with a lower deviation from the mean Albee it not in an appreciable sense we only point this out because of the misconception about Rison being quote smoother which isn't wholly true of course from a value standpoint the 3300 ex remains one of the best buys right now but this is where the 10700 ka Falls I read that to chart with medium settings is still very limited since we only added it after starting our initial foray into this new testing methodology to learn why we added medium later check out our dedicated testing methodology piece bearing in mind that we still need to add the 3600 the 3700 acts and everything else this chart at least shows relative scaling between Intel high-end parts and against a likely top performer for AMD the 3300 acts with a tune the result is the 10 700 K at 144 FPS average stock with it lows consistent and the 10 600 K about equal to it in both stock and overclocked metrics the 10700 5.1 pushes up to 153 FPS average which is about tied with a 10 900k a stock CPU at this point we're running into other limitations we won't bother showing the high charts for dx12 or vulcan everything's the same up and down the rankings due to a hard GP bottleneck and again see the methodology piece for more info finally before you get into the production benchmarks for an older more frequency bound games represent another class of gaming we can defer to steam top 10 title GTA 5 the 10 700 K stock CPU slightly outperforms the 9 after K stock CPU and the ten 600k a 5.1 gigahertz CPU pre tuned while also sitting below the ten nine by a few FPS average it's not a huge gap between the top performers but it's clear that the highest frequency parts are still winning the night I heard K and 9700 K at five point one both curiously outperform the ten seven at five point one and do so repeatedly we're not sure if this is a result of some mitigation and silicon of an immature bios or of yet another one of GTA 5's strange and numerous bugs but the results are repeatable and these CPUs top out around 113 FPS average including the likes of the 3950 3900 acts and the not-so-distant 3,700 X the 10-7 ends up holding about an 11 percent lead over the 3900 acts when both are stock now we can move into our production tests and next up is blender of 3d modeling and animation open source solution that has exploded in popularity over the years we use this tool in-house even to design all of our own products including our recent GN mouse mat with all those 3d PC components on it rendering can be done with Eevee in real time or with renderers and for CPUs we use cycles the GN monkeyhead render is a sepia stress test that we built for reviews the intel i7 10700 k stock cpu finishes the render in 14 minutes out matching the 9,900 k stock CPU by about 1.3 minutes or a reduction of 8.5% the 3700 X is about equal with the 10 700 K in this logo render but as we try to remind people regularly it's important to have multiple blender scenes to create a properly accurate depiction since CPUs behave different ways with different renders you'll see that in the next chart the 3900 X here obviously has a significant lead benefited directly by more threads to render more simultaneous tiles and finishes in 9.6 minutes for a 31% reduction where's the 10 700 K stock overclocking the 10 702 5.1 gigahertz gets it between the 10 core 2310 900k and 10 900 x stock CPUs for a 10% improvement here's the G M logo render which is our heaviest workload for this one for this the 10 700 K loses rank versus the a.m. the r7 3700 X which now completes the render in sixteen point nine minutes versus 18 point two on the ten 700k a reduction of 7.1% in total time required the ten 700k stock cpu is between the 9900 k stock and the OC results which makes sense and it's behind the 3700 X it's about one to two minutes faster than the r7 2700 at 4.1 gigahertz and the 10900 case stock and 3900 AK stock CPUs both have a significant lead ranked at 37% reduced time for the 3900 acts and about 15% reduced time for the 10 900k stock CPUs overclocking the 10 seven to five point one gigahertz helps to reach rough equivalence with a 10-9 but that's about the best we can do with normal cooling here's our set of adobe premiere results starting with our 1080p 60 render for this test the 10 700 K stock CPU completes the render in the same time as the 3700 X and 9900 Kate stock CPUs the 3900 acts and ten nine hundred K both complete the render ten and a half percent faster than the ten seven there is meaningful uplift against the 3600 but the price difference means a completely different category of CPU the 4k render is far more abusive with the 10 some 100k finishing an 11 minute stock or ten and a half overclocked the ten sub 100k stock is roughly the same as the 3700 ax here with the 3700 X being technically superior the 3900 X has a significant lead at 8.9 minutes now we can look at our sevens of compression and decompression tests which position