[SPON: Use "brokensilicon“ at CDKeyOffer’s Black Friday Sale to get Win 11 Pro for $23: www.cdkeyoffer.com/cko/Moore11 ] [SPON: Use “brokensilicon” to get $30 OFF the Minisforum V3 3-in-1 Tablet: shrsl.com/4rt3x ]
Was just thinking Nvidia's strategy with branding the 5080 and 5090 as professional cards might be them trying to segment their product stack into HEDT in the same vein as Threadripper. Allows them to charge obscenely more money for a product while not needing to provide the same level commitment and/or support as server grade. I guess technically the Titan was that card back in the day, which begs the question why they don't just return back to the Titan naming scheme, but I guess that's what think tanks are for.
I want Intel to not go out of business. Right now they are selling something that seemingly loses to the 7600 XT despite having a 33% larger die size and a MUCH more expensive node. They are losing money on each of these cards sold, and it is fully unsustainable. It's no coincidence that Pat was sacked the same week...this was his vanity project after all...
@@Speak_Out_and_Remove_All_Doubt they need to bet their company on whatever comes after the P and E cores. Just like AMD bet theirs on Zen. It has to happen otherwise it's certain death.
@@xpk0228 tbf Nova Lake which is still P & E sounds decent if they can deliver on it, 16p cores with a massive IPC uplift plus 32e cores would probably win against Zen 6 but they need to execute this well and no one thinks they can. As for the next thing after the p&e core, they need Jim Keller back for that!
@@Speak_Out_and_Remove_All_Doubt I am not sure NVL has that much cores tho. Tom said 16P+32E, but that is before royal core and rentable unit was cancelled. I think it might be just 8+32 like ARL was planned to be.
I 5556t 48:46 r5s t 48:4 4 8th 6t55555556566 48:46 666656666656665665556555565555r|to 48:the other day 48:I T4 48:4 5 48:4 5 48:4 5 48:4 5 48:4 5th of the 5 48:46 I 4444 5 48:46 48:but I 48:4 5 48:4 5 5 48:46 I 5 48:46 but I 5 48:46 I 5 48:46 but I 5 48:46 5 5 48:46 I 5 48:46 but I don't think 5 48:46 to 48:4 5 5 48:46 and to 5 48:46 and 5 48:46 and to 5 48:46 and then 5 48:46 to 5 48:46 and to 5 48:46 and then to 5 48:46 and 5 48:46 and 48:to 5 48:46 r|to be 5 48:46 and then 5 48:46 rrr4 5 48:46 and 5 48:46 to 5 48:46 and then 5 48:46 and 5 48:46 and 5r 48:46 rrr4 5 48:46 and 5 48:46 and 5 48:46 and I 48:and 5 48:46 and I 48:4 6 5 rt 5 48:46 to 48:4 5 5 5 48:46 I 5 48:46 5 48:5 5 5 48:46 I think 5 48:46 is going to 48:4 5 5 48:46 4 5 5 5 48:46 and 5 48:46 66 5 48:46 I 5 48:46 but 5 48:46 to 5 48:46 and 48:to 48:46 55r5 48:46 and I 5 48:46 5 48:46 5 48:46 I 5 48:4 4 48:4 5 55 485 48:46 5r 48:46 4 5th to 5th 5 48:46 to 5 48:46 and 5 48:46 and 5 48:46 and to 5 48:46 and 48:to the same time 48:46 55 48:46 and 5 48:46 and 5 48:46 to 48:the 56 485 48:46 and to the 48:to 5 48:46 and the 5 48:46 is to 48:4 5 5 48:46 and 5 48:46 and the 5 48:46 two of 5 48:46 5 48:and to 5 48:46 to 5 48:46 and then 5 48:46 48:4 5 55 48:46 and to the 5 48:46 and to 48:4 5 5 48:46 and then 5 48:46 and 56 48:4 5 48:4 and I 5 48:46 and 55 48:46 and I 5 48:5 85 485 48:46 55 4855 48:46 I 48:4 5 5 48:46 4 5 5 48:I rr to the 48:4 5th 5 48:46 to the same 5 48:46 I 5 48:46 but I 5 48:46 I 48:4 5 I 66 5 48:46 to 5 48:46 and 5 48:46 and 48:46 55 48:46 and 48:and I 5 48:46 and 48:to 5 48:46 and 5 48:46 and 5 48:46 and to be honest 5 48:46 I 48:5 48:4 5th of the 5th of the 5 48:46 I 48:4 5th 5 48:46 I 48:4 5 5 48:46 I 5 48:46 think 5 48:46 is 5 48:46 to the other rrr4 5 5 48:46 rrr4 5 5 48:and 48:to 5 48:46 the other day 5 48:46 and 48:46 55 48:46 4 66 48:I 5 48:46 to 48:46 55 48:and 5 48:46 T4 5 48:4 5 55 48:46 I 5 48:46 but I 48:46 55r5 48:46 I 5 48:4 I 48:think 5 48:46 is going to be a 5 48:46 but to 5 48:46 and to 55 48:46 66 48:66 5 5 48:46 4 5 I 5 48:46 but try 48:46 5 48:and 55 48:46 5th of the 5 48:46 r55 48:46 and 5 48:46 to 5 48:46 55 48:46 and I 5 have 48:4 5 5 48:46 and 5 48:46 and 5 48:46 to 5 48:46 and 48:4 5 and 48:4 5 48:46 55 48:46 and r 48:46 5 48:46 I 5 48:46 but I r95 485 48:46 I 5 48:46 48:to 5 48:46 the two 48:and the 5 48:46 two of 5 48:46 in the 48:46 to 5 48:46 5 48:46 48:4 66 5 48:46 5 48:46 4 66 5 5 48:46 48:4 5 I 5 5 48:4 5th 48:4 5 5 48:46 I 5 48:46 48:4 6 try 5 48:46 and then 55 48 48:46 48:46 55 48:46 4 5 48:46 55 48:46 48:4 5 48:4 5 48:4 5 48:I think 555 48:46 is 5 48:46 to 5th 5 48:46 and 5 48:46 485 48:46 5 48:46 and 5 48:46 and 48:to 5 48:46 5 48:46 and I 5 48:46 and 48:4 5 5th of the month 5 48:46 to 48:46 55 48:46 4 6 and I 48:5 5 48:4 5th 48:46 I 5 48:46 48:I 55 48:46 66 5 48:46 I 5 48:46 but I 5 48:46 55 48:46 rrr4 48:46 5 48:46 55 48:46 4 66 I 5 48:46 but I 48:4 5 I 5 48:46 5 I 48:4 5 I 5 48:46 but I 48:4 5 5 48:46 4 5 5 I 5 48:4 66 5 48:46 to 48:4 5 and 5 48:46 and 5 48:46 and 48:46 5rr|and I think they 5 48:46 48:46 55 48:46 to be honest to 48:46 48:4 5th 5 48:46 rt 5 48:46 I 5 48:46 |5th 5 48:46 48:4 5 I 5 48:46 48:4 5 5 48:46 4 5 5 5 48:4 5 5 48:46 I 48:46 5rr55 48:46 5 48:46 to Tytherington and 48:4 6 5th 48:46 rr5 48:46 5 48:46 I 48:46 5485 48:46 I 5 48:46 |5th r 48:I 48:46 to r5s 4 r5s 5 48:46 to 5 48:46 5th 5rr 48:46 I 5rr||rr5 48:46 r5 48:46 I 48:46 5r|5th of the 5r 48:46 5th to 5 48:46 to 5rr55r5r55rr5 48:46 48:46 5 485 48:46 and 48:4 5th of the 5r 48:4 4 8th 48:4 5 5 48:46 4 5 5 48:46 I 48:46 5r5 48:46 48:46 5 48:4 5th 48:46 55 485| 48:46 r 5r 48:46 48:46 555 485 485r 48:46 55 48:46 5 48:46 48:46 5 48:4 the 48:46 54855 48:46 5 48:46 55 485 485 48:46 5 48:46 48:46 I think 5 48:46 the 48:46 5 48:46 is to 5 48:46 and 48:and 48:4 5 and 5 48:46 and 5 48:46 48:4 5 5 48:46 48:46 55 48:46 I 48:4 5 I 48:4 5 55 48 48:46 55 48:46 I 5 48:46 but I 48:4 5 I 5 48:46 48:4 5th 5 48:46 55 48:46 5😅5th 55 485555 48:46 5 48:46 5😅😅55:46 :46 :46 :46 :46 :46 :46 :46 :46 :46 :46 48:46 48:46 :46 :46 :46 :46 48:46 :46 :46 :46 :46 :46 :46 :46 5:46 :46 :46
Us professional paint users working in a government position cannot justify the extra 10$ on a 2000$ GPU to get a wider bus Thank you for your understanding
At least in Finland, the ryzen 7 9800x3d is fully out of stock from all of the major retailers (gigantti, proshop, verkkokauppa) and it's the same case for all the smaller stores. On most retailers' websites it shows, that restocks would be available at 7.1. 2025. I couldn't find any other stores, for example ones from UK shipping to Finland, that would have them in stock. It's possible that the CPU is completely out of stock in EU.
