Nvidia’s In Trouble - AMD RDNA 5 Leaked Specs
HTML-код
- Опубликовано: 17 окт 2024
- Thank You to PCTechForum For Sponsoring This Video!
Join PCTechForum: www.pctechforu...
Nvidia’s In Trouble - AMD RDNA 5 Leaked Specs
AMD's RDNA 5 based RX 9000 GPUs are set to launch by the end of 2025 if leaks are to be believed meaning it will be releasing a full year before it was originally expected. With that being the case more leaks are popping up regarding its potential design and performance targets. Today we break down the latest leaks to find out just how fast the AMD RDNA 5 RX 9900 or 9950 XTX will be.
Sources
RDNA 5 Leak: • How AMD's RDNA 5 Leapf...
RX 7900 XTX Specs: www.techpoweru...
GPU Chiplet: wccftech.com/a...
FULL DISCLOSURE
Amazon has given me affiliate links which send me a percentage of each purchase when you use them. Additionally all Vip-cdkdeals links are sponsored.
END DISCLOSURE
Specs
Camera: Sony FX3: amzn.to/3VWS45H
RAM: 64GB Gskill DDR5: amzn.to/3F1Pa6j
CPU: Ryzen 7 7800X3D: amzn.to/4czpHAf
GPU: PNY XLR8 RTX 4090: amzn.to/3REwP6n
Storage: Kingston 4TB Renegade SSD: amzn.to/3OXvdlz
Case: Lian Li O11 EVO: amzn.to/3VUscF3
Thanks to Vip-cdkdeals for sponsoring this video. 30% off code: GPC20
▬ Windows 10 pro ($17):www.vip-cdkdea...
▬ Windows 11 pro($23):www.vip-cdkdea...
▬ Windows 10 home ($14):www.vip-cdkdea...
▬ Office 2016($28):www.vip-cdkdea...
▬ Office 2019($47):www.vip-cdkdea...
SUPPORT My Work & Get Discord Access: / thedisplayguy
Music
www.davidcutte...
Outro: Everyone by Monplaisir
Thank You to PCTechForum For Sponsoring This Video!
Join PCTechForum: www.pctechforum.net/
How fast do you think RDNA 5 will be?
Seriously it is NVIDIA. If they really want to make sure that AMD doesn't stand a chance no matter what they throw at them. Then NVIDIA can simply do some Name adjustments and price adjustments and be done with. Where the expected TITAN becomes RTX5090 and RTX5090 becomes RTX5080 and so on. NVIDIA absolutely got nothing to worry about. Because where they may have cut down on profit margins per card to get them stand competitive they will make up largely in sales volume.
talking about RDNA5 when 4 isn't even out yet. what is this a 2027 release?
from what i currently know . Amd wont launch high end gpus with RDNA 4 . they just want to drop a mid range competitor while working on FSR and drivers. while the GPU tech teams is trying to finish RDNA 5 before 60 series launches.
On the chip side , Intel has already made working prototype gens for Backside power delivery , and plants to use it in some next gen chips ( spliting the data and power layers , 1 side power, Silicon chip , 1 side data , currently they are mixed in 1 layer below the Chip).
TSMC isnt that far behind . Amd might just want to launch RNDA 5 with a TSMC Tech Advantage in efficiency and clock speed . Useing N2P node with backside delivery for HIgh clock speeds and power efficiency , as they perfect the Gpu Chaplet architecture in the meantime .
N2p node launches in 2025, and the backside power delivery launches in 2026. They can do a small 1 y gen and Express line Ship RDNA 5 in mid 2026 to early 2027.
@@snakeinabox7220 that makes no sense. how are yopu going to grap the "data lines" on the side where the cooler is suposed to be or the other way around? and the "mixed in one layer below the chip" are actuly many layers, same as the chip it selfe. a chip is not just one layer...
This guy “graphically challenged” is a clown. He’s really mentally challenged.
@@georgwarhead2801 I recommend watching Asianometry backside power delivery video hes going in to the detail how its working even tho I watched it two months ago I cant accurately say how it works.
