True true

And Still Its True Tech Geeks Find Him Anyway 👍

I'm don't think it's underrated, it's just that the more specific the cannel are, smaller will be your target audience. Other great channel like "Actually hardcore overclock" is well known in the OC community, as it is a more specific audience, the number of subscribers will also be smaller, but they can still be very well known within the target

MARK!

You can already buy them, it's what server farms use.

I want to see a raspberry pi with this on it

The fabrication process (like 2nm) is completely architecture agnostic.

@@GaryExplains ohhh okay sir I get it now thank you for explaining

ML tasks can run on CPUs, GPUs, and NPUs. These Cortex-A CPUs might be used in processors without a GPU or an NPU. Also, depending on the nature of the work the CPU might be a logical choice for the ML to run, especially now with SVE2.

Yeah, all these ML numbers quoted by Arm are for inference, not training.

@AstroCat oh boy don’t even get me started on training ml models… I’ve been trying to get an RTX 3090 because I can’t even load the models on my RTX 2060 due to VRAM issues 😂

big.LITTLE is an example of Heterogeneous Multi Processing (HMP) so I am little unclear what distinction you are making between the two.

@@GaryExplains I mean is there a difference in configuration between Snapdragon 845 and 730G? Because one is 4*4 and the other is 2*6?

Well, that is the difference. The number of big cores verses little cores. But the mechanisms and tech in terms of HMP is the same. Also it is 4+4 and 2+6, not 4*4 etc.

@@GaryExplains so the performance of the 845 is still better than 730G despite a two-year generation gap due to the difference in number of the big cores

Well that depends on how you measure performance. In terms of pure CPU performance (no GPU related workloads) then according to Speed Test G, the 730G is faster. This is the video comparing the performance of the two: ukposts.info/have/v-deo/jJ9zm2ePp4mWmaM.html

no, X1 is just not for smartphones, can't wait to see new laptop soc's using 4 X2 cores

@@carlod1605 Why stop at 4? There are plenty of mobile octa-core chips right now. Most are 2+6 or 1+3+4, but in a laptop with active cooling the sky is the limit.

I think you should also blame samsung process node(5LPE) 50-50 this explains sd 888 over heating

2+2+4 would be too power hungry for a smartphone. It is a possibility for a laptop, I guess, but if you have the bigger battery and the better thermals than 4+4 or just 8x X2 cores is better.

Tegra X1 already get stomped by older chips (like SD 855)

Doubtful. Historically there has been a 6+ month gap between announcement and launch of actual processors using the new designs. If "Whitechapel" is for the Pixel 6 this autumn then that is too early.

@@GaryExplains yep, y are totally wright, buy reality never stopped me dreaming, that my disadvantage. But to be honest it is to early for me to update form pixel 4a, I am aiming on pixel 7 XL. Hope it will have WC 2 on cortex x2.

I just hope qualcomm doesn't include it in its next iteration, just 710 would be great, because I'm a user of OnePlus 9 and the heat it generates is just unacceptable, it makes me feel something is wrong with my mobile everytime, let the X2 be reserved for some plus variant of Snapdragon which is intended for a gaming mobile with high grade cooling system, because in our normal mobiles in no way they would manage the heat as effectively as in a gaming mobile.

@@dr.allwyndsouza5246 maybe it's not a cortex X1 fault. Qualcomm back in the day made processors that heat up so much that some companies started to use their midrange processors in their flagships

@@HappyMarshmallowGamingComedy it maybe but I'm seeing which ever processor uses it, heats up so I associated with it. It may solely not contibute. Exynos 2100 despite giving up its mongoose cores still heats up, only flagship processor which is not heating up is kirin 9000, and it doesn't have X1, but sadly we can't use it except if you are in china. X1 is a very high power core so its obvious it will generate heat but the amount I'm facing is very disappointing in my OnePlus 9.

@@dr.allwyndsouza5246 i guess the reason why exynos heats up more is because of the mali gpu, since in cpu benchmarks the exynos wins over the snapdragon. Idk , all i know is with this new arm v9 , new cortex cores and new 5nm+ all this issues should be solved.

