To be honest, the pure performance of videos should not be a huge factor for whether you cover content. A good source should cover all angles of the market in an unbiased way. This is what makes you a reliable news source. Of course I don't expect you to make huge videos if there isn't anything huge, but at least covering the other side I believe is important. Good video tho
There is always a "but AMD will beat Intel" and a "pay special attention to the one part which is good news for AMD". You are getting more and more desperate in every video. Just sell your AMD stocks already!
You should make a slide that explains what all the codenames mean, for those who haven't memorized what all the "Lakes" and "Rapids" are supposed to represent.
Lake is client CPUs Falls is “HEDT” and workstation Rapids is P core server Forest is E core server Bridges is for server GPUs, but next gen Rialto has been canned in favor of Falcon Shores Shores is their server “APUs”, but for now it is for server GPUs for the next gen
Please make time lines for Intel, cause their code names are so confusing. At lease zen is named 1, 2 etc, but all these lakes and coves are so fucking confusing. Not for you because you work with it everyday, but for us who only follow this once in a while. Also please mention what generation the different lakes will likely be in, that way we can better relatively see how far it is
Totally agree. The complexity of the Intel offer is going to reduce their take-up by the public. Not the trade, they are used to this sort of thing, but the public will understand the zen5 offer as a straightforward 'next generation' one, whereas this Intel 'two competing generations on different architectures' is going to deter by its complexity.
@@simonbrown8509 That is why I gave up following Intel as their naming scheme just got too damn confusing, first with all the bridges, now all the lakes, which is newer, which is older? Who the hell knows. With AMD its Zen 1, 1+ which was just OCed Zen 1, 2 and 2+ ditto, Zen 3, Zen 4, and now Zen 5. Its simple, easy to tell what is newer vs what is older, Intel really needs to take a page from the AMD playbook and just call them Intel A,B,C or Intel 1,2,3 and drop the codenames. and the core vs Core ultra just makes it worse as there are some that will have a core and a core Ultra variant.
@@TheHangarHobbitSame issue for Xbox, no way to know what is newer, or just a variant. Sony straightforward like AMD. Also sockets are in order. Naming sockets by pin counts makes sense on a technical level, but for consumers it makes no sense. When Google named OS versions by sweets, they at least did them alphabetically.
@@asm_nop Density functional theory in nuclear physics is most bottlenecked by memory bandwidth not by compute power which is sort of a fringe domain of nuclear physics but in material science density functional theory is main stream so the majority of material science depends greatly on what AMD comes up each new generation because extra L3 cache is short of a revolution for compute times in that domain
@@lukabozic5 Very interesting! I suppose it depends on the amount of resources you have for your projects, and how multithreaded the workload is, but do you use the consumer level CPUs like the 7800X3D, or are you able to benefit from server parts like the EPYC 7373X, which have insanely high memory BW and L3 per core?
@@asm_nop Imagine a nucleus with protons and neutrons, the bigger the nucleus the more protons and neutrons hence more information a PC has to juggle since it needs to take into account all the intra nuclear interactions. In theory a 3d Threadripper would be ideal but since those do not exist I need to do with higher end consumer CPUs since server CPUs are insane price wise. There is a significant untapped market by AMD of experts who could use such a machine at home. Obviously at the institution people use HPC but that is beyond the scope of our conversation
I just cannot get excited over new lakes 4000 - Haswell 5000 - Nothing Lake 6000 - DDR4 lake 7000 - Same Lake 8000 - Mirror lake 9000 - Twin Lake 10000 - Stillsame Lake 11000 - Ohshit Lake 12000 - Mejump Lake 13000 - Jumphigher Lake 14000 - DejaVu Lake 15000 - OneArmed Lake ... For some reason most of these lakes just look exactly like other lakes I can't explain why ...
I predict ARL may be slightly faster in some instances, at a crazy power level, and will continue to have poor power scaling. And the X3D variant will dominate at half the power or less for gaming.
Intel just severely needs an X3D competitive technology, that's really the main thing holding them back. They can boost to the moon and their chips can sustain ungodly levels of tdp, and their e-cores are actually great for low-power tasks. But that v-cache technology has truly created a crater in performance gap relative to efficiency when you start pushing the chips. I really don't see how Intel can keep brute-forcing their mid-high end chips when we're already hitting the thermal limitations and power consumption levels where people don't just have to spend extra for the chip itself, but also for top-end cooling, cases that can fit those cooling solutions, and a psu that can sustain not just a high-power gpu, but also a cpu that may very well spike to similar heights.
Efficiency. For the next gen, they need +20% IPC jump at minimum AND cut the power consumption by half at the same time. And then we can talk about x3D and other shenanigans. In order to increase efficiency the first thing they need is better process tech, in other words, smaller feature size. It seems they are going to use TSMC manufacturing for this purpose. Not sure at what capacity, could be just for mobile chips or entire generation, we'll see.
have u seen the power consumption? there is no need for any cuts, it is already very low compared to intel products. Products speak for themselves. @@mmo0J
even ryzen non x3d cpus tend to have more cache than their intel counterparts. compare a 7700x to a 12700k. hey even a 7600x has more than a 12900k. intel is far behind in terms of cache. sometimes i just wish intel would add more cache to their cpus instead of brute forcing performance by going to 100 degrees or so
@@mmo0J tsmc 3nm would be good. even better if they add more cache. id go intel if they make a high core count e core only cpu with 2mb L3 cache per core. might never happen though....
This 40TOPS win12 requirement is something that makes me NOT to want to buy anything. Need to wait and see if that's true and sure want to make sure that my device will meet the requirement.
@@defeqel6537 It can not be mandatory for system integrators for at least 18 months as noted in the video - no high volume Intel based systems would be able to ship with Win12 installed and no low end systems from either AMD or Intel would be possible.
Honestly I think it will be years until those neural processors do anything worthwhile yet, and you'll probably be interested in upgrading at that point anyway. And the NPUs in the products available at that point will be WAY better. I have a feeling 40 tops is gonna be like RTX 2000 series RT. Does it, but the good stuff came later anyway, so if you held onto your 1070 until you got a 3070 you didn't really miss out. Plus, both Intel and AMD platforms should be forward compatible, so just upgrade your CPU in 4 years after buying what you want right now.
It's one thing to have the operating system mindlessly scrape some generic telemetry data, and another to have autonomous, adaptive system going through your stuff that is capable of adjusting itself to better match the user's behavior. There's no way i am installing any AIs on my system i can't access the source code of. Especially one that can access the internet or operates at root level. Maybe pulling my hair out for a month to get Pulse Audio working isn't so bad after all.
Dude I have a 13900k and a 13700k I can have both systems running pulling 1700watts. And it’s still not getting hot in my room here in CA… try running SLI or quad SLI.
Europeans in their insulated houses during summertime. I'm currently stillusing an i7 9700K for my gaming system, but i may upgrade it to an Ryzen 7700 with 65W TDP. It's the most efficient currently and since i'm not a fan of water cooling, the 7700 will also be much more noise compliant on a single tower 120mm cooler.
@@tringuyen7519 I wonder about this sometimes. My one 275W GPU turned my old room into a sauna when rendering or gaming. Many attest to these new hotter CPUs doing similar, bit then there are those who claim the opposite. Like they must have the AC set to industrial freezer mode or something, becasue even with temps below freezing outside and the window cracked, vents blocked, it's summer when the GPU's doing heavy lifting.
I'm still amazed that Pat is asking other companies to go and fab their chips on Intel's processes while at the same time they are going to TSMC in order to be competitive.
Think about it. Indirectly, they are taking capacity away from TSMC, right as they are pushing their fabs for external use. It could be both a safety measure AND a marketing move for their fabs to costumers
@@benc3825 I get it from a strategic point of view. What amazes me is that they are going with a straight face to try to convince their potential clients that they should go with them instead of the competition.
Opening fabs is a great solution if they are not able to compete with the most cutting-edge technology from TSMC. All chips dont need the best manufacturing node.
If you look at the history of TSMC, a big part of their success was that they are a fab only company. Chip designers came to them because they didn't have to worry about their ip being stolen, a huge problem in the tech industry. I just don't see Intel fabs being popular from outside companies unless they give up design.
As already shown by the A17 from Apple, N3B effectively has 0 advantage over N4. And TSMC themselves always said N3B would be a transitory higher early yielding node before the rest of the N3 line. Going N3B is a complete waste of time, you either go N4 or N3X/N3E. Don't expect Lunar Lake to be anything impressive.
I can't believe that there are haters of Tom claiming he is full of shit. I say, if you don't have haters, you're doing something wrong. IMO, Tom is the most thorough of anyone claiming to have this type of insider info.
@christophermullins7163 None of this stuff exists until they announce it on stage and magic wand it into existence 🪄 He should have looked into the proper future. Not the next reality over where that ONE detail was mildly off. If so, FRAUD. MADE UP FlUfF!
@@gozadrotit7426 is lying and keeps spamming this in the comments, they don't know anything or are really misinformed. EU and US price is basically the same.
@@N4CR After your Comment I decided to look for prices again. Yeah you're right I'm totally liar and I spammed it in a few comments. Sorry for that. I'll go delete other comments then.
The big feature for Lunar Lake will be the on-package memory ... reported to be Samsung LPDDR5X, 8.5 Gbps. Meteor Lake's best low power feature is video playback with CPU and GPU tiles disabled. Intel's research cranked out the Thunderbolt 5 and WIFI 7 support included on their Raptor Lake refresh chips.
2024 seems exciting for CPU buyers. I'm currently using a 9900K and will most likely be upgrading this year or early next year around the 5090's launch.
My machine is a 9900K as well; I agree with you that the timing will feel right for an upgrade with Arrow Lake and NVIDIA’s 5000-series. My friend and I refer to the i9 9900K as the “Herman Cain” CPU because of Cain’s 9-9-9 taxation plan and feeling like we paid “the Intel tax” when we built our machines.
