11900k can’t beat my 310 watt 4.3Ghz 5960x baby. Let’s call it failure to progress on power consumption or maybe we should just say Intel hasn’t stopped sucking power since 2015.
Now imagine if AMD hadn't gotten their act together; we'd have this level of "improvement", but the i3 would instead be named the i7-10700k and be priced like an i7 part as well.
@@ViolentMLG Yeah, but then AMD will just do the exact same thing. We're already seeing it. Ryzen 3000 was still trailing Intel, so AMD kept the prices low to attract market share from Ryzen 2000. The moment their market share saw spikes and they have the better CPU, boom, $100 (50%) increase from the 3600 to 5600X, when the performance is only 10% more. Similarly, 3700X to 5800X was the same story. And these are the like for like parts for both generations, I don't buy into that naming bull crap they're trying to pull. "Ooooh, you should be comparing the 3600X and 3800X price respectively, because that's the name of the line" as if anyone with any decent logic can't see they purposely skewed the SKU names to play this card. In the end, they're all the same. The market needs more competition for this kind of behaviour to stop, forever.
@@DanteLovesPizza Totally agree. They didn't even bother to upgrade their 5000 series APUs with a recent gen GPU architecture. No decently priced 5600 non X version. On the GPU side before the start of the current crisis, you just have to see how they priced their 6900xt, 6800xt, and 6800 GPUs compared to Nvidia's offerings.
@@SayedulHudaChowdhury Exactly. I suppose on the GPU side, at least Intel is joining the fray, so there's hope for more competition in the near future, although a triopoly isn't exactly a good market, but better than duopoly I guess. CPU side... Let's just hope Intel can do something so they have more competition between themselves, because if AMD keeps leading, they're just Intel number 2, and we're just going to get "Intel-level increments" from them generation on generation.
@@DanteLovesPizza there is some hope, Apple switching to ARM based systems was a huge blow to Intel, especially when they didn't suck. That is a 3rd player stepping up to the CPU race, and RISC-V is not too far behind them. In the next few years, things are going to get dicey, especially if Windows ejects X86 reliance on their platform.
@VERY EVIL PERSON FROM ILLUMINATI well, this comparison here wasn't cache normalised either. It was not video's intention, just a coincidence to have equal L3, but 120MB L4 was not exactly equal on both sides. Other than that you're right, but i'd say that older FX look back video really exhausted the topic - there was massive gap in performance=very little progress across the years, so i think this aint a great topic to create another video and dwell on this. I still loved my fx-6300 for many years it served me.
@VERY EVIL PERSON FROM ILLUMINATI To say nothing of how you evaluate the core count - pretty sure if you put an FX-8000 up against a modern 8 core it will look incredibly stupid. I'm still sticking with my 8370, but I'm not pretending it's any more than a very, very sad quad at this point.
@@youllnevertakemealive2833 Tbh if all you want is mostly an average over 45 and in many games over 60 average a FX 8370 probably does okay if you close most other things before gaming. Yeah even the lowest end chips in modern generations are gonna do better, but the FX 8370 is far older and it is still able to game decently. The same price at the time was a 2c/4t i3 CPU from Intel, which wouldnt really be doing much today that the FX 8370 isnt doing at the same or better levels. And the FX probably would be able to run stuff that i3 wouldnt even handle at all.
Because each year intel keep refreshing line up...that left nonsense...really, why do we need 7th, 9th, 11 gen when it basically a 6th vs 8th gen...i could said 10th gen did archieve something so it not really pointless...(at least it gave you something to compete with amd)...amd on other hand release when it need to...or at least make some sense.
@@saricubra2867 but why need 11th gen when there’s 10th gen exist. 5% in performance? I’m not talking about mobile cpu, they use 10nm so actually most of them have their own stenght like 11th 15wh have iris plus, 45wh have the best performance even better than ryzen 5000 45wh mobiles. But 11th gen on pc still 14nm+++++
Amusing how most of the improvement came from having more cache, rather than having a more sophisticated architecture. Looks like most of their R&D time was spent cramming more memory and achieving slightly more optimized performance. Ironic how their "tick-tock" model which takes after a clock, was perpetually stuck after 14nm, a sort of symbolism for them "being stuck in a time period".
