When it comes to Intel GPUs you should search for the Intel CPU overhead issues. The Intel GPUs can fall massively in performance if you are not pairing them with something insane like 9800x3d. Also Intel GPUs require rebar to be enabled or they will perform horribly. For example Hardware Unboxed did a piece on that recently. They are not reeeeally what they were made out to be in the reviews since many neglected to take in to a count that 9800x3d will very likely not be the CPU of choice with Intel GPUs.
Well, I feel for the AMD partners at CES who basically came to the conference and at most were allowed to show their products, but can't say anything about it.
Three years ago you couldn't generate a decent image of a ship. A year ago you couldn't generate a decent video of a ship. Today you can run Flux and Hunyuan on a consumer gaming GPU, and do both. The progress is incredible. But people who can't be bothered to vet results or touch up errors in post, because they only think cost-cutting and profits, will keep advertising these technologies in the worst possible ways.
We live in weird times where companies can completely ignore their consumers and 100% focus on investors. Prices and products don't matter anymore, only buzzwords. Profitability is irrelevant, the only goal is _line go up forever_
Weird why, companies have for a while had little interest in the workforce, having no respect for the customer is just a progression of that. A lot of people feel their governments ignore them also, so we should be used to it by now.
90% of Nvidia’s revenue is from AI. Profit is up 94% last year. These “consumers” are completely relevant and exactly where the company is focused. Somehow gamers don’t seem to realize this - you’re just no longer that important. This is like complaining that car companies are completely ignoring their manual transmission customers. Yes. They are. Not weird at all.
@@rdspam maybe not the best comparison, the auto industry is in complete turmoil just now because the usual brands failed to innovate (and ignored their "consumers").
Thanks Roman, and also Thanks Steve! :) @7:20 there was a good film in the year 2000 called Snatch. was actually a great film, and, the title was not a reference to female genetalia.
xD Thanks Steve!!! You made Der8auer make those compile videos of 1 word. About AMD acting weird, my understanding is that they poll everything internally and then what wins is used.
@@baoquoc3710 Yeah, my "knowledge" comes from the interactions between TinyGrad company and AMD. To be honest it feld weird exchange, George Hotz talked about it in one of he's streams. I have no reason to believe George Hotz was lying about it. Eventually Hotz corrected the AMD bug by him self, with almost no help from AMD. He also commented how weird it was from AMD to ignore a critical crash bug and why NVIDIA is winning cause the whole company would have mobilizised on 1 bug that treathens their software stack and their business model/profit. And this bug I believe is also a plague on MESA drivers and this is why the Mesa driver dev's had to make softboot work, cause the AMD card keeps crashing. Also a bug on Windows side.
14:10 In fairness, I think they just mean that tier in the product stack (ie xx70 series), not specifically the 4070. They did show the RDNA 3 cards in the same colours as the 40xx series cards, and left a spot for the 50xx series to go
Tim from Hardware Unboxed did a pretty good job grabbing some details on FSR 4 today considering the circumstances, but with pre-orders looming would be awful nice if AMD would give full details on the upcoming lineup and an in depth preview of the tech including FSR4. This is not the way and i really have no idea what they are doing, but it seems neither does AMD !
Don’t even get your hopes up. Some journalist managed to run a benchmark on 9070xt and saw tons of graphical artifacts. Either the card was overclocked by a tons or the current state of the GPU is unfit for shipping. Also the price can’t be lower than 550$ since that tag would ensure the 7900 cards will never sell
I mean we knew that was going to happen, with the advancements in robot technology and machine learning technology and large language models all colliding into one dystopia with social connection being more and more artificial and distant... But what a bold naming choice, just coming right out and saying it.
Huge miss for AMD to not have their top-line X3D part be dual cache; jack up the price if you have to, but having an uncompromised halo X3D part would be a great option in the product line given their current positioning.
Performance might be 2-3% higher, but manufacturing cost will be around 20% higher. (4 pieces of silicon to 5). Can you charge 20% more? Or heck even $50 more? Probably not, so it’s not worth it. Or even 1%, cause if AMD is limiting all game threads to a single CCD to preserve cache locality, then the second one isn’t going to be doing all that much anyways.
15:04. Wow look at the top of that table and how for 11 years, they were releasing graphics cards all at the same 500 nm process node size! Things have really sped up in that regard.
500nm process didn't even exist in '86; looks like a typo. So it's only '95-'97 on the same node -- just like Lovelace and Blackwell two years apart on 4.