9800X3d is non-existent here in Europe. And even the out-of-stock price is around 650 euros. And I've checked every single retailer that does EU-wide shipping.
Continental Europe here, Italy. The 9800x3d on amazon is very hard to find, and usually it's from third party sellers with quite a markup. Even mobos and coolers are not always available, it depends on what you're looking for. GPUs are less of an issue. I've started using third party tech online stores, since I'm planning a new build. 5080: if they market it as a professional GPU, then it must have vram accordingly to the definition. 16 GB won't cut it for professionals. 24gb should be the minimum.
Exactly. Ngreedia has gone too far. Look at gamers defending the steaming pile of shit that the 4060, 4060 Ti and 4070 are. The 4060 is AD107, and 107 is ALWAYS 50 class silicon.
I don't see AMD being allowed to have a monopoly on X86 nor do I see Lisa Su wanting a company with 50 billion in debt. Now that Pat is out maybe they can get a new CEO to bring back royal core and right the ship but it is still going to be a rough ride.
Anybody internal or coming externally from an adjacent organization willing to slim the company down another 20% (especially middle management!) with an eye for disruptive innovation in the core cash cow products while being an "under promise and over deliver" execution freak is fine by me! 🤞
What's funny is most TH-camrs hyped up Pat as a savior for Intel since he's an engineering insider and wanted to get back to the foundry business being top of the line. What all the TH-camrs should now understand is that it takes MANY years to gain back leadership in the semi-conductor business, and even with CPUs if you rest on your laurels for a stretch of 3 - 5 years like Intel did you not only let your competition catch up (AMD) but they've probably been working very hard to catch you and that means their roadmap and work towards reaching it probably past you up. What's saving Intel right now are OEMs and the Intel faithful, but no matter what gets said about AMD in the laptop space it seems all I see for reviews are Intel based laptops.
Now that Pats gone, Intel is really fucked. Spent his entire term cleaning up the prior 2 or 3 CEOs mess, his contributions to the company wont bear fruit until 2028.
@ I have had to resort to a pre-order with no fixed delivery date Very frustrating as it’s the one missing component stopping me from completing my pc build
20:06 Intel battlemaga can win with the 5060 in VRAM size. 5060 will still use only 8GB and BTM offer 12GB. So even if 5060 is faster it will be choked by this 8GB VRAM limitation. And so 50-100$ or even more for 5060 16GB. It's all over 8GB problem.
If the 5060 only has 8gb that’s a fucking slap in the face 😭😂 stupid ass gamers buying Nvidia no matter what even though the 4060 is 50 class silicon (AD107)
@@flimermithrandir What we know is that both Nvidia and AMD will have 128bit cards in that segment which means they will be either stuck at 8GB or use a clamshelled designs to get to 16GB which adds ~$30 extra cost to the cards beyond the memory chips themself.
@@lharsay Well ok. They could use 3GB VRAM as well though. That would make it 12GB as. Depending on when they launch it. Maybe that is reserved for Super Cards too. Sure. Still dont know if AMD can use those GDDR 7 3GB Chips though (and if they can, when they have access to it, because i am pretty sure Nvidia gets it first).
The argument that cross platform comparisons aren't fair because "Apple can program to the metal while windows needs to be open" just doesn't make sense. That's part of the appeal of a vertically integrated company. It's not Apple's fault that Microsoft can't do that 🤷♂ What matters is real world performance (fps, render times, file copy speed, loading times, etc.) and not how they achieve it. Either PCs can compete with Apple and their in-house advantage or not. No excuses needed.
Pat is not bad, but not good enough either. Hopefully a better CEO steps up and rework the whole Intel which looks likes now a typical government agency: expensive, bloated and underdelivering
If the 5080 is expensive because it is meant to be sold to "professionals", then why is it a part of the gaming lineup? I won't buy that marketing line and I hope the retailers will be responsible in their messaging. But it is probably in their best interest to upsell every customer they can. That is one of the nice parts of online shopping. You don't get talked into buying overpriced products, or prebuilt desktops with an i7 and a gimped 4060
I found your channel last year at the start of September and have loved keeping up with it since. I think one of my favorite things to do is just let these podcasts run in the background while I cook or do some mindless admin. Keep up the great work everyone!
What annoys me is USB-C is not just about VR, its very useful for monitors that have a KVM switch so you only need a single cable back to the PC, or just the sheer fact its neater to cable manage.
Im probably going to buy a couple of b580's if they are even close to 4060 performance. The reason is, the more intel sells these cards, the more competitive and cheaper the market will be.
I'd consider it. But lpddr5 is fast enough for everything, (depending on bus width), so... maybe just get a mac m3pro or amd strix(if you can find one). Unfortunately, everything that comes with a good APU and 64gigs of ram is either exorbitantly priced or impossible to buy.
@@btudrusPlenty of workloads that use 24 CPUs do not need a metric ton of IO. If there is a major uplift in pcie.. expect to see platform costs go up quite a bit.
Honestly Qualcomm should be playing to their strengths and work with Google to make Android Desktop or ChromeOS competitive with iPadOS & Windows and not some half-assed frustratingly lackluster experience with Microsoft.
Me: oh man it's past 12, I gotta sleep. I have to wake up at 6 tomorrow for work. MLID *Zen 6 CCDs , 5090, and Strix Halo video drops* Me: sleep is for the weak!
I do not use quick resume and I do not use sleep with my series x and it has been a fantastic console. Zero crashes. I believe the crashes are from over heating consoles and people never rebooting the console because it goes to sleep and tries to use quick resume.
Been using quick resume frequently on a refurb series X and I’ve had no problems. I do reboot my system but quick resume hasn’t bugged enough for me to disable it. These issues some people have had are so strange.
AMD already works with Mediatek for its wifi chips, so they can license MTK modems. ARM is providing complete solutions for partners who want them - ISPs, NPUs, etc, so it wouldn't be that much of a hassle for AMD to make a smartphone chip if they really wanted to do it.
i have only xbox series x and i can say i have never a single issue with it idk what problem with this guy unit but mine is super steady and never crash a game 2021 manufacturing
Have you two checked what kind of profits AMD is making off of low cost RDNA 2 and Zen 3? This is just a WILD guess but I have to imagine they're doing pretty well considering that's made off of TSMC N7 for which AMD is probably the main customer and I imagine TSMC has been giving them a really good price on it for the last 2 years or I don't think AMD would have been so willing to keep producing RDNA 2/Zen 3 as low cost options. I don't know this as a fact but when AMD announced these would be low cost options they had probably already worked out an agreement with TSMC since N7 by that time was no longer a cutting edge node. I mean right now we're at 2nd gen N3 for TSMC which is supposed to be an excellent node for TSMC AND their customers.
Around 11:30 two biggest in the UK are SCAN then Overclockers then there about 5 smaller ones I would normally buy from 1 of these 7 and only use amazon if your some reason they had better availability. On day one I checked Overclockers the sold out the 9800x3d and they had about 250 of them. It was the best selling CPU on that site by far and I suspect SCAN was in a similar position. Oddly the 3rd best Overclockers CPU is the 5900x because its such good value I bought it over the 5800x3d a year ago.
After the leak about Strix Halo "LP" TDP i'm worried about all the Strix Halo lineup efficiency. AMD is less efficient then Intel in mixed workloads with chiplets on desktop, where the E-cores can do what they are designed for, and Strix Point with 4 Zen 5 and 8 Zen 5c cores is already less efficient then Lunar Lake in a simple workload like office productivity and web browsing, let alone the comparison to any Arm counterparts. It would be a real let down if the situation with Strix Halo will be even worse then that. I hope i'm wrong and AMD actually found a way to make a mobile Soc with chiplets more efficient then their current monolithic design.
My guess is that idle power will be worse than monolithic SoCs but quite a bit better than desktop Zen 5. For that reason ST power draw will probably be relatively high as well but MT will be comparable to Strix. Unfortunately this really isn't a thin and light SoC and even less so for handhelds. I do expect that the 105W TDP for Strix Halo LP is probably the up-config and there will be laptop models utilizing it that have more sensible TDP configurations
@@2intheampm512 i think the 105W TDP, just like the leaked 120W Strix Halo 16 cores and 40 CUs inside the next Asus 13" laptop, include the total power consumption of the machine with the memory included.If i'm not mistaken the CPU was rated at 30W, 75W for the IOD and the rest for the memory, from 32 to 128 GB of Ram (obviously more ram more power required). If Strix Halo will be worse then Strix Point at efficiency i will consider it a "failure", because you're not always pushing the Soc to the maximum inside a laptop, and if performances are the same of a core ultra 200 + an RTX 4070 with the same power consumption, is not really a win if you use more energy for single core tasks and, consequently, you have the same of even worse battery life (with the same battery capacity). Also considering that we will have rtx 5000 soon after Strix Halo with presumably better efficiency then rtx 4000, idk, the more we get close to Strix Halo the more sounds a bit DOA at the beginning of 2025 as a laptop Apu. I hope i'm wrong since i need to upgrade my laptop and i don't know what to do 😂
I sometimes use ebuyer or CCL as well at a pinch but I much prefer Amazon - cheaper postage and easier to return faulty items. I had a nightmare with OC once where I returned a faulty graphics card and they sent it back to me and said it worked fine. I sent it back again with a letter explaining why it was broken and it took them nearly 6 weeks to send me a replacement during which time I couldn't use my PC. It had faulty VRAM which worked fine on windows at the desktop but if you launched a game it immediately started artifacting and then crashed after about 30 seconds. They said it passed all their tests. With Amazon I just call them and they give me a refund before I have even posted the item back a lot of the time. ebuyer use Yodel to deliver who are awful. Scan don't have live stock levels on their website so you can place an order of something it says is in stock but by the time you check out it is out of stock then you have to wait for it to come back into stock which can take weeks. Had that happen too many times. The only reason I use anybody other than Amazon is if I can't get the product I want or something just like it, eg a RAM kit that is on the QVL.