the 10 700 K stock CPU about equal to the 1900 at five point one gigahertz that's a good improvement generationally for Intel but only because it's the same architecture and sort of the same generation with a numerical iteration that's an improvement though about three and a half percent the 3700 X is roughly the same distance ahead and sits between the stock and overclocked 10 700 K numbers although the difference between all of these isn't too meaningful unless you're doing compression and decompression all day long constantly this application scales directly with cores but does run into memory limitations with top-end threader per part and it also benefits from increased core count clearly the ten-hut case stock CPU is an example of further scaling for decompression the Intel i7 10-7 hard case stock CPU scores 86,000 MIPS which isn't too impressive when considering the proximity of both these and over clocks are five 3600 CPE results that's at 73 k maps and 75 and a half-game ifs respectively the ten seven is functionally equivalent to ninety nine hundred k stock cpu here with the twenty seven hundred at 4.1 gigahertz or another eight core sixteen thread part for about a one percent lead these are mostly within or very near error but worst-case scenario is that they're functionally equal given the price advantage of the twenty seven hundred at least back when it was readily available that's a good look for AMD the ten seven hundred k OC does help it it gets to ninety three thousand MIPS and it brings it nearing equivalence with its ten nine hundred k stock cpu while the thirty seven hundred x outperforms the ten seven hundred k OC for the stock 3700 ax verse ten 7oc by seven point three percent the 3900 ax is in the next a salon of performance at an impressive 145 thousand MIPS or a lead of sixty eight percent when looking at these stock numbers for the 3900 X versus the 10 700 K our code compiled with chromium clean CL and ninja is next as tested on a Windows platform this is measured in minutes to complete similar to blender and is representative of recompile of an entire project lower is better here the 10 700 K stock CPU completes the compiled in 100 minutes allowing the 3700 X elite of 8 percent and the 3900 acts a lead of 35% where lead here means less time required to render all stock to stock numbers the 10 700 K roughly matches the 10 900 X not to be confused with the 10 nine hundred K where the former is overclocked to five point one gigahertz and the latter is the stock 10 900 X as for the 10 nine hundred K that's leading stock to stock against the 10 700 K by about 18% our next one is handbrake transcoding to h.265 the 10 700 K stock CPU finishes the work in fourteen point six minutes with an overclock improving to 13 and a half minutes that's flanking the 9700 K at five point one gigahertz on either side the our 737 hardac's meanwhile is outmatched by the 10 700 K here allowing the 10 700 K a stock the stock lead of 5.2% meanwhile the 3900 ax leads the 10-7 heart k with both stock by 12% forecasts groups v-ray another 3d modeling application with a baked in renderer the 10-7 RK aligned mostly between the 3700 acts and 99 hundred KS stock CPUs the Dalit between the 10 700 km 99 her case stock results is basically with an error the 10 7 hard kayo see also doesn't really gain as much it's about six point three percent here and it ends up with an error of the 99 RK at the same frequency the 10900 k benefits from a 21 percent uplift stock the stock thanks largely to its core account while the 3900 ex stock result is 42 percent ahead of the stock 10 700 k these are similar results to what we saw for scaling and blender finally for power consumption we'll start with the EPS 12 volt cable draw for a blender all for work look this one has the 10 700 K at 134 watts with the 10900 K at about the same power consumption keep in mind that we do our testing within the Intel pl/1 PL 2 and tau guidance for official spec so these are the proper stock numbers the tenth 700 K pulled about 268 watts with our overclock applied the overclock consumes less power than the 10900 K despite a higher voltage as a result of its lower core account for a senate Bancho our 20 NT before tau expiry the 10 700 k draws about 161 watts whereas the 10900 k was closer to 200 watts overall then like the other Intel CPUs in an objective sense from purely numbers this is a good CPU meaning that we're not seeing anything that's objectively bad at least when it's run under stock settings comparatively it's a bit tougher because comparatively now you're looking at price and competition and that standpoint which is the one that really matters for consumer we think that Intel was right to send out the AI 510 600 K instead of the 10 700 K when we first saw the specs that Intel posted with its slides when the 10th gen gen was announced our thinking was that the 10-7 hard K might actually be the most exciting just