My condolences to the minutes wasted in which we heard "AI". It's like a feature is screaming for a use case. And that makes me sad, because that die space could have been used for raster- or tesselation performance. But i am also slightly optimistic. Because everytime Amd was mum about something, the performance of that something was pretty good in the end (Ryzen) as an example. The Renaming shenanigans in the GPU-departement have a little whiff of Nvidia's GTX8800/8900 about it.
Holy actual F. There are *all* kinds of medical issues that can induce shaking. They range from diabetes-related issues to degenerative nerve diseases, metabolic disorders, liver disease, etc. The fact that you bleated this out and 25 sheep agreed is a damning indictment of the education system. Several medical conditions can cause shaking, known as tremors. These include: Neurological Disorders: Parkinson's disease, Multiple sclerosis, Stroke, Essential tremor, and Dystonia. Metabolic Disorders: Hypoglycemia (low blood sugar), Hyperthyroidism (overactive thyroid), and Liver disease. Medications: Certain antidepressants, Anti-seizure medications, and Chemotherapy drugs. Infections: meningitis and encephalitis. Other Causes: Caffeine withdrawal, Alcohol withdrawal, Tumors (e.g., pheochromocytoma), Stress, and Anxiety.
In all fairness, there are very few people that don't get the hand shakes when on stage presenting to the world... It proves they are human, and not AI Robot overlords. Most presenters learn tricks and techniques to help reduce it. Like, holding their elbow against their body, for better stability. Caffeine might actually help reduce nervousness. Kinda like alcohol. Done in moderation, "Dutch Courage" can be beneficial. But the zoomed-in wobble vision was disturbingly obvious. I would expect anyone doing that kind of presentation to be far more practiced and experienced.
I'm going to use a 5080 to create a tokenized transformer to implement a generative, neural enabled innovative deep learning natural process convolutional computer vision model for supervised inference waifu character creation.
There are some leaks this morning about AMD 9070 XT, one showing 3DMark benchmarks (14558 on TimeSpy Extreme and 6345 on Speedway). Then a rumour about the reference AMD card price being 479$ USD and release date of 22 Jan. If the benchmark leaks are true is really weird that they didn't show absolutely anything about it, because doesn't look bad.
About the AMD keynote RDNA4 branding picture: This table doesn't need to show some performance categorisation by itself. Maybe I'm wrong, but I won't interpret too much into it.
Thank you. You are always informative and inciteful. Rather than critiquing the presenters and other passable aspects of the keynotes, I like the new products and updates. AI is what it's all about. I expected a huge hype about it. Yes, too. What better place to advertise their products? I liked your thorough description of the products. Thanks, Rob
G'day Roman, Shiek & Makita, AI Performance/min chart 🤣🤣🤣🤣🤣 As the 9950X3D only has 1 3D Chiplet 🤔maybe Steve can get Amit & Bill to make you a special 9950X6D with 2 3D Chiplets "Thanks Steve" 😂
every content creator should mock those AI BS overloaded presentations, ty der8auer ..... .... you know what they do instead? put "5070 = 4090" in the thumbnails 🤡
Yeah Intel's doing great to remind me to get rid of my intel XPS running already burning hot at startup even with fresh thermal paste. I'll make sure to get one equipped with a Ryzen 7. Thanks Michelle.
Making a graphics adapter that is equal to something that has already existed for a long time, and at an inflated price relative to new solutions, is a dead end.
Arguably, motherboard manufacturers are producing excessive eWaste by not providing a debug LED. Wish the EU (and other regulatory bodies) would step in and regulate such an obviously necessary feature to reduce premature electronics recycling.
Apparently too much as far as share holders are concerned, that’s THEIR profit and damn it they want it. Too hell with what the damn customers want or need 🤷♂️😂
I think the comparison they mean, is based on the class. it's a 70 class gpu, 80 class. just a way to show where it is in their line up. not like they're comparing it to a specific gpu, just what class it competes in
Intel constantly saying "It's a great time to buy a new PC" is them really saying "We are hurting for money and need to make sales. Buy our stuff!". Also Intel not releasing the non-K CPU's and B series chipsets together is because again, they are hurting for money so they are rushing products to market. I am not an Intel guy right now but even if I was, I would actually wait a bit to see how the launches go and find out if there are any bugs (14th gen what?).
As an amd fan ( read: i preffer amd ober the others) its so sad, amd doesn't give a fuck about radeaon. Its the consumer electronics show. Has anybody seen the average person be excited for AI? Nobody gives a fuck. Id rather not pay for the die space. Disgusting, overpriced, over hype, unnecessary
Tragic how the new performance metric for AMD and Intel is "AIs per minute". It is almost as if AMD has discovered some catastrophic flaw in RDNA4 at the last second. Like, it causes your motherboard to explode when it hears "AI"... The product hasn't been pulled, but "later this quarter" sounds like they have some major rework to do. I hope we hear the truth sooner rather than later.
