The 40 core score actually rounds to about 70k or 75% more perf. I added two 32 ecore scores 😅 this one ain’t happening anyways but it was fun to dream…
as an AMD fan, I hope Intel really delivers. Because at the end of the day, AMD is only good if Intel is good. If there is a monopoly, like Intel had, then we as customers lose.
Subscribing! Your analysis is good. I don't think Arrow Lake will be a big uplift over Raptor Lake Refresh in MT, but if it's better in ST and doesn't eat itself alive while being able to heat your home I think it'll be a solid generation for Intel.
Hate to burst your bubble, but at 200 Watts CPU power, the 9950X gets a CB R23 of 46,324. How do I know this? Because it is my 9950X. If I don't give a crap about power, then at 230 Watts CPU power the score goes up to 47,166. This to my mind isn't worth the added power it takes, so I run with the 200 Watt option. But wait, that's not all. At the same performance as the maximum I could get from my 7950X which had been up 24/7 for around 20 months running a 10-minute CineBench R23 benchmark, with a score of 40,253, my 9950X got a CB R23 score of 40,402 after a 10-minute run. The kicker? It used 65% less power to achieve that score. At about 30% of the power draw, the 9950X outperforms what you are showing as a maximum score (probably just a single run) for a 13900K. Put in that context, your guestimate for the flagship Arrow Lake isn't looking all that great.
Yup. I say that I’m concerned about arrow lake, and it may not perform. But dude, we should all want arrow lake to do well, if it doesn’t then it basically means we are in a time of stagnation on both sides. And that means these companies will get to be comfortable rather than competitive. I think you missed the point of the vid, not about amd or intel, just about competition. The difference is amd already launched and it was a flop. It seems like you are getting good results out of your chip, but that’s not the norm.
@@SiliconSteak The reason why I am getting good results is that I am not stupid enough to run my Ryzen CPUs on WinTel 11. Also, for five years, nobody in the Tech Media/RUclips has bothered their backsides about finding out how to configure Ryzen. But the rot at Intel goes back decades, in that, during that time, they have spent more of their resources on marketing than on R&D and that is now coming back to bite them in the ass - HARD. Until that changes, Intel is doomed to go the way of IBM.
I have identified my mic issues… and should have it resolved going forward.. (except on the vid slotted for Wednesday cuz it’s already recorded) cheers 🎉
It’s a specific issue with my blue yeti mic. It’s crackles on this pc but not on my mini server pc. People speculate it’s because of a software running in the background. I’m thinking it’s either Lian li rgb software, or Logitech software…
FIXED IT! So for my newest videos i have just been recording with a different PC and that worked, but i wasnt satisfied. So i realized that i hadnt tried changing the usb port on my computer. This sounds obvious but i had my mic plugged into a usb hub. So i changed the port the hub was it was plugged into and viola! no more static!
These chips will likely be more focused on efficiency. Maybe they may push them harder for performance later. They cannot afford another 13/14700/900 processor problem.
@@JoeL-xk6bo Just because something can be pushed doesn't mean it will be. Based on current leaked information it is suggested that the chip is more efficient at doing its job so it will take less power to reach the same or better performance (at least in theory) Given the problems that Intel are having pushing the 13/14700/900 chips so hard it is likely (although not 100%) that Intel won't push these chips as hard at least until they can be more sure of stability. So better performance for the same or less power equates to better efficiency.
@@JoeL-xk6bo it has 24 cores; productivity apps use all cores; high amount of cores require power; but gaming doesn t use that amount of cores; so it can be way more efficient in gaming where the cpu boosts only a few cores, compared to the power required to boost 24 cores to high clockspeeds
13900k cinebench r23 score in default settings is just about 37000. Almost the same with 7950X. So if arrow lake in default power mode gets 40000 in R23, then it's not bad. I think it will at least be about 43000, which is too good. Especially if the power is under 230W average when running that benchmark.
in the cinebench leaks; the core 9 ultra QS( which stands for quality sample); basically the core 9 ultra that is approved for mass production and either identical or almost identical in specs with the final version of core 9 ultra K had a score of 43k multicore; based on the math i did; the jump should be around 22% compared to 13th gen; which is really similar to the jump that 12th gen had over 11th gen; the i5 12400f was slight faster than i9 11900k; so if we get 13900k gaming performance at the price of a low end i5 it will be really good
Mine got over 38K score on a 13900K with manual tuned settings with PL1/PL2 at 230W (because I’m limited by DH-15). I would never use default settings at all.
@@Dweller12Videos noted. The static I’m getting is due to something interfering with my mic on that particular pc. My server pc doesn’t show the same static for some reason
660Watts.. Satan: Damn and I thought it was hot in my office! Edit: Well after finishing watching the whole thing. I think a tie could be pretty good. Really Intel needs to Improve efficiency. In a lot of reviews that seems to really be where they're losing their ass off. A moderate gain in performance with a significant gain in efficiency will would be awesome. Orcs: Air coolers are back on the menu boys! I'm somewhat Intel biased, 20+ years of using Intel systems.. But I'm also an air cooling enthusiast. If I can't reasonably air cool it, I ain't buying it. Still running a 9700k desktop chip in my laptop. But here soon, building a new system, and going back to desktop. Once the new stuff drops anyways. If I can air cool an i9.. I'll be a happy man lol.
Yeah I rather have coolers being optional not a requirement. In fact its so much easier to cool a 30 watt CPU than some 300 watt CPU. The goal is to push them lower than 50 watts to compete with Apple and ARM. I hate people say it doesn't matter. Mobile devices is now the big thing and Intel failed at that. Its would be nice to run a powerful PC that could run on 60,000 mah battery like a Steam Deck setup.
There is no way that hyperthreading adds 43% to the multicore performance of the P-cores even on something like cinebench. that seems very unrealistic to me. I would be shocked if it has gotten that good, it certainly wasn't anywhere near that good a handful of generations ago. AMD's SMT was typically quite better than hyperthreading for adding multithreaded performance.
We are entering another period of at least a couple generations of stagnancy. Maybe longer. Sucks but here we are. Zen 5 was the least impressive release of zen so far and took a full 2 years, longer than it used to between the last release. I think arrow lake will be ok, but nothing special even in a best case.