@@HappyMarshmallowGamingComedy it would be great if this issue gets resolved with the advent of V9. Let's hope for the best. Also now qualcomm should again make chips with TSMC rather than samsung.

Maybe SnapDragon 1000 😳

800 has been on v8 from my understanding. I'm thinking these will be the 900 or 1000 series.

Your comment doesn't make sense, both Apple and Qualcomm use ARM. They choose different configuration and tweek. Both companies don't use ARM GPU which gives them an edge over Samsung and Huawei. Samsung is about to change that, hope they succeed so the user benefit from competition.

Brother arm designs the architecture other company's tweak em and use it

I think @tabaghdissar you have missed understood something. The original comment makes perfect sense. Apple uses the Arm architecture but it doesn't use Arm's designs, no Cortex-A chips in the iPhone. Arm designs it own CPU that are compatible with the Arm architecture. That is what the OP mentioned custom cores.

Not quite @Balaji Prithviraj . Arm defines the architecture (like Armv9) and it also designs it own CPU cores that use that architecture. Companies like Qualcomm, Samsung, MediaTek uses those designs directly without any tweaking. However there are other companies that design their own cores which are compatible with the architecture. That is what Apple does.

A14 Bionic is still more powerful. No contest with the A15.

Optane is costly af, there's no way it will used in smartphones and also optane only works in intel systems to my knowledge.

My thoughts exactly, it’s been the A55 for many years now while the bigger A line has been changing.

I can give a as technical/detailed answer as Gary could of coure but I'll try. One reason I've read that Apple's ARM SOC's have such high performance is they implement a large cache size, esp the size of the level 3 cache. The latest Android flagship SOC's are starting to catch up in this regard thankfully. Plus Apple and any other company that pays the liscence fee to ARM can create a ARM chip as custom as they want as long as it runs the ARM instruction set. Similar to why AMD and Intel CPU's can run the same apps because they both use the X86-64 instruction set. So there are all sorts of design decisions Apple's silicon engineers can make (and Apple are known to have some very talented silicon engineers). Obviously the combo of their decision to give their SOC's a large amount of cache plus whatever design decisions they are making for their SOC's has been working extremely well for them keeping them pretty much a full generation ahead of the fastest SOC you can get in a Android phone. I say all this as someone that hasn't owned iPhone (or any apple device) since the iPhone 4S. Not because I hate the company or hardware, I just don't like iOS. It's way too locked down for my taste. But I got no problem giving credit where credit is due, the engineers that work on making their SOC's are very very good at what they do.

Their SOC designs are nothing special. They can invest in wider cores and larger caches because their devices go to more expensive premium devices with higher mark ups. So they have larger due sizes and they’re also usually first priority on a new fabrication process. Another reason why their performance is better than android and in the x86 space is because they design these socs exclusively for a small number of devices hence it’s easier with optimization.

Nothing "happened". As far as we know Qualcomm will use the new tech it has acquired. But the question is when and in which products. I don't think Qualcomm will build a Snapdragon 400 processor with a Nuvia core.

Yes, but not Arm's designs.

M1 goes upto 28W on load so light use is around 10W.

@@kac5527 yes that's what I'm asking if a new processor has to be made on these cores, would a similar performance of M1 attract a similar power or not?

@@thekothari it also depends on the node used, but most likely the X2 wont be able to catch up in terms of perf/watt to apples high power cores used in the M1, even if fabricated on the same TSMC node. Judging from the marketed improvements over the X1.

According to Anandtech, the Pcluster used 13.8W peak load, the Ecluster used 1.3W peak load and the DRAM 1.5W. So total peak CPU load used 16.6W. However, this doesn’t include 8c GPU and 16c NPU, both of which are pretty big parts of the M1 die in their own right. The GPU at its peak used 10W so that’s there. The NPU wasn’t tested for power draw but judging by the size comparison with the GPU, I’d assume it would draw 5-7W. Now obviously, even at peak load, you won’t use all the components to fullest. The M1 when running the Game Rise of tomb raider used 16.6W total. I’d put the peak chip TDP at 25W, which is about right with what other comments say.