I'm surprised there are no comparisons with AMD during the bulldozer days. AMD got out of it with far fewer resources than Intel. AMD stock was $2 during that time...
People saying that an Intel 4060ti with 12gb will sell well at 250$ are out of their minds. Remember that the 4060ti was equivalent to the 3060ti, and that card was 400$ in 2020. Getting a 150$ discount 4 years later with inferior drivers and software compatibility will NOT sell well:
What else can you get under $300 with more than 8GB VRAM? It basically just needs to beat the RTX3060 12GB by 30% (including RT and upscaling) and it will have it's place on the market.
It will be te best selling GPU in the EU because other options suck. This time you wont need to emulate DX11 and 9 which was a hardware fault with the Alchemist line. Intel already unfucked their drivers.
As someone who actually used an alchemist card for a while intel still is significantly behind in software. Some games don't work at all. So any intel card above 200 bucks is wasted, better get a red or green card instead if you want to spend more than 200.
I bought an A380 (on the second or third wave!) because I like collecting weird hardware and I wanted to see how the architecture performed... in a secondary PC, of course. I haven't run into anything that just straight does not work on recent drivers, but it took them over a year and SO MANY DRIVERS (including TWO that broke my system!) to... not actually improve performance very much? DX9 results are "less deficient" now, but DX11 and DX12 have actually regressed according to my notes. They're margin of error kind of losses, but everyone was saying how Arc cards were gonna go up a tier as they aged, and that sure didn't happen. (Arc fans have insisted to me that ACM-G10 did, but I have doubts.) In a best case scenario where the A380 is paired with a fast Zen 2 CPU with PCIe 4.0 and ReBar and dual channel memory, it still can't match the RX 5500XT I had lying around, and it suffers a greater bottleneck than the AMD card does when things start to get less and less optimal.
@@kazuviking Alchemist started off using the D3d9on12 wrapper in Windows before moving to using DXVK within the driver, which was a lot better. I'm seeing that they added some optional instructions for DX12 that should help, but I'm not seeing any indication they've ditched API translation via DXVK in favor of purely native DX9, though. And the state of API translation isn't relevant to DX11/12/Vulkan issues.
When you have over 85% of the market you can effectively act like your competitors don't exist until the market proves otherwise. Radeon 7000 vs RTX 4000 already had multiple price segments with slam dunk AMD wins, and it didn't result in mass exodus from NVIDIA cards. If there was a 4070 Ti available for $600 tomorrow it would sell out in a day. That's just a fact. So a 5070 for $600 makes complete sense since they should only price their products against their own offerings. People upgrading from a 3070 will be pleased with the boost they get, and older holdouts (think 1080 Ti/2070 Super) will see a massive performance leap. AMD unfortunately has to go above and beyond to change this reality.
I don't think they'll make a 6 core Zen 6 CPU, just like how they stopped producing 4 core chiplets a number of years ago. I think they'll use previous gen products for budget CPU's again. That strategy worked well so far.
I agree, I don't think yields will be bad enough to need to use 50% disabled cores, at least not for a while then they'll only have enough for a Microcenter exclusive or something.
I was thinking about zen 6 desktop segmentation with 12 core ccd's right when you ended up theorizing about it yourselves. Rewatching that segment, you split it up into: - 10500 6c (1 ccd -50%) - 10600 8c (1 ccd -33%) - 10700 10c (1 ccd -17%) - 10800x3d 10c (1 ccd -17%) - 10900x3d 12c (1 ccd full) - 10950 20c (2 ccd -17%) - 10960 24c (2 ccd full) Considering we're already seeing multiple full single ccd mainstream models (700/800) in all recent gens, I doubt they'd cut down the lineup this way. I'd expect more something like this: - 10500 6c (1 ccd -50%) - 10600 8c (1 ccd -33%) - 10700 12c (1 ccd full) - 10800x3d 12c (1 ccd full) - 10900 18c (2 ccd -25%) - 10950 24c (2 ccd full) This does assume there is also threadripper with better IO along the gen. But I doubt they'd have to cut down so many dies that it would benefit them to sell them in every of their main volume sku's (600 to 800). Also going 6 - 8 - 10 - 12 is a really crowded lineup, and then the gap from 12 to 20 would be kinda weird, with the 20 basically making little sense for people looking for that multi-core performance since there's a 24-core right next to it, meaning whichever offers better value would cannibalize the other - and I'm leaning to the 24-core winning that one. Instead I'd expect cut down yields to go into 500 and 600, with no need to cut down 700 and 800, with the 800 being just the better binned x3d variant of the 700 just like we see in zen 5. Then, to offer a more sensible cut down, cheaper alternative, they could dump some relatively garbage yields into a 900 with 18 cores, and offer it as this better than previous gen 950 'budget' multi-core value king, with the 24 core still offering a decent step up as the true desktop multi-core king. For pricing I'd guess from 500 to 950: $200 - $300 - $450/480 - $550 - $650 - $900. Maybe an 8-core x3d variant at $400 as a more budget second-to-the-gaming-crown full 12-core $550 800x3d sku later in the generation, with the 500 also probably later in the generation to build up inventory of ccd's that actually need to be cut down 50%. Either that or they'd employ zen 5+ as the zen 6 low end. But who knows, without real intel competition maybe they'll milk hard, price 10%-20% higher than the numbers above even to truly get rid of zen 4 and zen 5 stock as a budget filler for this lineup and only dropping prices when they've ran out. The previous broken silicon with James Prior shone some light on how these higher-than-expected pricing structures come to be, would be in line with that
I’ve bought both alcemist founders editions for Intel, one copy of all sparkle gpu cards, & one pc with a i5 12400 and a core ultra 265k. Big fan. Hoping they succeed and definitely will be buying battlemage 💯💯
Tom, remember Intel can sell "their" x86 license to whoever they want. However, it will only get you 8, 16, & 32bit x86, sure you also get Intel _64 AKA Itanium(a failure). Intel does not own the rights to "modern" x86 because it's 64bit AKA AMD_64. When cash strapped AMD couldn't sell a license for AMD_64 because it's built on x86 so legally Intel gets the final OK. Likewise AMD gets the final say of any license deal using AMD_64 AKA any modern x86. Sure Nvidia could buy a license from Intel but they would be limited to the unsupported and failed Itanium Intel version of 64bit x86 or be limited to 32bir or smaller.
The RX 8800XT insanity has to stop! Cherrypicking the game with the *weakest* RT implementation as some kind of "comparison" with a 4080 is ludicrous. CP2077, Hogwarts Legacy, Plague Tale Requiem and others have the 7800XT at abysmally slow, unplayable, framerates (typically _worse_ than a 3070!), a "40%" uplift would still be unplayable at 1440p and _maybe_ as good as a 4070 non-super! The 8800XT will *NOT* be anywhere close to a 4080 in games that actually *use* RT in any meaningful way. Also, the 8800XT's raster performance will be probably about 10%-15% _worse_ than a 7900XT (not XTX!), meaning it will be a decent generational uplift, not a miracle!
The Switch 2 had to wait at least 2 years to launch, they don't want to accidently have modern tech in their console. They have to be at least a couple generations old.
Your arguments about battlemage are valid, as long as Nvidia and AMD offer 250 euro cards to compete. I mean that is the whole point, green and red don't supply for the entry/mainstream market. That group of buyers doesn't care if they buy a 6 nm, 4 nm or 3 nm die. The only thing that counts is fps per Dollar. And I'm happy about Intel releasing a 250 euro card. been gaming for more than 20 years and I couldn't care less about 800 Euro cards for shitty and repetitive AAA games in 4K. Doesn't make the games any better for all I care and the time to play is very little as an adult, so small cards are again where the magic happens.
It’s crazy b580 supposed to be 250 comes with 12 gb but nvidia 300 dollar 4060 but it supposed to be as good or better everywhere in recent release data. We can’t get anywhere in reducing overall prices without competition.
The Zen 6 lineup should have more pcie lanes at that point where you speculated/imagine the Ryzen 9s would be. Those are pretty much in-between high end mainstream and workstation-level. X970/E motherboards may have two 16x PCIe 5.0 lanes or more lanes dedicated for storage. If I am paying nearly $1000 for 20-24 core CPUs, might as well throw in some more pcie lanes.