I'm guessing driver problems, like Intel had on their A770 Discrete Graphics Processing Unit Add In Board, at launch. Point 2, it appears AMD is keeping the RDA4 DGPU ever so slightly below the GRE Edition, so they are not fighting embargoes and Tariffs, both at once. "gravitic propulsion"? If real "G. P." Is claimed to only be possessed by the Pentagon and the CCP. IF true, it could be cold war act two, this time between the CCP and the Pentagon. Which, while awful... Is better than humanity so destroying itself and the planet, and everything that a Neolithic Era begins. I really hope that the NeoLithic era is not the goal of Gates and the WEF, et al.
you cant do any hardware changes on this type of timescale - the chips are all already made - so if there is a problem they gotta figure out how to software fix it
3:50 I don't know if overclocking is the big deal it used to be. At least with the clocks as high as they are on stock processors as long as we can still unlock the TDP to reach the max stock clocks we should be good. Maybe I'm crazy but when I see a 14700 (non-k) pulling 250w I don't know what for overclocking. Also considering the turnover and competition in pricing I was finding Z690 boards for much cheaper than a B660 or B760. I assume due to the market saturation enthusiasts will find Z890 boards a better deal anyway. I guess what I mean is if you get to the i7 level you shouldn't need overclocking.
AMD naming 9070/XT for easier comparison with NVidia cards doesn't _have_ to be an indication of their cards performance, it could simply be to hint at MSRP. After all, they can't be sure what performance Nvidia is bringing to the table with their xx70 lineup. But they could be fairly sure that Nvidia would not shake things up too much when it comes to price. The graphic showing 9070 series topping out at same level as 4070Ti might be more of a hint on price for 9070 than performance. Rumors say 9070 is not a chiplet design, and if true there's really no reason to expect horrible performance per Watt. Then considering that the 9070XT comes with 3x 8-pin PCIe Power connectors. The performance might indeed be significantly higher than a 4070Ti.
I recall when Byte Magazine in the 80's - remember them? - ran an editorial trashing AI marketing, with a graphic showing a box for a business app with the sticker on it "NOW WITH ADDED AI!". This nonsense has been a plague for a long time.
The funniest thing of all is that all large companies are talking about "AI" in consumer products, but mobos couldn't be worse for AI: Not enough pcie lanes, not enough pcie slots, no bifurcation and the worse layout ever made.
AMD started the confusing since X370 / B350, they skipped 100 - 200 to match with Intel, so where's a lot of confusion then. However when Intel got to 700 chipset series they had nothing as the AM5 was on X670/B650. So now Intel with 800 so they again, skipped the naming to match.
I honestly have no idea of the CPUs Intel is planning to release as they have changed their naming convention and I am lost on how to figure out which gen they are at.
15:55 They might even just call it the 1070, And no one should confuse it with the nvidia 1070 due to how old it is, and google shouldn't give incorrect search results because the searches would be for AMD RX 1070 or AMD 1070 for short. *shrugs*
Fully agreed with the debugs LED or even better a 2 digit debug code. Cost optimisation are making for worst products unfortunately but I really think every mobo should have a precise way of knowing what is causing any issue in their boot process.
Speaking of weird naming schemes, as someone who isn't in advertising, I'd love to see them just having their product names start with the model year. E.g. RX2570 - A 70-class GPU released in 2025. RX2670 - A 70-class GPU released in 2026. RX2660 - 60 class GPU released in 2026. Hell, put any number of letters and digits after the model year. 25750XT. Released in 2025, 750 class, XT variant. Whatever.
Honestly - if they hadn't ditched hyperthreading on core ultra 200 series and I could find a motherboard with a single CPU M.2 slot, an 8 lane GPU slot and three 4 lane CPU slots then I'd probably be getting one as it'd be like pseudo hedt. Unfortunately, I've not found any boards where they opted for a normal slot instead of second M.2 slot or boards with three physical slots from the 16 lanes.
For a regular consumer, is there any use of AI hardware capabilities yet? I haven't seen any app or whatever that makes me want to upgrade to an AI pc. I guess the eventual plan is for Chat GPT and stuff to run locally?
Even though this year CES keynotes are messier than others, luckily we always have Roman as an Eierschalensollbruchstellenverursacher to crack into it.
Uh oh, my tired brain clicked this video excited about intel GPU stuff... B860 is a chipset 😭
Chip manufacturer not making two overlapping naming schemes for different product lines challenge, difficulty: impossible
When it comes to Intel GPUs you should search for the Intel CPU overhead issues. The Intel GPUs can fall massively in performance if you are not pairing them with something insane like 9800x3d. Also Intel GPUs require rebar to be enabled or they will perform horribly.