I think its something thats incredibly hard to achieve, a ton of work from the top brains on the planet. We expect & demand CONSTANT LEAPS, but hey it doesnt always go to plan, fuk ups occur. (e.g the 10nm screw up). AMD languished in the mire for ages, and was so close to bankruptcy and thankfully had amazing turnaround/results. Intel had an excellent , almsot flawless run for ages. Anything less is now derided and heavily criticized. We should have our fingers crossed & blessings for every company at this, the better they do the better for all of us. Shit's hard yo
@@oldtimergaming9514 AMD simply didnt have enough competitive products to make ends meet, but they got their act together with new leadership/plan/direction and pulled off AMAZING results. Outstanding work, plus luck where it was needed (although good work creates more rolls of the dice/higher chance of success). Blaming intel or any other company is ridiculous. For tech/medicine etc, at least, we want results. We want breakthroughs and novel design. Think how much hard work/logistics is done even when they fail. The unappreciated efforts by the myriad of workers. Let alone the funding. Im in construction management, man the hours we work & effort we put in, and this isnt rocket science or chip design/fabrication at atom level, yet we still fuk up from time to time.
@@fredfinks blaming Intel is not ridiculous in the slightest. Look at the number of anti trust violations they've incurred. They literally pay oems not to use amd. They're use disgusting tactics all the time
Well considering that Skylake, Kabylake and Coffeelake are all the same architecture on more optimized process node for higher clock speeds, it's really not 5 generations, more like 2. Intel need to get their crap together.
@Solendore Dude they could still do that all this time, it's not the engineers, it's the executives that just milked us for so many years and they fired people while having record income. And I'm happy that they are back on track. A doupoly is bad, but a monopoly is worse. It was a child dream of mind to design CPUs as well, I was even self-educating in middle and early highschool, but then i found out that I have to move to the USA or China to work that so I stopped and started following my second passion software development, now I want to make interesting software or video games. XD
@Solendore please. Historically Intel was outperformed during many decades by its competitors and at merely every single generation gap. The company's strategy was more about shady copyright lawsuits to ruin smaller concurrents and monopolistic deals with Microsoft and OEMs than perf improvements.
@Solendore Pat is a clown and will end up ruining what's left after BK and Murthy, mark my words. Their ongoing sugar-coating of their delays will break them their necks. Neither will their 7nm arrive on time nor will their 5nm, 14nm was here to stay and will be pretty much their last actual node apart from the dumpster-fire called 10nm™ ...
@@Smartcom5 Definetly you are living under a rock. 11900KB from Beast Canyon is a 11900K that has half of the power consumption. Tiger Lake i7-11800H is a 45 watt version of the 10th gen i7-10700K. Same perfomance, ridiculous difference in efficency.
I am seriously amazed by the sheer amount of work you guys put into your research and testing routines, the only other channel who I can even think of doing this kind of work being Gamer's Nexus. By this time "Hardware Unboxed" has turned into "Hardware Unraveled".
Insane if you think about the fact that there was a time when Intel was ahead of TSMC. They stagnated so hard that TSMC is now ahead of them by a few years.
Which is exactly why I'm still not impressed with AMD. It took them until last year to catch up to years and years of almost 100% stagnation in Intel's line up. I'm still rocking a Xeon 1650v2 6 core / 12 thread CPU @ 4.6Ghz all core, 12Mb cache (based on Ivy Bridge-E on an x79 chipset). I can play ANY game with zero issues, there is no CPU bottleneck, and my DDR3 Quad channel setup has more bandwidth and less latency than any Ryzen CPU with two channel DDR4.
"Everyone" AKA the only other company designing desktop mainstream CPUs. Nobody denies that progress was slow at Intel, even before skylake. But we should not bend the truth: They are still competitive. "Everyone" else was really slow too. And why? Because Moore's Law is dead. Not because it is impossible to make faster processors but rather because the demand for more performance on the software side just is not there like it used to. Cutting redering times in half is pretty huge when the rendering of a 720p video takes the better part of a day. But in times where 4K videos render in minutes, it is a cool thing to have but nowhere near as urgent as back then. And when demand slows down, manufacturers will offer less improvement for more money. Which is what we see happening over at Nvidia also as demand for faster graphicscards slows down since everyone has what it takes to play at 1080p. Last but not least, let's not forget about the fact that the 5775C was pretty much the fastest damn processor that most consumers could afford back then while the 10105F is pretty much the cheapest and slowest offering that still somewhat has a bit of a good price to performance offering. So literally every CPU you can buy these days is faster and the more you spend, some of them have literally 4 times the core count. I can see progress there. It may not be as huge as in the 90s but I already explained why that is.
@@thingi to be fair AMD was almost bankrupt, they didnt have the financial power that intel had, hell they still dont, intel still has a lot more money even now. also AMD stock prices are now 10 times higher than what they were in 2017, which is also pretty impressive
Something to consider for the "Not how to run a business" and "The 5% improvements YoY was a lie" crowds: The 3-5% YoY improvements are real from the buyer’s perspective. Which, while technically unimpressive (done by adding cores/cache/using more power/etc), shows that you can easily run an existing business "successfully" for 5+ years even without much innovation. Having said that, holy fucking poo on fire was it time for a more technically-minded CEO to take over. Because running a business like that for more than the 5 years will get you in trouble fast.