I still think Console Makers should make new "Gens" and every Gen should last for 5 Years now, each while they need to support the old Gen still. Until 10 Years at least, then Companys are free to do whatever. It would help everyone.
maybe the zen 6 24 core could be reserved for the "AM6" socket (with ddr6) while AM5 only having 20cores zen6 with some higher ddr5 sweet spot than with ZEN5?
I can see an AMD future with more *different* cores, the low hanging fruit would be 8 "full fat" cores, then several "compact" cores like the "Zen-C" chiplets. if you have an application or workload that can utilise threads and is highly parallelised, then it would be unwise to only have full fat cores over the compact ones, for workstation and enthusiast use. For server I can see having only full fat or full compact SKUs because they partition the cores into virtualised machines, Threadripper though I see being better as a mix, as a workstation. The kinda low hanging fruit after this would be accelerators, like video encoder and decoders, file compression, vector/matrix/graph compute, neural compute e.t.c. The not so low hanging fruit would be to have other architectures like RiscV chiplets (for specialised cases), and FPGA chiplets (from AMD Xilinx) for having a "application as hardware" cases.
Regarding Apple M-chips, my prediction for the future headlines around Apple M5 and M6 chips are: "Apple's new M-chip architecture has no more low hanging fruits" :) I haven't heard anything regarding of a total redesign for any future M-chips, which I guess would be needed in the long run? I mean you can only iterate so far on same fundamental design.
I wonder if AMD could put Vcache or something like it on a GPU. With how popular the X3D CPUs are it might make Radeon more interesting for some people who wouldn't buy AMD GPUs otherwise.
@@adeptuspotatocus6451 be better for marketing to use infinity cache and brand it as vcache lol. Imagine the 9800XT3D or something. If they were smart it'd be really be 9900XT level but by dropping a tier it'd look like it had brutal performance
The current sweet spot is defined by the memory controller on the IO Die that is from Zen 4 and reused on Zen 5. Any further AM5 chips are likely to have an updated IO die on a better node, and would therefore have a new sweet spot for memory speed/timings. The 8000 series can actually run higher controller and fabric speeds but no one is using those chips for peak performance since they have cut down cache and PCIe
B580 should be $250 now but I expect it to drop down to 199 dollars anytime soon and that would be a good price for something that beats RTX 4060 while making a huge RT performance
The issue with the IO die is the link between the cores and io die if we can either clock the fabric to like 4GHz along with a better memory controller on a io die the performance would come back. Or they could just widen the link but that sounds expensive
if they moved past the fabric through the organic substrate and started putting it all on a silicon interposer or move past the zen2 configuration alltogether and do something like stacking the CCDs on the IO die - then widening the fabric should be easily doable. Also would reduce trace lengths which makes lower die-to-die latencies possible.
Overclockers UK is where I got my 9800x3D, on the first batch of deliveries ... And they were the only retailer/e-tailer that managed to keep it at the suggested MSRP from AMD £449, all the other ones said they couldn't do it as it was little to no margin at that price so they were all at or above £469
Matrix hardware support for XeSS2 Supersampling, frame insertion and the new XeLL latency improvement, all focused to provide 1440 gaming. Looks like Intel has a good plan for Battlemage. They are built on TSM N5.
@@MooresLawIsDead Looks like Battlemage has caught up with several AMD and NVDA GPU features. Pretty impressive update in one generation ... improvements in ray tracing, upscaling, frame rates, latency, ai processing performance. Perhaps in another year any differences in gaming features will be so-what, and all anyone will care about is the support for the latest AI models.
Tarrifs are AMAZING and they're purpose is NOT for competitive reasons contrary to popular belief. That's not a benefit they have I 💯 agree. But that's the publicly stated reason, because it's popularly believed.
I still don't understand why intel cancelled the rentable unit single thread beast project. And then released the wet fart of what arrow lake was. They could have just grabbed the gaming crown and sold like crazy like 9800x3d
Lunar Lake for laptop is good, handles 95% of use cases, the only downside I see is the limited number of cores A 16c/16t have the potential to beat Apple and AMD in performance, 8c/8t Lunar Lake is OK but does not make you excited
![ILLSLICK X YOUNGOHM - We're The same [Official Video]](http://i.ytimg.com/vi/yUJNkobYR0E/mqdefault.jpg)
[SPON: Use "brokensilicon“ at CDKeyOffer’s Black Friday Sale to get Win 11 Pro for $23: www.cdkeyoffer.com/cko/Moore11 ]
[SPON: Use “brokensilicon” to get $30 OFF the Minisforum V3 3-in-1 Tablet: shrsl.com/4rt3x ]
Pretty sure 25 percent of people buy amd cards
Guessing Nvidia will announce their new GPUs at 4am day of ces Amd at CES
Dan looks better than usual in this episode.
Was just thinking Nvidia's strategy with branding the 5080 and 5090 as professional cards might be them trying to segment their product stack into HEDT in the same vein as Threadripper. Allows them to charge obscenely more money for a product while not needing to provide the same level commitment and/or support as server grade. I guess technically the Titan was that card back in the day, which begs the question why they don't just return back to the Titan naming scheme, but I guess that's what think tanks are for.
you are really gonna die on that hill with your views on intel arc eh?
Hour or two ago Pat resigned/retired/fired from Intel CEO position. Suprise stream for that :p?
Wait really? lol
@@cosmic_gate476 "Retired". Make of that what you will.
@@cosmic_gate476 Yep.
maybe apple should buy them afterall
Does this mean royal core may come back???
I personally would like Intel to keep making discrete GPUs, and I believe they DO have a place in the low to mid tier market even nowadays
I want Intel to not go out of business. Right now they are selling something that seemingly loses to the 7600 XT despite having a 33% larger die size and a MUCH more expensive node. They are losing money on each of these cards sold, and it is fully unsustainable.
It's no coincidence that Pat was sacked the same week...this was his vanity project after all...
@@MooresLawIsDead but what isn't a vanity project for Intel right now? They aren't really competitive anywhere in any department are they?
@@Speak_Out_and_Remove_All_Doubt they need to bet their company on whatever comes after the P and E cores. Just like AMD bet theirs on Zen. It has to happen otherwise it's certain death.
@@xpk0228 tbf Nova Lake which is still P & E sounds decent if they can deliver on it, 16p cores with a massive IPC uplift plus 32e cores would probably win against Zen 6 but they need to execute this well and no one thinks they can.
As for the next thing after the p&e core, they need Jim Keller back for that!
@@Speak_Out_and_Remove_All_Doubt I am not sure NVL has that much cores tho. Tom said 16P+32E, but that is before royal core and rentable unit was cancelled. I think it might be just 8+32 like ARL was planned to be.
Gelsinger yeeted
Still can't believe people think intel is "doing fine"
Intel just got $7.86 billion from the federal government. It’s too big to fail. Right?😂
@@tringuyen7519no they didnt
@@tringuyen7519Whos gonna make uncle sam's military chips? Taiwan? God forbid 😂
Which people? Imaginary?
@ nahh lol I’ve seen plenty of people online saying “Intel is too big to fail, they’ll be fine”
Now there are two of them! This is getting out of hand!
Always two, there are. No more. No less. A Master and an apprentice.