For example Hardware Unboxed did a piece on that recently.
They are not reeeeally what they were made out to be in the reviews since many neglected to take in to a count that 9800x3d will very likely not be the CPU of choice with Intel GPUs.
same... -_-
@@ZyxSpeed mine seems ok on an ES tigerlake, but i only play at 4k
@@ZyxSpeedyou are aware most of that was driver issues right? Which has since been fixed like over a year ago by now
I think there's enough AI snatch already on the internet
and sadly, it's pretty much the only field where the AI version is better than the real deal...
hahaha
I am actually very impressed that a non-native English speaker picked up on the ‘snatch’ slang so well.
Nah bro. There can never be enough AI snatch. And soon we will have autonomous full service robots. It will be glorious and I can't wait.
I want AI bussy.
AI per minute is a great metric. I'd love to see it more
AMD used AI aboud 154 times while Nvidia 204 or 206 times.
For clarity though, you'd have to make sure you're measuring the real number of AIs and not just the fake 'generated' AIs /s
Is that how many times per minute these snake oil salespeople say AI?
@@Bobarrella MULTI fake generated AI, don't be old fashioned.
Well, I feel for the AMD partners at CES who basically came to the conference and at most were allowed to show their products, but can't say anything about it.
@@stennan Yeah, I really wanted to see what Sapphire had on offer with their new Nitro cards but as far as I know they haven’t been shown yet?
@@KitsuneKierathere was a picture of it, it looks a little weird, Google it 😅
Next generation Intel keynotes are rumoured to feature a groundbreaking 4.5 AI per minute.
Intel: AI Inside
Wonder they don't put it in there actual company name. Intel AI. Nvidia AI. AIMD.
What the hell is 4.5 ai per min
CES: 20-25 buzzwords per presentation
Edit: at least*
@@bot7845 Watch the video to find out
It's nice to see the thanks steve lady on stage, thanks, thanks steve lady.
except now she's in-charge.. good luck intel.
@@sirmonkey1985 what did she do?
@@RobertFletcherOBE
Thank Steve
Thanks Michelle
1:38 "incredible AI experiences" watching a wave going through the ship
That is peak AI
Three years ago you couldn't generate a decent image of a ship.
A year ago you couldn't generate a decent video of a ship.
Today you can run Flux and Hunyuan on a consumer gaming GPU, and do both.
The progress is incredible. But people who can't be bothered to vet results or touch up errors in post, because they only think cost-cutting and profits, will keep advertising these technologies in the worst possible ways.
We live in weird times where companies can completely ignore their consumers and 100% focus on investors. Prices and products don't matter anymore, only buzzwords. Profitability is irrelevant, the only goal is _line go up forever_
wall street was a mistake
That’s a fact. Unfortunately it will mold the way people think, and the products they like.
Weird why, companies have for a while had little interest in the workforce, having no respect for the customer is just a progression of that. A lot of people feel their governments ignore them also, so we should be used to it by now.
90% of Nvidia’s revenue is from AI. Profit is up 94% last year. These “consumers” are completely relevant and exactly where the company is focused. Somehow gamers don’t seem to realize this - you’re just no longer that important. This is like complaining that car companies are completely ignoring their manual transmission customers. Yes. They are. Not weird at all.
@@rdspam maybe not the best comparison, the auto industry is in complete turmoil just now because the usual brands failed to innovate (and ignored their "consumers").
RDNA5 will be no doubt named RX AI 70 XT 🙃
lol...I'm saving up of an RX AI X70 XTX XXX Edition
There will be no such thing as RDNA, it's gonna be called UDNA.
So you are right, the naming scheme is gonna be completely different for sure.
Rxx AI 9890TKH3R5Y with the DAOT editions for special market release in limited numbers...
The first 200 sold get signed by vashtor himself.
Gotta share that Ai's per minute benchmark with Steve; needs an in-depth collaboration.
Thanks Roman, and also Thanks Steve! :)
@7:20 there was a good film in the year 2000 called Snatch.
was actually a great film, and, the title was not a reference to female genetalia.
Snatch and Lockstop, two of the best films ever.
Snatch is one of the greatest crime films ever made.
Knew and lived with two brothers from the UK, they basically have Snatch rented from a DVD video store put on repeat...
Oh, and I recommend Hot Fuzz.
Brick Top calmly explaining how to get rid of a body is the best scene in any movie EVER.