Right. I think Rocket Lake being a backported 10nm design shows how absolutely squeezed for every drop of performance 14nm was. Thank God Intel finally got 10nm working, or they could have been _hurting_ this next gen.
With the amount of cash Intel has at their disposal, I would be more afraid of the customers hurting, even. Intel can drop a bunch of cash into a bidding war for TSMC/Samsung capacities, perhaps only second to what Apple can pay. Which, I think, would mean higher prices for the processors that are being manufactured _and_ shortages of silicon of companies that lost out because of the bidding war.
you forgetting that intel goal were to shrink cpu to slimmer laptops with better performance. the cpu's is powerful enough for many years now, more performance just not giving same results anymore.
you forgetting that intel goal were to shrink cpu to slimmer laptops with better performance. the cpu's is powerful enough for many years now, more performance just not giving same results anymore.
@Phoenix 𝙾𝚙𝚎𝚗 𝙼𝚢 PROFILE intel had no competition, thats how business works, now they have and as far as you can see latest intel cpu's not all that bad, future will show
Before 2017 Intel were competing against themselves so they had no reason to make any significant improvements. I think they decided to milk as much cash as possible from the customers. That's why competition is such a good thing.
Actually, the issue was more like trying to chew more than it could bite. Intel management gave too high goals and fast timetables and it caused process development to become stuck trying to solve issues caused by impossible goals and trying to rush things.
Earlier than that in fact, 14nm 15W mobile parts aren't faster than Sandy Bridge desktop ones core for core. Dual cores still are meh for Windows of course but it's not terrible yet.
What make me kind if upset. Is that Intel for so many years shortchanged the consumer marked, And said 4 cores are more than enough. Take it or leave it. If you need more cores you have to open your wallet wide open and get a HEDT CPU. Another puzzling history is that Intel slashed the prices of their HEDT CPUs with up to 50 percent from one generation to another. And still earn money. What a rip-off
From trash tier to god tier in 6 years. I'm going to laugh my royal ass off if AMD's V-Cache is enough to catch up to Alder Lake despite Alder Lake being on DDR5. If that happens Intel is going to be a meme forever....
@@andersjjensen it will. And I don't see how that's surprising to you. Nowadays cache is the biggest contributer to performance. The more the better. But those 3D cache chips will also cost a shit ton. Like a theoretical "5600X 3D" or something like that would easily be $400+. And Intel can always just drop prices to combat it. Not sure why the obvious happening would be funny.
And this explains why my 4770k is still delivering a good 1440p gaming experience all these years later. Intel has been treading water for a long time releasing new chips which are no better than last year's model.
@@davidandrew6855 even games like cold war lol... I went from 50-70 ish fps to 120... same settings.... same with my friend who went from the 6700 non k to 9600k and tarkov was the same... got a huge fps boost.
Steve, hey! I have to say I'm so pleased you're doing this type of content during the shortage rather than basically anything else! By far the best use of non-new-hardware benchmark time!
Zen 2 was just as much of a jump from zen1/zen+ .. AMD went from being stuck at 4790k/rtx2070 gaming performance with ryzen 3000 and rx5000 to the best CPU and joint best GPU. Hats off to Miss Lisa Su ! Before she took over AMD were pretty much stuck in 2010.
@@sabnox9869 The 11900K is technically better for gaming because it has 5.3GHz max clockspeed at similar IPC. If AMD launches a Ryzen 9 with 5.3GHz instead of 4.9-4.7 in games, then they are the gaming king because the singlethread speed can beat that i9 and you have more cores.
@@sabnox9869 The best gaming GPU is the RTX3090, not Radeon. The best overall CPU is the Ryzen 9 5950X, it's as good as the i9 in games but way superior for raw perfomance (twice the core count).
@@saricubra2867 Even core for core the 5800x beats the 11900k, the 10900k beats the 11900k in some games. And for raw FPS the rx6900xt matches the rtx3090. The 3090 does however have better productivity, raytracing and more features so it is better in some ways but in no way does the 11900k beat the 5800x.
@@sabnox9869 11900K has higher clockspeed than 5800X, the singlethread perfomance is higher and has integrated graphics. The 10900K is a Skylake chip and Skylake chip is behind Zen 3 and Rocket Lake in games, higher IPC or singlethread speed in gaming always wins over more cores.
I would love to see this same comparison to the last several generations on AMD desktop parts going from the old FX series through the Ryzen 5000 parts.
The biggest question is, whether all of the software based fixes for the Intel related security hazards were active during the testing. Most of the BIOS based fixes obviously were not applicable, as the latest Z97-Pro BIOS is 3,5 years old.
Sadly Steve didn't talk about that but it might explain why such an old chip with lower cache bandwidth, DD3 and many small improvements from moving to Skylake arch is not that much slower than its actual equivalent.