Honestly, I think the issue with your intel videos is that almost every title has some variation of "AMD in trouble", but throughout the video you list reasons why Intel's next gen is having issues. If you were more honest in the title, the videos would do much better. For this vid, something like "Intel's Arrow and Lunar Lake punches: Enough to threaten Zen 5?" Would be more accurate, but still vague enough to make people want to watch it
AMD has couple of advantages that most people miss. Power efficiency gives noise reduction for people who actually use PC:s for something that needs concentration. Second advantage that most people miss is AVX-512 support. It's not about width, it's all about handling problems that break vectorization.
Yes. And the power efficiency (and reduced waste heat generation) are also relevant for those concerned with energy bills and/or living in warmer areas.
Intel is working on a new instruction set that will be an evolution of AVX512, called AVX10. It’s unclear if ARL or LRL will include it, or if, in the worst case, it will be made exclusive to their Xeon lineup.
@@GeekNerdNoir Intel''s main problem with AVX-512 is that their little cores do not support AVX-512 and for scheduler they needed to harmonize the instruction set between big cores and little cores. So their consumer lineup will only support the instruction set of their little cores. But if there is not consumer software that will use the AVX-512 then adding support for it would be too costly for the little cores.
The last time Intel did a presentation about something of that regard, they essentially advertised AMD for free since they couldn't stop talking about it. Performance-wise, Intel still has to solve the heat issue before they add more performance. It's clear they don't have the same hand into making asymmetric core CPUs the same way ARM does.
You should though. Regardless of what I think, the vast majority of people aren't willing to pay for ad-free online services and they definitely don't want to go back to the days of paying substantial amounts of money for the next version of Windows. Just like ADs subsidized product costs over the past decade, edge-client NPU processing ability is definitely going to be the biggest business *and* government target for the rest of the 2020's (especially now with the near complete proliferation and public acceptance of always-online sensors). Since pre-processing done for free (i.e. without using power at the datacenter) on the client side can result in more than linear savings (such as only sending a compressed Speach-to-Text transcript of conversations and raw data when flagged for re-evaluation), it also makes this one of the only practical ways for non-state actors to classify billions of real-time physical signals, not to mention handling all of the video streams.
@@mortlet5180 Fuck me.. how's THAT for a response? I didn't care that much about NPU either, but now I do. Even thought I didn't understand much of what you wrote lol
@@MrHav1k How about being able to solve *ALL* cybercrime and convict almost all other criminals? Just think of the children! I mean, everyone has already decided that privacy isn't important. Actually, it's a marker of either bad people with something to hide, or of the patriarchy that want people to be ashamed of showing their naked bodies.
@@mortlet5180 Yeah but for some basic voice recognition and stuff I think it will have more than enough power. Maybe for some more advanced fancy features you need a more powerful NPU but ... I don't see what I would personally use a powerful NPU for right now. Also I think companies are not doing edge AI not because the machines don't have enough power to do so but because it's just easier for them to just upload the data and have it done on their servers. Having all this valuable data on your server is of course just a side effect and we all know that not one of the big corps would ever use this data for their own benefit. All in all ... brind me CPU power and screw the NPU power. There's just no use case for me now.
Big/Little is useful for gaming if you have APO. But, stupidly, it took until 14th gen for Intel to release APO after 2 generations of having their efficiency cores do nothing useful for gaming, and only 14 games (as of the most recent list I was able to find) actually support APO, so it's too little, too late.
Did enjoy, thanks for asking my take. Quite information dense video, thanks. You mentioned windows 12, I don't have any information on 12, if you're referencing the erroneous uniformed content after hearing Qualcomm say "the next version of windows": That doesn't mean 12 at all. MS has a new build released regularly, which is probably what Qualcomm intended. Again, thanks for this video, quite enjoyable
I think around 20-30% but unsure on the power usage and demands? Of course the new lakes will need the new socket size motherboard for all future upgrades going forward now
@@Lone_GamerUK Problem is, that he kept comparing Arrow Lake to Meteor Lake, that is why I'm not sure about those 20-30%, since we don't even have Meteor Lake for desktop. And if I recall correctly, there was a leak not too long ago saying that the next generation i9 desktop chip will perform the same as the 14900k in gaming 🤔 That is why I'm confused here.....
Great video, I love that MLID went with the "ughhh I have to do this, but I won't be fun about it" just to get the good juicy info out in a short format. It's a responsibility at this point, for the torchbearer. So, thank you brudda.
Intel needs to commit themselves to support 3 different architectures (not mere generations) with each socket they release. If they said that LGA 1851 will support at least 3 different architectures it would become very interesting to me compared to getting a Zen 5 system. The 1 - 2 years of true or fake generations on the same socket does not cut it anymore whether you are either the underdog or the market leader after what AM4 has achieved. And Intel management needs to stick that to their head and require a socket change only if a new architecture would either be limited by the pin count of the socket or by the standard they are able to support without changing socket.
Agree, Intel needs to pull an AM4, I don't mind paying slightly higher price for the motherboard to be overbuilt and buy 15 gen if 5 years down the line I can buy a 19 gen CPU and drop it in after a bios update.
I have been seeing twice as many Meteor Lake laptops drop onto the market this year with over 10 hours of tested battery life (by Notebookchek) than Ryzen 7000/8000 SKUs so far. At least buying a Windows ultrabook this year with Apple M1 levels of efficiency will be standard instead of something you have to check tons of reviews to come up with a short list of models to consider before buying.
@@ofon2000 Intel is opening their fabs for other manufacturers to use. For example right now they are actually in a discussion with AMD so AMD might be using intel nodes for some of their products.
The big issue Intel has always had has been platform cost, with Intel they seem to change sockets nearly every other year whereas AMD is still releasing new AM4 chips in 2024. With my current build I went from a Ryzen 3 1200 to a 3600 to a 5700x with nothing more than a CPU change, with Intel I would have had to replace my board...what 3 times at least? That is just crazy.
Intel cannot miss this launch window that they've set themselves, hopefully they don't. Jam packed with info as usual, keep up the amazing work Tom and co.
Serious Question if AMD is on TSMC N4 and Intel is on TSMC N3 , how does AMD then have the first mover advantage since TSMC N3 is better than TSMC N4? Is not N3 the smaller and better node from TSMC? Should not Intel then have the first mover advantage since they are on the smaller N3 node? Also in terms of efficiency?
I bet, after so many zen killers in a row, amd is in such a poor state it continues capturing the high margins market. If amd wasn't capacity constrained, intel would have gone bankrupt by now. They absolutely can't survive on desktop margins. Patty has this "my precious" look, like half-way through the transformation.
It makes sense to target more tops with the mobile variant as it's more likely to be employed without a GPU, where Arrow Lake is far more likely to be paired with a dGPU. The fight over who has the most tops in CPU seems irrelevant for desktops when even current GPUs have vastly more AI compute, and that's just off tensor cores that are usually idle anyways. It's not like your GPU is going to ramp up to 100% power draw to churn out a 30-40 tops workload. Beyond all that, I could care less about AI or iGPU performance in my CPU as we still have no app that can even use it. This just sounds like a distraction to make you not look at raw performance that actually matters in a CPU. Even if I'm buying a laptop for gaming or some future AI app I'm just going to make sure it has a competent GPU.
Intel is completely pointless for DIY buyers anyway until they fix their stupid problem with their need for new motherboard every 2 years. And I dont think intel is interested in solving this problem because theyre doing it intentionally to boost motherboard sales.
@@damara2268most people go about 5 years between upgrades even am4 was ending by that point anyways. Most people it doesn’t matter there is only 1 and cpu I would consider over my 12900k is 7800x3d is really the only competitor imo. The 7950x3d kinda sucks. 7800x3d doesn’t have enough cores. Only way I would consider changing to amd is if the 7950x3d had both CCDs with vcache. I don’t have gpu assisted rendering so extra cores is great. 12900k is $500 CAD or less 7950X is $700 13900k is $750 IMO Intel is still the better buy if you do any sort of work and not just play games Okay why do my comments keep disappearing
@@damara2268As StrixWar said, I'm upgrading after 6 years of using my i7 9700k and RTX 2080 and 2024 seems like the best year for that as we're getting new GPUs AND CPUs, not only that, it's also a new platform for both AMD and Intel, meaning I'll have upgradability with both brands (thanks to that new Intel platform, seeing how the LGA 1700 did), all I have to do is watch Gamer's Nexus and Hardware Unboxed to know if I jump ship or stay hyperthreadingless, which didn't seem to affect my day to day tasks much with my 9700k's 8 cores and 8 threads
5 months ago: Is Zen 5 doomed? 4 months ago: Intel i9-14900K Analysis: This JOKE is Zen 5's Competition 3 months ago: AMD just won 2024 - Zen 5 Strix, Hawk Point, MI300X Analysis Now: Can AMD Zen 5 Survive Q4? Stop the ridiculous flip-flopping clickbait.
Intel died for me. I don't even consider their products as a possible option because I'm a musician / producer. The 13900K had disastrous latency problems making it impossible to use low buffer sizes / low latency in your DAW. On top of that, it gets worse. Intel's Big.Little has been a nightmare for DAWs trying to balance the DSP graph, because if you have a bunch of heavy chains that use similar CPU time, there is no way to balance them across cores, because some of the cores can't finish in time. So the DAWs keep moving the chains around every few seconds to try to accommodate the Big.Little, and that's been causing unreliable performance with sudden dropouts when you don't expect it. I'll never buy another Intel CPU, I'm out for the next decade at least.