All these high end chips seem to be putting a ton of E cores on top of 8 P cores, just like 14th gen. What if.... they had a SKU with 10 P cores and no E cores? Why don't they do that? TDP? Some people want raw power at all cost. Controlling your background tasks to be as little as possible so all performance is focused on the task at hand, be it gaming, big data or scientific work would seem like something that people would buy. There were no E cores until 12th gen so I am not excited about E cores whatsoever (I have 2x 13900k systems and 1x 14900k system so coming from a friendly POV).
I get the only p core argument. But we don’t really need more than 8 imo. That’s enough threads to saturate a high demand app that is in need of high single thread, if you need more threads and more multi core, why not just add more smaller cores that are more efficient in terms of silicon space. Just my thoughts
@SiliconSteak This is very true regarding Socket restrictions. For eg. with Zen4 and RPL, TSMC’s 5nm is extremely good and far ahead of Intel 10nm. With Zen 4 AMD should’ve beat intel into the ground, but they ended up losing slightly when it comes to all round performance. Some factors which I presume contributed to their loss: AMD tried to keep compatibility with AM4 coolers, so ended up making this chip with a smaller contact surface area, and a much thicker IHS. The thicker IHS can be compensated for to some extent like how they did it with the 8700-G by extending only the area in with the substrate and removing material from the other areas to keep it thin everywhere else except the portion in contact with the substrate. But it's not clear how they can make it much better with the chiplets themselves being waay off center. There's no workaround for this at the moment. There is also no workaround for the surface area issue. A larger plate makes for slightly better heat transfer characteristics when there’s an AIO plate to match. It would also help center the compute dies a bit with respect to the coldplate of the cooling solution, which would help. Maybe they can try to center the compute dies and fill out the spider leg area and place capacitors underneath the IHS overhang? Then there's also heat density, they could've just made the features larger or a bit more spaced-out during lithography/etching. I'm pretty sure that you'd end up with slightly larger dies by a few mm2, but it might've been possible to push it further before it hit the ceiling. I'm guessing, they pushed things very close to the transistor density limit of what the node is capable of? Did it contribute to the situation? We can only speculate. But as the situation stands RN, currently the high heat density of the dies contributes to the situation where Zen 4 is difficult to keep cool at much lower wattages comparatively. Ie. The CPUs hits the temperature ceiling at much lower wattages. And the falloff situation arises much faster, if you were to plot it on a graph, you'd see how rapidly it approaches the thermal limit. This is also another situation which could've arisen due to the choice to keep cooler compatibility. Since there's a hard limit on available space for the silicon. If they didn’t make these compromises and any other ones which I may have missed out, the result could have been different. PS: I'm getting notifications for one of my earlier comments, but I can't seem to find underneath the comment I replied to. What's up with that?
@@jenxrj it’s the I/O die. They didn’t update it and the cores are constrained for bandwidth and there’s too much latency. Don’t get me wrong, those other factors contribute aswell.
@@SiliconSteak With a new IO die? Maybe. But I think their primitive packaging technology is also hamstringing them significantly. Intel have done something incredible with Foveros. Their idle & standby numbers are so strong, they're using it in Mobility products. AMD still haven't used their chiplet tech in anything similar yet, even though it has been long since they brought it to desktop & server. Idle/standby numbers are excellent for MTL, on par, even beating monoliths. For. eg. Zen 4 uses like ~30-35 watts doing nothing, monoliths use about 4-5 watts. youtube(dot)com/watch?v=Obtc24lwbrw youtube(dot)com/watch?v=D1mbMg2hjXs
Good job on the videos ... a few suggestions in my opinion .... change the BGM to something else .. maybe no lyrics .. kind of distracting, also your mic has some sort of digital anomaly, maybe something with your audio interface ?
Ah I see that was the article title. Arrow lake cpus are already out there, based on the leaks of it being tested, it doesn’t have any instability issues
@@SiliconSteak Thanks, but I think until the major reviewers get them we'll not know for sure the stability, benchmarks, power utilization, etc. But I like the premise of your video, Intel really needs to hit a home run with Arrow Lake. Honestly AMD needed to hit a home run with Ryzen 9000 but they did not.
Yeah for sure, that’s just a leak. I don’t think I emphasized the instability part. I was more focused on performance and whether Arrow lake will be viable or not
Some things don't really add up in your estimations. Let's assume +10% IPC at the same clockspeeds (unlikely, but let's assume so). Then, let's assume 30% performance loss due to not having HT (unlikely too in a design that's built with not using ht in mind). So, we'll have 17k for P Cores only in cb. Now, intel says Skymont has +2% IPC over Raptor Lake (it doesn't but that's in an ideal scenario that intel will not use in ARL). Let's assume RPL Level IPC for an optimistic approximation and let's assume the same clock speeds. In a 14900k, the e Cores are running at 4.4GHz in a all core scenario and the P Cores will run at 5.5GHz (P Cores only). Assuiming the e Cores will run at the same clock speed, one e Core will be the 80% the performance of one RPL P Core (4.4/5.5). So, the e Cores might add 24000/8*16*0.8 Points in a best case scenario. Therefore, we could assume an up to 38.4k points for e Cores only. Adding the 17k from the P Cores, we would have about 55k points in cb r23 which would be a 37% improvement over RPL. I don't think that improvents will be close to this, given that Skymonts IPC will be significantly lower than RPL ipc (according to intel) but when you're trying to make estimations on theoretical performance you should keep that in mind.
Yes not all the estimates are accurate, I had 3 different scenarios with an extra one being for the 40 core version. It’s up to the viewer to decide what’s right. I also welcome any more speculation about this because I’m interested.
But I think the way your looking at is Unnecessarily complicated. You aren’t factoring in ecores only have 1 thread compared to raptor coves 2 (hyper threading) that’s why I explicitly said that in a single threaded application they would perform the same at the same clock speed, multithreaded is a different story.. Separately calculating performance of the pcores and ecores based on their disclosed ipc from lunar lake, is the easiest and most accurate way to do this. I don’t see the point of going backwards and trying to figure out chadmount perf from raptor cove perf. Just add the ipc to the raptor lake p core and ecore and your done… also skymount and raptor cove have the same ipc, this has been officially disclosed by Intel in the slide I was reading in the video. One last thing. 10% ipc uplift for lion cove does seem too low to me too. But for the first conservative estimate, I went with the worst case scenario. In my mind going to 7nm to 2nm and a brand new architecture would get you something like 30% ipc but the leaks aren’t pointing to that
@@SiliconSteak well, but that’s the easiest way of calculating given that Intel does not disclose IPC improvements between RPL e Cores and ARL e Cores. You’re right about Skymont not having HT but in a benchmark like CB it doesn’t have to much effect anyway. In most tests, ht adds about 5-10% performance in cinebench and therefore e Cores might will hit up to 35k points on their own. No, Skymont and Raptor cove do not have the same ipc, at least not in practice. Intels numbers were based on a theoretical design where a Skymont core has the same access to L3 cache and Ring as a Raptor cove core. Both drastically improve performance at the same clock speeds (and therefore ipc). In ARL, e Cores will be clustered again, so each four core cluster will have to share its access to L3 Cache/Ring.