Sadly, none of the guy Richie films were as good as layer cake, from one of his proteges coming into his own.
amd marketing team doing its thing as per usual
A.I. - the new snake oil.
2025 is snake year! Therefore must be GREAT year for AI 😂
Just don't tell Intel 🤫
The new 3D TV. The experience is awful, no one wants that shit, and it only costs us half our GDP and remaining rainforests.
Then why is AI taking away jobs. Keep coping
@@nathsabari97 What jobs is it taking?
these motherboard prices are getting absurdly expensive
B850 starts at $110, and B840 (though you should avoid it) is surely cheaper.
Money loses value
@puciohenzap891 That does tend to happen when you keep printing more of it.
Micro ATX expansion options on an ATX or EATX sized board, poorest investment in in any boards above the baseline
They just overload so many useless features which enables them to increase the price. Wish the newer CPU supported older models.
xD Thanks Steve!!! You made Der8auer make those compile videos of 1 word. About AMD acting weird, my understanding is that they poll everything internally and then what wins is used.
it's like what sticks in the wall works kind of thing.
@@baoquoc3710 Yeah, my "knowledge" comes from the interactions between TinyGrad company and AMD. To be honest it feld weird exchange, George Hotz talked about it in one of he's streams. I have no reason to believe George Hotz was lying about it. Eventually Hotz corrected the AMD bug by him self, with almost no help from AMD. He also commented how weird it was from AMD to ignore a critical crash bug and why NVIDIA is winning cause the whole company would have mobilizised on 1 bug that treathens their software stack and their business model/profit. And this bug I believe is also a plague on MESA drivers and this is why the Mesa driver dev's had to make softboot work, cause the AMD card keeps crashing. Also a bug on Windows side.
Ai fallus was the alternative choice
AI is a new electricity? Wut?
AI needs to be censored in those presentations with a *beep* :D
You need Coding skills madley !
WHAT IS IT ?
14:10 In fairness, I think they just mean that tier in the product stack (ie xx70 series), not specifically the 4070. They did show the RDNA 3 cards in the same colours as the 40xx series cards, and left a spot for the 50xx series to go
Tim from Hardware Unboxed did a pretty good job grabbing some details on FSR 4 today considering the circumstances, but with pre-orders looming would be awful nice if AMD would give full details on the upcoming lineup and an in depth preview of the tech including FSR4. This is not the way and i really have no idea what they are doing, but it seems neither does AMD !
amd never starts pre orders before reviews you have a full day to catch up to any flaws
Don’t even get your hopes up. Some journalist managed to run a benchmark on 9070xt and saw tons of graphical artifacts. Either the card was overclocked by a tons or the current state of the GPU is unfit for shipping. Also the price can’t be lower than 550$ since that tag would ensure the 7900 cards will never sell
Could you tell me the name of the journalist please? @@CaptainKipo-n6g
maybe AI snatch is related to the "physical AI" Jensen was talking about 😂
that would suddenly make AI more interesting 😏
STAR-BABES & STARSHIPS
You better believe it.
Soon full service autonomous bots will be here. And then it is AI snatch for everyone who can afford one.
That guy that can't iron his shirt properly, is telling me this year is the year to buy a PC.
RED ffin FLAG!
"AI per minute" is right up there with "number of X in product names" that Steve did for best chart/metric.
Aorus AI Snatch. Thank you, Gigabyte.
AMD - Never miss an opportunity to miss an opportunity.
Snatch can also refer to a womans lower regions. 😂
@@viperpit-k5j What do you mean also? What else does it mean?
Why do you think I thanked Gigabyte??
Grab ‘em by the Snatch 2.0
I mean we knew that was going to happen, with the advancements in robot technology and machine learning technology and large language models all colliding into one dystopia with social connection being more and more artificial and distant... But what a bold naming choice, just coming right out and saying it.
8:55 Is the top left quarter of the M.2 heatsinks a different tint of white?
Huge miss for AMD to not have their top-line X3D part be dual cache; jack up the price if you have to, but having an uncompromised halo X3D part would be a great option in the product line given their current positioning.
Performance might be 2-3% higher, but manufacturing cost will be around 20% higher. (4 pieces of silicon to 5).
Can you charge 20% more? Or heck even $50 more? Probably not, so it’s not worth it.
Or even 1%, cause if AMD is limiting all game threads to a single CCD to preserve cache locality, then the second one isn’t going to be doing all that much anyways.
The only highlight of that entire CES presentation from AMD was just the revelation of their new 40 CU APUs
15:04. Wow look at the top of that table and how for 11 years, they were releasing graphics cards all at the same 500 nm process node size! Things have really sped up in that regard.