@@shresthsonkar9207 yeah and what do you think will be the power consumption of a chip made by Qualcomm based on new X2 having similar raw power if at all it is possible?

I guess you didn't see this video of mine: ukposts.info/have/v-deo/aZt5ZaGBmG-Jq3U.html

@@GaryExplains I do but I heard some said they just need ARMV8 and they good to go no more Licenses instruction Set

I don't quite know what you mean by "just need ARMV9 and they good to go".

@@GaryExplains sorry ARMv8

No, Apple has been in fact one of the main influencers over Armv9.

How do you know that the 8 gen 1 has heating issues?

@@GaryExplains i have seen a realtime antutu benchmark test of moto edge x30

@@GaryExplains temps increased by 12 deg C

@@GaryExplains Sd888 increases like 10 degrees

I don't think a 10 or 12 degree increase can be called over heating. When I tested the iPhone 12 the temp went up by 22 degrees. ukposts.info/have/v-deo/nZ2TmXmjpnh_uqc.html

I have no thoughts at the moment because all we have is rumor and speculation. When it is announced then we can look at it properly.

@@GaryExplains Cool.

No. If anything it would be based on the A78, but since this is an Armv9 processor then it is probably more new than just improved.

@@GaryExplains If it has some base on a78 it woudn't be based on a76? Because the a78 is based on a77 which is based on a76 .

@@Analisede_Tudo Why it would matter when we go that far back ? If we go that route, then everything is based on Intel's 4004 chip, right ?

@@Winnetou17 no. What I'am saying is that , there is some core that are made from 0 . Is not that far, for example , the Cortex a57 is made clean , not based in Cortex a15. The Cortex a76 is a design clean , not based on a75. That was my question , a question about a clean design .

@@Analisede_Tudo late replay but, we can't be sure as arm v9 cores didn't drop yet. There is no way of determining their connection unless ARM decides to tell us. But my personal prediction is that they shouldn't be based on 76 or 78 cause they are part of the whole new architecture

ARM is the way to go, laptops will no longer have serious battery issues that's the biggest benefit.

The new cores wont really force apple to up their game by much, apple is still ahead by quite a bit. Personally i have far bigger hopes on the nuvia aquisition by qualcomm.

Apple will be far ahead as usual because they are gonna benifit too by this new armv9 cores

@@manulovesjesus Apple is ahead, yeah. But I don’t want them getting complacent. And this development will hopefully help create a market for windows on arm devices. The prospect of an m1 like chip on a thin and light windows ultrabook is the stuff of my dreams.

tbh intel can probably use risc v to make laptop cpus

No Pixel 7 (maybe)

I doubt it will! 😂

@@GaryExplains Your video's are always very good, dunna put yourself down lad!

I think it's hard to get TOO excited about that kind of a video. It would be (unaviodably) pretty statement-of-facty about this happening, informational about the benefits and potential downsides and things to take care/into consideration and I guess a bit historical about how we got here and maybe how extremely long the 64bit processors' architecture will last.

@@GaryExplains I believe the A710's can still run 32bit can't they, so I would say 32-bit was 'fatally wounded' rather than dead. Let's play predictions: What year will we see the first mainstream 128-bit CPU?

Well he hyped the x1 also and that thing has been underdelivering since release.

The 801 was an overheating pos

@@RadialSeeker113 They are certainly not as great as Apple chip on iPhone 5S but combine with One M8 chassis and HTC Sense UI they do deliver a good overall performance where you can’t get on any other Android especially Samsung during that era.

Yes

Apple silicons are basically arm based. The first arm v8 based chip was the Apple A7.

@@___Amal_ they are using arm architecture, not the arm designed cores.

@@prianshubhatia2759 that's what I said based.

No. M2 still wouldn’t match needs of HEDT and server which have double or even triple digit cores. Xeons and Threadrippers run in excess of 16-64cores. And above that, support multithreading which can boost each core performance by 25% No matter how fast the single core performance of the M2 may be, it wont be any match for these HEDT/server chips. That’s also a big market. Another big market is gaming and there the issue is more about software compatibility than hardware being inferior. M1 is decent hardware and can match a 1050Ti in GPU and i7-8th gen desktop. However due to lack of native support for games, one needs to use VM for windows and Rosetta for x86-arm which leads to further reduction in performance.