Those top 2 SKUs were probably cancelled because there would be no realistic way to cool a monster like that, over 600 watts is insanity, they probably realized this and cancelled it. Also it may just be an experimental only SKU.
@@SiliconSteak well, currently, if you think about it, Intel CPUs can't fully take advantage of DDR5 anyways, the IMC isn't good enough to allow for higher speeds using four sticks of ram. So with that limitation, it makes sense to not bother with the higher core counts, also, beyond server loads, not many programs would even utilize that high of a core count. Mostly it's just used for being able to run multiple programs at the same time.
Arrow lake will have PowerVia and ribbonfet .. been hearing about how backside power delivery has been tough to crack but will create less power needed and more room for transistors for future use.. they are on track with the 5 nodes in 4 years.. will see if they reach their goal of 1 trillion transistors on a chip by 2030
@thpunish3r735 I think I saw reports saying that ARL is gonna be using N3B. @@SiliconSteak this video was well put together and I don't think I've seen anyone come up with such an analysis of ARL leaks complete with reasoning and arguments to tow. Nice job.
Intel could offer all affected customers a MSRP equivalent (I.e. $649) rebate coupon on future Intel purchases, as alternative to RMA. They could then MORTGAGE the current value of the total potential rebate claim versus the actual future claim. This would reduce the total RMA cost, amortize the loss, and keep customers locked into Intel all at the same time.
If they are same as 13900k performance but 100w less i would consider that a big win. If its the same and 15-20w less, I will stick with AMD. Waiting this one out to pick my winner.
Wake up call, this already happened 10 years ago. Since then the performance advancement slowed down significantly. But thing is that it is also less important then in the '90-00 era this giant where really much needed for software advancement. Now not that much anymore.
@@kristof6372 let's assume you have the expertise to make the claim that x86 instruction sets are a limitation. What is it about the x86 instruction set that is the limitation?
@@Matrix1Gamer I disagree. Typically for (80%+) of applications you only need a handful of really fast threads (pcores) but when it comes to multithreaded optimized apps you need all the cores and threads as possible. This is a smart design. Give us a handful of blazing fast threads for the single thread apps and then a ton of smaller cores which are basically a multicore accelerator. Smart design imo, but if you are talking about the segmentation, ya it’s weird. Ultra 9 maybe should have 10p core with ultra 7 having 8.
Panther lake will most likely be mobile only like meteor lake. We will prob have a arrow lake refresh, (they might put in a 40 core version there) and then we will have nova lake on lga 1851
If 4 E-cores take up the same space as 1 P-core, I really wish intel would release a full E-core desktop version like instead of 8+16, I would buy a full 48 e-cores version. I don't need 200-300fps in gaming, so I don't need those P cores, but I do need to multitask better in my system, productivity + multiple gaming in background+ recording + audio editing, watching videos. I have 128GB of ram but a lot of time my CPU (13700K) is bottlenecking me at the moment.
Watch every impressive intel launch; they were always on a new platform; you ll not find an impressive intel cpu that was already on an existing platform; all intel cpus were beasts on their first gen on the platform; then filler cpus for the same socket until new one comes out again with impressive jumps
The lga 1851 should have the same Z height. There's no reason to break cooler compatibility since the pin count is not massively different and can be accommodated in the same area.
@@SiliconSteak also; cinebench is a bit tricky bcs scores in cinebench don t equal gaming performance; the memory controller might be much better than raptor's lake with way higher stable ddr5 frequency which will also help in gaming; the architecture itself is new and internal cpu lag might be lower; and again, intel doesn t have to provide extreme high end performance uplifts; look at 12400f and 11900k; 12400f is 1-2% faster than 11900k on avg; if the core 5 ultra is in +-3% of 14900k, that s a really good uplift on the cpus that matter for gamers; bcs even now and always i5s will be the cpus that provide most value in gaming; the higher the tier of the cpu the more you pay for the extra cores that won t provide much gaming performance, only productivity; and as in ur video the E cores bump especially the 1% lows quite well; now that arrowlake has way stronger E cores, the performance lift on arrowlakes might be higher; for now, unless other leaks appear, based on math i expect arrow lake to be around 20% faster in gaming than raptor lake which is really similar to how 12th gen was compared to 11th gen; also as u said in the video, the E cores will match the raptor lake P cores performance at same clockspeed; so basically every E core will be a i5-13500 P core bcs those have 4.6 ghz; the core 3 ultra will be a 4P+4E, much stronger beast than i3-13100f which has 4P cores at 4.5ghz; basically only arrow lakes E cores will be faster than the i3 which is already a really impressive budget cpu; idk why but i think that the core 3 ultra with it s price and performance will be the highlight of arrowlake; and on 2nd place the core 5 ultra that will probably be slightly faster than 13900k
I really think Intel Arrow Lake should change the game Intel and AMD have always provided multiple options, with 1 company going into dust we will have a lower competition and also much less CPUs to choose from if Intel 14th gen were not cooked AMD 9th would be a$$, its not really that great plus it does not even do good in productivity intel does both productivity and Gaming so if arrow lake does well, Intel is back in game and AMD is Screwed
@@SiliconSteak to do so, they'll have to find a future proof replacement for their IF. That will be their most important point to improve. All Zen architeuctures suffer from the huge if bottleneck
@@harsha6937 Which AMD are you talking about? The 9950X is the strongest CPU on the market in terms of productivity where clock frequency is concerned, leaping past Intel's 14700K, which for productivity is better value than the 14900K from an economical stance. "Productivity" in my case refers to those tasks wherein single-core performance matters a lot more than core count: audio production, signal processing, music production, and some aspects of videographic production and programming. When ordered, for example, strongest-to-weakest CPUs for production (still all good) would go: 9950X > 14700K/14900K > 13700K > 12900K. If Intel wants to take down the current best performer, it need do but one of two things if not both: (1) reduce their upcoming chip's price to undercut AMD's currently massively overpriced 9950X, or (2) either increase the single-core performance beyond that of what the 9950X currently offers or guarantee significantly --- and I mean *_significantly_* --- greater power/performance efficiency, sort of like AMD has done, impressively. More performance output at lower power draw or voltage demonstrates a mastery of engineering or manufacturing, contrary to what Intel's 12900K, 13900K, 14900K, 13700K, and 14700K, and several of AMD's previous chips have demonstrated, which implies a level of infantile stagnation. *EDIT:* _Marketing-wise,_ the price of the 9950X is enough to send everyone here to the intensive care unit in critical condition after laughing at the sheer absurd levels of pricing. It's a joke. Basically CAD$1,000 for a CPU that won't be moving as many units as its 3D counterpart --- why? The 14th Gen is also overpriced severely, and the 13700K is currently nearing decent territory. The 12900K is decently priced for what it offers. That 9950X, though...? Jesus Christ. Drugs are bad.