Disable hyperthreading and the issue goes. Here's my worst case dpc and isr latency on a 13600k after two minutes of testing. DPC worst case was 67.8us, ISR worst case was 1.74. I think at stock, techyes city was getting like 400us of dpc latency or smth with a 13900k, Either way not exactly a small improvement from doing what i just suggested.
I feel your pain. However, the software stack needs to find the way to utilize heterogeneous CPU cores properly eventually. And yes, I'm a fairly low-level programmer (C++) and no, I don't know how to achieve this yet ;-)
Excited to see what their evolution of X86 looks like but the part where you implied that they're throwing money at their launch problems is worrying. I just had to remind myself that after Arrowlake is Arrowlake Refresh... maybe with a different NPU die?
I don't see how there could be an Arrow Lake refresh between Arrow Lake in late 2024 and Panther Lake in late 2025, would they really ship a refresh 6 months after Arrow Lake, where Panther Lake is set to arrive 6 months later ? If there is to be a refresh, it has to be Panther Lake's imo
"25-30%" more performance per physical core on ARL vs MTL (which had performance regressions vs RPL outside of specific power usage bands on top of IPC regressions) being a victory over zen5 doesn't really add up (literally), unless the leak was referring to those specific power bands in laptops where MTL actually is kinda ok? It also doesn't bode well for desktop if those are the most optimistic number & comparison too, I would have hoped to see a RPL comparison, not an MTL comparison.. It feels like when Intel was comparing MTL to ADL instead of RPL in official marketing materials.
MTL is actually like 10-20% behind 14th gen so it might happen that arrow lake will only be up to 10% faster than 14th gen which would be a huge disappointment for intel considering the absolutely enormous difference in manufacturing cost and basically a jump over 3 generations of nodes. Personally I think this will be the reality though.. also some games will definitely have problems with FPS drops because they will get confused that there is no multithreading and start pushing load on e-waste cores which have skylake-level performance.
I re-watched an MTL performance leak Broken Silicon section and the Intel sources were claiming similar big gains last summer. Tom has a problem reporting anything reliable if Intel itself is short on realism.
The hard thing to weigh up with the ARL performance claims is the lack of power envelope info. It's a bit meaningless to be "25-35% > MTL" when MTL couldn't match RPL above approx 100W. But MTL was supposed to thrash Phoenix and didn't. It seems Zen5 will be established for the main '24 buying season with ARL struggling to make launch by holidays. Currently it's a bit ridiculous seeing Amazon top sellers dominated by Zen3 except for #1 7800x3D (#2 7600), but most people are happy with CPU performance. The cost to make factor likely affects margins more than retail price, at least in gaming it's the 6c/8c speed and cache not high thread count which matters. Low priority background stuff runs fine as virtual threads. Totally understand why Tom's presented the quotes with little analysis but it's hard to guage the true strength of ARL without power data. Overall AMD seems to have jam today and aren't rushing Zen5 while Intel hopes for 2025.
@@damara2268 well ARL needs to be better than that while reducing power usage!! Zen5 IPC for AVX512 should boost certain applications but probably explains Tom's sober IPC estimate compared to the wilder claims. I think AMD chose 4nm for time to market and cost reasons, balancing out performance & power efficiency gains. They don't have to win vs ARL on benchmark performance just be close enough. I am wondering if the MTL & Sapphire Rapids fails, lead to greater Hawk Point production and a later Zen5 Ryzen launch which may have an extra stepping polishing it. That would explain the Q3 Ryzen & EPYC launch close together. Overall Zen4 was received poorly due to costs but as a platform introduction it appears the AM5 backed by AM4 for the memory transition was good strategy. Intel have been forced to make risky plays, P+E papered over cracks but screwed the ISA strategy (adding AVX512). Later after Intel pipe clean all the software issues, AMD just add compact low frequency cores to their monolithic laptop dies.
@@damara2268 I wrote a longish reply discussing that, don't see it show up. Suffice to say ARL MUST top Zen5 without RPL power hunger or Intel are in trouble.
@@RobBCactive That happens to me way to often. I have started to make sure I always make copy my reply before I post them now. And then check back to see if it actually gets posted if I really want to make sure it gets posted. RUclips auto censors algo get snagged in a lot of words, that really should not have any business being censored nowadays.
Maybe someone can correct me if I'm wrong, but after hearing all about AMDs power savings, I came across a number of articles and test that show in standard every day use of your computer, Intel is actually more power efficient. Intel is even more efficient at adobe work/cpu heavy workloads. In gaming amd is better, especially the 7800x3d. The max power draw under non-realistic cases the AMD is better, but in reality you never see it. This is comparing 12600k-13900k vs the AMD 7600-7950.
I'm mostly here to learn about the newest mobile type chips for use in miniPCs, both intel and AMD. Mostly want to see how Strix, Strix Halo, arrow lake are going, and also curious how much AI in chips will matter in the future. (I still think a lot will be done in the cloud and not locally)
I think the biggest issue for intel is highlighted correctly by you. I do not think that Arrow Lake will beat Zen 5 X3D. I anticipate that Zen 5 being 15%+ faster than Zen 4 essentially means that X3D will bring a boost over that 15%+ so realistically Arrow Lake will have no performance advantage. To me this means that all people should just buy AMD. I was on a 13900k and switched to a 7800X3D and boy do I love that decisions, 70w of power during gaming, same or better fps than 13900k, way cheaper, like there is no reason not to choose AMD at this point.
The best thing is we will be able to slap a Zen 6 chip in the B650 also. I just upgraded to 7800x3d because the price for that chip became quite affordable. Im skipping Zen5 but hope it does well as that will bode great things for my Zen 6 upgrade.
RGT, last I checked in, spouts so many different things over time that he is statistically likely to be right on some of them. But sentences like "Zen 3 could be up to 45% faster in single threaded than Zen 2" and "RDNA3 might be 2.5X faster than RDNA2" does smear smelly poo on those few instances where he got something right.
@@MoonshineOctopusAh, yes. Sorry. But there are like 10 other commenters who are actively defending RGT elsewhere in the comment stack... No, I shit you not.
And besides the obviously meaningful price difference between 3 nm and 4 nm wafers, AMD also likely has a decent discount on any TSMC nodes for being a major and long lasting partner. So the cost gap between the allegedly $15k 4 nm and $20k 3 nm likely widened further to maybe ~$12k for the "dependable customer" AMD's 4 nm, while rises to ~$22k for the "disloyal and even market opponent" Intel's 3 nm.
Oh, I have one of those flat and wide power banks. They are portable in the bag, but take up a ton of space on desk. I wished I could use it as a mousepad or touchpad.
Intel going from Raptor Lake being able to heat your home in the winter to at least being in the ballpark of AMD in terms of power consumption for a given unit of performance means Intel is getting even more competitive against AMD. So the whole "Intel is doomed" narrative really doesn't hold water. Now is it so much better that it's worth buying over AMD, dumping your older Zen system, etc? Time will tell, but it's at least in the ballpark.
I’m not sure how the npu in the cpu tile can be used if the memory bandwidth is really low. I guess you can run 7b parameter models but it’s going to be slow. What other use cases do they have that don’t require as much bandwidth? 🧐
If Intel don't get a clean win on TSMC N3 against AMD on TSMC N4 even the most adamant Intel fan will have to admit that AMD has beaten Intel at their very core strength: raw CPU design. A lot can be swept under the rug with "Intel lost the manufacturing game" that excuses the absurd power consumption, but with a node advantage on the same manufacturer nothing can really be said that excuses Intel if they lose. Not that any of it really concerns me, as my 7950X3D will do everything I need for quite some time to come. The only thing that could make me consider Zen 5 is if AMD does an 8 Zen 5 V-Cache + 16 Zen 5c desktop part with a default TDP of 150W or less.
Does the Dual Channel interface of Standard Desktop platforms limit the total CPU Power at high core counts. When I think at Apples M CPUs, it’s seems possible to me - but that‘s ARM….Quad Channel as Standard would be nice…As well as an Arrowlake with 16P cores 😅 at Mainstream
Former MLID community member here. I’m hoping this channel will move to more efficiently present leak information in dedicated videos. Leak process information and preamble is not interesting to me. I was bored until 5:00.
Waiting for LGA1851 and Intel AL, I already have sufficient cooling for my 9900ks, I've been waiting for Intel to finally release a 5nm or less Desktop CPU and it looks like it's not too far away. I honestly hope both companies release great products, it only helps us consumers regardless of our preferences.
A lot of people, such as myself, are going to be upgrading off of AM4, so we'll need a new platform either way. Cost isn't something that concerns me too much, Intel generally sticks to around the same prices, even considering inflation. A 14700k doesn't cost much more than something like an 8700k did. If anything, I'd expect AMD to have lower prices this time than Intel having higher prices. I also hope that boards aren't as expensive as AM5, board cost is what really killed AM5's launch and is why a lot of us didn't even consider Zen 4. Power draw will be interesting. The jump from Intel 7 to 20A is huge, so I'd expect that they won't need to push 300W through the Ultra 9 to compete with the X3D chips, but I don't know what power draw will be considered acceptable. Hopefully the stock TDP will be around the 140W that my 3900x uses under load. My guess will be that they'll try to target the power draw of Zen 4 R9s with the Ultra 9.
They in theory can this time. They aren’t changing mode families. A 7nm family node stacking on 5nm family node is ready. There will still be a slight delay, but it shouldn’t been much.
If arrow lake is much more power efficient than raptor lake while being better than the 7800x3d it will sell. Intel needs to tame power consumption and enhance platform longevity.
Maybe Microsoft has paid a premium in order to get 18A fab usage first for their own NPU. That would be a reason for Intel to move LNL to tsmc. Another source said the top sku of LNL is still on 18A.