They already have p550 (a75) now so they are about 4 years late

Really? Maybe you are referring to the SD8G1. What about the SD8G1+ or the Dimensity 9000? Are you also taking into account the higher clock speeds?

Well the actual data is confidential and not shared publicly. However it doesn't matter as the important thing is the relative performance not the absolute.

@@GaryExplains Wow i didn't really expect a reply, especially so quickly, thanks. But still I believe anyone would be able to manipulate such data to his advantage and really prove anything he wants and companies do that often because no one pays attention to such essential thing eh

Absolutely, but the damage such tricky would do to a brand is enormous. So while all data presented by these big companies must not be accepted without some level of critical analysis, there must also be some level of trust.

No, too early.

LOL... But in all seriousness, no, that is why Linaro exists.

@@GaryExplains uh... sir, i'm currently on 5.13-rc4 and ryzen's new sensor still troublesome 😅 Edit: imagine if it new architecture, Linux maintainer could be very busy handling new features despite with new toolchain, wow!

Hmmm... Sorry for the confusion... Linaro exists to add new Arm architectures to Linux. It was setup by Arm for that very purpose. Sad that AMD don't have the same kind of setup.

Thank god

These cores won’t appear in devices until 2022. The X1 was launched in 2020 but only came to devices in 2021. By the time Apple would have A15 for mobile and M2 for Macs. Apple would also likely have a very powerful multicore chip for MacBook Pro 16” called M1X.

Architecture (.e.g. Armv8, Armv9) changes can't bring performance boosts, except when new types of instructions are introduced like SVE2. It is the microarchitecture that defines the performance.

@@GaryExplains But when the version 8 is a decade old should version 9 bring a major changes after a decade. Also if what you said is true then y haven't they introduced new instructions in Arm v9. as per your statement it's just 30% faster and that 30% is upto not overall. So it just on par with Apple A13 in single core score now don't you think this a bit disappointing . Also will this enable the video quality of iphones in an android phone.

So Armv9 did introduce new instructions, SVE2. As for performance, as I said that is about the microarchitecture and Apple's microarchitecture is better, simple as that. But I must say that using the word "just" in front of 30% is a little harsh. I can't think of a single part of my life where I have managed a 30% improvement in anything over the period of ne year. Maybe you are different.

@@GaryExplains no no you got me wrong when I said 30% I was comparing with apple. So is it that Arm has an microarchitecture which can rival apple but they have not released bcs of the cost

Well I don't want to quibble about semantics but your wrote, "as per your statement it's just 30% faster"... just 30%... As for Arm not rivalling Apple, it is more complex than that. Cost is part of it, that is why the Cortex-X2 exists, but Apple has a 1 to 2 year lead in CPU design, so Arm is always playing catch up.

Second lel

Third lul

If you watch carefully, Arm didn't say the X2 will be 30% faster than the X1. It said that a flagship phone in 2022 should be 30% faster overall compared to a flagship phone in 2021. This is including the X2 but also the A710 and also an expect tweak in clock frequencies. So... your comment is fake.

@@GaryExplains thank you :)

What has Samsung and AMD got to do with it? These are the CPU cores that Samsung will be using.

While I absolutely like to look at numbers, points and seconds and get a really nice warm feeling when they get better, I really don't understand the obsession with performance in nowadays smartphones. There are less and less applications/usages that REALLY need performance improvements (I'd even argue that the number is 0 for non-enthusiasts) Also if you were to have a flagship apple phone vs a flagship phone using, say, Qualcomm's CPU and actually use them, not run benchmarks, do you even notice the difference ? If anything, I think that people might benefit more from a flagship configuration (1 X core + 3bigA cores + 4littleA cores) all capped in frequency to be at that point where the power/performance slope changes. The performance loss won't be noticed, I think, but the increase in battery life would.