This chip's about to fry your rig SO FAST you'll regret not having fire insurance after your entire block burns down due to rapid degradation in your sleep. *Thanks, Intel.*
It doesn't look to good. single thread performance (game performance) looks bad, about 4% uplift. Multithreading is worse than the i7 and i9 since it lacks hyper threading. the power limit is STILL 250W with an unlimited mode on top of that. They're aiming to push it to the limits of breaking AGAIN because they raised the safe operating temp by 10 degrees. Everyone keep that same "Zen 5%" energy.
@@JoeL-xk6bo multithreaded performance might not impress due to loss of hyperthreading. I don’t see a scenario where single thread is only 4%. I think single threaded will at the minimum be 10% faster, I could see even up to 20% faster max oc. Ultimately we will see perf/efficiency gains soon, seems like they are launching in October. I know there will be clock speed regression on the p cores, and we are going to a chiplet architecture. If these things really impact performance negatively then maybe we could only see 4% single threaded uplift. (Very unlikely imo) You have to understand, zen 4 to zen 5 was from 5nm to 4nm, with the same 6nm I/o die. Raptor lake to arrow lake will be going from intel 7nm to intel 2nm (which for all intents and purposes are 7nm and 2nm because they are equal to tsmcs 7 & 2) this is a much bigger jump, and I would expect to see much better gains than 4%.
@@SiliconSteak It's actually Intel 10nm SuperFin (known as Intel 7) the underlying architecture is still comparable to 10nm, which is slightly inferior to TSMC 7nm in density. (2nm, 7nm and 10nm aren't actually the actual transistor measurements, rather just names given to a rough transistor size, so 2nm better than 4nm, 4nm better than 7nm, 7nm better than 10nm and so on)
people obsess over performance too much when typically they don't need nearly that much cpu for games, and only professionals really need it for various real world uses. The home user should concentrate on better software. Everyone should concentrate on better software.
@@EverSpaceTime I have a m1 air for mobile work. No regrets sips power and does what I need it to do. If I’m out and about I can even remote into my windows server that has a 12600k 64gb of ram and a a770 16gb. Best of both worlds
@@EverSpaceTime I use Mac and windows very differently. On windows I am more of a power user and like to have all the options available to me. On Mac I basically just use it as a web browsing/word typing machine for paperwork. If I need some serious grunt or compatibility I route into my windows machine. Basically I’m saying to get both so you have the best of both worlds, if I just had Mac I wouldn’t be satisfied because they just don’t give the users enough options like windows does. The software is well optimized and most things just work, but I wouldn’t trade my windows machine for anything
The 40 core score actually rounds to about 70k or 75% more perf. I added two 32 ecore scores 😅 this one ain’t happening anyways but it was fun to dream…
as an AMD fan, I hope Intel really delivers. Because at the end of the day, AMD is only good if Intel is good. If there is a monopoly, like Intel had, then we as customers lose.
Being a fan of a silicon company is weird as hell to begin with
yeah because 9950x already suckass
@@SPG8989 Tell me you're a fan of nothing...
Let's hope it won't be *Bottom of the Lake* , like the 13th and 14th gen Core i9s
The Core Ultra 9 285K needs to provide 10-20% faster performance than the 14900KS if AMD is to start competing again
Not to mention significantly lower power draw. Cuz the 14900KS is one hot and hungry boy...
Subscribing! Your analysis is good. I don't think Arrow Lake will be a big uplift over Raptor Lake Refresh in MT, but if it's better in ST and doesn't eat itself alive while being able to heat your home I think it'll be a solid generation for Intel.
Hate to burst your bubble, but at 200 Watts CPU power, the 9950X gets a CB R23 of 46,324.
How do I know this?
Because it is my 9950X.
If I don't give a crap about power, then at 230 Watts CPU power the score goes up to 47,166. This to my mind isn't worth the added power it takes, so I run with the 200 Watt option.
But wait, that's not all. At the same performance as the maximum I could get from my 7950X which had been up 24/7 for around 20 months running a 10-minute CineBench R23 benchmark, with a score of 40,253, my 9950X got a CB R23 score of 40,402 after a 10-minute run.
The kicker? It used 65% less power to achieve that score.
At about 30% of the power draw, the 9950X outperforms what you are showing as a maximum score (probably just a single run) for a 13900K.
Put in that context, your guestimate for the flagship Arrow Lake isn't looking all that great.
Yup. I say that I’m concerned about arrow lake, and it may not perform. But dude, we should all want arrow lake to do well, if it doesn’t then it basically means we are in a time of stagnation on both sides. And that means these companies will get to be comfortable rather than competitive.
I think you missed the point of the vid, not about amd or intel, just about competition. The difference is amd already launched and it was a flop. It seems like you are getting good results out of your chip, but that’s not the norm.
@@SiliconSteak The reason why I am getting good results is that I am not stupid enough to run my Ryzen CPUs on WinTel 11.
Also, for five years, nobody in the Tech Media/RUclips has bothered their backsides about finding out how to configure Ryzen.
But the rot at Intel goes back decades, in that, during that time, they have spent more of their resources on marketing than on R&D and that is now coming back to bite them in the ass - HARD.
Until that changes, Intel is doomed to go the way of IBM.
No rentable units or royal core = no faith in intel
I have identified my mic issues… and should have it resolved going forward.. (except on the vid slotted for Wednesday cuz it’s already recorded) cheers 🎉
What was the issue? Have a similar audio issue and have been unable to pin down where it comes from.