I 5556t 48:46 r5s t 48:4 4 8th 6t55555556566 48:46 666656666656665665556555565555r|to 48:the other day 48:I T4 48:4 5 48:4 5 48:4 5 48:4 5 48:4 5th of the 5 48:46 I 4444 5 48:46 48:but I 48:4 5 48:4 5 5 48:46 I 5 48:46 but I 5 48:46 I 5 48:46 but I 5 48:46 5 5 48:46 I 5 48:46 but I don't think 5 48:46 to 48:4 5 5 48:46 and to 5 48:46 and 5 48:46 and to 5 48:46 and then 5 48:46 to 5 48:46 and to 5 48:46 and then to 5 48:46 and 5 48:46 and 48:to 5 48:46 r|to be 5 48:46 and then 5 48:46 rrr4 5 48:46 and 5 48:46 to 5 48:46 and then 5 48:46 and 5 48:46 and 5r 48:46 rrr4 5 48:46 and 5 48:46 and 5 48:46 and I 48:and 5 48:46 and I 48:4 6 5 rt 5 48:46 to 48:4 5 5 5 48:46 I 5 48:46 5 48:5 5 5 48:46 I think 5 48:46 is going to 48:4 5 5 48:46 4 5 5 5 48:46 and 5 48:46 66 5 48:46 I 5 48:46 but 5 48:46 to 5 48:46 and 48:to 48:46 55r5 48:46 and I 5 48:46 5 48:46 5 48:46 I 5 48:4 4 48:4 5 55 485 48:46 5r 48:46 4 5th to 5th 5 48:46 to 5 48:46 and 5 48:46 and 5 48:46 and to 5 48:46 and 48:to the same time 48:46 55 48:46 and 5 48:46 and 5 48:46 to 48:the 56 485 48:46 and to the 48:to 5 48:46 and the 5 48:46 is to 48:4 5 5 48:46 and 5 48:46 and the 5 48:46 two of 5 48:46 5 48:and to 5 48:46 to 5 48:46 and then 5 48:46 48:4 5 55 48:46 and to the 5 48:46 and to 48:4 5 5 48:46 and then 5 48:46 and 56 48:4 5 48:4 and I 5 48:46 and 55 48:46 and I 5 48:5 85 485 48:46 55 4855 48:46 I 48:4 5 5 48:46 4 5 5 48:I rr to the 48:4 5th 5 48:46 to the same 5 48:46 I 5 48:46 but I 5 48:46 I 48:4 5 I 66 5 48:46 to 5 48:46 and 5 48:46 and 48:46 55 48:46 and 48:and I 5 48:46 and 48:to 5 48:46 and 5 48:46 and 5 48:46 and to be honest 5 48:46 I 48:5 48:4 5th of the 5th of the 5 48:46 I 48:4 5th 5 48:46 I 48:4 5 5 48:46 I 5 48:46 think 5 48:46 is 5 48:46 to the other rrr4 5 5 48:46 rrr4 5 5 48:and 48:to 5 48:46 the other day 5 48:46 and 48:46 55 48:46 4 66 48:I 5 48:46 to 48:46 55 48:and 5 48:46 T4 5 48:4 5 55 48:46 I 5 48:46 but I 48:46 55r5 48:46 I 5 48:4 I 48:think 5 48:46 is going to be a 5 48:46 but to 5 48:46 and to 55 48:46 66 48:66 5 5 48:46 4 5 I 5 48:46 but try 48:46 5 48:and 55 48:46 5th of the 5 48:46 r55 48:46 and 5 48:46 to 5 48:46 55 48:46 and I 5 have 48:4 5 5 48:46 and 5 48:46 and 5 48:46 to 5 48:46 and 48:4 5 and 48:4 5 48:46 55 48:46 and r 48:46 5 48:46 I 5 48:46 but I r95 485 48:46 I 5 48:46 48:to 5 48:46 the two 48:and the 5 48:46 two of 5 48:46 in the 48:46 to 5 48:46 5 48:46 48:4 66 5 48:46 5 48:46 4 66 5 5 48:46 48:4 5 I 5 5 48:4 5th 48:4 5 5 48:46 I 5 48:46 48:4 6 try 5 48:46 and then 55 48 48:46 48:46 55 48:46 4 5 48:46 55 48:46 48:4 5 48:4 5 48:4 5 48:I think 555 48:46 is 5 48:46 to 5th 5 48:46 and 5 48:46 485 48:46 5 48:46 and 5 48:46 and 48:to 5 48:46 5 48:46 and I 5 48:46 and 48:4 5 5th of the month 5 48:46 to 48:46 55 48:46 4 6 and I 48:5 5 48:4 5th 48:46 I 5 48:46 48:I 55 48:46 66 5 48:46 I 5 48:46 but I 5 48:46 55 48:46 rrr4 48:46 5 48:46 55 48:46 4 66 I 5 48:46 but I 48:4 5 I 5 48:46 5 I 48:4 5 I 5 48:46 but I 48:4 5 5 48:46 4 5 5 I 5 48:4 66 5 48:46 to 48:4 5 and 5 48:46 and 5 48:46 and 48:46 5rr|and I think they 5 48:46 48:46 55 48:46 to be honest to 48:46 48:4 5th 5 48:46 rt 5 48:46 I 5 48:46 |5th 5 48:46 48:4 5 I 5 48:46 48:4 5 5 48:46 4 5 5 5 48:4 5 5 48:46 I 48:46 5rr55 48:46 5 48:46 to Tytherington and 48:4 6 5th 48:46 rr5 48:46 5 48:46 I 48:46 5485 48:46 I 5 48:46 |5th r 48:I 48:46 to r5s 4 r5s 5 48:46 to 5 48:46 5th 5rr 48:46 I 5rr||rr5 48:46 r5 48:46 I 48:46 5r|5th of the 5r 48:46 5th to 5 48:46 to 5rr55r5r55rr5 48:46 48:46 5 485 48:46 and 48:4 5th of the 5r 48:4 4 8th 48:4 5 5 48:46 4 5 5 48:46 I 48:46 5r5 48:46 48:46 5 48:4 5th 48:46 55 485| 48:46 r 5r 48:46 48:46 555 485 485r 48:46 55 48:46 5 48:46 48:46 5 48:4 the 48:46 54855 48:46 5 48:46 55 485 485 48:46 5 48:46 48:46 I think 5 48:46 the 48:46 5 48:46 is to 5 48:46 and 48:and 48:4 5 and 5 48:46 and 5 48:46 48:4 5 5 48:46 48:46 55 48:46 I 48:4 5 I 48:4 5 55 48 48:46 55 48:46 I 5 48:46 but I 48:4 5 I 5 48:46 48:4 5th 5 48:46 55 48:46 5😅5th 55 485555 48:46 5 48:46 5😅😅55:46 :46 :46 :46 :46 :46 :46 :46 :46 :46 :46 48:46 48:46 :46 :46 :46 :46 48:46 :46 :46 :46 :46 :46 :46 :46 5:46 :46 :46
Actual productive employees get 15% job cuts
Upper management gets 100% job creation
Business as usual
Source - CNBC:
Intel CEO Pat Gelsinger ousted by board after disastrous underperformance
Published Mon, Dec 2 20248:34 AM ESTUpdated 30 Min Ago
Watch the 5080 be marketed as a "1440p gpu"...
We are well and truly cooked
We would be cooked.. but I doubt though
1440p 240hz maybe lmfao
3080 was marketed for 4k and capable of 8k, 3090 was an "8k" gpu. 🤣
@@Dante_S550_Turbo The 980 Ti was marketed as 4k capable.
What is wrong with 1440p? 1080p film looks millions times better than 4k video games.
Ah yes. Those mythical professionals that need a narrow bus, and only 16gb of VRAM.
what!!! AM Professional nvidia fanboy i need 8gb vram on 64bit'
“Pro” means above $1k
No self respecting professional would want to pay less than $1k for their gear. That wouldn’t be pro enough.
Us professional paint users working in a government position cannot justify the extra 10$ on a 2000$ GPU to get a wider bus
Thank you for your understanding
Im gonna be using my 3080 12G and 7950X for the next FIVE YEARS because i have like 200 games in the backlog released any where from 2008 to 2021 lol
I keep finding I don't have enough time to finish the games I already own lol
I keep buying games as a form of wishful thinking
At least in Finland, the ryzen 7 9800x3d is fully out of stock from all of the major retailers (gigantti, proshop, verkkokauppa) and it's the same case for all the smaller stores. On most retailers' websites it shows, that restocks would be available at 7.1. 2025. I couldn't find any other stores, for example ones from UK shipping to Finland, that would have them in stock. It's possible that the CPU is completely out of stock in EU.
Same in Sweden, we have 1 store that states 9 dec, 1 states 18 Dec, proshop states 7 Jan and the rest are unverified
Paperlaunch
In my local Microcenter in California they get in regular shipments of 500 to 600 9800X3D's and they sell out in 2 days!
@@GenX_in_the_wild something selling out doesn't make it a paper launch.
Same in Denmark, i saw one store estimating availability in Febuary
9800X3d is non-existent here in Europe. And even the out-of-stock price is around 650 euros. And I've checked every single retailer that does EU-wide shipping.
800€ if you want one new on-line...
😂 just wait for January GPUs.
just wait if you want it, absolute madness
I can find it for 550~€ in France tho ?
Its in stock in hungary for 605€ BUT its price started at 565€.
Continental Europe here, Italy. The 9800x3d on amazon is very hard to find, and usually it's from third party sellers with quite a markup.
Even mobos and coolers are not always available, it depends on what you're looking for.
GPUs are less of an issue.
I've started using third party tech online stores, since I'm planning a new build.
5080: if they market it as a professional GPU, then it must have vram accordingly to the definition. 16 GB won't cut it for professionals. 24gb should be the minimum.
Exactly. Ngreedia has gone too far. Look at gamers defending the steaming pile of shit that the 4060, 4060 Ti and 4070 are. The 4060 is AD107, and 107 is ALWAYS 50 class silicon.
@@morpheus_9yet they always end up buying the 4050... I mean the 4060 in droves. We get what we fking deserve 😂
@@AthamAldecua I didn’t buy that steaming pile of shit, I got a 6750xt for $280
I don't see AMD being allowed to have a monopoly on X86 nor do I see Lisa Su wanting a company with 50 billion in debt. Now that Pat is out maybe they can get a new CEO to bring back royal core and right the ship but it is still going to be a rough ride.
Moore's Law is UnDead DLC
The timing of this is even more odd than Pats "retirement" :)
There's only one man that can bring Intel back, Jim 'the god' Keller!
While him being CEO of Intel would be great. He is busy with his own company Tenstorrent.
@@theemperorofmankind3739& Tenstorrrent is 10x better than Intel. No politics. Just RISC-V research, development, & innovations.