According to RedGamingTech a Battlemage Xe core will only have 8EUs vs 16EU on Alchemist, so going from 4 Xe Cores on Meteor Lake-U to 8 Xe cores on Lunar Lake shouldn't be that big of an uplift. Interestingly Tiger Lake and Alder Lake had the same iGPU on the small "U" series die and the "H" series die on laptops but Meteor Lake had cut it in half for the "U" series and Lunar Lake will follow MTL-U.
He also said that RDNA 3 was going to be 2.5 to 3.5 times faster than RDNA 2 depending on the time you checked. He also claimed there would be an Alchemist refresh in mid to late 2023, 30%+ IPC for Zen 5, Battlemage Q1 or Q2 of this year, etc. Anything he claims I wouldn’t trust
@@MooresLawIsDead I'm pretty sure he's the guy you are talking about when you say things like "Some people are claiming these huge increases for Zen 5, and it's going to be decent but it's not going to be a 40% uplift. That's bullshit." RGT has been wrong more often than the early days of WCCFtech...
I was subscribed to his channel for a couple of months. Besides the daily videos that would contradict each other on a daily basis, he put out a video titled something like "The Ps6 will be revolutionary". The big reveal........one of his "sources" told him.......the Ps6 will be revolutionary. The man is playing the RUclips game, daily content at around 15 minutes does well in the algorithm. However, there simply isn't enough info in this space to support daily uploads at 15 minutes. There is a popular word to describe youtubers like this, and it starts with a "g".
A lot of people and Tom himself don't seem to understand that RL easily beats non-3D Zen4 in raw performance, and that a Max OC 13900K is easily 15-20% ahead of Max OC 7900X in gaming, base Zen 5 first needs to catch up with 13th gen lol. And if AL turns out to be such an increase, non 3D Zen 5 is going to have a hard time. X3D might fix it though in gaming. I also remember Tom saying that 9900K is stronger than 2700x by 20%, it just got me rolling on the floor. It annihilates in avgs and lows even 3700X and is basically = 5700X released years after 9th gen. Sometimes I'm wondering if that dude even knows basic stuff, he just gives wrong numbers all the time.
I probably missed this but are we gonna get an arrowlake refresh after arrowlake or a new architecture? if only a refresh is what's coming, then RIP intel.
am i the only one who cannot keep track of all the lakes, plz tom, give us a chart showing if they are mobile/Desktop/server and what gen core they belong too, i kinda gave up on intel-codenames and (new/changed numberings).
I've rewatched your videos of last year almost "claiming" that Raptor Lake Refresh should get at top a 20% difference gain over Raptor Lake, and at the end we got just overclocked and golden samples of Raptor Lake. With Meteor Lake unable to surpass Raptor Lake performance, i wouldn't bet on Intel stockings this year nor the next on the markets. If we also loose the hability of Hyperthreading, we're now back to the 9th gen era were these chips will be useless at all for smart dedicating resources for VMs, for example. I bet they won't remove this feature for the server counter parts until Rentable Units show that can replace it at all for virtualization purposes (as Data Centers manage their hardware resources efficiently, and effectively).
I think the combo of Zen 5 with Blackwell in 2025 will make laptops a really good option for gaming on the go. I hope RTX 5060 has at least 12GB Vram on mobile
Indeed, the fact I can swap my 7800X3D from my huge gaming box into an ITX system when I upgrade it, is a HUGE deal. Plus like you said, if anyone actually uses the new Windows AI functions, having more TOPS could be a massive selling point.
I'm positive that ARL will at least be competitive. Wouldn't be surprised if 20A will disappoint us as did its predecessors, but Intel is in position to order N3 capacity to conceal whatever fiasco the 20A might turn into. MTL has provided Intel a platform to stop being dragged by their own fab.
Need a 3970X replacement in around zen6 era so I am Hoping intel has Something competetive to threadripper pro in the 5000€ range. Closer is a Laptop right now. (Still stiting on a 2017 MSI one with 32GB, 1060M 6G and a i7 7700HQ Want to see a 15W CPU with Maximum battery life while offering enough RAM to open my stuff on the go without it crashing. dGPU not needed anymore. So lunar Lake may is the thing..
I want to know what Intel is going to do to compete with the 8800X3D the 7800X3D is the perfect chip for a mini ITX build it doesn't need a lot of cooling is power efficient and beats intel chips that cost double the amount.
The simple truth that Intel has to source all of ARL from TSMC in order to even stay competitive with AMD should raise all the red flags there are. Now I can already hear the screams that AMD had a manufacturing advantage for years that isn't their doing - well, maybe. But if this time Intel doesn't manage to decisively outperform Zen5 even with a better node by the same manufacturer, that'll probably look even worse. And that's not even taking into account the financial side of things.
In the last podcast episode you asked/sad something about some kind of “all in one” software for gpus that includes dlss, frame gen, anti lag etc all of that nice stuff into one software that would change settings live. I thought you mean something like hyper-rx from AMD? That’s what I thought😂 This new “Hyper-AI”Software would read your pc stats and bring you the best possible performance? I would love to see AMD working on that and something that would eliminate 1% fps drops. AMD is building up. Mini pc, laptop, handheld, desktop cpu, gpu, AI. AMD is everywhere. You have to give it to them, they are doing a good job, most of the time. Where is anti lag+ btw? 😂
![Felix "Unfair" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/Oswujxm2Ag0/mqdefault.jpg)
[SPON: Use “30BLADE65W” to save 30% on the 65w Baseus Blade 2: amzn.to/3V0uoO6 ]
#Baseus #BaseusPowerBank #BaseusBlade
Its a shame i really wanted that pb for personal and office useage. Its currently not available 😢 maybe it will get back in stock soon
To be honest, the pure performance of videos should not be a huge factor for whether you cover content.
A good source should cover all angles of the market in an unbiased way. This is what makes you a reliable news source.
Of course I don't expect you to make huge videos if there isn't anything huge, but at least covering the other side I believe is important.
Good video tho
"Currently unavailable.
We don't know when or if this item will be back in stock."
Guess stock is coming soon™
Congrats on the baseus sponsor, it's the next affordable quality Chinese brand, what anker used to be, I have some of their products
There is always a "but AMD will beat Intel" and a "pay special attention to the one part which is good news for AMD".
You are getting more and more desperate in every video. Just sell your AMD stocks already!
You should make a slide that explains what all the codenames mean, for those who haven't memorized what all the "Lakes" and "Rapids" are supposed to represent.
Naming is abitrary. You will have to study yourself on which product is on which platform, because code name don't tell you anything.
Lake is client CPUs
Falls is “HEDT” and workstation
Rapids is P core server
Forest is E core server
Bridges is for server GPUs, but next gen Rialto has been canned in favor of Falcon Shores
Shores is their server “APUs”, but for now it is for server GPUs for the next gen
@@benc3825thank you for this
yeah, THAT video will be a hit fr /s
@@ConsecDesignno one said anything about a video
Please make time lines for Intel, cause their code names are so confusing. At lease zen is named 1, 2 etc, but all these lakes and coves are so fucking confusing. Not for you because you work with it everyday, but for us who only follow this once in a while.
Also please mention what generation the different lakes will likely be in, that way we can better relatively see how far it is
Totally agree. The complexity of the Intel offer is going to reduce their take-up by the public. Not the trade, they are used to this sort of thing, but the public will understand the zen5 offer as a straightforward 'next generation' one, whereas this Intel 'two competing generations on different architectures' is going to deter by its complexity.
@@simonbrown8509 That is why I gave up following Intel as their naming scheme just got too damn confusing, first with all the bridges, now all the lakes, which is newer, which is older? Who the hell knows. With AMD its Zen 1, 1+ which was just OCed Zen 1, 2 and 2+ ditto, Zen 3, Zen 4, and now Zen 5. Its simple, easy to tell what is newer vs what is older, Intel really needs to take a page from the AMD playbook and just call them Intel A,B,C or Intel 1,2,3 and drop the codenames. and the core vs Core ultra just makes it worse as there are some that will have a core and a core Ultra variant.
@@TheHangarHobbitSame issue for Xbox, no way to know what is newer, or just a variant. Sony straightforward like AMD. Also sockets are in order. Naming sockets by pin counts makes sense on a technical level, but for consumers it makes no sense. When Google named OS versions by sweets, they at least did them alphabetically.
Intel's got way better code names than AMD. still can't remember what Mendocino did
Man I cannot wait for AMD leaks, need me some 3DVcache news for my work in physics. Thank you for all you do MLID!
If you don't mind me asking, what are you working on that's hammering the cache so hard?
@@asm_nop Density functional theory in nuclear physics is most bottlenecked by memory bandwidth not by compute power which is sort of a fringe domain of nuclear physics but in material science density functional theory is main stream so the majority of material science depends greatly on what AMD comes up each new generation because extra L3 cache is short of a revolution for compute times in that domain
@@lukabozic5 Very interesting! I suppose it depends on the amount of resources you have for your projects, and how multithreaded the workload is, but do you use the consumer level CPUs like the 7800X3D, or are you able to benefit from server parts like the EPYC 7373X, which have insanely high memory BW and L3 per core?
@@asm_nop Imagine a nucleus with protons and neutrons, the bigger the nucleus the more protons and neutrons hence more information a PC has to juggle since it needs to take into account all the intra nuclear interactions.
In theory a 3d Threadripper would be ideal but since those do not exist I need to do with higher end consumer CPUs since server CPUs are insane price wise. There is a significant untapped market by AMD of experts who could use such a machine at home. Obviously at the institution people use HPC but that is beyond the scope of our conversation
@@lukabozic5what about 7950X3D??
I just cannot get excited over new lakes
4000 - Haswell
5000 - Nothing Lake
6000 - DDR4 lake
7000 - Same Lake
8000 - Mirror lake
9000 - Twin Lake
10000 - Stillsame Lake
11000 - Ohshit Lake
12000 - Mejump Lake
13000 - Jumphigher Lake
14000 - DejaVu Lake
15000 - OneArmed Lake ...