It’s a specific issue with my blue yeti mic. It’s crackles on this pc but not on my mini server pc. People speculate it’s because of a software running in the background. I’m thinking it’s either Lian li rgb software, or Logitech software…
yeah the sound is crackling
FIXED IT!
So for my newest videos i have just been recording with a different PC and that worked, but i wasnt satisfied. So i realized that i hadnt tried changing the usb port on my computer. This sounds obvious but i had my mic plugged into a usb hub. So i changed the port the hub was it was plugged into and viola! no more static!
These chips will likely be more focused on efficiency. Maybe they may push them harder for performance later. They cannot afford another 13/14700/900 processor problem.
it has the same 250W PL2 and also has unlimited mode for the i9. efficiency where?
@@JoeL-xk6bo Just because something can be pushed doesn't mean it will be. Based on current leaked information it is suggested that the chip is more efficient at doing its job so it will take less power to reach the same or better performance (at least in theory) Given the problems that Intel are having pushing the 13/14700/900 chips so hard it is likely (although not 100%) that Intel won't push these chips as hard at least until they can be more sure of stability. So better performance for the same or less power equates to better efficiency.
@@JoeL-xk6bo it has 24 cores; productivity apps use all cores; high amount of cores require power; but gaming doesn t use that amount of cores; so it can be way more efficient in gaming where the cpu boosts only a few cores, compared to the power required to boost 24 cores to high clockspeeds
Let's hope this is not a broken arrow for intel... heh
That's a good 1 lol.
All Pat has to do is keep reciting bible verses and doing pushups on stage, Intel shall prevail then...
13900k cinebench r23 score in default settings is just about 37000. Almost the same with 7950X. So if arrow lake in default power mode gets 40000 in R23, then it's not bad. I think it will at least be about 43000, which is too good. Especially if the power is under 230W average when running that benchmark.
in the cinebench leaks; the core 9 ultra QS( which stands for quality sample); basically the core 9 ultra that is approved for mass production and either identical or almost identical in specs with the final version of core 9 ultra K had a score of 43k multicore; based on the math i did; the jump should be around 22% compared to 13th gen; which is really similar to the jump that 12th gen had over 11th gen; the i5 12400f was slight faster than i9 11900k; so if we get 13900k gaming performance at the price of a low end i5 it will be really good
Mine got over 38K score on a 13900K with manual tuned settings with PL1/PL2 at 230W (because I’m limited by DH-15). I would never use default settings at all.
Well made video Mr. Steak, definitely earned my subscription
Check out the song I just released about arrow lake!
660w+? 70k+? At that point they might as well launch dual socket motherboards and call it a day.
basically epyc
Yea, that sku is now rumored to be cancelled
Let hope it’s as they say. Still using my 13900k and score in cinebench 40600 at stock which still powerful I have been using so it came out
Don't use loud music or any music with lyrics as background music. It is overpowering your voice when played on any speakers with compressed audio
@@Dweller12Videos noted. The static I’m getting is due to something interfering with my mic on that particular pc. My server pc doesn’t show the same static for some reason
@@Dweller12Videos the lyrics are funny tho
Lmao that opener was over the top and I love it.
660Watts..
Satan: Damn and I thought it was hot in my office!
Edit:
Well after finishing watching the whole thing. I think a tie could be pretty good. Really Intel needs to Improve efficiency. In a lot of reviews that seems to really be where they're losing their ass off. A moderate gain in performance with a significant gain in efficiency will would be awesome.
Orcs: Air coolers are back on the menu boys!
I'm somewhat Intel biased, 20+ years of using Intel systems.. But I'm also an air cooling enthusiast. If I can't reasonably air cool it, I ain't buying it. Still running a 9700k desktop chip in my laptop. But here soon, building a new system, and going back to desktop. Once the new stuff drops anyways.
If I can air cool an i9.. I'll be a happy man lol.
Yeah I rather have coolers being optional not a requirement. In fact its so much easier to cool a 30 watt CPU than some 300 watt CPU.
The goal is to push them lower than 50 watts to compete with Apple and ARM. I hate people say it doesn't matter. Mobile devices is now the big thing and Intel failed at that. Its would be nice to run a powerful PC that could run on 60,000 mah battery like a Steam Deck setup.
I'm not convinced that we see this kind of uplift with AL without hyper-threading. They must have something up their sleeve?
@Bass.Player chadmont
@@Bass.Player also, the conservative estimate I gave was basically 0% multi perf uplift too. I think we will get a little more that though
There is no way that hyperthreading adds 43% to the multicore performance of the P-cores even on something like cinebench. that seems very unrealistic to me. I would be shocked if it has gotten that good, it certainly wasn't anywhere near that good a handful of generations ago. AMD's SMT was typically quite better than hyperthreading for adding multithreaded performance.
Its more in the range of 20-30% perf from HT
We are entering another period of at least a couple generations of stagnancy. Maybe longer. Sucks but here we are. Zen 5 was the least impressive release of zen so far and took a full 2 years, longer than it used to between the last release. I think arrow lake will be ok, but nothing special even in a best case.
All these high end chips seem to be putting a ton of E cores on top of 8 P cores, just like 14th gen. What if.... they had a SKU with 10 P cores and no E cores? Why don't they do that? TDP? Some people want raw power at all cost. Controlling your background tasks to be as little as possible so all performance is focused on the task at hand, be it gaming, big data or scientific work would seem like something that people would buy. There were no E cores until 12th gen so I am not excited about E cores whatsoever (I have 2x 13900k systems and 1x 14900k system so coming from a friendly POV).
I get the only p core argument. But we don’t really need more than 8 imo. That’s enough threads to saturate a high demand app that is in need of high single thread, if you need more threads and more multi core, why not just add more smaller cores that are more efficient in terms of silicon space. Just my thoughts
@SiliconSteak This is very true regarding Socket restrictions. For eg. with Zen4 and RPL,
TSMC’s 5nm is extremely good and far ahead of Intel 10nm. With Zen 4 AMD should’ve beat intel into the ground, but they ended up losing slightly when it comes to all round performance.
Some factors which I presume contributed to their loss:
AMD tried to keep compatibility with AM4 coolers, so ended up making this chip with a smaller contact surface area, and a much thicker IHS.
The thicker IHS can be compensated for to some extent like how they did it with the 8700-G by extending only the area in with the substrate and removing material from the other areas to keep it thin everywhere else except the portion in contact with the substrate.