He is a tech channel, he is commenting on leaks by other sources and downplaying the exaggerated specs of Rdna5 and making educated speculations on what target they are going to hit realistically. Rdna5 is hitting 2026. 18 months is reasonable time to be discussing new hardware.
RUclipsrs: "Nvidia is in BIG trouble now!"
Meanwhile: Nvidia worth is more than all competitors combined..
Well , MI300 from amd is eating at Nvideas Ai market , its not like they are That Far ahead in tech, They are ahead . but the gap isn't as big as the marketshare and Worth suggests .
sounds like what intel fanboys were saying before zen 1 came out………also zen 2 wiped the floor with intel.
i still say rdna 5 is definitely 💯 going to compete with nvidia’s top end.
@@chrispauldupre Nvidia doesn't sleep like Intel. LOL.
@@nossy232323 well 2000 series was 100% skylake moment for nv. They just did not sleep for 5 generations like intel did, waiting for fab to work.
@@chrispauldupre Yeah, no. Blackwell will still outsell rdna 5 with a huge margin. AMD cannot outsell Nvidia for many years. And that trend is not slowing down with Blackwell on the horizon.
NVidia is never concerned. It never hurts their sales no matter how good AMDs GCs are
cause they are selling gpus to ai researchers and not gamers
@@gs-nq6mw well even if we are talking about gaming alone, AMD gets stomped on in market share lol. Probably more 20 series owners than all AMD gpu owners combined.
it would if they were still marketing to gmers
Impressive how often you Think NVIDIA is in trouble ☝🏼😅😂
800$ amd 4090
Apperantely nvidia is in trouble every generation amd is threatened you know...
He even lied about the "many problems" he had with his RTX 4090 card!
@@gamingcomputers7485😂😂😂😂 yeah it will be funny to see but good for us
if being the most profitable company in the whole is a problem hes right
How many times have we heard this in the last 6-7 years? '1080Ti in trouble', '3090 Killer', it never ends....; and is all BS
they cant even beat a 4060 wym
@@iikatinggangsengii2471 the 7800XT is nice....
To be fair, I recall a particular model of 7900xtx that, when pushed, it could match 4090s in performance (iirc it was the Aqua model from Asrock) but then nvidia has all the bells and whistles. All AMD has to do is figure out how to make those bells and whistles play nicely with their own hardware.
The RX6950XT did beat the 3090 in most titles .
I reminder when AMD had the 9700 Pro. It was the 1080 TI of its time. AMD had the lead for 2 years.
Yeah... So have you and alot of other content lying creators said since 2019.
It's rumors and speculation not lying stop interrupting everything as facts 😂😂
You don't interrupt as facts. You interpret as facts.
Actually AMD had the better product many times now like 6900 XT. NVIDIA is not in trouble not because they have better products. It is because they have a tight grip on the industry as a whole. Ex scamming consumers, and blackmailing companies to promote their tech or else. Will give some examples. Self destructing GPUS on MAC's , XBOX, PLAYSTATION, and PC (do your search i won't provide links). Connectors which catch fire on supposedly top of the range 4090 GPU. Scamming consumers on 970 GPU with less usable memory than advertised. Forcing closed technology to cut out competition (GSYNC, PHYSX, CUDA). And many others i forget right now. So when AMD released the 6900 XT NVIDIA released an overclock ready to burn out product so they keep the first place. But in reality if you want quality GPU that will last you should get AMD.
But do you interrupt with counter spell, shield slam, or kidney shot? Who’s side are you all on exactly?
@dfloper AMD makes the gpus for xbox and ps5 you goober and the 6900xt was never better than a 3090 at anything.
Can't believe how often Nvidia gets in trouble
"Nvidia should be concerned", cannot tell you how many times i've heard this but always hope Nvidia will be pushed. I just got my 7800X3D to replace my 14900k last night. I love to see what AMD got ready to slay Nvidia, give me a reason to not use my waterblocked 4080.
How you like the 7800X3D? I’m considering making the same switch but not sure if it’s worth it in my case.