@@PainterVierax Interesting, doesn't Windows take precedent though for some of these security flaws, like Spectre and Meltdown? I thought that even if you had didn't have a bios microcode available, Windows would still have it and take lead. Wonder if that would also mean any performance hit would still be there even for the 5775c.
@@davidandrew6855 well sure. Microcode is how my old Linux rig manage to have mitigations despite the fact the BIOS wasn't updated by the manufacturer. The question is more if Steve disabled it or not.
@@PainterVierax Plus there are probably a number of flaws, that can only be handled prior to the operating system level. The operating system does not "see" bellow a certain level of the CPU functions. For those cases there were the regular BIOS updates, that fixed/excluded the faulty operations inside the CPU workflow management.
@@alogonomus6486 Not really since those are pure hardware architecture flaws only mitigated by randomizing data or disabling certain functions. Wherever the microcode comes from (BIOS or the booting drive) we still need an updated kernel to avoid them. To be clear, the microcode itself is a blob that runs before the OS initialization, it's not really part of it, it's just provided through the OS update tools. What is more harder to update is the firmware for the Intel ME and its AMD equivalent packed inside AGESA. The day those "security" falls (like it already happened with Qualcomm's implementations of Trustzone) we will be in deep shit as OEMs and motherboard manufacturers' support is terribly short.
When I saw the title, I was quietly hoping for a comparison of the CPU's in between as well, 6th, 7th, 8th and 9th gen, just to really show the lack of generational improvements.
3 to 5 years? I'm from a time when PC magazines would constantly tell you computers with high-end parts were expected to be relevant up to 10 years in terms of performance and most computing features, and that was from the late 90s up to late 2000s, when there was clearly significantly more progress gen-over-gen in terms of CPUs and GPUs! And just to note, during those 10 years we went from single core 700MHz Pentium II-III CPUs to dual-core Core 2 CPUs running at 2.6 GHz, and from AGP graphic cards such as the 3dfx. Voodoo2, Nvidia RIVA TNT, ATI Rage Pro with 64-128mb VRAM all the way to PCIe 2.0 GeForce 900 Series with 1GB VRAM! Also, the Pentium III was discontinued just in 2007, which proves the 10-year relevance people in the industry would then tell you about.
@@you2be839 are you drunk? What 10 years are you talking about? Late 90s and early 2000s when buying an PC even the guy selling it to you warned that it will be for upgrade within a year.
@@u3pyg Plenty of reputable PC magazines, national and from foreign press were saying that... I think I still have some lying around! Upgradable with something better within a year? Yes. Outdated after 5 years? Naturally. Totally irrelevant and unusable after 10 years? Certainly not! Intel would not have kept the Pentium III in production up to 2007 if they weren't selling enough of them to justify the production costs!! And I do happen to enjoy myself being sober 99% of the time, thank you.
This was interesting to see the per core performance scale of Intel over the last decade. I know it would be a lot of work. Could we get a similar comparison to AMD with either 4 core offerings just for the growth comparison.
lntel's history is just a confusing mishmash of core / cache configurations with an even more confusing naming scheme. I commend Steve for having the patience to benchmark these chips.
Would love to see a 2600K/2700K @ 4.2 GHz paired with D3 2400 C10/11 against 10300 @ 4.2 GHz paired with D4 2400 C10/11. Both parts have 8 MB of L3 cache, essentially an apples to apples IPC showdown! Might have to downclock a faster RAM kit for DDR4 because none of the 2400 sticks available seem fast enough, I guess because it's too close to original JEDEC spec.
@@GatsuOnBerserk Yeah its literally half the price of a i7 4790k and about on par in performance. Not a bad deal if you are searching for a cheap upgrade from a i5 4690k, you only pay about 20 bucks extra if you resell the 4690k and get much higher performance.
It's almost like the only real performance improvements we've been seeing for decades were actually the result of node shrinks and nothing magical architecture wise.
@KuraiShidosha Yeah that is where most of intels performance leaps were. But you also have stuff like Zen 2 to Zen 3, same process and with a 20% performance increase in a single year. Competition is the only factor pushing either company to improve.
do keep in mind that you are comparing an i7 to an i3. of course todays cheapest is only slightly better to a few years ago's most expensive. that's like saying the 1650 is bad because it doesn't beat the 980 ti. (obv a weird example now considering current market). It's a shame no publications have the size of the i3 die, so we can't compare how efficiently die space is being used. Another thing to consider is how much of the i7's die was being used for iGPU (182/259mm^2), which seems stupid considering most would pair this with a dedicated GPU.