But it's not clear how they can make it much better with the chiplets themselves being waay off center. There's no workaround for this at the moment.
There is also no workaround for the surface area issue. A larger plate makes for slightly better heat transfer characteristics when there’s an AIO plate to match. It would also help center the compute dies a bit with respect to the coldplate of the cooling solution, which would help.
Maybe they can try to center the compute dies and fill out the spider leg area and place capacitors underneath the IHS overhang?
Then there's also heat density, they could've just made the features larger or a bit more spaced-out during lithography/etching.
I'm pretty sure that you'd end up with slightly larger dies by a few mm2, but it might've been possible to push it further before it hit the ceiling.
I'm guessing, they pushed things very close to the transistor density limit of what the node is capable of? Did it contribute to the situation? We can only speculate.
But as the situation stands RN, currently the high heat density of the dies contributes to the situation where Zen 4 is difficult to keep cool at much lower wattages comparatively.
Ie. The CPUs hits the temperature ceiling at much lower wattages. And the falloff situation arises much faster, if you were to plot it on a graph, you'd see how rapidly it approaches the thermal limit.
This is also another situation which could've arisen due to the choice to keep cooler compatibility. Since there's a hard limit on available space for the silicon.
If they didn’t make these compromises and any other ones which I may have missed out, the result could have been different.
PS: I'm getting notifications for one of my earlier comments, but I can't seem to find underneath the comment I replied to. What's up with that?
@@jenxrj it’s the I/O die. They didn’t update it and the cores are constrained for bandwidth and there’s too much latency. Don’t get me wrong, those other factors contribute aswell.
@@jenxrj watch, the next time they update the I/O die there will be a huge improvement in perf. Mark my words
@@SiliconSteak With a new IO die? Maybe. But I think their primitive packaging technology is also hamstringing them significantly. Intel have done something incredible with Foveros. Their idle & standby numbers are so strong, they're using it in Mobility products. AMD still haven't used their chiplet tech in anything similar yet, even though it has been long since they brought it to desktop & server. Idle/standby numbers are excellent for MTL, on par, even beating monoliths. For. eg. Zen 4 uses like ~30-35 watts doing nothing, monoliths use about 4-5 watts.
youtube(dot)com/watch?v=Obtc24lwbrw
youtube(dot)com/watch?v=D1mbMg2hjXs
Very true
@@SiliconSteak Can't find my latest comment again after I refreshed the page..lol. Something's up with YT.
Good job on the videos ... a few suggestions in my opinion .... change the BGM to something else .. maybe no lyrics .. kind of distracting, also your mic has some sort of digital anomaly, maybe something with your audio interface ?
Yea I’ve fixed the audio. Thanks for the comment!
How do you know there are no instability issues if its not out yet?
Where did I say that in the vid (i genuinely dont remember saying that)
@@SiliconSteak Sure, its was in the very first few frames, its printed into the background.
Ah I see that was the article title. Arrow lake cpus are already out there, based on the leaks of it being tested, it doesn’t have any instability issues
@@SiliconSteak Thanks, but I think until the major reviewers get them we'll not know for sure the stability, benchmarks, power utilization, etc. But I like the premise of your video, Intel really needs to hit a home run with Arrow Lake. Honestly AMD needed to hit a home run with Ryzen 9000 but they did not.
Yeah for sure, that’s just a leak. I don’t think I emphasized the instability part. I was more focused on performance and whether Arrow lake will be viable or not
Some things don't really add up in your estimations.
Let's assume +10% IPC at the same clockspeeds (unlikely, but let's assume so). Then, let's assume 30% performance loss due to not having HT (unlikely too in a design that's built with not using ht in mind). So, we'll have 17k for P Cores only in cb.
Now, intel says Skymont has +2% IPC over Raptor Lake (it doesn't but that's in an ideal scenario that intel will not use in ARL).
Let's assume RPL Level IPC for an optimistic approximation and let's assume the same clock speeds.
In a 14900k, the e Cores are running at 4.4GHz in a all core scenario and the P Cores will run at 5.5GHz (P Cores only).
Assuiming the e Cores will run at the same clock speed, one e Core will be the 80% the performance of one RPL P Core (4.4/5.5). So, the e Cores might add 24000/8*16*0.8 Points in a best case scenario. Therefore, we could assume an up to 38.4k points for e Cores only. Adding the 17k from the P Cores, we would have about 55k points in cb r23 which would be a 37% improvement over RPL.
I don't think that improvents will be close to this, given that Skymonts IPC will be significantly lower than RPL ipc (according to intel) but when you're trying to make estimations on theoretical performance you should keep that in mind.
Yes not all the estimates are accurate, I had 3 different scenarios with an extra one being for the 40 core version. It’s up to the viewer to decide what’s right. I also welcome any more speculation about this because I’m interested.
But I think the way your looking at is
Unnecessarily complicated. You aren’t factoring in ecores only have 1 thread compared to raptor coves 2 (hyper threading) that’s why I explicitly said that in a single threaded application they would perform the same at the same clock speed, multithreaded is a different story..
Separately calculating performance of the pcores and ecores based on their disclosed ipc from lunar lake, is the easiest and most accurate way to do this.
I don’t see the point of going backwards and trying to figure out chadmount perf from raptor cove perf. Just add the ipc to the raptor lake p core and ecore and your done… also skymount and raptor cove have the same ipc, this has been officially disclosed by Intel in the slide I was reading in the video.
One last thing. 10% ipc uplift for lion cove does seem too low to me too. But for the first conservative estimate, I went with the worst case scenario. In my mind going to 7nm to 2nm and a brand new architecture would get you something like 30% ipc but the leaks aren’t pointing to that
@@SiliconSteak well, but that’s the easiest way of calculating given that Intel does not disclose IPC improvements between RPL e Cores and ARL e Cores.
You’re right about Skymont not having HT but in a benchmark like CB it doesn’t have to much effect anyway. In most tests, ht adds about 5-10% performance in cinebench and therefore e Cores might will hit up to 35k points on their own.
No, Skymont and Raptor cove do not have the same ipc, at least not in practice.
Intels numbers were based on a theoretical design where a Skymont core has the same access to L3 cache and Ring as a Raptor cove core. Both drastically improve performance at the same clock speeds (and therefore ipc).
In ARL, e Cores will be clustered again, so each four core cluster will have to share its access to L3 Cache/Ring.