because of the numbers on the spec sheet but in reality the 10 600 K is so damn close to the 10 7 in both realistic gaming scenarios and in scenarios where we've intentionally removed the restraints on the GPU to push it all to the CPU it's just the 10-6 is too close to really justify the 10-7 for a gaming only scenario and once you look at production and these two competitive there especially with the 3900 acts now roughly the same price and the 3700 acts a bit cheaper for mostly equivalent performance in a lot of the applications so where this leads us is we think if you're someone who needs fairly high FPS but not the highest FPS the 10 600 K makes a lot of sense it saves a lot of money it performs roughly the same as the 10 700 K and if you're willing to do this and please please be willing to do it because it's a case Q part and you're buying is e49 your motherboard if you're willing to do it please spend about an hour trying an overclock on it a simple one set the cash ratio especially to something like maybe 48 acts and then try to get the core 25.0 all core at least maybe 5.1 you can see our previous video on the 10-6 tuning for more on that because once you do those things it becomes a night at 10 900 km performance in most gaming scenarios so the 10 6 is extremely compelling in that regard if you're someone who needs more balance you care a lot about production applications you care at least a little bit about things like video editing Adobe Premiere performance and you want to save some money the 3600 is still the absolute best all-around performer at the price nothing comes close to it in terms of price to all-around performance it is a well-rounded CPU we still keep the 3600 in the top regard it's basically the 36 is how we view what Sandy Bridge used to be we think the 3600 is sort of the new Sandy Bridge it's gonna hold on a long time so those are your two choices there if you want absolute highest framerate and you're an FPS snob and you play competitive games only well shell out for a 10 900k maybe overclock it a bit because it can outdo the 10 600 in a like-for-like scenario the 10 700 then has really nowhere left to stand because we've now recommended a 3900 axe or a 3700 axe for people who are a little bit more production or workstation focused and we've recommended the intel parts for people who are less price conscious but very gaming focused and then the 3600 for everyone in between where your price conscious and you want to be of everything where does the ten 7rk fall in this scenario it's kind of like the 3,700 X was in our opinion we know it was a popular CPU but our opinion on the 3700 X part was that you should buy a 3900 X or a 3600 and the part in between the extra hundred bucks in most scenarios but not all didn't make a lot of sense now obviously there are scenarios where it made sense and we explain those in the 37 X review but the 10700 falls in basically the same position as far as we're concerned and either of the other two on either side make a lot more sense so that's it for this one thanks for watching subscribe for more go to store documents access dotnet to support these types of efforts especially when we shell out the amount of money we did for the CPU for a review or go to patreon.com/scishow and his access to help site directly and on the store of course you can backorder our new Mouse mats if you want a functional product for us or the mouse the mod mats for another functional product we'll see you all next timewe bought the elusive Intel i7 10 700k on launch day and then overnighted it then Patrick and I worked literally around the clock to get the review turned around in under 24 hours from the time we received the cpu through the mail so the price we pay it was four hundred ten dollars for the ten 700k that puts it between expectedly the ten 600k at about 260 280 range and theory and the ten 900k which will be over five hundred dollars the pricing on Amazon for the nearest competition would be and these are our nine 3900 X currently 420 dollars and it's similar elsewhere on the web that's supposed to be the new permanent price the 3,700 X might be another apt comparison if you wanted to save some money on something that competes potentially with a ten 700k that one is about two hundred ninety dollars these days so we're gonna be looking at the tensile mark a obviously four quarter core Delta's for thermals we looking at the frequency behavior the expected frequency behavior and stock specification from Intel we're looking at 3d rendering video rendering compression decompression applications code compile gaming obviously overclocking and more all in under 24 hours this was done before that this video is brought to you by Squarespace Squarespace is what we've been using for years to manage our own gamers Nexus store and we've been incredibly happy with the choice Squarespace makes ecommerce easy for those interested in starting stores