@@kuyans3889 No disagreement, but Intel still follow this mantra with every K based SKU that they release. So 10900K and 10700K are no different to the 5775C but don't use that L4/ED-Ram, and even then, disabled down to 4 cores with all that extra L3 cache, were on average 3-4% per year better than the 5775C, with the same IPC/memory improvements than the 5775C, which isn't the best, and that's a direct apples to apples i7 comparison. 🤷
I think people were enamored with the 5775C because of it's massive L4 cache, which would be indicative of what we can expect in terms of results with AMDs newer CPUs that will feature a huge 3D V-cache. They were looking at comparing it to the other 4 core parts at the time, like Haswell, not necessarily comparative to chips today. Although still cool to see.
6 years ago we could've had so many more cores than 4, with this same 14nm. I bought at 4 core desktop replacement laptop in early 2016 and it didn't age all that well. While it was roughly as powerful as my preceding ~2012 desktop at the time it was a bit silly that for far less money I could move to a Ryzen desktop just a year later and the difference was quite noticeable even if I didn't really have a need for all the cores. The only thing I'm happy about is that I "only" paid 1600 € for the laptop instead of the regular 2700 € list price.
all intel has been doing in the past 6 years is just rename high-end cpus to mid-range and low-end, they haven't created anything new in a while i3-10100 = i7-7700 i5-10600k = i7-8700k i7-10700k = i9-9900k
@@obesespringroll3997 They're nearly exactly the same, what are you on about? The only reason the 10600k outperforms the 8700k at all is because it has a 30w higher tdp, hence the higher base and boost clocks (400Mhz and 100Mhz respectively)
@@christianmccollough5005 Nevermind that guy... His stock must have ran out and it's hard to grasp reality now. The 10th and 1th Intel generations are a flop and if he was paying attention he would quickly realized that. 3 generations of cpu's after 8700K and they offer like 5-10% of more performance? That's barely noticeable. With huge TDP's increases. So yeah, they are "equivalent" indeed. Idk about video rendering but in games? 10-15FPS is nothing really.
I would have liked to see the integrated graphics comparison. I remember Anandtech reviewing it and it not actually showing benefits in CPU tasks aside from a few edge cases. But the integrated graphics went like hell fire.
One thing I notice on the 5775C on Asus motherboards is if you mess with the cache multiplier or eDRAM clock it might kill the eDRAM performance and you'd need to reset the BIOS to default to regain it back. The only way to see this is if you check using aida64 cache and memory bench. When it bugs out the L4 cache speeds are going to be exactly memory speed. So just be careful if any of you guys are messing with a 5775C yourself.
Last intel CPU I bought was a 2500k. Used that until first gen Ryzen and been team red ever since. If Intel is ahead next time I need an upgrade I’ll switch back, but I tend to hold onto cpus for a long time so no predicting the market then, probably a few years from now
You showed cache/memory bandwidth. I would have also liked to have seen a cache/memory latency comparison as well. (I suspect the 5775C 128MB L4 cache might be lower latency than the 10th gen's DDR4 ram).
I love when you guys talk about old CPUs and my 4670k is too old to mention. Gaming wise, it still has no problem running "Mo Man's Sky" at 60 FPS with a small push and an RX 590.
It's a decade old m8. I've upgraded my CPU two times since then 😂 3700X then 5800X. But I got a new gpu and ultrawide so different needs Also NMS is from 2014 even if it's still being updated. Heavier games absolutely will stutter on 4C/4T but won't do so on 4C/8T
@@joeykeilholz925 A new build is in project right now. All that is left is board and CPU. A 5800x is planned, but waiting to see the new intel chips and boards. Cheers.
It's interesting to see that the Intel CPUs from the last 2-3 generations have been a very curious mix of miracle-level CPU design done on an obsolete manufacturing node. It's like having the best motor in a wooden chassis with cotton tyres... makes up to have a mediocre product in the end. Props to the chip design team! Also, this kind of blatantly shows one thing: the vast majority of the CPU performance that we got over the years was from shrinking the manufacturing node and adding more transistors. This isn't Intel being uncompetitive, that's how it is. While at it, I'm curios, did the i7 5775C (it really needs its name to be shown in the title or in the description) have the mitigations for spectre and meltdown vulnerabilities active ? I remember that it affected the performance by a few percent, while the newer (9th gen on?) CPUs have the mitigations directly in the hardware. Also, I'm even more curious about another thing... productivity benchmarks. Gaming, in general, isn't that CPU intensive. More like latency intensive. I suspect that the newer generations have more increases in productivity than in benchmark. Edit: I mean, even compared with the latest Zen3 CPUs, the i7 5775C might look good, and then you could say that the AMD CPUs of 2020 are only 15-25% better than a CPU from 2015, at the same core and clock count and with a much better manufacturing node. Which, of course, might be true for gaming, but would be so not true in general. Lastly, ... I know that there won't be any more videos on this topic, as much as I like #forscience but @Steve can you also add Doom Eternal for these benchmarks ? It's a shame not to. The game can run at 1000 FPS, given a powerful enough GPU, so it should work very well as a benchmark tool. Given that it is the best optimized AAA game in the last several years (by far, might I add), it deserves to be used as a benchmark tool. Ok, I'll stop rambling.