For some reason most of these lakes just look exactly like other lakes I can't explain why ...
The longer the number is, the more desperate intel is.
Lake? More like swamp
The 5000 series was a -Well not a -Lake.
Yeah, what is their obsession with lakes?
5000 - eDRAM Lake
11000 - Waste Lake
AMD's lower cost advantage is real here, if Intel wins with Arrow Lake AMD can simply drop prices on Zen 5 and still offer way better value
Not running a 500w cpu is a thing too
AMD save so much money with chiplets they still got similar profit margin even at lower costs..
@@htoomyatlin123don't they get like 100% yields anyway?
@@GewelReal not 100% but it is always higher than large chips
@@htoomyatlin123Intel uses chiplets (TILEs) too. But Intel just sucks at it! MTL uses TILEs but cost more than ARL & performs worse than ARL.
Sounds very promising. I hope both Zen5 and ARL will turn out great.
I predict ARL may be slightly faster in some instances, at a crazy power level, and will continue to have poor power scaling. And the X3D variant will dominate at half the power or less for gaming.
@@xerideaI fear the same but I hope ARL won't be that bad.
Intel just severely needs an X3D competitive technology, that's really the main thing holding them back. They can boost to the moon and their chips can sustain ungodly levels of tdp, and their e-cores are actually great for low-power tasks. But that v-cache technology has truly created a crater in performance gap relative to efficiency when you start pushing the chips.
I really don't see how Intel can keep brute-forcing their mid-high end chips when we're already hitting the thermal limitations and power consumption levels where people don't just have to spend extra for the chip itself, but also for top-end cooling, cases that can fit those cooling solutions, and a psu that can sustain not just a high-power gpu, but also a cpu that may very well spike to similar heights.
gamers are still sleeping on how ungodly good the X3D is.... I cant in good faith tell someone to buy a NEW gaming pc, if it doesnt have X3D chip.
Efficiency. For the next gen, they need +20% IPC jump at minimum AND cut the power consumption by half at the same time. And then we can talk about x3D and other shenanigans. In order to increase efficiency the first thing they need is better process tech, in other words, smaller feature size. It seems they are going to use TSMC manufacturing for this purpose. Not sure at what capacity, could be just for mobile chips or entire generation, we'll see.
have u seen the power consumption? there is no need for any cuts, it is already very low compared to intel products. Products speak for themselves. @@mmo0J
even ryzen non x3d cpus tend to have more cache than their intel counterparts. compare a 7700x to a 12700k. hey even a 7600x has more than a 12900k. intel is far behind in terms of cache.
sometimes i just wish intel would add more cache to their cpus instead of brute forcing performance by going to 100 degrees or so
@@mmo0J tsmc 3nm would be good. even better if they add more cache. id go intel if they make a high core count e core only cpu with 2mb L3 cache per core. might never happen though....
I was about to go to sleep, guess not anymore!
lol me too, stumbled upon youtube while going to be and had to watch all of this
This 40TOPS win12 requirement is something that makes me NOT to want to buy anything. Need to wait and see if that's true and sure want to make sure that my device will meet the requirement.
It will be like the TPM requirement for win 11, optional for upgrade, mandatory for system integrators
@@defeqel6537 It can not be mandatory for system integrators for at least 18 months as noted in the video - no high volume Intel based systems would be able to ship with Win12 installed and no low end systems from either AMD or Intel would be possible.
Honestly I think it will be years until those neural processors do anything worthwhile yet, and you'll probably be interested in upgrading at that point anyway. And the NPUs in the products available at that point will be WAY better.
I have a feeling 40 tops is gonna be like RTX 2000 series RT. Does it, but the good stuff came later anyway, so if you held onto your 1070 until you got a 3070 you didn't really miss out.
Plus, both Intel and AMD platforms should be forward compatible, so just upgrade your CPU in 4 years after buying what you want right now.
Mmm yes, AI based spyware. I can't wait!
It's one thing to have the operating system mindlessly scrape some generic telemetry data, and another to have autonomous, adaptive system going through your stuff that is capable of adjusting itself to better match the user's behavior.
There's no way i am installing any AIs on my system i can't access the source code of. Especially one that can access the internet or operates at root level.
Maybe pulling my hair out for a month to get Pulse Audio working isn't so bad after all.
0 views in 10s. Bro fell off
lol
🤣🤣🤣
@@MooresLawIsDeadlol let the failure sink in dude
Seems like intel vids don’t get the same clicks as amd and nvidia. I want to see intel be super competitive, just like I did when AMD sucked.
😂😂😂😂
Arrow Lake, launching just in time for the winter heating season 😂
Summer time in Australia
Dude I have a 13900k and a 13700k I can have both systems running pulling 1700watts. And it’s still not getting hot in my room here in CA… try running SLI or quad SLI.
Europeans in their insulated houses during summertime.
I'm currently stillusing an i7 9700K for my gaming system, but i may upgrade it to an Ryzen 7700 with 65W TDP. It's the most efficient currently and since i'm not a fan of water cooling, the 7700 will also be much more noise compliant on a single tower 120mm cooler.
@@Typhon888You must have a monster AC unit next to the PCs!
@@tringuyen7519 I wonder about this sometimes. My one 275W GPU turned my old room into a sauna when rendering or gaming.
Many attest to these new hotter CPUs doing similar, bit then there are those who claim the opposite. Like they must have the AC set to industrial freezer mode or something, becasue even with temps below freezing outside and the window cracked, vents blocked, it's summer when the GPU's doing heavy lifting.
I'm still amazed that Pat is asking other companies to go and fab their chips on Intel's processes while at the same time they are going to TSMC in order to be competitive.
Think about it. Indirectly, they are taking capacity away from TSMC, right as they are pushing their fabs for external use. It could be both a safety measure AND a marketing move for their fabs to costumers
@@benc3825 I get it from a strategic point of view. What amazes me is that they are going with a straight face to try to convince their potential clients that they should go with them instead of the competition.
Just depends on the price and what your requirements are for the fab
Opening fabs is a great solution if they are not able to compete with the most cutting-edge technology from TSMC. All chips dont need the best manufacturing node.
If you look at the history of TSMC, a big part of their success was that they are a fab only company. Chip designers came to them because they didn't have to worry about their ip being stolen, a huge problem in the tech industry.
I just don't see Intel fabs being popular from outside companies unless they give up design.
As already shown by the A17 from Apple, N3B effectively has 0 advantage over N4. And TSMC themselves always said N3B would be a transitory higher early yielding node before the rest of the N3 line.
Going N3B is a complete waste of time, you either go N4 or N3X/N3E.
Don't expect Lunar Lake to be anything impressive.
This is one of few channels I can trust about Intel performance.
I can't believe that there are haters of Tom claiming he is full of shit. I say, if you don't have haters, you're doing something wrong. IMO, Tom is the most thorough of anyone claiming to have this type of insider info.
@@christophermullins7163he tells it how it is and how it is makes fanboys upset lol
@christophermullins7163
None of this stuff exists until they announce it on stage and magic wand it into existence 🪄
He should have looked into the proper future. Not the next reality over where that ONE detail was mildly off.
If so, FRAUD. MADE UP FlUfF!
@@puffyips facts.
I just ordered a new pc, and yes 7800x3d large cache and low power was the main reason for skipping Intel for the first time in 20 years.
Enjoy it, I have it too, excellent performance and it usually only draws 50 watts when I game! 😁😁
Sadly in Europe it's more expensive option than Intel
@@gozadrotit7426 No its not, its cheaper than 14700K usually
@@gozadrotit7426 is lying and keeps spamming this in the comments, they don't know anything or are really misinformed. EU and US price is basically the same.
@@N4CR After your Comment I decided to look for prices again. Yeah you're right I'm totally liar and I spammed it in a few comments. Sorry for that. I'll go delete other comments then.
The big feature for Lunar Lake will be the on-package memory ... reported to be Samsung LPDDR5X, 8.5 Gbps.
Meteor Lake's best low power feature is video playback with CPU and GPU tiles disabled.
Intel's research cranked out the Thunderbolt 5 and WIFI 7 support included on their Raptor Lake refresh chips.
2024 seems exciting for CPU buyers. I'm currently using a 9900K and will most likely be upgrading this year or early next year around the 5090's launch.
My machine is a 9900K as well; I agree with you that the timing will feel right for an upgrade with Arrow Lake and NVIDIA’s 5000-series. My friend and I refer to the i9 9900K as the “Herman Cain” CPU because of Cain’s 9-9-9 taxation plan and feeling like we paid “the Intel tax” when we built our machines.
I’m on 9700K and I usually do upgrades every four years. Can’t wait!
Honestly, I think the issue with your intel videos is that almost every title has some variation of "AMD in trouble", but throughout the video you list reasons why Intel's next gen is having issues. If you were more honest in the title, the videos would do much better. For this vid, something like "Intel's Arrow and Lunar Lake punches: Enough to threaten Zen 5?" Would be more accurate, but still vague enough to make people want to watch it
AMD has couple of advantages that most people miss. Power efficiency gives noise reduction for people who actually use PC:s for something that needs concentration. Second advantage that most people miss is AVX-512 support. It's not about width, it's all about handling problems that break vectorization.
They also support ECC memory on every SKU.
Yes. And the power efficiency (and reduced waste heat generation) are also relevant for those concerned with energy bills and/or living in warmer areas.
Intel is working on a new instruction set that will be an evolution of AVX512, called AVX10. It’s unclear if ARL or LRL will include it, or if, in the worst case, it will be made exclusive to their Xeon lineup.
@@GeekNerdNoir Intel''s main problem with AVX-512 is that their little cores do not support AVX-512 and for scheduler they needed to harmonize the instruction set between big cores and little cores. So their consumer lineup will only support the instruction set of their little cores. But if there is not consumer software that will use the AVX-512 then adding support for it would be too costly for the little cores.