500nm process didn't even exist in '86; looks like a typo. So it's only '95-'97 on the same node -- just like Lovelace and Blackwell two years apart on 4.
Remember the locked VCCSA on B-series? Effectively you can't properly OC memory either.
Very nice, Intel, very nice.
is the gpu on test bench amd 9000 series?
ah, the Ariola Snatch, sounds good to me
My condolences to the minutes wasted in which we heard "AI". It's like a feature is screaming for a use case. And that makes me sad, because that die space could have been used for raster- or tesselation performance. But i am also slightly optimistic. Because everytime Amd was mum about something, the performance of that something was pretty good in the end (Ryzen) as an example. The Renaming shenanigans in the GPU-departement have a little whiff of Nvidia's GTX8800/8900 about it.
Show us the MORA.
I want to see the full MORA setup and benchmark!
Man that board has no PCI-e slots. That's insane.
9:49 Wow, this guy's more shaky than Radeon's marketing department, lay off the caffeine my dude.
Holy actual F.
There are *all* kinds of medical issues that can induce shaking.
They range from diabetes-related issues to degenerative nerve diseases, metabolic disorders, liver disease, etc.
The fact that you bleated this out and 25 sheep agreed is a damning indictment of the education system.
Several medical conditions can cause shaking, known as tremors. These include:
Neurological Disorders: Parkinson's disease, Multiple sclerosis, Stroke, Essential tremor, and Dystonia.
Metabolic Disorders: Hypoglycemia (low blood sugar), Hyperthyroidism (overactive thyroid), and Liver disease.
Medications: Certain antidepressants, Anti-seizure medications, and Chemotherapy drugs.
Infections: meningitis and encephalitis.
Other Causes: Caffeine withdrawal, Alcohol withdrawal, Tumors (e.g., pheochromocytoma), Stress, and Anxiety.
He's on cocAIne
In all fairness, there are very few people that don't get the hand shakes when on stage presenting to the world... It proves they are human, and not AI Robot overlords.
Most presenters learn tricks and techniques to help reduce it. Like, holding their elbow against their body, for better stability.
Caffeine might actually help reduce nervousness. Kinda like alcohol. Done in moderation, "Dutch Courage" can be beneficial.
But the zoomed-in wobble vision was disturbingly obvious. I would expect anyone doing that kind of presentation to be far more practiced and experienced.
When can we expect W880 boards with ECC?
I'm going to use a 5080 to create a tokenized transformer to implement a generative, neural enabled innovative deep learning natural process convolutional computer vision model for supervised inference waifu character creation.
Oh crap. Now I need to change it's name from Waifu Snatch Creator.
There are some leaks this morning about AMD 9070 XT, one showing 3DMark benchmarks (14558 on TimeSpy Extreme and 6345 on Speedway). Then a rumour about the reference AMD card price being 479$ USD and release date of 22 Jan.
If the benchmark leaks are true is really weird that they didn't show absolutely anything about it, because doesn't look bad.
About the AMD keynote RDNA4 branding picture:
This table doesn't need to show some performance categorisation by itself. Maybe I'm wrong, but I won't interpret too much into it.
Thank you. You are always informative and inciteful. Rather than critiquing the presenters and other passable aspects of the keynotes, I like the new products and updates. AI is what it's all about. I expected a huge hype about it. Yes, too. What better place to advertise their products? I liked your thorough description of the products. Thanks, Rob
how about b580 on oldr 2018 sys fails?
It's not possible to buy a prefab delid 14Gen at Der8auer anymore anywhere?
Is there any powermods you can do to your 7900xtx? I would like to overclock and add a waterblock
How wide do you gape when you read the word fist ?
I feel AMD was indeed going to talk about RDNA 4. Nvidia did their keynote. AMD probably panicked.
G'day Roman, Shiek & Makita,
AI Performance/min chart 🤣🤣🤣🤣🤣
As the 9950X3D only has 1 3D Chiplet 🤔maybe Steve can get Amit & Bill to make you a special 9950X6D with 2 3D Chiplets "Thanks Steve" 😂
is it just me or there is only new moıbile or laptop ryzen chips?
best chips are just mobile or laptop...
there is no box pc ryzen ai cpus ?
ai cpus for diy pc is not needed as your gpus already beat them all by considerable margin
I love it when Roman is roasting so many brands in comprehensive kind of way, it's satisfying.
I really appreciate ASRock not mentioning AI at all on their B860 motherboard pages (at least the couple i looked at)
14:40 also, if you look at that graphic, the performance levels line up better to NVIDIA using the old names than they do with the new ones
Gr8 vid my Friend, fun to watch and nice info about the crazyness of the AI crap 😂😊
every content creator should mock those AI BS overloaded presentations, ty der8auer
.....