Anybody internal or coming externally from an adjacent organization willing to slim the company down another 20% (especially middle management!) with an eye for disruptive innovation in the core cash cow products while being an "under promise and over deliver" execution freak is fine by me! 🤞
@@theemperorofmankind3739 Maybe a merger?
@@tringuyen7519 He can probably do both.
Excited to see a *potential* 12core, single CCD, X3D, chip
What's funny is most TH-camrs hyped up Pat as a savior for Intel since he's an engineering insider and wanted to get back to the foundry business being top of the line.
What all the TH-camrs should now understand is that it takes MANY years to gain back leadership in the semi-conductor business, and even with CPUs if you rest on your laurels for a stretch of 3 - 5 years like Intel did you not only let your competition catch up (AMD) but they've probably been working very hard to catch you and that means their roadmap and work towards reaching it probably past you up.
What's saving Intel right now are OEMs and the Intel faithful, but no matter what gets said about AMD in the laptop space it seems all I see for reviews are Intel based laptops.
Now that Pats gone, Intel is really fucked. Spent his entire term cleaning up the prior 2 or 3 CEOs mess, his contributions to the company wont bear fruit until 2028.
His 'contributions' include selling broken CPUs to customers and staying silent while they flounder around with instability.
Cut the BS
It’s been incredibly difficult to get a 9800x3d at any retailer in the uk
In the world. 7800X3D launch was far better.
@ I have had to resort to a pre-order with no fixed delivery date
Very frustrating as it’s the one missing component stopping me from completing my pc build
@@Protector1rkAMD is just trying to build up US inventory before Trump’s tariffs are enacted.
Nah,it's everywhere in the US.
Amazon, Neweegg microcenter literally can buy it anywhere
I got an email that OCUK recently got stock but they're selling it for higher than MSRP. I think I'll just wait.
20:06 Intel battlemaga can win with the 5060 in VRAM size. 5060 will still use only 8GB and BTM offer 12GB. So even if 5060 is faster it will be choked by this 8GB VRAM limitation. And so 50-100$ or even more for 5060 16GB. It's all over 8GB problem.
Yeah, i can see that. But i dont know about AMD. (Actually just dont know)
If the 5060 only has 8gb that’s a fucking slap in the face 😭😂 stupid ass gamers buying Nvidia no matter what even though the 4060 is 50 class silicon (AD107)
@@flimermithrandir What we know is that both Nvidia and AMD will have 128bit cards in that segment which means they will be either stuck at 8GB or use a clamshelled designs to get to 16GB which adds ~$30 extra cost to the cards beyond the memory chips themself.
@@lharsay Well ok. They could use 3GB VRAM as well though. That would make it 12GB as. Depending on when they launch it. Maybe that is reserved for Super Cards too. Sure. Still dont know if AMD can use those GDDR 7 3GB Chips though (and if they can, when they have access to it, because i am pretty sure Nvidia gets it first).
@@flimermithrandir I think RDNA4 is limited to G6 and the RTX5060 will also likely launch with G6.
There is still so much Ampere left to find it's unbelievable.
The argument that cross platform comparisons aren't fair because "Apple can program to the metal while windows needs to be open" just doesn't make sense. That's part of the appeal of a vertically integrated company. It's not Apple's fault that Microsoft can't do that 🤷♂ What matters is real world performance (fps, render times, file copy speed, loading times, etc.) and not how they achieve it. Either PCs can compete with Apple and their in-house advantage or not. No excuses needed.
Great to see you both together in the same room. This proves that Dan is not an AI generated vtuber
Won`t 10 000 series be APU`s as usual like the 6000/8000
And 11 000 actually the 12 core per CCD Zen 6 ?
Great to see you guys in the same room for once! 😁
Are You Goin live to talk about Pat? Have you heard anything beyond the official statement?
is pat leaving ?😮
@@_yuriLeft... official as of yesterday. Ian Cuttress went live addressing it earlier today.
Pat is not bad, but not good enough either. Hopefully a better CEO steps up and rework the whole Intel which looks likes now a typical government agency: expensive, bloated and underdelivering
If the 5080 is expensive because it is meant to be sold to "professionals", then why is it a part of the gaming lineup? I won't buy that marketing line and I hope the retailers will be responsible in their messaging. But it is probably in their best interest to upsell every customer they can.
That is one of the nice parts of online shopping. You don't get talked into buying overpriced products, or prebuilt desktops with an i7 and a gimped 4060
And Pat Leaving
more like pushed out but yea.....not expected.
@@olo398 Technically we don't know that, but yes, it does look this way.
I found your channel last year at the start of September and have loved keeping up with it since. I think one of my favorite things to do is just let these podcasts run in the background while I cook or do some mindless admin. Keep up the great work everyone!
What annoys me is USB-C is not just about VR, its very useful for monitors that have a KVM switch so you only need a single cable back to the PC, or just the sheer fact its neater to cable manage.
Ohhhh, shit, the Broken Sili-bros in the same room?!?! Noice.
Im probably going to buy a couple of b580's if they are even close to 4060 performance.
The reason is, the more intel sells these cards, the more competitive and cheaper the market will be.
One day we will have our true HBM APU. Basically, the consumer laptop version of the MI 300X.
I'd consider it. But lpddr5 is fast enough for everything, (depending on bus width), so... maybe just get a mac m3pro or amd strix(if you can find one). Unfortunately, everything that comes with a good APU and 64gigs of ram is either exorbitantly priced or impossible to buy.
go, Intel, go! we need competition
Whoa 24C/48Th Zen6 sounds delicious. Wonder how high the clocks would be able to boost all cores? for $799 it's gonna be a steal.
As long as there will be too little PCIe lanes in AM5 it is all trash...
@@btudrusPlenty of workloads that use 24 CPUs do not need a metric ton of IO. If there is a major uplift in pcie.. expect to see platform costs go up quite a bit.
Honestly Qualcomm should be playing to their strengths and work with Google to make Android Desktop or ChromeOS competitive with iPadOS & Windows and not some half-assed frustratingly lackluster experience with Microsoft.
Andoid is a way more usable desktop OS versus IOS... since Android Honeycomb (version 4.0, we're at version 13 now).
Pat has left Intel 🎉
Hopefully intel will start to make Good CPUs.
Me: oh man it's past 12, I gotta sleep. I have to wake up at 6 tomorrow for work.
MLID *Zen 6 CCDs , 5090, and Strix Halo video drops*
Me: sleep is for the weak!
I do not use quick resume and I do not use sleep with my series x and it has been a fantastic console. Zero crashes. I believe the crashes are from over heating consoles and people never rebooting the console because it goes to sleep and tries to use quick resume.
Been using quick resume frequently on a refurb series X and I’ve had no problems. I do reboot my system but quick resume hasn’t bugged enough for me to disable it. These issues some people have had are so strange.
quick resume is a cool feature tho. I wonder if we can make it work on PC.
ARC Battlemage revealed today but it's still canceled :'(
AMD already works with Mediatek for its wifi chips, so they can license MTK modems. ARM is providing complete solutions for partners who want them - ISPs, NPUs, etc, so it wouldn't be that much of a hassle for AMD to make a smartphone chip if they really wanted to do it.
i have only xbox series x and i can say i have never a single issue with it
idk what problem with this guy unit
but mine is super steady and never crash a game
2021 manufacturing
Same. It’s just one person’s experience, not a big deal
Thank you for making work on a Monday following a nice extended weekend more enjoyable. Cheers, Tom, and wishing you and your family Happy Holidays 🍻
Have you two checked what kind of profits AMD is making off of low cost RDNA 2 and Zen 3?
This is just a WILD guess but I have to imagine they're doing pretty well considering that's made off of TSMC N7 for which AMD is probably the main customer and I imagine TSMC has been giving them a really good price on it for the last 2 years or I don't think AMD would have been so willing to keep producing RDNA 2/Zen 3 as low cost options. I don't know this as a fact but when AMD announced these would be low cost options they had probably already worked out an agreement with TSMC since N7 by that time was no longer a cutting edge node. I mean right now we're at 2nd gen N3 for TSMC which is supposed to be an excellent node for TSMC AND their customers.
Around 11:30 two biggest in the UK are SCAN then Overclockers then there about 5 smaller ones I would normally buy from 1 of these 7 and only use amazon if your some reason they had better availability. On day one I checked Overclockers the sold out the 9800x3d and they had about 250 of them. It was the best selling CPU on that site by far and I suspect SCAN was in a similar position.
Oddly the 3rd best Overclockers CPU is the 5900x because its such good value I bought it over the 5800x3d a year ago.
After the leak about Strix Halo "LP" TDP i'm worried about all the Strix Halo lineup efficiency. AMD is less efficient then Intel in mixed workloads with chiplets on desktop, where the E-cores can do what they are designed for, and Strix Point with 4 Zen 5 and 8 Zen 5c cores is already less efficient then Lunar Lake in a simple workload like office productivity and web browsing, let alone the comparison to any Arm counterparts. It would be a real let down if the situation with Strix Halo will be even worse then that. I hope i'm wrong and AMD actually found a way to make a mobile Soc with chiplets more efficient then their current monolithic design.