but it also has powerful tools to build all types of websites photo galleries for photographers resume and portfolio sites and small business sites are all easily done through Squarespace having built a lot of client websites the old way before running GN full-time we can easily recommend Squarespace has a powerful fast solution go to squarespace.com slash gamers nexus to get 10% off your first purchase with Squarespace the 10700 kind of reminds us of the 90s have a hard K where it's sort of in a weird position in an Intel's pricing structure but the 9700 K although it felt terrible to buy because it didn't have hyper threading enabled on an i7 at least had the advantage of having a couple extra threads whereas the 9600 K without hyper threading a lot of the time it was dead in the water despite being objectively better in a lot of games than am these closest competitor the 3600 the 9600 K was occasionally abysmal and it just depended on what game you were looking at now dames that exceeds six active threads to the extent that it tanks frame times aren't too common normally the CPU can deal with it a little bit better than that in the form of just lower performance but not game ruining performance that wasn't always the case though and those games that did achieve ruinous levels of performance made it hard to justify the 9600 K over just losing 13% of your performance for a 3,600 instead it's different now though the 10 700 K has brought hyper-threading back it's a 9900 except it's about $100 cheaper and the frequencies are changed around a little bit so that's where we are for the 10-7 it's a 99 heart a drop-in replacement for cheaper which is how the markets supposed to work as Thane's age but Intel still getting back into it after getting pretty thoroughly clobbered by AMD for the last couple of years so let's just get straight into the numbers we've got a lot to go through today we need to start with some thermals so the court scored deltas if you want to learn about what cooling solution we're using you can check our in-depth CPU testing methodology for more information on the coolers we use and why we chose them for this it's an ex 60 to 80 mil CLC at 100% fan and palm speeds we illustrated with both the 10900 k and 10600 k that intel's approached to die sanding has had actual meaningful improvement on core to core Delta's at least it has so far as we continue to build our sample size will better understand how much of that is solder variants versus improvement from the 300 micron reduction in the thigh height and the subsequent increase in the IHS thickness in blender all core workloads the 10 700 kg maximum quarter core Delta was approximately 6 degrees Celsius at full stock for 134 watts this always gets more exaggerated as the power that goes up but we still max out at 9 degrees Celsius core to core this is much better than we've seen on previous Intel CPUs for core to core Delta's so it appears that following the 10 600 K nine hundred K and 700 K results we've now collected Intel's die sanding approach has legitimately improved the thermal performance as we've done for the previous two 10 series CPUs we need to start next by defining the expected stock behavior of the CPU not doing this means that you could end up running the CPU way out of spec which some motherboards enable by default asus gets a lot of credit for following Intel guidance rather than cheating to try and beat the other motherboard makers the 10 700 K has a pl one of 125 watts PL 2 of 229 and tau of 56 it should never exceed 5.1 gigahertz on three active cores with the rest set to 48 48 and then 47 down the stack so our CPU like in the previous reviews was running at the proper Intel guidance if you don't know what that means check our other video on it and also note that Intel guidance is going to be conservatively low as compared to the motherboard makers who just blast the frequencies without any regard to spec to better understand how the CPU behaves with regard to boosting in frequency here's a benchmark of cinnamon charge 20 single threaded runt this shows the maximum clock at a given interval we hit 50 100 megahertz twice during the run checked across all cores with the average at 4900 to 5000 mega Hertz for a given interval we don't want to make this video even longer so we won't break it out per core but the numbers match the previous image that we showed before getting into the gaming charts one more look at a frequency plot this one shows all core frequency under a load that hits tau and then expires the behavior is as expected for all cores loaded it starts closer to 40 900 megahertz but immediately drops to 40 700 megahertz then hits the tau expiration period and drops to about 4300 243 75 mega Hertz all core average Total War three kingdoms campaign will start us off for games this one is interesting because the very top end is actually memory bound rather than GPU about something that we