@@siggspeed1665 so far I’ve had no issues. Being a custom cooling system I pull only like 47w under gaming, the frame latency is fluid and smooth. I play on a OLED1440p 240hz panel and I see any type of judder or shudder.
Overall I felt the same way changing to 7800X3D but benchmarks and talking to other people including now being a 7800X3D owner is all confirmed facts.
All my fans run at slow speed even at 100% load. Never saturates my cooling loop and room is cooler. Less heat, less stress and FPS WAYYY better consistently.
So honestly me to you, yes it’s a big change. The design is basically a run away diesel motor. It will boost and push constantly until it hits a TJ max on temperature. Believe 90 but max is 95c.
On a custom cooling loop, well worth. Half the price, I feel double the stable experience. Intel gonna have to do some god sent magic to win me back.
Have you heard the latest on the RX 10990XTX-X? It's a gazillion times faster than a R9-270.
Nvidia will only be concerned if AMD releases a card with 48 to 64GB. Now that is what I would expect to pay $2000 for.
The problem with monolithic die designs....the larger you go, the fewer die's you can fit on a single wafer....AND the worse your yields will be, which is cost that's pushed onto the consumer. The 7900xtx's GCD has an area of 203 mm sq, a 30 % redection in die size would put that at 140mm sq for 48 WGPs....the die size of the 4090 is 609 mm sq...140 into 609 is 4.35, which would mean that just by moving to a smaller node, they could fit 208 WGPs into an area the size of a 4090.
The MCD's in the 7900xtx were manufactured on a 6nm node. Moving to 3nm, offers a logic density increase of 1.7x, which would shrink the MCD area down to 197 mm sq. 806 mm sq of silicon isn't that much when it's split across chiplets....but ok maybe it is. So subtract that MCD area from that 609 limit, and you're left with 412 mm sq. A 2.9x increase over the 7900xtx, which would still put in a 140 WGPs.
But wait! What about Zen 4/5 vs Zen 4/5c? Same architecture, identical performance clock for clock, 25% reduction in size....one would think they'd employ the same design philosophy into their next fence swing of a GPU, right? Which would put it back up to 175 WGPs in that same 609 mm sq area.
But all of that is theoretical, right?
Lets take a look at what we actually know.
So how about this, go take a look at the die of Strix point that shows off how each component fits into the die. The total die area of strix point is 232 mm sq (a 54mm increase over hawkpoint due to the inclusion of 4 additional CPU cores, and 2 additional WGPs) Strix point has 8 WGPs....and takes up 11% of the die area, or 25.52 mm sq....meaning you could fir 72 WGPs....into that 232 mm sq package...609 / 232 = 2.625 x 72 WGP - 189....and strix point is manufactured on the 4nm node! 3nm offers a density increase of 10-15% (and AMD has committed to sticking with the newest node for each new product release going forward) which would mean they could get between 208 and 218 WGPs into that 609 mm sq die area, just based off the manufacturing tech being used today.
And keep in mind, strix halo is set to move from 12 cores to 16, and from 8 to 20 WGPs, double the cache and a larger NPU....because where strix point is a monolithic design, Halo is moving to chiplets.
C cores are much smaller because they are condensed down and l2 cache is halved. That condensing put hot parts a lot closer together limiting speed. And gpu's do cache very differently. I don't think doing the same thing for gpus would be a win. at best some things might be faster because more simultaneous calcs matter more than how fast they're each done. But the drop back to 2ghz would likely hurt performance in to many things. And because there is so little cache at that level you wouldn't get that much area savings, 10-20% if you're lucky. I'd choose a different route.
Ain't reading all that
@@m3rcf1ve53 Don't imagine you do much reading at all.
@@kaseyboles30 Zen5c has a 20% clock reduction compared to the full fat dies, but the actual Zen5c CPU cores take up 25% less area compared with the full fat zen5. Cache size is not relevant to the point.