I'm always wondering why it took Intel this long to finally backport the Cove microarch to 14nm. They saw that 10nm had severe problems for many years which could not be solved and AMD was really putting pressure on Intel. Something like Rocket Lake could have been released much earlier. Sleepy management? Revenues still high enough?
@@johnscaramis2515 I'm gonna go with "revenue still high enough". Intel got complacent and only started to take AMD seriously when Zen3 launched and was faster in games - something which was Intel's bastion for over a decade. Also I have couple of friends and I see tons of people in comments that are buying every new Intel processor to upgrade performance - on many Rocket Lake reviews you can still see people saying they will upgrade their 10700k to 11700k. Though to be fair you'll also find a lot of people saying that "this time" it isn't worth it (but they still upgraded from 9th Gen to 10th Gen). To be fair - I simply don't understand this people. I don't have that much money - I buy new PC every 5-8 years. Sometimes with mid-life upgrade for GPU. In the past AMD used to be cheaper for similar computing power (not only games but overall), so I mostly bought AMD. Which is why I managed to avoid the worst of it ;) (bulldozer). The amount of people who still now buy "Intel no matter what" is staggering. Did AMD had problems with Software? Yes - my newest computer which is on Zen+ still needed to have "drivers" installed for CPU, so it could be utilized properly. But now Microsoft doesn't require that and is supporting them just like Intel CPU's. And their graphics drivers have been at least good for couple of years. When it comes to such obvious failures people have long memories. But when it comes to 14nm+++++++ people seem to forget and forgive in months. I do hope that Intel will succeed with Alder Lake and that AMD will not slow down it's generational improvements. Looking forward to see how Zen3D with Vcache will compete against Alder Lake, until Zen4 arrives. Especially since we know how important cache is. Though ATM managing to satiate demand is the most important thing - otherwise we will not see the prices go at least partially back to normal.
@@johnscaramis2515 Dr. Ian Curtess commented on that (Tech Tech Potato channel here on YT). If I remember correctly it took at least 2 years to do the porting. I mean, it does seem that they did went not too late to be truly obsolete.
Makes you wonder where intel could ve been, had they kept on going with their succes... Would we have weighed a faster 4 C intell cpu for the slower multi core Zen CPUs?
I went from a 2600k to an 8700k because the 8700k was the first CPU that felt like a big enough performance upgrade to make the move. Nvme SSDs were also a selling point for me. I'm now running a 9900K and will probably keep it for a couple more years.
@@stridezz7686 Not "pretty much".. it IS a 10700k. Both will do 5.0GHz all core at the same power level. Same cache size too. The only difference is the realization in the marketing department: "AMD is kicking our asses hard, so we guess an i7 now has 8 cores... you guys can shove more cores in for the i9, right?... right?!?".
Running an i9 9900k here at 5Ghz all-core on air. I'll probably keep this as my primary work and gaming station for another generation or two so I can see if Intel's 10nm and 7nm are truly viable. Otherwise my next workstation will be Ryzen based. In the meantime, I desperately want to upgrade my GTX 1080Ti to a 3080. I've been trying to grab a 3080 for 9 months now. 😡
If you have a 1080p monitor there is no AAA game you can't play at +75FPS with it still looking really good, so why bother with a GPU in these times? Especially considering all the rumours that Nvidia is actually taking AMD seriously, and that RDNA2 and Ampere are expected to be short lived generations...
@@andersjjensen I'm running a 38" ultrawide in 3840x1600, flanked by two 27" widescreens in 2560x1440... Hence the reason why I want to upgrade the GPU. We aren't going to be getting the next-gen GPUs for at least another year and a half.
Considering your results, and that the 4790K outperformed the 5775C by a good margin, you can actually see how few reasons there have been to upgrade for 7 generations so far. The 12th generation SoC architecture change seems like it may be the first game-changer worth upgrading to after all these years.
According to last years anandtech review, with good GPU broadwell eats 4790k (and 6th, 7th and in some games, like ff14 or civ6 trades blows with 10th gen) in games thanks to edram (and is worse for non-gaming due to lower clocks)
Yeah, what is with Intel still being on 14nm (+++ etc)?! i3-10105F is about 75 GBP at the moment, so pretty much the best budget choice there is, at least whilst 4C/8T is still ok for gaming. On a side note: I really enjoyed watching the new office come together by the way - you're a super handy dude/family, and i picked up some knowledge!