My guess is that idle power will be worse than monolithic SoCs but quite a bit better than desktop Zen 5. For that reason ST power draw will probably be relatively high as well but MT will be comparable to Strix. Unfortunately this really isn't a thin and light SoC and even less so for handhelds. I do expect that the 105W TDP for Strix Halo LP is probably the up-config and there will be laptop models utilizing it that have more sensible TDP configurations
@@2intheampm512 i think the 105W TDP, just like the leaked 120W Strix Halo 16 cores and 40 CUs inside the next Asus 13" laptop, include the total power consumption of the machine with the memory included.If i'm not mistaken the CPU was rated at 30W, 75W for the IOD and the rest for the memory, from 32 to 128 GB of Ram (obviously more ram more power required). If Strix Halo will be worse then Strix Point at efficiency i will consider it a "failure", because you're not always pushing the Soc to the maximum inside a laptop, and if performances are the same of a core ultra 200 + an RTX 4070 with the same power consumption, is not really a win if you use more energy for single core tasks and, consequently, you have the same of even worse battery life (with the same battery capacity). Also considering that we will have rtx 5000 soon after Strix Halo with presumably better efficiency then rtx 4000, idk, the more we get close to Strix Halo the more sounds a bit DOA at the beginning of 2025 as a laptop Apu. I hope i'm wrong since i need to upgrade my laptop and i don't know what to do 😂
In the uk, scan is far better than overclockers
Overclockers sent me an empty box instead of a 9800X3D.
I sometimes use ebuyer or CCL as well at a pinch but I much prefer Amazon - cheaper postage and easier to return faulty items. I had a nightmare with OC once where I returned a faulty graphics card and they sent it back to me and said it worked fine. I sent it back again with a letter explaining why it was broken and it took them nearly 6 weeks to send me a replacement during which time I couldn't use my PC. It had faulty VRAM which worked fine on windows at the desktop but if you launched a game it immediately started artifacting and then crashed after about 30 seconds. They said it passed all their tests. With Amazon I just call them and they give me a refund before I have even posted the item back a lot of the time. ebuyer use Yodel to deliver who are awful. Scan don't have live stock levels on their website so you can place an order of something it says is in stock but by the time you check out it is out of stock then you have to wait for it to come back into stock which can take weeks. Had that happen too many times. The only reason I use anybody other than Amazon is if I can't get the product I want or something just like it, eg a RAM kit that is on the QVL.
Had an XBox series X since launch, and play it everyday without issue thankfully.
Intel: pays rich people millions and billions to fail with no strings attached
Also Intel: We nEeD fUnDiNG...
Overclockers sold out, restocked and sold out in a few hours
I never buy from understockers, not since the 3000 series saga... Amazon is pretty huge here, and I'll buy from there.
Didn't you say Battlemage was dead like a year ago
And alchemist is cancelled 2 days after launch.
Y’all did a great job with the audio
I'm surprised there are no comparisons with AMD during the bulldozer days. AMD got out of it with far fewer resources than Intel. AMD stock was $2 during that time...
People saying that an Intel 4060ti with 12gb will sell well at 250$ are out of their minds. Remember that the 4060ti was equivalent to the 3060ti, and that card was 400$ in 2020. Getting a 150$ discount 4 years later with inferior drivers and software compatibility will NOT sell well:
Yes it will when you add the nvidia aura. 6700xt are around 250-300
What else can you get under $300 with more than 8GB VRAM? It basically just needs to beat the RTX3060 12GB by 30% (including RT and upscaling) and it will have it's place on the market.
We have hours at this point to find out true performance of the cards
It will be te best selling GPU in the EU because other options suck. This time you wont need to emulate DX11 and 9 which was a hardware fault with the Alchemist line. Intel already unfucked their drivers.
As someone who actually used an alchemist card for a while intel still is significantly behind in software. Some games don't work at all. So any intel card above 200 bucks is wasted, better get a red or green card instead if you want to spend more than 200.
I bought an A380 (on the second or third wave!) because I like collecting weird hardware and I wanted to see how the architecture performed... in a secondary PC, of course.
I haven't run into anything that just straight does not work on recent drivers, but it took them over a year and SO MANY DRIVERS (including TWO that broke my system!) to... not actually improve performance very much? DX9 results are "less deficient" now, but DX11 and DX12 have actually regressed according to my notes. They're margin of error kind of losses, but everyone was saying how Arc cards were gonna go up a tier as they aged, and that sure didn't happen. (Arc fans have insisted to me that ACM-G10 did, but I have doubts.)
In a best case scenario where the A380 is paired with a fast Zen 2 CPU with PCIe 4.0 and ReBar and dual channel memory, it still can't match the RX 5500XT I had lying around, and it suffers a greater bottleneck than the AMD card does when things start to get less and less optimal.
Remember that alchemist needed emulation while battlemage wont.
@@kazuviking Alchemist started off using the D3d9on12 wrapper in Windows before moving to using DXVK within the driver, which was a lot better.
I'm seeing that they added some optional instructions for DX12 that should help, but I'm not seeing any indication they've ditched API translation via DXVK in favor of purely native DX9, though.
And the state of API translation isn't relevant to DX11/12/Vulkan issues.
Inconsiderate Intel not getting rid on a Friday so he could have made it into the show!
12 Core Single CCD X3D chip would be rad.
You guys should do this live in together in one place more often but do the setup like a podcast. A table or two couches with two mics etc.
Super hyped for 2nm chips from TSMC 💯
When you have over 85% of the market you can effectively act like your competitors don't exist until the market proves otherwise. Radeon 7000 vs RTX 4000 already had multiple price segments with slam dunk AMD wins, and it didn't result in mass exodus from NVIDIA cards. If there was a 4070 Ti available for $600 tomorrow it would sell out in a day. That's just a fact. So a 5070 for $600 makes complete sense since they should only price their products against their own offerings. People upgrading from a 3070 will be pleased with the boost they get, and older holdouts (think 1080 Ti/2070 Super) will see a massive performance leap. AMD unfortunately has to go above and beyond to change this reality.
they need to get ai upscaler and boost rt (this specially for high end who would pay 1000 dollar for an amd card) and they would get people in
I don't think they'll make a 6 core Zen 6 CPU, just like how they stopped producing 4 core chiplets a number of years ago. I think they'll use previous gen products for budget CPU's again. That strategy worked well so far.
I agree, I don't think yields will be bad enough to need to use 50% disabled cores, at least not for a while then they'll only have enough for a Microcenter exclusive or something.
I was thinking about zen 6 desktop segmentation with 12 core ccd's right when you ended up theorizing about it yourselves.
Rewatching that segment, you split it up into:
- 10500 6c (1 ccd -50%)
- 10600 8c (1 ccd -33%)
- 10700 10c (1 ccd -17%)
- 10800x3d 10c (1 ccd -17%)
- 10900x3d 12c (1 ccd full)
- 10950 20c (2 ccd -17%)
- 10960 24c (2 ccd full)
Considering we're already seeing multiple full single ccd mainstream models (700/800) in all recent gens, I doubt they'd cut down the lineup this way. I'd expect more something like this:
- 10500 6c (1 ccd -50%)
- 10600 8c (1 ccd -33%)
- 10700 12c (1 ccd full)
- 10800x3d 12c (1 ccd full)
- 10900 18c (2 ccd -25%)
- 10950 24c (2 ccd full)
This does assume there is also threadripper with better IO along the gen.
But I doubt they'd have to cut down so many dies that it would benefit them to sell them in every of their main volume sku's (600 to 800). Also going 6 - 8 - 10 - 12 is a really crowded lineup, and then the gap from 12 to 20 would be kinda weird, with the 20 basically making little sense for people looking for that multi-core performance since there's a 24-core right next to it, meaning whichever offers better value would cannibalize the other - and I'm leaning to the 24-core winning that one.
Instead I'd expect cut down yields to go into 500 and 600, with no need to cut down 700 and 800, with the 800 being just the better binned x3d variant of the 700 just like we see in zen 5. Then, to offer a more sensible cut down, cheaper alternative, they could dump some relatively garbage yields into a 900 with 18 cores, and offer it as this better than previous gen 950 'budget' multi-core value king, with the 24 core still offering a decent step up as the true desktop multi-core king. For pricing I'd guess from 500 to 950: $200 - $300 - $450/480 - $550 - $650 - $900. Maybe an 8-core x3d variant at $400 as a more budget second-to-the-gaming-crown full 12-core $550 800x3d sku later in the generation, with the 500 also probably later in the generation to build up inventory of ccd's that actually need to be cut down 50%. Either that or they'd employ zen 5+ as the zen 6 low end.
But who knows, without real intel competition maybe they'll milk hard, price 10%-20% higher than the numbers above even to truly get rid of zen 4 and zen 5 stock as a budget filler for this lineup and only dropping prices when they've ran out. The previous broken silicon with James Prior shone some light on how these higher-than-expected pricing structures come to be, would be in line with that
I’ve bought both alcemist founders editions for Intel, one copy of all sparkle gpu cards, & one pc with a i5 12400 and a core ultra 265k. Big fan. Hoping they succeed and definitely will be buying battlemage 💯💯
You guys should always attempt to be in person. I love this content.