You’re 100% right. Maybe me and you could have a talk for a video about this or similar subjects sometime.
well I'm 100% AMD Fanboys who still use FX8320 For home lab！
Respect for sticking with your hardware for that long
Actually +16 IPC from what I've heard.
Would be great to see
Those top 2 SKUs were probably cancelled because there would be no realistic way to cool a monster like that, over 600 watts is insanity, they probably realized this and cancelled it. Also it may just be an experimental only SKU.
@@christopherbartleson8918 either that and/or there just isn’t enough ddr bandwidth for that much cores
@@SiliconSteak well, currently, if you think about it, Intel CPUs can't fully take advantage of DDR5 anyways, the IMC isn't good enough to allow for higher speeds using four sticks of ram. So with that limitation, it makes sense to not bother with the higher core counts, also, beyond server loads, not many programs would even utilize that high of a core count. Mostly it's just used for being able to run multiple programs at the same time.
@@christopherbartleson8918 true, I would like to see a 40 core version at least on nova lake on lga 1851 just to see what it could do
Arrow lake will have PowerVia and ribbonfet .. been hearing about how backside power delivery has been tough to crack but will create less power needed and more room for transistors for future use.. they are on track with the 5 nodes in 4 years.. will see if they reach their goal of 1 trillion transistors on a chip by 2030
Indeed we will..
@thpunish3r735 I think I saw reports saying that ARL is gonna be using N3B. @@SiliconSteak this video was well put together and I don't think I've seen anyone come up with such an analysis of ARL leaks complete with reasoning and arguments to tow. Nice job.
@@SiliconSteak Also, @9:35 the math should equal out to 70k.
@@jenxrjur right, it’s 70k or 75% more perf. I’ll see if I can update that in the RUclips editor
Intel could offer all affected customers a MSRP equivalent (I.e. $649) rebate coupon on future Intel purchases, as alternative to RMA. They could then MORTGAGE the current value of the total potential rebate claim versus the actual future claim. This would reduce the total RMA cost, amortize the loss, and keep customers locked into Intel all at the same time.
That's what I asked Intel. They told me I had to contact my reseller.
If they are same as 13900k performance but 100w less i would consider that a big win. If its the same and 15-20w less, I will stick with AMD. Waiting this one out to pick my winner.
Probably only ~10% faster in gaming. That’s my guess
I feel like we are approaching the limits of x86 advancement.
Keep that text in your copy paste list. That way, you can post it every year and some day you will be right.
LOL
Wake up call, this already happened 10 years ago. Since then the performance advancement slowed down significantly. But thing is that it is also less important then in the '90-00 era this giant where really much needed for software advancement. Now not that much anymore.
@@kristof6372 let's assume you have the expertise to make the claim that x86 instruction sets are a limitation.
What is it about the x86 instruction set that is the limitation?
Good video, 1st time here
If they don't make it right for 13th and 14th CPU owners Intel's cooked already in so many ways.
they offer 50% off if you wish to upgrade from 14 gen to Arrow Lake
need
@@jmtake85 not a optiona at all i would have to buy a new motherboard and my 128 GIG ram too usless. It's not a option at all :((
@@pc-ninja I take it, 🤩only new mob
@@jmtake85 🤣
Arrow lake will be better than this joke of zen 5
I hope so!
Intel has a weird architecture scheme. How many 8 core chips are they going to bring to market.
@@Matrix1Gamer I disagree. Typically for (80%+) of applications you only need a handful of really fast threads (pcores) but when it comes to multithreaded optimized apps you need all the cores and threads as possible.
This is a smart design. Give us a handful of blazing fast threads for the single thread apps and then a ton of smaller cores which are basically a multicore accelerator.
Smart design imo, but if you are talking about the segmentation, ya it’s weird. Ultra 9 maybe should have 10p core with ultra 7 having 8.
32 core canceled not looking good for Arrow Lake. I would skip Arrow Lake and wait for Panther Lake.
Panther lake will most likely be mobile only like meteor lake. We will prob have a arrow lake refresh, (they might put in a 40 core version there) and then we will have nova lake on lga 1851
@@SiliconSteak Thank you. I couldn't remember all the Lakes lol.
If 4 E-cores take up the same space as 1 P-core, I really wish intel would release a full E-core desktop version like instead of 8+16, I would buy a full 48 e-cores version.
I don't need 200-300fps in gaming, so I don't need those P cores, but I do need to multitask better in my system, productivity + multiple gaming in background+ recording + audio editing, watching videos. I have 128GB of ram but a lot of time my CPU (13700K) is bottlenecking me at the moment.
e-cores would have too high latency for audio editing
as an intel fan i am looking to try out AMD and see who it goes just gonna wait to the 50s come out
Let us know how it goes. I’d go with a X3D chip
I use a 35w cpu on my desktop
13600t with 50w can make 14747 pt on r23
With 65w can beat the new 9600x
:)
doesn't matter, never buying an intel product again
You will, you just won’t know it. Intel makes far more than CPU’s. Think WiFi, usb, and most likely future chips for other companies
What pisses me off is that intel changed socket again AND some cooler will have issues as the new CPU will have different Z axis.
@@TheAcadianGuy really? I thought lga 1851 will be backwards compatible with existing coolers?
Watch every impressive intel launch; they were always on a new platform; you ll not find an impressive intel cpu that was already on an existing platform; all intel cpus were beasts on their first gen on the platform; then filler cpus for the same socket until new one comes out again with impressive jumps
The lga 1851 should have the same Z height. There's no reason to break cooler compatibility since the pin count is not massively different and can be accommodated in the same area.