@@BrunodeSouzaLino Note quite. For some reason, they don't support ECC on APU's that aren't part of the Pro lineup.
The last time Intel did a presentation about something of that regard, they essentially advertised AMD for free since they couldn't stop talking about it. Performance-wise, Intel still has to solve the heat issue before they add more performance. It's clear they don't have the same hand into making asymmetric core CPUs the same way ARM does.
I can't say I care all that much about NPU performance right now
You should though.
Regardless of what I think, the vast majority of people aren't willing to pay for ad-free online services and they definitely don't want to go back to the days of paying substantial amounts of money for the next version of Windows.
Just like ADs subsidized product costs over the past decade, edge-client NPU processing ability is definitely going to be the biggest business *and* government target for the rest of the 2020's (especially now with the near complete proliferation and public acceptance of always-online sensors).
Since pre-processing done for free (i.e. without using power at the datacenter) on the client side can result in more than linear savings (such as only sending a compressed Speach-to-Text transcript of conversations and raw data when flagged for re-evaluation), it also makes this one of the only practical ways for non-state actors to classify billions of real-time physical signals, not to mention handling all of the video streams.
@@mortlet5180 Fuck me.. how's THAT for a response? I didn't care that much about NPU either, but now I do. Even thought I didn't understand much of what you wrote lol
It needs a killer app/game that justifies it's usage before everyone will go buy one.
@@MrHav1k How about being able to solve *ALL* cybercrime and convict almost all other criminals? Just think of the children!
I mean, everyone has already decided that privacy isn't important. Actually, it's a marker of either bad people with something to hide, or of the patriarchy that want people to be ashamed of showing their naked bodies.
@@mortlet5180 Yeah but for some basic voice recognition and stuff I think it will have more than enough power. Maybe for some more advanced fancy features you need a more powerful NPU but ... I don't see what I would personally use a powerful NPU for right now.
Also I think companies are not doing edge AI not because the machines don't have enough power to do so but because it's just easier for them to just upload the data and have it done on their servers. Having all this valuable data on your server is of course just a side effect and we all know that not one of the big corps would ever use this data for their own benefit.
All in all ... brind me CPU power and screw the NPU power. There's just no use case for me now.
Baseus product sponsorship is a good direction. I think you’re finding a great niche in sponsorships!
This channel has grew fast. Love the content.
Big/Little is hard to get excited about from a gaming perspective, my next upgrade will be AMD.
Big/Little is useful for gaming if you have APO.
But, stupidly, it took until 14th gen for Intel to release APO after 2 generations of having their efficiency cores do nothing useful for gaming, and only 14 games (as of the most recent list I was able to find) actually support APO, so it's too little, too late.
@@nathangamble125 There is good evidence that suggests that the reason the list is so short is that many games will not benefit no matter what you do.
@@nathangamble125 Intel is having to optimize each game internally. This is not scalable. Biglittle is terable for gamers.
Did enjoy, thanks for asking my take.
Quite information dense video, thanks.
You mentioned windows 12, I don't have any information on 12, if you're referencing the erroneous uniformed content after hearing Qualcomm say "the next version of windows":
That doesn't mean 12 at all.
MS has a new build released regularly, which is probably what Qualcomm intended.
Again, thanks for this video, quite enjoyable
I'm confused. So how much more performance will Arrow Lake desktop have vs. the 14900K(S)? 🤔
I think around 20-30% but unsure on the power usage and demands? Of course the new lakes will need the new socket size motherboard for all future upgrades going forward now
@@Lone_GamerUK Problem is, that he kept comparing Arrow Lake to Meteor Lake, that is why I'm not sure about those 20-30%, since we don't even have Meteor Lake for desktop.
And if I recall correctly, there was a leak not too long ago saying that the next generation i9 desktop chip will perform the same as the 14900k in gaming 🤔
That is why I'm confused here.....
@@freesS3 MTL has very little ipc uplift anyway.
Finally, been waiting for this video!
I need to go to bed and here you go and drop another banger!
Great video, I love that MLID went with the "ughhh I have to do this, but I won't be fun about it" just to get the good juicy info out in a short format.
It's a responsibility at this point, for the torchbearer. So, thank you brudda.
Intel needs to commit themselves to support 3 different architectures (not mere generations) with each socket they release.
If they said that LGA 1851 will support at least 3 different architectures it would become very interesting to me compared to getting a Zen 5 system.
The 1 - 2 years of true or fake generations on the same socket does not cut it anymore whether you are either the underdog or the market leader after what AM4 has achieved.
And Intel management needs to stick that to their head and require a socket change only if a new architecture would either be limited by the pin count of the socket or by the standard they are able to support without changing socket.
Agree, Intel needs to pull an AM4, I don't mind paying slightly higher price for the motherboard to be overbuilt and buy 15 gen if 5 years down the line I can buy a 19 gen CPU and drop it in after a bios update.
I have been seeing twice as many Meteor Lake laptops drop onto the market this year with over 10 hours of tested battery life (by Notebookchek) than Ryzen 7000/8000 SKUs so far. At least buying a Windows ultrabook this year with Apple M1 levels of efficiency will be standard instead of something you have to check tons of reviews to come up with a short list of models to consider before buying.
So Intel isn’t even using their own foundry services for their newest cutting edge desktop product? Is that huge red flag or what
it's sad and perplexing...so who is even gonna use their nodes if they don't use it themselves?
@@ofon2000 there are plenty of customers for fabs
@@ofon2000 Intel is opening their fabs for other manufacturers to use. For example right now they are actually in a discussion with AMD so AMD might be using intel nodes for some of their products.
😂😂 noobtel
It's a smart move. Their own foundry nodes are what is holding them back.
Still power and price to overcome... fingers crossed.
The big issue Intel has always had has been platform cost, with Intel they seem to change sockets nearly every other year whereas AMD is still releasing new AM4 chips in 2024. With my current build I went from a Ryzen 3 1200 to a 3600 to a 5700x with nothing more than a CPU change, with Intel I would have had to replace my board...what 3 times at least? That is just crazy.
Intel cannot miss this launch window that they've set themselves, hopefully they don't. Jam packed with info as usual, keep up the amazing work Tom and co.
Have a great weekend mate!
Thanks, you too!
The patreon really is great I binged half of em already
Serious Question if AMD is on TSMC N4 and Intel is on TSMC N3 , how does AMD then have the first mover advantage since TSMC N3 is better than TSMC N4?
Is not N3 the smaller and better node from TSMC? Should not Intel then have the first mover advantage since they are on the smaller N3 node? Also in terms of efficiency?
yeah that thing needs more explanation or clarification from mlid as your point is really good.
I bet, after so many zen killers in a row, amd is in such a poor state it continues capturing the high margins market. If amd wasn't capacity constrained, intel would have gone bankrupt by now. They absolutely can't survive on desktop margins. Patty has this "my precious" look, like half-way through the transformation.
Yep, this is more lies from intel marketing disguised as leaks. How many years was MTL hyped for it to be slow as balls.
It makes sense to target more tops with the mobile variant as it's more likely to be employed without a GPU, where Arrow Lake is far more likely to be paired with a dGPU. The fight over who has the most tops in CPU seems irrelevant for desktops when even current GPUs have vastly more AI compute, and that's just off tensor cores that are usually idle anyways. It's not like your GPU is going to ramp up to 100% power draw to churn out a 30-40 tops workload.
Beyond all that, I could care less about AI or iGPU performance in my CPU as we still have no app that can even use it. This just sounds like a distraction to make you not look at raw performance that actually matters in a CPU. Even if I'm buying a laptop for gaming or some future AI app I'm just going to make sure it has a competent GPU.
5+ years and still no killer app for ai. I will not even be considering tops when looking at my next several upgrades.
My biggest concern is z890 pricing. Well, concern for intel buyers since I'm already on am5 lol.
Hehe 😉
Intel is completely pointless for DIY buyers anyway until they fix their stupid problem with their need for new motherboard every 2 years. And I dont think intel is interested in solving this problem because theyre doing it intentionally to boost motherboard sales.
@@damara2268most people go about 5 years between upgrades even am4 was ending by that point anyways. Most people it doesn’t matter
there is only 1 and cpu I would consider over my 12900k is 7800x3d is really the only competitor imo. The 7950x3d kinda sucks. 7800x3d doesn’t have enough cores. Only way I would consider changing to amd is if the 7950x3d had both CCDs with vcache. I don’t have gpu assisted rendering so extra cores is great.
12900k is $500 CAD or less
7950X is $700
13900k is $750
IMO Intel is still the better buy if you do any sort of work and not just play games
Okay why do my comments keep disappearing
@@damara2268As StrixWar said, I'm upgrading after 6 years of using my i7 9700k and RTX 2080 and 2024 seems like the best year for that as we're getting new GPUs AND CPUs, not only that, it's also a new platform for both AMD and Intel, meaning I'll have upgradability with both brands (thanks to that new Intel platform, seeing how the LGA 1700 did), all I have to do is watch Gamer's Nexus and Hardware Unboxed to know if I jump ship or stay hyperthreadingless, which didn't seem to affect my day to day tasks much with my 9700k's 8 cores and 8 threads
@@StrixWar yeah but amd is better either way, platform longevity is just a bonus.
Thanks Tom, always appreciate your insight
5 months ago: Is Zen 5 doomed?
4 months ago: Intel i9-14900K Analysis: This JOKE is Zen 5's Competition
3 months ago: AMD just won 2024 - Zen 5 Strix, Hawk Point, MI300X Analysis
Now: Can AMD Zen 5 Survive Q4?
Stop the ridiculous flip-flopping clickbait.