Leaks on strix point, with it's 25.5mm sq of gpu power, put the gpu clock speed over 3ghz, on par with a 3050 laptop gpu (with a die size of 203 mm sq) and 10% more processing power than the ps5, which has an apu that's 308 mm sq, and consumes 350w when going full tilt.
They're already doing it, and it's working.
Interesting points, sounds all good in theory, but I'm not buying it based on history.
Ну 600 ватт энергопотребления, это уже ни в какие ворота не лезет. Сколько будет требовать система с флагманским CPU и такой видеокартой? 1500 ватт? У нас в России в старых домах проводка не выдержит.
These AMD videos are a meme at this point, like flat earthers declaring they finally found proof.
I'm pretty saure RGT said that they weren't going to be making anything to compete at the high end and that they weren't going to beat the 7900xtx. He's been saying that over and over again.
That's for RDNA 4. This video is talking about RDNA 5, which is ridiculous because that's at least 2.5 years away.
@@edragyz8596 not really... RNDA 5 is the clean slate design which is already overdue. It maybe come out either next year or 2026. GNC only lasted 5.5 years... Terascale also lasted 6 years. So the have an established cadence to this change.... 5-6 years. And we just passed 5 years now... so either next year announce at computex and launch in the summer because we can expect delays. Or they will stretch RNDA another year past its due date. RNDA 4 is a relatively minor chance probably so the driver team can be free to work on RNDA5 etc.... which is probably the focus internally right now. GDDR7 is supposed to be available Q1 2025... which means GPUs start showing up around Q3-Q4. So based on that the latest we can expect RNDA 5 is Q1 2026. AMD is being tight lipped on this because people would definitely stop buying GPUs if they knew a 2.5-4x faster GPU was coming down the pike.
2:43
Why even mention this?
THIS JUST IN: GPU GENERATION BUILT TO COMPETE WITH THE RTX 6000 SERIES IS FASTER THAN THE RTX 5000 SERIES OMG GUYS WOW
That's why I unsubbed from this dummy. Absolute worthless and baseless info for clicks.
Let’s put that mustache on the line? lol
When has Radeon's flagship ever beaten Nvidia's flagship?
7970 I think
Why talk about RDNA 5 when we could be scouring the internet for RDNA 6 rumor releases of 2030 🙃
that TDP tho...
imagine the power bill
Nvidia have nothing to worry about 🙄
Amd fanboys have been saying this for years. Still waiting.......
RX 9900xtx 32GB released, hell freezing over, guess which one will come first.
NVIDIA won’t care. They’re too busy making billions serving the AI market. AMD may grab gaming market share and that’s probably a good thing for market health.
The only way AMD can compete and eat away market share is if they put performance close to a NVIDIA card, but price it several hundred less.
They already do that. They just need to keep getting better at integrated GPU with their killer CPU, continue do dominate the console market and try to eat at the professional AI market-share and they will be ok.
Gaming market is no longer a big player is not even where NVIDIA is making bread.
The more I think of about it. The lower the clock speed necessitated by the infinity fabric the more cores they will have to use to compensate. Further each CCD (GCD?) will have to have cache memory on board to alleviate some of the latency issues. I think we will get a good hint from how Strix Point and Strix Halo perform.
As a disgruntled RTX 4090 owner with a fire extinguisher next to his PC. I am happy to wait for something that might beat the RTX 5090, but I also want some AI capability, how long will it take for that to work with cards that are still just rumours.
All the potential in the Gpu market is centered around Intel Arc and XeSS. Radeon is going to be in 3rd place soon enough.
I don't care about the 9950XT, I care about the 8700XT and 9700XT... Everything above them shouldn't exist...
And 95%+ of people buy GPUs that are 500$ or less...
Give us an efficient, small, fast 250-500$ GPU line up that tops out at 200W and a dual slot, standard length and width card.
I heard Nvidia's in trouble about 5 times this year. yet I don't know what it is.
LMAO Nvidia will always be better than AMD. People been saying this for 6 years and AMD still hasn't caught Nvidia.