rarely see in gaming applications given the prevalence of heavy graphics loads first and foremost we discovered this in our 10 600 K tuning benchmark which you can find on the channel separately if you want to learn more the 10 700 K stock CPU doesn't get too close to that 152 FPS average memory bandwidth barrier and so it's scoring has it between the 9900 case stock and 9700 k 5.1 gigahertz results the overclock gets it to 147 FPS average about tied with the stock 10 900 K and these closest CPU is the 1350 a stock CPU down at 127 FPS average which allows the 10 700 K a lead of about 13% if you wanted to look at it another way because stat math is a great way of portraying whatever scenario you want the 3300 ax 4.4 gigahertz result at 124 FPS average is able to achieve a five percent of the performance shown by the ten 700k 5.1 gigahertz results the are three 3300 acts really is enough for gaming for most users and if all you're doing is building a gaming machine without much regard for the absolute highest frame rate performance or care for other applications the 3300 acts would still be a major cost saver that allows GPU usage pretty much all the way up the stack and you can see our bottleneck testing for that but not everyone wants something lower-end and might have value for some of these mid-range and high-end chips and for that we've got the numbers on the screen for you hitman two has shown expressive CPU scaling and our benchmarks lately the ten 700k stock CPU is within error of the 9900 K at five point one gigahertz it's just ahead of the ten 600 K at five point one and it's just behind the ten 600 K with arraigned overclock the 10900 case stock CPU outdoes the ten 700 K stock CPU by about three point three percent here so nothing exciting for an ante comparison the chart leader for AMD section is the 3900 X doc CPU at 124 FPS average this positioning allows the 10700 stock CPU a lead of about 10% F 1 20 19 gives us a relatively unconstrained look at CPU performance without regard for GPU limitations the top of the chart is occupied by the 10 600 K with a heavy tune and the 10900 K at 5 points each gigahertz showing scallion from core and memory overclocking the 10-7 hard case stock CPU is about equal to the 9700 case stock CPU so the extra threads don't mean a whole lot here we saw this same behavior with the 9700 9900 K both at five point one as well the 10900 k stock CPU leads the 10 700 K stock by about two and a half percent so the Delta is meaningless in a practical sense and the lows are also within error the 10 700 K a five point one scales directly responding with a 283 FPS average from a 5% uplift against stock which ranks it as between the 9900 K at five point one and the 10900 K at five point two gigahertz and these CPUs are mostly led by the 3300 acts with a memory tune and infinity fabric tune at 242 FPS average peak but as for comparisons using our standard test memory the 3700 ax is closest in stock performance at 239 FPS average that's about a 13 percent lead again for the 10 700 K versus the 3700 ex stock now force 40p all the results become truncated by a GPU limitation we include these charts to show realistically what's expected for higher resolution gaming where GPU limitations are real consideration but the framerate is still very high f1 is great for that and these results we still see a direct advantage for the higher-end in two parts with higher frequencies but that advantage becomes more constrained the 10700 K for example is now about equal with the 3700 X and the 10 600 K stock CPUs overclocking still pushes it toward the top but they're all about the same at that point the division 2 is another great example of a realistic GPU bind this is at 1080p medium and that's the limit of what we're willing to run for a benchmark although we could force more scaling by further driving the settings into the ground at some point we have to draw the line for what's reasonable or useful to the end user sure we could show more scaling here by dropping the resolution comically low but ultimately outside of speculation on what that might mean for the future which is always a dangerous game to play there's not a whole lot of practical information that we can derive from that everything between the 10 700 K and 99 ook at 5.1 gigahertz is roughly with an error or close enough since it's all limited by the GPU no meaningful differences emerge until we get to a.m. these parts statistically speaking and even those differences aren't necessarily appreciable to most users but they'd be important for people who are frame snobs and we don't mean that in a critical way from the standpoint of a strict CPU review these numbers are invalid for measurement of differences because the 10s days would peak higher and so you'd have a bigger gap however they're valid for illustrating that real limitations outside the CPU are often encountered first shuttle the Tomb Raider with dx12 