@@KoItai1 what ppl don’t realize is removing the socket restriction really lets the architects cook however they want
@@SiliconSteak also; cinebench is a bit tricky bcs scores in cinebench don t equal gaming performance; the memory controller might be much better than raptor's lake with way higher stable ddr5 frequency which will also help in gaming; the architecture itself is new and internal cpu lag might be lower; and again, intel doesn t have to provide extreme high end performance uplifts; look at 12400f and 11900k; 12400f is 1-2% faster than 11900k on avg; if the core 5 ultra is in +-3% of 14900k, that s a really good uplift on the cpus that matter for gamers; bcs even now and always i5s will be the cpus that provide most value in gaming; the higher the tier of the cpu the more you pay for the extra cores that won t provide much gaming performance, only productivity; and as in ur video the E cores bump especially the 1% lows quite well; now that arrowlake has way stronger E cores, the performance lift on arrowlakes might be higher; for now, unless other leaks appear, based on math i expect arrow lake to be around 20% faster in gaming than raptor lake which is really similar to how 12th gen was compared to 11th gen; also as u said in the video, the E cores will match the raptor lake P cores performance at same clockspeed; so basically every E core will be a i5-13500 P core bcs those have 4.6 ghz; the core 3 ultra will be a 4P+4E, much stronger beast than i3-13100f which has 4P cores at 4.5ghz; basically only arrow lakes E cores will be faster than the i3 which is already a really impressive budget cpu; idk why but i think that the core 3 ultra with it s price and performance will be the highlight of arrowlake; and on 2nd place the core 5 ultra that will probably be slightly faster than 13900k
I really think Intel Arrow Lake should change the game
Intel and AMD have always provided multiple options, with 1 company going into dust
we will have a lower competition and also much less CPUs to choose from
if Intel 14th gen were not cooked
AMD 9th would be a$$, its not really that great plus it does not even do good in productivity
intel does both productivity and Gaming
so if arrow lake does well, Intel is back in game and AMD is Screwed
Idk if AMD would be screwed, but they’d have to actually try again
@@SiliconSteak to do so, they'll have to find a future proof replacement for their IF. That will be their most important point to improve. All Zen architeuctures suffer from the huge if bottleneck
@@lukass8506 ^^^
@@harsha6937 Which AMD are you talking about? The 9950X is the strongest CPU on the market in terms of productivity where clock frequency is concerned, leaping past Intel's 14700K, which for productivity is better value than the 14900K from an economical stance.
"Productivity" in my case refers to those tasks wherein single-core performance matters a lot more than core count: audio production, signal processing, music production, and some aspects of videographic production and programming. When ordered, for example, strongest-to-weakest CPUs for production (still all good) would go: 9950X > 14700K/14900K > 13700K > 12900K. If Intel wants to take down the current best performer, it need do but one of two things if not both: (1) reduce their upcoming chip's price to undercut AMD's currently massively overpriced 9950X, or (2) either increase the single-core performance beyond that of what the 9950X currently offers or guarantee significantly --- and I mean *_significantly_* --- greater power/performance efficiency, sort of like AMD has done, impressively.
More performance output at lower power draw or voltage demonstrates a mastery of engineering or manufacturing, contrary to what Intel's 12900K, 13900K, 14900K, 13700K, and 14700K, and several of AMD's previous chips have demonstrated, which implies a level of infantile stagnation.
*EDIT:* _Marketing-wise,_ the price of the 9950X is enough to send everyone here to the intensive care unit in critical condition after laughing at the sheer absurd levels of pricing. It's a joke. Basically CAD$1,000 for a CPU that won't be moving as many units as its 3D counterpart --- why? The 14th Gen is also overpriced severely, and the 13700K is currently nearing decent territory. The 12900K is decently priced for what it offers. That 9950X, though...? Jesus Christ. Drugs are bad.
@@lukass8506 They already have plan to do I/O die on the bottom layer and cores on top of it. It will reduce latency.
Chadmont
That's a mini Threadripper!
Arrowlake will be better than Zen 5
Seems that way now
yeah lets just hope the voltage doesn't go to 200 volts and burns millions of cpus, yay
@@jacobgaysawyer337 same could be said of am5 cpus last year
@@SiliconSteak Nope. Motherboard vendors cranked up the voltage, not the processor itself.
BangForBuck gamer bro WOOP WOOP
I can't believe people are so concerned about their GAMES.
What is the problem ? You don't like playing games ?
This chip's about to fry your rig SO FAST you'll regret not having fire insurance after your entire block burns down due to rapid degradation in your sleep.
*Thanks, Intel.*
This is the way!
It doesn't look to good. single thread performance (game performance) looks bad, about 4% uplift. Multithreading is worse than the i7 and i9 since it lacks hyper threading. the power limit is STILL 250W with an unlimited mode on top of that. They're aiming to push it to the limits of breaking AGAIN because they raised the safe operating temp by 10 degrees. Everyone keep that same "Zen 5%" energy.
@@JoeL-xk6bo multithreaded performance might not impress due to loss of hyperthreading. I don’t see a scenario where single thread is only 4%. I think single threaded will at the minimum be 10% faster, I could see even up to 20% faster max oc. Ultimately we will see perf/efficiency gains soon, seems like they are launching in October.
I know there will be clock speed regression on the p cores, and we are going to a chiplet architecture. If these things really impact performance negatively then maybe we could only see 4% single threaded uplift. (Very unlikely imo)
You have to understand, zen 4 to zen 5 was from 5nm to 4nm, with the same 6nm I/o die. Raptor lake to arrow lake will be going from intel 7nm to intel 2nm (which for all intents and purposes are 7nm and 2nm because they are equal to tsmcs 7 & 2) this is a much bigger jump, and I would expect to see much better gains than 4%.
@@SiliconSteak It's actually Intel 10nm SuperFin (known as Intel 7) the underlying architecture is still comparable to 10nm, which is slightly inferior to TSMC 7nm in density. (2nm, 7nm and 10nm aren't actually the actual transistor measurements, rather just names given to a rough transistor size, so 2nm better than 4nm, 4nm better than 7nm, 7nm better than 10nm and so on)
this guy is not creditable
Need to beat X3D. Period.
people obsess over performance too much when typically they don't need nearly that much cpu for games, and only professionals really need it for various real world uses. The home user should concentrate on better software. Everyone should concentrate on better software.
I agree, unfortunately devs like just throwing more hardware at their problems
@@SiliconSteak I'm getting a Mac where this seems like less a problem than the windows world
@@EverSpaceTime I have a m1 air for mobile work. No regrets sips power and does what I need it to do. If I’m out and about I can even remote into my windows server that has a 12600k 64gb of ram and a a770 16gb.
Best of both worlds
@@SiliconSteak In your opinion is the software generally better on macos? I get that distinct impression having looked at it excessively.
@@EverSpaceTime I use Mac and windows very differently. On windows I am more of a power user and like to have all the options available to me. On Mac I basically just use it as a web browsing/word typing machine for paperwork.
If I need some serious grunt or compatibility I route into my windows machine.
Basically I’m saying to get both so you have the best of both worlds, if I just had Mac I wouldn’t be satisfied because they just don’t give the users enough options like windows does. The software is well optimized and most things just work, but I wouldn’t trade my windows machine for anything