GPUs were never a priority for AMD. GPU chips are 2-3x larger than CPUs and so expensive to make. If nearly all fab capacity goes to making Ryzen, AMD stands to gain maximum profit. I'm fairly certain that only 10-20% of AMD's fab allocation at TSMC is used for making GPUs.
You forget the MI chips
we had crossfire / SLI / NVLink. it means that having multiple GPUs can increase game performance. if you move those GPUs closer to each other it decreases latencies and will lead to even higher gaming performance.
The thing is that stopped because even when it was done on a single card the micro stuttering was still annoying and getting even 1.8 performance bump over single die was amazing with it often being as low as 1.4x.
It's the latency. On die latency is a cycle or two. The two chips one card latencies were in the tens of cycles. Now two dies one substrate needs to get down to just a few cycles and maintain a huge bandwidth, bigger than the memory bandwidth feeding them by a healthy amount IIRC.
This is non-trivial. Why do you think Intel is still on monolithic dies this long after zen1? And zen's multi die approach is much simpler and easier than it will be for gpu's almost all the extra latencies are where latencies are fairly high to begin with, between the cpu's proper and the io subsection. Zen is Man on the moon, gpu is man on Mars. Neither is trivial.
Great vid thanks mate! I reckon your best yet loved it! Thanks
I love how everyone is “team red vs team green” when realistically you should be supporting the company that offers the best performance per dollar (which currently is AMD, 90% of the performance for 1/2 the price) and encourage the competition. When Nvidia rules the high end market like they do currently, they can charge twice the price as a similar AMD card just because the demand is higher. Same reason Apple can charge an arm and a leg for a similar or inferior product (I’m an iPhone/applewatch/airpod guy and still admit this)
How is this AMD GPU going to be for gaming vs the 5090 or Titan
nvidia has been in trouble for 6 years now
Dude AMD is always playing catch up on nvidia high end cards, these click bait videos are cringe
0:38 That was a little bit uncalled for, Pc enthusiasts aren't *that* sad.
192 CUs makes way more sense than WGP that would be an absurd jump in compute units even with a massive node shrink
Hold your horses - -30% for new architecture, -90% fps for some games. Jensen might still invent even new path tracing that halves perf on AMD. I believe in 360CU leak 384bit, 9CU blocks, 120cu per block, for ps6 and integrated, same chip for discrete and integrated.
It doesn’t make sense accelerating RDNA5 unless they’re also accelerating RDNA6. Little to no benefit if the 60-series comes out 6 months later with RDNA6 two years away…
AMD is just innovating so much more then these other companies, They took on intel which they are now winning and now going hard on NVIDIA
Even if AMD comes out with an RDNA 5 ... it wouldn't really compete with the 50 series but the 60 Series (Vera Rubin)
Is that GPU config for cards or for APUs.
GPU
AMD may need an uber Frankenstein card to shut down Nvidia marketing. Just provide it to a few selected influences and no GeForce card would ever appear again in benchmarks.
In order for AMD to Seriously outperform the next Gen NVIDIA GPUs they first have to beat the current, which they havent as of yet. AMD has consistently dragged behind NVIDIA with roughly a full year a part. So the chances of the RDNA5 to even be at about the same level as the New NV chiplets is a very odd ball. It could be but is unlikely.
To be exact, this year and the next NVIDIA can afford doing absolutely nothing and still sell the 4000 GPUs and skip the 5000 series as of whole to use resources to work on the 6000 Series , this is how far a head NV is.
That's years away yet.
And Nvidia aren't really concerned about graphics. They could easily sell every chip as an AI chip.
just brought a 6750xt should i return it ?
nah
I dont like the way hardware is advancing as it causes hardware and software to be obsolete for current gpus sooner and thus you have to be new gpu earlier than normal
I hope. I got stocks in AMD now 😅. Competition again would be really nice, but do you want to wait for rdna 5 when the 5000 series will be out a year before?