is next the 10 700 K stock CPU ran at 166 FPS average stock with frame x pastes consistently behind as compared to other CPUs on the charts while the overclocked posts no meaningful gain note that this is a title where we saw a good scaling out of cash ratio overclocking on our 10 600 K in that content so that would be the next move to improve 10 700 K performance the 10 700 K ends up functionally the same for performance as the 10 6 in gaming it has a tactical advantage as in a measurable and repeatable one but not an appreciable advantage the ten nine stock CPU runs about seven percent better here but also runs into a GP limit at the top illustrated by the tied results of the ten nine hundred K stock and the OC benchmarks that gap is wider in some of the earlier games we showed if you wanted to see true scaling the ten six hundred K is also competitive and although this is an arms race where you could also OC the cash of the ten seven the ten six can be made competitive at least with these results as for Rison that establishes its top end at about 150 FPS average in this game so the 3700 X is led by the ten 700 K by about twelve percent stock to stock this is the same Delta that we saw in some of the earlier games and is about the max range that you can expect to see in less constrained scenarios that don't have as much of a GP limitation as the previous game the tens of 100k also runs technically better frame time consistency with a lower deviation from the mean Albee it not in an appreciable sense we only point this out because of the misconception about Rison being quote smoother which isn't wholly true of course from a value standpoint the 3300 ex remains one of the best buys right now but this is where the 10700 ka Falls I read that to chart with medium settings is still very limited since we only added it after starting our initial foray into this new testing methodology to learn why we added medium later check out our dedicated testing methodology piece bearing in mind that we still need to add the 3600 the 3700 acts and everything else this chart at least shows relative scaling between Intel high-end parts and against a likely top performer for AMD the 3300 acts with a tune the result is the 10 700 K at 144 FPS average stock with it lows consistent and the 10 600 K about equal to it in both stock and overclocked metrics the 10700 5.1 pushes up to 153 FPS average which is about tied with a 10 900k a stock CPU at this point we're running into other limitations we won't bother showing the high charts for dx12 or vulcan everything's the same up and down the rankings due to a hard GP bottleneck and again see the methodology piece for more info finally before you get into the production benchmarks for an older more frequency bound games represent another class of gaming we can defer to steam top 10 title GTA 5 the 10 700 K stock CPU slightly outperforms the 9 after K stock CPU and the ten 600k a 5.1 gigahertz CPU pre tuned while also sitting below the ten nine by a few FPS average it's not a huge gap between the top performers but it's clear that the highest frequency parts are still winning the night I heard K and 9700 K at five point one both curiously outperform the ten seven at five point one and do so repeatedly we're not sure if this is a result of some mitigation and silicon of an immature bios or of yet another one of GTA 5's strange and numerous bugs but the results are repeatable and these CPUs top out around 113 FPS average including the likes of the 3950 3900 acts and the not-so-distant 3,700 X the 10-7 ends up holding about an 11 percent lead over the 3900 acts when both are stock now we can move into our production tests and next up is blender of 3d modeling and animation open source solution that has exploded in popularity over the years we use this tool in-house even to design all of our own products including our recent GN mouse mat with all those 3d PC components on it rendering can be done with Eevee in real time or with renderers and for CPUs we use cycles the GN monkeyhead render is a sepia stress test that we built for reviews the intel i7 10700 k stock cpu finishes the render in 14 minutes out matching the 9,900 k stock CPU by about 1.3 minutes or a reduction of 8.5% the 3700 X is about equal with the 10 700 K in this logo render but as we try to remind people regularly it's important to have multiple blender scenes to create a properly accurate depiction since CPUs behave different ways with different renders you'll see that in the next chart the 3900 X here obviously has a significant lead benefited directly by more threads to render more simultaneous tiles and finishes in 9.6 minutes for a 31% reduction where's the 10 700 K stock overclocking the 10 702 5.1 gigahertz gets it between the 10 core 2310 900k and 10 900 x stock CPUs for a 10% improvement here's the G M logo render which is our heaviest workload for this one for this the 10 700 K loses rank