....
you know what they do instead? put "5070 = 4090" in the thumbnails 🤡
um this is about Intel and AMD ?
Jensen said AI or artificial intelligence 125 times live on stage at his key note... no joke. Once again, nvidia leads the way! 😂
“please check a dictionary for what snatch can stand for” 😂😂😂😂😂😂😂😂😂😂😂
Audio seemed to be a bit low on this one FYI
Nice shirt. Where did you get it?
is it 18A or 18Ai ?
3:10 sure Intel I’ll tell people to buy new PCs, I’ll just make sure that I specify AMD PCs.
Yeah Intel's doing great to remind me to get rid of my intel XPS running already burning hot at startup even with fresh thermal paste. I'll make sure to get one equipped with a Ryzen 7.
Thanks Michelle.
Making a graphics adapter that is equal to something that has already existed for a long time, and at an inflated price relative to new solutions, is a dead end.
Arguably, motherboard manufacturers are producing excessive eWaste by not providing a debug LED. Wish the EU (and other regulatory bodies) would step in and regulate such an obviously necessary feature to reduce premature electronics recycling.
really, how much could it possible cost
Apparently too much as far as share holders are concerned, that’s THEIR profit and damn it they want it. Too hell with what the damn customers want or need 🤷♂️😂
I am so happy der8auer is sharing the Desktop Gang experience with all of us, yes including you AMD 🤦♂
ai's per minute is so good. my fav part was watching the stock drop in every keynote after the first time any company mentioned Ai
Let's see how many AI's per minute they can manage for Computex... Will they break the 100 AIPM barrier?
I think Nvidia wins, the AIPM was impressive on that one!
You the wrong time unit - minutes are for noobs. We have nanoseconds for the normal folks
Do you think AI can teach you how to use apostrophes?
AI... is per?
AI's what per?
I wonder where would be a good place to get me some of that AI. Any suggestions?
great video! thank you :)
I think the comparison they mean, is based on the class. it's a 70 class gpu, 80 class. just a way to show where it is in their line up. not like they're comparing it to a specific gpu, just what class it competes in
Intel constantly saying "It's a great time to buy a new PC" is them really saying "We are hurting for money and need to make sales. Buy our stuff!". Also Intel not releasing the non-K CPU's and B series chipsets together is because again, they are hurting for money so they are rushing products to market. I am not an Intel guy right now but even if I was, I would actually wait a bit to see how the launches go and find out if there are any bugs (14th gen what?).
Ryzen 10600X and RX10060 For next Gen i forgot B1060 for chipset
As an amd fan ( read: i preffer amd ober the others) its so sad, amd doesn't give a fuck about radeaon.
Its the consumer electronics show. Has anybody seen the average person be excited for AI? Nobody gives a fuck. Id rather not pay for the die space. Disgusting, overpriced, over hype, unnecessary
Do you think 3D cache will become the norm in the future?
The audience was there to hear 1 thing and what we got was droning on about AI and laptops instead.
Tragic how the new performance metric for AMD and Intel is "AIs per minute".
It is almost as if AMD has discovered some catastrophic flaw in RDNA4 at the last second. Like, it causes your motherboard to explode when it hears "AI"...
The product hasn't been pulled, but "later this quarter" sounds like they have some major rework to do. I hope we hear the truth sooner rather than later.
I'm guessing driver problems, like Intel had on their A770 Discrete Graphics Processing Unit Add In Board, at launch.
Point 2, it appears AMD is keeping the RDA4 DGPU ever so slightly below the GRE Edition, so they are not fighting embargoes and Tariffs, both at once.
"gravitic propulsion"?
If real "G. P." Is claimed to only be possessed by the Pentagon and the CCP.
IF true, it could be cold war act two, this time between the CCP and the Pentagon.
Which, while awful...
Is better than humanity so destroying itself and the planet, and everything that a Neolithic Era begins.
I really hope that the NeoLithic era is not the goal of Gates and the WEF, et al.
you cant do any hardware changes on this type of timescale - the chips are all already made - so if there is a problem they gotta figure out how to software fix it
3:50 I don't know if overclocking is the big deal it used to be. At least with the clocks as high as they are on stock processors as long as we can still unlock the TDP to reach the max stock clocks we should be good. Maybe I'm crazy but when I see a 14700 (non-k) pulling 250w I don't know what for overclocking. Also considering the turnover and competition in pricing I was finding Z690 boards for much cheaper than a B660 or B760. I assume due to the market saturation enthusiasts will find Z890 boards a better deal anyway. I guess what I mean is if you get to the i7 level you shouldn't need overclocking.