I dont know if even AMD herself has plans about RDNA5, but here we are talking about it and hoping it will outperform the previous nVidia gen
Plans? They already have primary designs and simulations and will probably have some early in house test silicon in less than a year. RDNA 5 started before the rdna 3 came out most likely. Todays cpu's and gpu's have several years lead time. They don't finish a before starting b. more like they finish a while doing earliest testing on b as they do preliminary work on c and start brainstorming ideas for d. Back about 15-20 years ago Intel had a leapfrog team set. one team would be designing the next new arch while the other team was creating a rework and refresh of the current gen. This was their Tick-Tock strategy and then bulldozer made them lazy.
I have no interest in video cards over 300 watts. Remember back in the day the highest card never went over 300 watts?
I have a hard time even going over 200 personally. I want 4080 performance for a bit over 200w when undervoltes and maybe power limited slightly
@@ofon2000 Yeah I prefer high end to be 180-250 watts. I had a 780 TI that was great for 5-6 years before it glitched out and it was 250 watts. My current 1070 TI runs everything I play great in 1080p and it's only 180 watts. 300 watts should be for 90-90ti cards stock and let them be o.c. able to 350 limit. Cards stock at 400-600 watts is just insanity.
@@Jolly-Green-Steve yessir I'm on a 3060 ti atm. Not a great GPU...but it's OK for now at 1440p.
@@ofon2000 What games do you play that it struggles with? I don't play newer high end games because I'm on Windows 7.
Some people would buy a gpu that's $2000 that would run 8k well, cause I think a gpu like the biggest one supposed here would be able to run 8K at 60 fps.
Will it have a better video encoder and software suite?? Like will amd fix their horrible encoder starting at h264 then h265 then av1..
AI bros want VRAM more than compute, at least for LLMs not art generations. Just make a slow but efficient GPU, quite cheap, with 48-128 GB VRAM.
Nvidia will likely never have anything to worry about from AMD graphics cards. It’s just wishful thinking on the creators part at this stage.
GC Guy is now breaking the sound barrier with his underdesk-up head-swoosh ... Only Einstein's laws can limit the progressive acceleration with every video...
Will wait...though the 7900 xtx is at 62 TF and my RX 5700 is around 8 TF which theoretically is almost 8 times as much...though in benchmarks it is barely over 2 times faster at 1440p?
When is AMD going to publish consistently stable drivers for VR?
What about RDNA 9? Since we are just making things up here? Lets focus on the next one which still isn't releasing for another year. A lot can change. I hope AMD gives Nvidia a run for their money tho.
X5d, 8 chiplet matrix and 64gb gddr10x+ 7gb/s AI quantum cores
I remember you talking about RDNA 4 like this right when the 6950 XT was released. Just, stop misleading people. RDNA 4 will be DOA and RDNA 5 looks like it will be the same.
I will believe it the day I see it..
Do I think Lisa Su could pull it off? In time, yes.. But definitely not now..
AMD does not care about gaming, this is probably a CDNA5 leak not RDNA5
You have to remember AMD's current die is only 300mm. while this is the same die class as the 3060/4070TI you have to remember they broke the cache and memory controllers into their own seperate dies to get that die so small, but just because this is chiplet doesnt mean physics doenst exist any more, they can nearly tripple the die size of the current card up to ~830mm, but they'd NEVER do this for a consumer part.
I honestly would not be surprized if the 9950XTX was 250mm with 70WGUs/140CUs on 3 nm still with 384 bit bus
are they skipping the 8000 series?
N51 has 3 GCDs (each GCD has 64 WGPs) and 6 MCDs so it has a 384 bit bus and comes with 48 GB of GDDR7 memory. each GCD will be smaller than a 7800XT GPU. it supports hardware based ray tracing. it is faster than the RTX 60 fully unlocked flagship and it will be released after RTX 50 but before RTX 60. it will cost over 1500 USD with some AIB models costing 2000 USD.
At this point I feel like you're click baiting...
"NVIDIA IS IN TROUBLE", meanwhile on track to become the richest company on the planet. Hard to take you seriously with garbage like that.