I run screaming from Intel 4th & 5th Gen - 4th on desktop and 5th in the NUC. Performance for hardware (FPGA) design software was noticably slower than 2nd and 3rd gen at higher clock rates and that was with HDDs on the 2/3'rd Gen and SSD on the 4/5'th Gen. Logins were a lot slower on 4th Gens - people thought they were the oldest machines we had. I tried BIOS tweaks, all sorts of different drivers; but then noticed that when I flipped to CentOS Linux - the 4th Gen ones had respectable performance with the FPGA design software.
50MB L3 cache and +100MB L3 cache days are starting next year. And further more next gen GPUs will be 2x from this generation. DDR5 platform launches are very soon. Huge gaming performance boost will be here to enjoy in less than 6 months! 4K 144FPS will not me a dream anymore.
It makes you think when it took Intel from 4th to 11th gen to only gain a 40% ish gain in IPC when it only took AMD from Zen+ to Zen 2 then to Zen 3 to get a 34% IPC boost minimum. That is quite shocking. What were Intel even doing for it to go so wrong.
It is interesting to look at the clock-for-clock comparison. But I think the bigger takeaway for me here is that a (much cheaper) 10th gen i3 gets you similar performance to the best mainstream i7 from 5 generations earlier. Clock for clock performance hasn't improved much at the same core count, but overall performance per dollar is much better than it was just a few generations previous.
True, but couldn't Intel do both architectural and core count increases? AMD sure has from it's Bulldozer/Piledriver days. I am not shitting on Intel like it seems to be cool to do these days, I actually would love to build an Intel system with a 10700f. However it seems like for some reason, whether it was no competition or a technical issues, architectural improvements over the last 5 to 6 years for Intel have been minimal. Alder Lake, may actually change all that, will be interesting to see.
@@How23497 I wasn't lied to as I never purchased a Bulldozer/Piledriver CPU, and still does not change what I stated. AMD has upped core count and IPC performance drastically, no matter how many true cores AMDs old architecture had or didn't have.
Pretty insane to see the IPC increase so little. Guess they were mostly busy fixing their thermal paste and replacing it with liquid metal to achieve higher clockspeeds, which is where all their gains seem to come from along with newer cache.
They actually aged a lot better than they released because software just wasn't made to take advantage of the architecture like it is now. Steve from Gamers Nexus still uses one in his personal rig and Jay just did a video testing how well they run today.
I've had a 4790K and a 5775c. The 5775c was - in terms of gaming experience - way better, while running much cooler without delidding. I got 4.3ghz out of mine and overclocked the L4 cache a bit too; I was running the 4790K at 4.6ghz. I bought a 10900F setup in January; obviously that is way better than either of them but, if you are a gamer and want the best CPU that will fit on a Z97 board, the 5775c is it. Unless you find a unicorn 4790K that will do 5ghz+ all day and the fastest DDR3 RAM you can ind etc.
I only recently went from my 4790k to a 8700k. I'm still pretty happy with the move, but really I just went from 80% utilization in game to 25%. DDR3 was holding me back in titles like Total War.
also isn't rocketlake on 14 nm enhanced superfin? just because they have the same number in the name, doesn't mean that the transistor density of a 5775 and a 10105 are the same.
Could you guys look at the improvements of quicksync over the years? I rip my blurays to h.265 and saw a huge leap moving from a 6700k to a 9600k. I'd love to know if the newer chips offer much more improvement.
Been sitting on my 8700k at 5ghz for about 4 (?) years now, playing at 4k with a 3090, there's been absolutely zero incentive for me to upgrade. Who would've thought investing in a 4k display and a heavy-lifter GPU would save you money in the long run? :D
Its important to OC L4 cache, it gives huge performance boost to 5775c in games. On ASUS mobos it can be done the same way as for cores and ring: with ratio multiplier.
I wonder what would've happened if Lisa Su never came to AMD and introduced Ryzen. Even with the gigantic competition from the red team Intel has no real improvements to show. I bet we would still use 4c 8t CPUs and all with 8th gen Intel or so.
@@TrueThanny Well if I'm not mistaken she was the one at the presentation holding the first gen CPU in her hand and showing it to the public. But sure she wasn't the one engineering it by herself I get the point. But she did bring AMD to whole new standards and I'm thankful for that 🤝🏼
I'd say what I personally took away from this is there are gonna be a LOT of people that stick with Windows 10 because despite MSFT saying you need the latest and greatest the older CPUs? Yeah they work just fine. I've seen that in my own family with my hand me downs Ryzen 1200 and even my old FX-8320e as both when paired with R9 380X GPUs, 16Gb of RAM and SSDs are happily playing games like World of Warships over 60fps with great visuals. So it honestly doesn't surprise me that the older Intel chip is still able to get playable performance as I've been saying for years when we switched from the MHz wars to the core/thread wars that chips quickly got "fast enough" for the vast majority and that most aren't even pushing their CPUs hard enough to worry about needing a new chip,especially since a new GPU that could really make the older CPU bottleneck will end up costing you more than a flipping used car in many areas.