Intel died for me. I don't even consider their products as a possible option because I'm a musician / producer. The 13900K had disastrous latency problems making it impossible to use low buffer sizes / low latency in your DAW. On top of that, it gets worse. Intel's Big.Little has been a nightmare for DAWs trying to balance the DSP graph, because if you have a bunch of heavy chains that use similar CPU time, there is no way to balance them across cores, because some of the cores can't finish in time. So the DAWs keep moving the chains around every few seconds to try to accommodate the Big.Little, and that's been causing unreliable performance with sudden dropouts when you don't expect it. I'll never buy another Intel CPU, I'm out for the next decade at least.
What do you use? I use FL Studio myself, much more of a hobby then a living at present - running on a 7950X.
Disable hyperthreading and the issue goes. Here's my worst case dpc and isr latency on a 13600k after two minutes of testing. DPC worst case was 67.8us, ISR worst case was 1.74. I think at stock, techyes city was getting like 400us of dpc latency or smth with a 13900k, Either way not exactly a small improvement from doing what i just suggested.
I feel your pain. However, the software stack needs to find the way to utilize heterogeneous CPU cores properly eventually. And yes, I'm a fairly low-level programmer (C++) and no, I don't know how to achieve this yet ;-)
Seems like you should go for Intel's workstation platform which has lower latency than AMD and all P cores
Excited to see what their evolution of X86 looks like but the part where you implied that they're throwing money at their launch problems is worrying.
I just had to remind myself that after Arrowlake is Arrowlake Refresh... maybe with a different NPU die?
I don't see how there could be an Arrow Lake refresh between Arrow Lake in late 2024 and Panther Lake in late 2025, would they really ship a refresh 6 months after Arrow Lake, where Panther Lake is set to arrive 6 months later ? If there is to be a refresh, it has to be Panther Lake's imo
More l3 cache tho (36mb vs 48mb)
"25-30%" more performance per physical core on ARL vs MTL (which had performance regressions vs RPL outside of specific power usage bands on top of IPC regressions) being a victory over zen5 doesn't really add up (literally), unless the leak was referring to those specific power bands in laptops where MTL actually is kinda ok? It also doesn't bode well for desktop if those are the most optimistic number & comparison too, I would have hoped to see a RPL comparison, not an MTL comparison.. It feels like when Intel was comparing MTL to ADL instead of RPL in official marketing materials.
MTL is actually like 10-20% behind 14th gen so it might happen that arrow lake will only be up to 10% faster than 14th gen which would be a huge disappointment for intel considering the absolutely enormous difference in manufacturing cost and basically a jump over 3 generations of nodes.
Personally I think this will be the reality though.. also some games will definitely have problems with FPS drops because they will get confused that there is no multithreading and start pushing load on e-waste cores which have skylake-level performance.
@@damara2268e-waste cores.😂
I re-watched an MTL performance leak Broken Silicon section and the Intel sources were claiming similar big gains last summer.
Tom has a problem reporting anything reliable if Intel itself is short on realism.
Wow. That power bank is genius. It's hard to find any that put out more than 20 watts. And that form factor is ideal.
The hard thing to weigh up with the ARL performance claims is the lack of power envelope info. It's a bit meaningless to be "25-35% > MTL" when MTL couldn't match RPL above approx 100W. But MTL was supposed to thrash Phoenix and didn't.
It seems Zen5 will be established for the main '24 buying season with ARL struggling to make launch by holidays. Currently it's a bit ridiculous seeing Amazon top sellers dominated by Zen3 except for #1 7800x3D (#2 7600), but most people are happy with CPU performance.
The cost to make factor likely affects margins more than retail price, at least in gaming it's the 6c/8c speed and cache not high thread count which matters. Low priority background stuff runs fine as virtual threads.
Totally understand why Tom's presented the quotes with little analysis but it's hard to guage the true strength of ARL without power data.
Overall AMD seems to have jam today and aren't rushing Zen5 while Intel hopes for 2025.
Personally I dont believe that ARL will even get to 90% of zen5 performance.
@@damara2268 well ARL needs to be better than that while reducing power usage!!
Zen5 IPC for AVX512 should boost certain applications but probably explains Tom's sober IPC estimate compared to the wilder claims.
I think AMD chose 4nm for time to market and cost reasons, balancing out performance & power efficiency gains.
They don't have to win vs ARL on benchmark performance just be close enough.
I am wondering if the MTL & Sapphire Rapids fails, lead to greater Hawk Point production and a later Zen5 Ryzen launch which may have an extra stepping polishing it. That would explain the Q3 Ryzen & EPYC launch close together.
Overall Zen4 was received poorly due to costs but as a platform introduction it appears the AM5 backed by AM4 for the memory transition was good strategy.
Intel have been forced to make risky plays, P+E papered over cracks but screwed the ISA strategy (adding AVX512).
Later after Intel pipe clean all the software issues, AMD just add compact low frequency cores to their monolithic laptop dies.
@@damara2268 I wrote a longish reply discussing that, don't see it show up. Suffice to say ARL MUST top Zen5 without RPL power hunger or Intel are in trouble.
@@RobBCactive That happens to me way to often. I have started to make sure I always make copy my reply before I post them now. And then check back to see if it actually gets posted if I really want to make sure it gets posted. RUclips auto censors algo get snagged in a lot of words, that really should not have any business being censored nowadays.
@@cajampa good idea, but it used to tell you if the censor kicked in. I don't think I used any words any filter would trigger on
Great video! Looking forward to more info on the AI accelerators by Intel.
I dig the power bank. Will use the code! Thanks Tom
Maybe someone can correct me if I'm wrong, but after hearing all about AMDs power savings, I came across a number of articles and test that show in standard every day use of your computer, Intel is actually more power efficient. Intel is even more efficient at adobe work/cpu heavy workloads. In gaming amd is better, especially the 7800x3d. The max power draw under non-realistic cases the AMD is better, but in reality you never see it. This is comparing 12600k-13900k vs the AMD 7600-7950.
Link please.
ruclips.net/video/JHWxAdKK4Xg/видео.htmlsi=arWYMBbE3CSdl7Vp
ruclips.net/video/ei1zgUr6aF8/видео.htmlsi=f6WyOf3gGsFpP56K
@@PanduPoluan
I'm mostly here to learn about the newest mobile type chips for use in miniPCs, both intel and AMD. Mostly want to see how Strix, Strix Halo, arrow lake are going, and also curious how much AI in chips will matter in the future. (I still think a lot will be done in the cloud and not locally)
I think the biggest issue for intel is highlighted correctly by you. I do not think that Arrow Lake will beat Zen 5 X3D. I anticipate that Zen 5 being 15%+ faster than Zen 4 essentially means that X3D will bring a boost over that 15%+ so realistically Arrow Lake will have no performance advantage. To me this means that all people should just buy AMD. I was on a 13900k and switched to a 7800X3D and boy do I love that decisions, 70w of power during gaming, same or better fps than 13900k, way cheaper, like there is no reason not to choose AMD at this point.
Enjoy the AMDip fanboy
The best thing is we will be able to slap a Zen 6 chip in the B650 also. I just upgraded to 7800x3d because the price for that chip became quite affordable. Im skipping Zen5 but hope it does well as that will bode great things for my Zen 6 upgrade.
@@kingyogesh441 Acts like a fanboy, calls other people fanboys
It's like some people actually lack self awareness.
Congratulations for using reason to do that change, most people tend to keep and defend their choices. Amazing CPU/platform
Weird sidegrade. Traded in cores for lower power.
Hasn’t RGT scooped you on all these rumors?!? At least your camera is in focus. 😉
Love your channel and keep up the good work! 😎
RGT, last I checked in, spouts so many different things over time that he is statistically likely to be right on some of them. But sentences like "Zen 3 could be up to 45% faster in single threaded than Zen 2" and "RDNA3 might be 2.5X faster than RDNA2" does smear smelly poo on those few instances where he got something right.
@@andersjjensen 100% agree 😎 I should have added a sarcasm warning
@@MoonshineOctopusAh, yes. Sorry. But there are like 10 other commenters who are actively defending RGT elsewhere in the comment stack... No, I shit you not.
And besides the obviously meaningful price difference between 3 nm and 4 nm wafers, AMD also likely has a decent discount on any TSMC nodes for being a major and long lasting partner. So the cost gap between the allegedly $15k 4 nm and $20k 3 nm likely widened further to maybe ~$12k for the "dependable customer" AMD's 4 nm, while rises to ~$22k for the "disloyal and even market opponent" Intel's 3 nm.
IIRC TSMC did away with partner discounts a couple of years ago
Why does a desktop chip need an NPU? It makes more sense to run the AI workloads on nvidia's tensor cores on your dedicated GPU.
Heard similar suppoosed gains thanks to DLVR in Raptor Lake not that long ago. I'll believe it when i see it
Oh, I have one of those flat and wide power banks.
They are portable in the bag, but take up a ton of space on desk. I wished I could use it as a mousepad or touchpad.
Do you have any leaks about FSR 3 and AI upscaling or Cyberpunk and FSR3 release date?
Intel going from Raptor Lake being able to heat your home in the winter to at least being in the ballpark of AMD in terms of power consumption for a given unit of performance means Intel is getting even more competitive against AMD. So the whole "Intel is doomed" narrative really doesn't hold water. Now is it so much better that it's worth buying over AMD, dumping your older Zen system, etc? Time will tell, but it's at least in the ballpark.
I’m not sure how the npu in the cpu tile can be used if the memory bandwidth is really low.
I guess you can run 7b parameter models but it’s going to be slow. What other use cases do they have that don’t require as much bandwidth? 🧐
@@MooresLawIsDead The 1440p footage looks like 1080p crummy quality, why did you go with such a low recording bitrate?