I mean... Nvidia makes GPUs that sell for $2k+ there is no reason AMD cannot do the same. I mean who are we kidding Nvidia sells the 4080 for $1400... AMD could double the silicon on next gen GPU and still turn a hefty profit at $1500.
By the time RDNA 5 comes out, nVidia will be up to their next generation. I personally think that AMD will close the gap a lot on that generation, and may even slightly edge it in some areas, but it's not a slam dunk wipeout of nVidia.
I would drop Nvidia in a heartbeat if AMD could get near the 4090 Without the Envidia fire hazard. I have a 4090 right now and I can’t walk away from my computer for more than five minutes with fear of fire because yes, I burnt one already.
imagine if the 9900 xtx becomes a reality, could that possibly push the prices down for the 5090 to 800 dollar. a 5090 would then be midrange hehe
Unlikely, I believe Intel Arc has a better chance at spanking Nvidia, eventually they will be dethroned. The hubris of Nvidia will be their downfall.
@@xlr555usa you are probably correct.
nvidia has become to big and to corporate and are on a downwards spiral, AMD has yet to prove themself in my eyes.
intel has some great software benefits and are aiming for good hardware too.
but also intel cpu division issues might draw a lot of recourses out of arc division and they can not fail at both and survive.
the only downfall of MCM being......it's a little tricky to undervolt, 100 % usages stable. Depends of binning surely, but not only.
Apart this,......i guess no real flaws, good and "affordable" (started pc gaming in 1990....😢) cards
But since Nvidia's pricing policy,.........why buying them ? next step 3000€/$ a gpu for "gaming" ?
Jensen is crying right now and wiping his tears with hundred dollar bills.
yeah sure one of the wealthiest companies in the world is in trouble because some dude constantly overhyping the upcoming AMD products said so once again
We've heard this before. rDNA 4 halo product ... Canceled. Amd is more likely to find other ways to increase performance. I look forward to better ray tracing from them.
I think it would be cool if something like this came out. But i feel like AMD never hits the targets people talk about. I bet if anything its gonna be 50% More then a 7900xtx.
he forgot about RTX6000 seires release same time as RDNA5....while RTX70 is been work on same time that supposed to be compete with AMD 11990 XTX ??
Seems less than 1yr ago that AMD publically stated they would not compete with Nvidia in the Fastest GPU market. Of course, I'd love for that not to be true!
decided to opt for the RDNA4 highest model GPU they offer next year not gonna bother with RDNA5 i'll probably get w/e comes after it tho so theirs that but decided to focus around Mid Tier no longer interested in 4080/W.E. is next up the change type of cards.
edit: so basically anything less then 20gb vram is good enough for me going forward which in theory should be their next highest mid tier card for hopefully under if not around the price range of the 7800XT launch price
Fool me once, shame on you; fool me twice, shame on me.
I think AMD should say screw it for once and create the biggest and fastest GPU they possibly can. At least give it a try.
Yeah, and almost no-one would buy it because it would be too expensive. The thing AMD needs to do, is to not try and get NVidia's profit margin on GPUs. They designed GPUs with chiplets on different nodes to make manufacturing cheaper. And instead of using this to sell GPUs for less, they used it to raise profit margin.
@@korinogaro AMD needs RDNA 5 to be a monstrous performance increase over RDNA 4. Otherwise, they risk becoming irrelevant to performance-focused gamers such as high-end and enthusiast-class gamers, and that's what will stop them from gaining a considerable amount of market share in the gaming GPU market. They need to aim for the performance crown for once. This is why most consumers don't take AMD seriously and why they are not always considered proper competition against Nvidia.
@@korinogaro you know that "apperantely" a cousin will follow his other cousin that's what you should know because amd obviously don't want to compete so they'll keep go with this practices no matter they are thinking high in their head with a lot of arrogance and pride.