versus the a.m. the r7 3700 X which now completes the render in sixteen point nine minutes versus 18 point two on the ten 700k a reduction of 7.1% in total time required the ten 700k stock cpu is between the 9900 k stock and the OC results which makes sense and it's behind the 3700 X it's about one to two minutes faster than the r7 2700 at 4.1 gigahertz and the 10900 case stock and 3900 AK stock CPUs both have a significant lead ranked at 37% reduced time for the 3900 acts and about 15% reduced time for the 10 900k stock CPUs overclocking the 10 seven to five point one gigahertz helps to reach rough equivalence with a 10-9 but that's about the best we can do with normal cooling here's our set of adobe premiere results starting with our 1080p 60 render for this test the 10 700 K stock CPU completes the render in the same time as the 3700 X and 9900 Kate stock CPUs the 3900 acts and ten nine hundred K both complete the render ten and a half percent faster than the ten seven there is meaningful uplift against the 3600 but the price difference means a completely different category of CPU the 4k render is far more abusive with the 10 some 100k finishing an 11 minute stock or ten and a half overclocked the ten sub 100k stock is roughly the same as the 3700 ax here with the 3700 X being technically superior the 3900 X has a significant lead at 8.9 minutes now we can look at our sevens of compression and decompression tests which position the 10 700 K stock CPU about equal to the 1900 at five point one gigahertz that's a good improvement generationally for Intel but only because it's the same architecture and sort of the same generation with a numerical iteration that's an improvement though about three and a half percent the 3700 X is roughly the same distance ahead and sits between the stock and overclocked 10 700 K numbers although the difference between all of these isn't too meaningful unless you're doing compression and decompression all day long constantly this application scales directly with cores but does run into memory limitations with top-end threader per part and it also benefits from increased core count clearly the ten-hut case stock CPU is an example of further scaling for decompression the Intel i7 10-7 hard case stock CPU scores 86,000 MIPS which isn't too impressive when considering the proximity of both these and over clocks are five 3600 CPE results that's at 73 k maps and 75 and a half-game ifs respectively the ten seven is functionally equivalent to ninety nine hundred k stock cpu here with the twenty seven hundred at 4.1 gigahertz or another eight core sixteen thread part for about a one percent lead these are mostly within or very near error but worst-case scenario is that they're functionally equal given the price advantage of the twenty seven hundred at least back when it was readily available that's a good look for AMD the ten seven hundred k OC does help it it gets to ninety three thousand MIPS and it brings it nearing equivalence with its ten nine hundred k stock cpu while the thirty seven hundred x outperforms the ten seven hundred k OC for the stock 3700 ax verse ten 7oc by seven point three percent the 3900 ax is in the next a salon of performance at an impressive 145 thousand MIPS or a lead of sixty eight percent when looking at these stock numbers for the 3900 X versus the 10 700 K our code compiled with chromium clean CL and ninja is next as tested on a Windows platform this is measured in minutes to complete similar to blender and is representative of recompile of an entire project lower is better here the 10 700 K stock CPU completes the compiled in 100 minutes allowing the 3700 X elite of 8 percent and the 3900 acts a lead of 35% where lead here means less time required to render all stock to stock numbers the 10 700 K roughly matches the 10 900 X not to be confused with the 10 nine hundred K where the former is overclocked to five point one gigahertz and the latter is the stock 10 900 X as for the 10 nine hundred K that's leading stock to stock against the 10 700 K by about 18% our next one is handbrake transcoding to h.265 the 10 700 K stock CPU finishes the work in fourteen point six minutes with an overclock improving to 13 and a half minutes that's flanking the 9700 K at five point one gigahertz on either side the our 737 hardac's meanwhile is outmatched by the 10 700 K here allowing the 10 700 K a stock the stock lead of 5.2% meanwhile the 3900 ax leads the 10-7 heart k with both stock by 12% forecasts groups v-ray another 3d modeling application with a baked in renderer the 10-7 RK aligned mostly between the 3700 acts and 99 hundred KS stock CPUs the Dalit between the 10 700 km 99 her case stock results is basically with an error the 10 7 hard kayo see also doesn't really gain as much it's about six point three percent here and it ends up with an error of the 99 RK at the same frequency the 10900 k benefits from a 21 percent uplift stock the stock