Seeing this I'd say they could have kept the same motherboard chipset for all the years if they wanted to. Well of course they would have made much less money doing so.
They couldn't with the increased power requirements. What's more, Intel don't make motherboards, they just design the chipset. Except for certain OEM uses.
Yup. They even had to thin down the quartz layer to get rid of the heat to get there. They've probably explored every physical aspect that can be altered at this point just to show SOME improvement. The 9900K and the 10700K were practically identical in performance....
There's two ways to look at this 1: Intel's per core performance has aged well in gaming performance that a 2015 processor is keeping up with the latest and greatest games and GPUs 2. Intel's per core performance has stagnated so much that for 6 years that there haven't been any meaningful performance uplift in gaming so they are going the Moar Cores route just to make their chipsets look faster.
@@Spido68_the_spectator AMD and Intel's core design are not comparable, there's a reason Intel opted for that much cache on their CPUs and so did AMD. but what I understood about the latter is that, by design Zen architecture benefits from lots of cache, much better if there's little to no hops in between the core complexes.
@@ffwast They did have new stuffs on the pipeline, the problem is that they got both their hands tied up with their yield problems with 10nm. But if you mean designing a new architecture while sticking with the same 14nm process, then that's what Rocket Lake is all about.
It's surprising that a lot of folks seem to think only Zen2 and 3 existed, and insist that Zen (2017) on release knocked it out of the park. It didn't, because while it had some advantage in certain use cases due to more cores on mainstream, IPC and frequency wasn't anywhere near Intel's best, so much that an Intel 6/12 was still rivalling/equalling an 8/16 from AMD in MT workloads while devastating them in most games/ST apps. Even with Zen+ (2018), the contest was still fairly close, with Intel coming mostly ahead even with higher power. Zen2 (2019) was the real watershed moment - AMD finally ties or even pulls ahead, at lower cost and power as Intel equivalents, along with having more cores on average. A combination of cache, process advantage/refinement, and architectural improvements finally overcame the Intel's 'refined' 14nm process in all counts. Zen3 simply buried whatever was left, with even higher perf at lower power. Just keeping the timelines real.
The 4790k runs at 4.2GHz all core turbo and Steve clocked the 5775C to 4.2GHz. Given that the 5th gen is pretty much a die shrink of the 4th gen the result would be the same.
@@TrueThanny Oh, no I get that the video is about 14nm, i'm fine with the 4790k not being here. I'm just telling what I think the result would be since @Chinmay Tiwari asked.
man my pc feels old i7 2600k with gtx 960 running 12gb of ddr3 1333 was top back in days now still does job on games one day ill get good pc that dont make heat as its dam hot here where i am over 45deg in summer
I'm actually more curious to see what Intel and AMD could do if they temporarily forgot about the whole race to the smallest nm-sized processor, and instead actually tried making a "Cube" shaped processor. It would have new requirements when it comes to cooling, yes, but it would definitely become a beast not seen before.
Its still silicon, you still have to go through all the effort and Lithography all those transistors. Just grab 4 of the most dense CPUs now and stick them in a 4 socket motherboard. :D
Don't worry. AMD is working hard on it. V-Cache is basically 2.5D stacking, in the sense that they're still bound by planes, but can interconnect them efficiently. True 3D stacking spreads the logic across as many layers as they need, putting the least energy intensive logic at the bottom and the go-fast bits at the top. From what I can see in the patent applications they're going to use TSVs (the copper "towers" they're currently using for plane interconnect) as cooling towers to help transfer heat upwards.
You mentioned power consumption spiralling... what I felt this video missed was a power consumption comparison between the 10th and 5th gens in the tests where they performed similarly. Can you give a 1-line text summary on that?
All these results are not really that surprising. They'd be a lot worse for AMD if you were comparing it to an old AMD chip. What these tests tell us is that you don't fix something that's not broken. You just improve it little by little and that's what Intel did. It was a smart business move. It's quite phenomenal that Intel dominated the market for so long on basically one architecture and previous one. That is something they should be praised for. Gamers certainly were happy with that and so were a lot of other people. Yes you can still argue that the chips were expensive but they were the best at the time and still hold up. Now AMD chips are the best and they're more expensive. That's business.
This pretty much outlines how behind AMD was with Zen, Zen+, and even Zen 2. Intel was able to squeeze that level of gaming performance out of 14nm while AMD had to drop to 7nm to finally get ahead. Hopefully both of them can remain competitive going forward otherwise we're gonna get stuck with the same shit for another decade.
They should study Intel in the business management classes so they know how NOT to run a company - prioritizing short term profits rather than long term competitiveness.