360 AIO seems cool, even when it is not needed. But when it is needed it doesn't feel as a waist as much.
If ARL ultra 7 needs a 360mm AIO to function properly, 7800X3D owners will be laughing their a** off!
lga 1851 doesn't need to really be any more expensive. having a tops advantage really wont matter until it has a legitimate main stream use
If Intel don't get a clean win on TSMC N3 against AMD on TSMC N4 even the most adamant Intel fan will have to admit that AMD has beaten Intel at their very core strength: raw CPU design. A lot can be swept under the rug with "Intel lost the manufacturing game" that excuses the absurd power consumption, but with a node advantage on the same manufacturer nothing can really be said that excuses Intel if they lose.
Not that any of it really concerns me, as my 7950X3D will do everything I need for quite some time to come. The only thing that could make me consider Zen 5 is if AMD does an 8 Zen 5 V-Cache + 16 Zen 5c desktop part with a default TDP of 150W or less.
Does the Dual Channel interface of Standard Desktop platforms limit the total CPU Power at high core counts. When I think at Apples M CPUs, it’s seems possible to me - but that‘s ARM….Quad Channel as Standard would be nice…As well as an Arrowlake with 16P cores 😅 at Mainstream
NGL, I'm waiting to see what lunar lake will be like, especially the low power CPU ones. I'm impressed with n5105.
Lunar Lake will be super expensive bc you have to buy the memory with the CPU. & then AIB will charge 10% extra just to get their profit margin.
Former MLID community member here. I’m hoping this channel will move to more efficiently present leak information in dedicated videos. Leak process information and preamble is not interesting to me. I was bored until 5:00.
Waiting for LGA1851 and Intel AL, I already have sufficient cooling for my 9900ks, I've been waiting for Intel to finally release a 5nm or less Desktop CPU and it looks like it's not too far away. I honestly hope both companies release great products, it only helps us consumers regardless of our preferences.
I’m excited for whey they bring on a better node than amd, love competition that drives innovation -7800x3d owner
opening music sounds like animal as leaders
Love this band
Sounds a lot like The Brain Dance
Honestly, getting 25-35% performance with less threads than MTL is what matters here. 13TOPS NPU will be niche.
A lot of people, such as myself, are going to be upgrading off of AM4, so we'll need a new platform either way. Cost isn't something that concerns me too much, Intel generally sticks to around the same prices, even considering inflation. A 14700k doesn't cost much more than something like an 8700k did. If anything, I'd expect AMD to have lower prices this time than Intel having higher prices. I also hope that boards aren't as expensive as AM5, board cost is what really killed AM5's launch and is why a lot of us didn't even consider Zen 4.
Power draw will be interesting. The jump from Intel 7 to 20A is huge, so I'd expect that they won't need to push 300W through the Ultra 9 to compete with the X3D chips, but I don't know what power draw will be considered acceptable. Hopefully the stock TDP will be around the 140W that my 3900x uses under load. My guess will be that they'll try to target the power draw of Zen 4 R9s with the Ultra 9.
Honestly AMD should just throw out the X3D chips from the start.
They in theory can this time. They aren’t changing mode families. A 7nm family node stacking on 5nm family node is ready. There will still be a slight delay, but it shouldn’t been much.
I liked it. Glad you waited Tom. Intel been uninteresing for a while
If arrow lake is much more power efficient than raptor lake while being better than the 7800x3d it will sell. Intel needs to tame power consumption and enhance platform longevity.
Maybe Microsoft has paid a premium in order to get 18A fab usage first for their own NPU. That would be a reason for Intel to move LNL to tsmc. Another source said the top sku of LNL is still on 18A.
Another great video!
According to RedGamingTech a Battlemage Xe core will only have 8EUs vs 16EU on Alchemist, so going from 4 Xe Cores on Meteor Lake-U to 8 Xe cores on Lunar Lake shouldn't be that big of an uplift. Interestingly Tiger Lake and Alder Lake had the same iGPU on the small "U" series die and the "H" series die on laptops but Meteor Lake had cut it in half for the "U" series and Lunar Lake will follow MTL-U.
Where's Zen 3+ Desktop and Alchemist+ dGPU?
He also said that RDNA 3 was going to be 2.5 to 3.5 times faster than RDNA 2 depending on the time you checked. He also claimed there would be an Alchemist refresh in mid to late 2023, 30%+ IPC for Zen 5, Battlemage Q1 or Q2 of this year, etc.
Anything he claims I wouldn’t trust
@@benc3825I trust myself because I 💯 fund myself.
@@MooresLawIsDead I'm pretty sure he's the guy you are talking about when you say things like "Some people are claiming these huge increases for Zen 5, and it's going to be decent but it's not going to be a 40% uplift. That's bullshit." RGT has been wrong more often than the early days of WCCFtech...
I was subscribed to his channel for a couple of months. Besides the daily videos that would contradict each other on a daily basis, he put out a video titled something like "The Ps6 will be revolutionary". The big reveal........one of his "sources" told him.......the Ps6 will be revolutionary.
The man is playing the RUclips game, daily content at around 15 minutes does well in the algorithm. However, there simply isn't enough info in this space to support daily uploads at 15 minutes. There is a popular word to describe youtubers like this, and it starts with a "g".
A lot of people and Tom himself don't seem to understand that RL easily beats non-3D Zen4 in raw performance, and that a Max OC 13900K is easily 15-20% ahead of Max OC 7900X in gaming, base Zen 5 first needs to catch up with 13th gen lol. And if AL turns out to be such an increase, non 3D Zen 5 is going to have a hard time. X3D might fix it though in gaming.
I also remember Tom saying that 9900K is stronger than 2700x by 20%, it just got me rolling on the floor. It annihilates in avgs and lows even 3700X and is basically = 5700X released years after 9th gen. Sometimes I'm wondering if that dude even knows basic stuff, he just gives wrong numbers all the time.
I probably missed this but are we gonna get an arrowlake refresh after arrowlake or a new architecture? if only a refresh is what's coming, then RIP intel.
How are the rumors of Arrow Lake against the Zen 5 X3D CPUs?
Always here for an Intel news...
am i the only one who cannot keep track of all the lakes, plz tom, give us a chart showing if they are mobile/Desktop/server and what gen core they belong too, i kinda gave up on intel-codenames and (new/changed numberings).
Intel GPU news vs Intel CPU news I'm guessing is relevant to the buzz it brings.
I've rewatched your videos of last year almost "claiming" that Raptor Lake Refresh should get at top a 20% difference gain over Raptor Lake, and at the end we got just overclocked and golden samples of Raptor Lake. With Meteor Lake unable to surpass Raptor Lake performance, i wouldn't bet on Intel stockings this year nor the next on the markets.
If we also loose the hability of Hyperthreading, we're now back to the 9th gen era were these chips will be useless at all for smart dedicating resources for VMs, for example.
I bet they won't remove this feature for the server counter parts until Rentable Units show that can replace it at all for virtualization purposes (as Data Centers manage their hardware resources efficiently, and effectively).
I think the combo of Zen 5 with Blackwell in 2025 will make laptops a really good option for gaming on the go. I hope RTX 5060 has at least 12GB Vram on mobile
Issues for intel - Power draw, heat, socket changes. If you already own AM5 you can just throw in Zen5.
Lunar/Arrow either a banger competition to Zen 5 or delayed because bad production and hardware bugs
finally the HT will go. loved the 9700k, because games was running smoother on this cpu. never had an better gaming cpu.
Indeed, the fact I can swap my 7800X3D from my huge gaming box into an ITX system when I upgrade it, is a HUGE deal.
Plus like you said, if anyone actually uses the new Windows AI functions, having more TOPS could be a massive selling point.
As an AMD user is really exciting that Intel keeps getting better so AMD don't stop innovating as well 😁
Please Sir, can I have some more RDNA4 leaks? Or the new AI upscaling that has been rumoured lately
He will get to it when he has something tangible. Funnily enough Radeon stuff does really well in views.
dang it bobby, thats too many lakes, I tell you hwhat
I'm positive that ARL will at least be competitive. Wouldn't be surprised if 20A will disappoint us as did its predecessors, but Intel is in position to order N3 capacity to conceal whatever fiasco the 20A might turn into. MTL has provided Intel a platform to stop being dragged by their own fab.
Need a 3970X replacement in around zen6 era so I am Hoping intel has Something competetive to threadripper pro in the 5000€ range.
Closer is a Laptop right now. (Still stiting on a 2017 MSI one with 32GB, 1060M 6G and a i7 7700HQ
Want to see a 15W CPU with Maximum battery life while offering enough RAM to open my stuff on the go without it crashing. dGPU not needed anymore. So lunar Lake may is the thing..
Core i7-3970x
I want to know what Intel is going to do to compete with the 8800X3D the 7800X3D is the perfect chip for a mini ITX build it doesn't need a lot of cooling is power efficient and beats intel chips that cost double the amount.
The simple truth that Intel has to source all of ARL from TSMC in order to even stay competitive with AMD should raise all the red flags there are.
Now I can already hear the screams that AMD had a manufacturing advantage for years that isn't their doing - well, maybe. But if this time Intel doesn't manage to decisively outperform Zen5 even with a better node by the same manufacturer, that'll probably look even worse. And that's not even taking into account the financial side of things.
i watch every Video and every Podcast because its interesting and informative
In the last podcast episode you asked/sad something about some kind of “all in one” software for gpus that includes dlss, frame gen, anti lag etc all of that nice stuff into one software that would change settings live. I thought you mean something like hyper-rx from AMD? That’s what I thought😂 This new “Hyper-AI”Software would read your pc stats and bring you the best possible performance? I would love to see AMD working on that and something that would eliminate 1% fps drops. AMD is building up. Mini pc, laptop, handheld, desktop cpu, gpu, AI. AMD is everywhere. You have to give it to them, they are doing a good job, most of the time. Where is anti lag+ btw? 😂