Man i love the new AI/minute performance metric💀
HAHAHAHA I heard A/O Snatch in a tech video and almost choked. Love it... Don't change the name please.
AMD naming 9070/XT for easier comparison with NVidia cards doesn't _have_ to be an indication of their cards performance, it could simply be to hint at MSRP. After all, they can't be sure what performance Nvidia is bringing to the table with their xx70 lineup. But they could be fairly sure that Nvidia would not shake things up too much when it comes to price. The graphic showing 9070 series topping out at same level as 4070Ti might be more of a hint on price for 9070 than performance. Rumors say 9070 is not a chiplet design, and if true there's really no reason to expect horrible performance per Watt. Then considering that the 9070XT comes with 3x 8-pin PCIe Power connectors. The performance might indeed be significantly higher than a 4070Ti.
All right, Brian, this is it. Get in there and see if you can overclock the sRAM with ASUS AI snatch.
I recall when Byte Magazine in the 80's - remember them? - ran an editorial trashing AI marketing, with a graphic showing a box for a business app with the sticker on it "NOW WITH ADDED AI!". This nonsense has been a plague for a long time.
You really marinated so long on that snatch part lol
The funniest thing of all is that all large companies are talking about "AI" in consumer products, but mobos couldn't be worse for AI: Not enough pcie lanes, not enough pcie slots, no bifurcation and the worse layout ever made.
what motherboard manufacturers being miss leading .. I am shocked
1.5 AIs per minute is rather impressive. It's a spectacular use of a buzz word.
New Intel IHS looks weird though
I suspect the lower-end non-K variants use thermal paste instead of solder between the die and IHS , thus explains the difference.
I'm a little disappointed you didnt give yourself an AI score XD
hahaha good point
“At some point early in the 21st century, all of mankind was united in celebration. We marveled at our own magnificence as we gave birth to AI."
2:50 I totally agree! (checks shipping status of 9800x3d...)
AMD started the confusing since X370 / B350, they skipped 100 - 200 to match with Intel, so where's a lot of confusion then.
However when Intel got to 700 chipset series they had nothing as the AM5 was on X670/B650. So now Intel with 800 so they again, skipped the naming to match.
I honestly have no idea of the CPUs Intel is planning to release as they have changed their naming convention and I am lost on how to figure out which gen they are at.
15:55 They might even just call it the 1070, And no one should confuse it with the nvidia 1070 due to how old it is, and google shouldn't give incorrect search results because the searches would be for AMD RX 1070 or AMD 1070 for short. *shrugs*
Asus is notorious for adding bios sub menus that work around vendor locked features
As a 7900XTX owner, I'm really sad about the FSR4 news 😭😭
Okay Gigabyte, we love the enthusiasm but, the world isn’t ready for AI snatch yet.
Adaxum Defi and E-commerce combo is what sets it apart for me. Time to join the presale ASAP.
Fully agreed with the debugs LED or even better a 2 digit debug code. Cost optimisation are making for worst products unfortunately but I really think every mobo should have a precise way of knowing what is causing any issue in their boot process.
Speaking of weird naming schemes, as someone who isn't in advertising, I'd love to see them just having their product names start with the model year.
E.g. RX2570 - A 70-class GPU released in 2025. RX2670 - A 70-class GPU released in 2026. RX2660 - 60 class GPU released in 2026.
Hell, put any number of letters and digits after the model year. 25750XT. Released in 2025, 750 class, XT variant. Whatever.
Honestly - if they hadn't ditched hyperthreading on core ultra 200 series and I could find a motherboard with a single CPU M.2 slot, an 8 lane GPU slot and three 4 lane CPU slots then I'd probably be getting one as it'd be like pseudo hedt. Unfortunately, I've not found any boards where they opted for a normal slot instead of second M.2 slot or boards with three physical slots from the 16 lanes.
15:10 come on dude, all before 97 is ATI not AMD, some of us still remember the best GPU manufacture that did laps around Nvidia
Already participated in the Adaxum presale, and I’m feeling pretty good about the future of ADX. Time to get in while the price is still low!
I'm really looking forward to the Nvidia AI per minute performance!
For a regular consumer, is there any use of AI hardware capabilities yet? I haven't seen any app or whatever that makes me want to upgrade to an AI pc. I guess the eventual plan is for Chat GPT and stuff to run locally?
Even though this year CES keynotes are messier than others, luckily we always have Roman as an Eierschalensollbruchstellenverursacher to crack into it.
Adaxum token is built for moments like these-alt runs drive adoption, and this project is primed to take off!