thanks largely to its core account while the 3900 ex stock result is 42 percent ahead of the stock 10 700 k these are similar results to what we saw for scaling and blender finally for power consumption we'll start with the EPS 12 volt cable draw for a blender all for work look this one has the 10 700 K at 134 watts with the 10900 K at about the same power consumption keep in mind that we do our testing within the Intel pl/1 PL 2 and tau guidance for official spec so these are the proper stock numbers the tenth 700 K pulled about 268 watts with our overclock applied the overclock consumes less power than the 10900 K despite a higher voltage as a result of its lower core account for a senate Bancho our 20 NT before tau expiry the 10 700 k draws about 161 watts whereas the 10900 k was closer to 200 watts overall then like the other Intel CPUs in an objective sense from purely numbers this is a good CPU meaning that we're not seeing anything that's objectively bad at least when it's run under stock settings comparatively it's a bit tougher because comparatively now you're looking at price and competition and that standpoint which is the one that really matters for consumer we think that Intel was right to send out the AI 510 600 K instead of the 10 700 K when we first saw the specs that Intel posted with its slides when the 10th gen gen was announced our thinking was that the 10-7 hard K might actually be the most exciting just because of the numbers on the spec sheet but in reality the 10 600 K is so damn close to the 10 7 in both realistic gaming scenarios and in scenarios where we've intentionally removed the restraints on the GPU to push it all to the CPU it's just the 10-6 is too close to really justify the 10-7 for a gaming only scenario and once you look at production and these two competitive there especially with the 3900 acts now roughly the same price and the 3700 acts a bit cheaper for mostly equivalent performance in a lot of the applications so where this leads us is we think if you're someone who needs fairly high FPS but not the highest FPS the 10 600 K makes a lot of sense it saves a lot of money it performs roughly the same as the 10 700 K and if you're willing to do this and please please be willing to do it because it's a case Q part and you're buying is e49 your motherboard if you're willing to do it please spend about an hour trying an overclock on it a simple one set the cash ratio especially to something like maybe 48 acts and then try to get the core 25.0 all core at least maybe 5.1 you can see our previous video on the 10-6 tuning for more on that because once you do those things it becomes a night at 10 900 km performance in most gaming scenarios so the 10 6 is extremely compelling in that regard if you're someone who needs more balance you care a lot about production applications you care at least a little bit about things like video editing Adobe Premiere performance and you want to save some money the 3600 is still the absolute best all-around performer at the price nothing comes close to it in terms of price to all-around performance it is a well-rounded CPU we still keep the 3600 in the top regard it's basically the 36 is how we view what Sandy Bridge used to be we think the 3600 is sort of the new Sandy Bridge it's gonna hold on a long time so those are your two choices there if you want absolute highest framerate and you're an FPS snob and you play competitive games only well shell out for a 10 900k maybe overclock it a bit because it can outdo the 10 600 in a like-for-like scenario the 10 700 then has really nowhere left to stand because we've now recommended a 3900 axe or a 3700 axe for people who are a little bit more production or workstation focused and we've recommended the intel parts for people who are less price conscious but very gaming focused and then the 3600 for everyone in between where your price conscious and you want to be of everything where does the ten 7rk fall in this scenario it's kind of like the 3,700 X was in our opinion we know it was a popular CPU but our opinion on the 3700 X part was that you should buy a 3900 X or a 3600 and the part in between the extra hundred bucks in most scenarios but not all didn't make a lot of sense now obviously there are scenarios where it made sense and we explain those in the 37 X review but the 10700 falls in basically the same position as far as we're concerned and either of the other two on either side make a lot more sense so that's it for this one thanks for watching subscribe for more go to store documents access dotnet to support these types of efforts especially when we shell out the amount of money we did for the CPU for a review or go to patreon.com/scishow and his access to help site directly and on the store of course you can backorder our new Mouse mats if you want a functional product for us or the mouse the mod mats for another functional product we'll see you all next time\n"