@@Gamer-q7v they not only must be faster in raster, they must be at least as good in RT and upscalling as top NVidia's GPU and on top of it be cheaper. If they actually want to un-charm Nvidia's grasp on high-end customers. And if they don't care about it... ok. But then to make market gains they would need to sell mid and low end for waaaaaaay less than NVidia. Problem is that AMD has *etarded marketing strategy. They allow way too much freedom in pricing to retail and not compete with it in any way (AMD products on AMD website are more expensive than in store). This causes them to lose market because retail tries to milk the market as much as possible when new product comes. Then their rebate politics kick in and AMD pretty mcuh starts to sell CPUs for DIY market with no profit. This is why they suck at GPU pricing.
They already did. And failed.
Ahhh… that sucks for me if the rx 9950 xtx is better than the rtx 5090, because I would love to have a AMD gpu that’s powerful, but waiting a year from now or more is to long, so I’m gonna get the 5090, I have a 4060, and I could hold out but… no one knows me for my patience
Would it make sense as an enterprise AI product they could sell for $30 grand?
AMD diesnt have to compete with nVidia at the high-end.. onlybinnthe Mid-range Value segment.. and enough t always keep a strong lead on APU Integrated Raphics CPUs.
You misunderstand the actual AMD Graphics Strats.
amd needs to gets stronger with price to performance
The only Nvidia killer may be their 12 pin connectors 😢
Would just need a new gpu to power g9 57 ultrawide so either 9900 xtx or blackwell titan water cooled with alphacool 600 (9x200mm fan radiator, fan, tank and pump
not sure if I agree with how it would look like.
there is a chance AMD might go for smaller "chiplet" if ratios would be better. IMHO we might see 2/3/4 x 24 designs, as they all will take 4 dies of ram each. at least one of them will be soaking the worse yields, so it will go to 20 or 16 WGP per chiplet. (maybe 32, but then just add +8/4 to all numbers mentioned)
Chiplets with 64 WGP's would take 8 dies of ram each, and that would be absurd amount of connections for small chip.
smaller is easier to route, cool and control, and they already have experience from EPYC, with running swarms of smaller bois.
whole difficulty is not the WGP chiplet, but insane bandwidth "core" that would have to feed them.
number of lines that's needed for it, means host board for such gpu is as thicc as a bible.
Yep, I'd say what AMD needs most of all is a massive jump in their fabric tech. This would also support higher ddr5 speeds for their processors at 1:1 ratio and reduce inter-ccd latencies. I'd be surprised if that wasn't a priority for them.
If AMD ever came out with a card that could crush nvidias it would be probably cause a rip in space time.
9900xtx in mini itx build what I wanted but I need to think twice.
The 7900 xtx has 96 compute units. The 6900 XT has double the compute units of the 6700xt and is 53% faster. So the 7900 xtx has 20% more compute units than the 6900 xt and it is 47% faster. So let’s say rdna5 has the same performance scaling per cu as rdna3. Doubling from 96 cus to 192 would make the 9900 xtx around 220% faster than the 7900 xtx or roughly 30% faster than the RTX 5090. The RTX 6090 would be 20-30% faster than the 9900 xtx.
This assumes no ipc gains for rdna5
I'd believe 192 CU over 192 WGP. That number seems absurd.
192 CU would make sense, that's still 3x CUs of Navi 48 which is rumored to be faster than a 7900XT. Then again with rumors of 3 GCDs going around and how small N48 is said to be and the fact that it will be based on N3 so each GCD will be way smaller than N48 with only 64 CUs each, so 96 CUs per GCD is more likely if AMD wants to crush blackwell refresh, so 288 CUs is more likely, 192CU will only be around Titan performance (tripling CUs usually only gives 2X performance in non-compute tasks because of power limits and scaling gets worse the more shaders there are.)
Can we get RDNA 4 first? Why are we skipping generation? 🤷🏽♂️
How much will the electric ⚡ bill , reach at the end of the month?
Believe it when I see it. I do hope there is competition though.
But look at the TDP number if they are true who would want to by this? Unless you plan on keeping your house warm during winter lol
Didn't you tell us last week rdna 5 won't even be better then the last gen flagship. Makes me think we can't trust your sources! Cancel this guy
amd never got the idea of under promise and over deliver
For the 1 millionth time, CUDA.