Loading:
Nvidia rtx 6060 8gb, $1000 msrp, for professionals.
Tom, remember Intel can sell "their" x86 license to whoever they want. However, it will only get you 8, 16, & 32bit x86, sure you also get Intel _64 AKA Itanium(a failure). Intel does not own the rights to "modern" x86 because it's 64bit AKA AMD_64. When cash strapped AMD couldn't sell a license for AMD_64 because it's built on x86 so legally Intel gets the final OK. Likewise AMD gets the final say of any license deal using AMD_64 AKA any modern x86. Sure Nvidia could buy a license from Intel but they would be limited to the unsupported and failed Itanium Intel version of 64bit x86 or be limited to 32bir or smaller.
The RX 8800XT insanity has to stop! Cherrypicking the game with the *weakest* RT implementation as some kind of "comparison" with a 4080 is ludicrous. CP2077, Hogwarts Legacy, Plague Tale Requiem and others have the 7800XT at abysmally slow, unplayable, framerates (typically _worse_ than a 3070!), a "40%" uplift would still be unplayable at 1440p and _maybe_ as good as a 4070 non-super! The 8800XT will *NOT* be anywhere close to a 4080 in games that actually *use* RT in any meaningful way. Also, the 8800XT's raster performance will be probably about 10%-15% _worse_ than a 7900XT (not XTX!), meaning it will be a decent generational uplift, not a miracle!
The Switch 2 had to wait at least 2 years to launch, they don't want to accidently have modern tech in their console. They have to be at least a couple generations old.
You would have gotten better audio if Dan just sat on your lap.
Your arguments about battlemage are valid, as long as Nvidia and AMD offer 250 euro cards to compete. I mean that is the whole point, green and red don't supply for the entry/mainstream market. That group of buyers doesn't care if they buy a 6 nm, 4 nm or 3 nm die. The only thing that counts is fps per Dollar. And I'm happy about Intel releasing a 250 euro card. been gaming for more than 20 years and I couldn't care less about 800 Euro cards for shitty and repetitive AAA games in 4K. Doesn't make the games any better for all I care and the time to play is very little as an adult, so small cards are again where the magic happens.
17:12 BDM as a metaphor is creative NGL
It’s crazy b580 supposed to be 250 comes with 12 gb but nvidia 300 dollar 4060 but it supposed to be as good or better everywhere in recent release data.
We can’t get anywhere in reducing overall prices without competition.
Ryzen 7 12 Core - 400$
Ryzen 5 8 Core - 250$
Ryzen 3 6 core - 170$
Do it AMD, DO IT.
The Zen 6 lineup should have more pcie lanes at that point where you speculated/imagine the Ryzen 9s would be. Those are pretty much in-between high end mainstream and workstation-level. X970/E motherboards may have two 16x PCIe 5.0 lanes or more lanes dedicated for storage. If I am paying nearly $1000 for 20-24 core CPUs, might as well throw in some more pcie lanes.
I still think Console Makers should make new "Gens" and every Gen should last for 5 Years now, each while they need to support the old Gen still. Until 10 Years at least, then Companys are free to do whatever. It would help everyone.
This guy doesn't know what he's talking about. Guy lives off wccftech
With the new 9800X3D
The next gen UDNA will be amazing 😉😉
maybe the zen 6 24 core could be reserved for the "AM6" socket (with ddr6) while AM5 only having 20cores zen6 with some higher ddr5 sweet spot than with ZEN5?
I can see an AMD future with more *different* cores, the low hanging fruit would be 8 "full fat" cores, then several "compact" cores like the "Zen-C" chiplets.
if you have an application or workload that can utilise threads and is highly parallelised, then it would be unwise to only have full fat cores over the compact ones, for workstation and enthusiast use. For server I can see having only full fat or full compact SKUs because they partition the cores into virtualised machines, Threadripper though I see being better as a mix, as a workstation.
The kinda low hanging fruit after this would be accelerators, like video encoder and decoders, file compression, vector/matrix/graph compute, neural compute e.t.c.
The not so low hanging fruit would be to have other architectures like RiscV chiplets (for specialised cases), and FPGA chiplets (from AMD Xilinx) for having a "application as hardware" cases.
Regarding Apple M-chips, my prediction for the future headlines around Apple M5 and M6 chips are: "Apple's new M-chip architecture has no more low hanging fruits" :)
I haven't heard anything regarding of a total redesign for any future M-chips, which I guess would be needed in the long run? I mean you can only iterate so far on same fundamental design.
There was already a Mercedes X Mario Kart collab 10 years ago. You can drive some Mercedes vehicles as free DLC in Mario Kart 8
Mercedes is IN Mario kart 8 -like, actually. You can unlock the Mercedes.
I wonder if AMD could put Vcache or something like it on a GPU. With how popular the X3D CPUs are it might make Radeon more interesting for some people who wouldn't buy AMD GPUs otherwise.
They're already using cache like that in MI300, so it's not out of the question. They'll likely brand it as Infinity Cache.
@@adeptuspotatocus6451 be better for marketing to use infinity cache and brand it as vcache lol. Imagine the 9800XT3D or something. If they were smart it'd be really be 9900XT level but by dropping a tier it'd look like it had brutal performance
Will 24c zen 6 have dual ccd 3D VRAM
I’ve never had an issue with XBX yet, though I haven’t been on it that much lately. Not a huge fan of the UI but no real crash issues.
My napkin math on zen 6 structure:
10960x -24c +X3D
10950x -20c +X3D
10900x -16c
10800x- 12c + X3D
10700x -10c + X3D
10600x -8c
10500x -6c (OEM only)
6c? I don't think so, it will be 8c without HT. Still missing 12c+32c+x3D
you're forgetting that they will have changed naming schemes, probably twice by then
I bought a 7900xtx back in Oct during the prime days for $788
Hi if CL16 3600mhz is a sweet spot for ryzen on AM4 what is the sweetspot for AM5? CL30 6000 or CL32 7200/7400?
6000 cl28. I have the gskill royal neo
@butonghit thanks man, helps a lot
@@butonghit he is asking for sweetspot , not the best of the best. 6000 cl 30 is way better price/performance.
The current sweet spot is defined by the memory controller on the IO Die that is from Zen 4 and reused on Zen 5. Any further AM5 chips are likely to have an updated IO die on a better node, and would therefore have a new sweet spot for memory speed/timings.
The 8000 series can actually run higher controller and fabric speeds but no one is using those chips for peak performance since they have cut down cache and PCIe
Definitely Cl30 6000
B580 should be $250 now but I expect it to drop down to 199 dollars anytime soon and that would be a good price for something that beats RTX 4060 while making a huge RT performance
Imagine it's going to beat 3070 and next year Celestial C580 will beat RTX 4070
The issue with the IO die is the link between the cores and io die if we can either clock the fabric to like 4GHz along with a better memory controller on a io die the performance would come back. Or they could just widen the link but that sounds expensive
if they moved past the fabric through the organic substrate and started putting it all on a silicon interposer or move past the zen2 configuration alltogether and do something like stacking the CCDs on the IO die - then widening the fabric should be easily doable. Also would reduce trace lengths which makes lower die-to-die latencies possible.
Overclockers UK is where I got my 9800x3D, on the first batch of deliveries ... And they were the only retailer/e-tailer that managed to keep it at the suggested MSRP from AMD £449, all the other ones said they couldn't do it as it was little to no margin at that price so they were all at or above £469
all i want is a sapphire nitro 8800xt
Matrix hardware support for XeSS2 Supersampling, frame insertion and the new XeLL latency improvement, all focused to provide 1440 gaming. Looks like Intel has a good plan for Battlemage. They are built on TSM N5.
"N5” isn't a node, it's a family of nodes...and I didn't realize sub-7600 XT performance is good for 1440p lol
@@MooresLawIsDead Looks like Battlemage has caught up with several AMD and NVDA GPU features. Pretty impressive update in one generation ... improvements in ray tracing, upscaling, frame rates, latency, ai processing performance. Perhaps in another year any differences in gaming features will be so-what, and all anyone will care about is the support for the latest AI models.
@@MooresLawIsDead Battlemage outperforming 7600 XT in Vulkan benchmarks according to wccf tech.
Tarrifs are AMAZING and they're purpose is NOT for competitive reasons contrary to popular belief. That's not a benefit they have I 💯 agree. But that's the publicly stated reason, because it's popularly believed.
I still don't understand why intel cancelled the rentable unit single thread beast project. And then released the wet fart of what arrow lake was. They could have just grabbed the gaming crown and sold like crazy like 9800x3d
Not a single headset in a bin somewhere?, or no I/O? No matter, great cast as usual, happy hindsight turkey daze.
I've been enjoying my 9800x3d for over a week now ROFL.
Will there be a Zen 6 LPDDR6 192bit version? Or will LPDDR6 be available in Zen 7?
Lunar Lake for laptop is good, handles 95% of use cases, the only downside I see is the limited number of cores
A 16c/16t have the potential to beat Apple and AMD in performance, 8c/8t Lunar Lake is OK but does not make you excited