Yes, an announcement of a product announcement presentation, haha. Very exciting!!!
Hopefully the trend of bring quality cheap stuff continues, I got my 8 core AMD CPU that's 50% the price of an Intel one. Lets see if their GPUs also 50% cheaper.
It won't be anywhere near half price going by current pricing. NVIDIA had their last gen RTX artificially high due to lack of competition, but AMD brought their current cards out to be barely competitive with those high prices.
The next gen AMD cards have to compete with what looks to be much, much better NVIDIA cards. That 3070 in particular looks to be solid 4k gaming card. We'll be lucky if AMD has something with that much power, and even luckier if AMD can match, or top that price to performance ratio.
But would you buy AMD if it was only 10% cheaper? I'm not a charity, it would have to be at least 30% or more for me to consider it. The NVIDIA ecosystem is definitely worth more than whats' just the sales tax difference in most states depending on online store.
You're saying that AMD would have to sell their products at a level that wouldn't provide them with any profit margins in order for you to consider them. Just say you're not interested 🤷♂️
He's not wrong though. I don't think 10% cheaper will win many people over. I think if they want market share they'll need to be 10% cheaper and/or 10% faster and have 16gb of ram instead of 8 or 10. But I don't see how they can be cheaper or faster and have more ram and still have respectable margins.
The top-end Big Navi will almost certainly have 16 GB of RAM, according to reasonably trustworthy leaks. Only the top-end though. In performance it will probably be targeted against the 3070; *if* it can even compete with it, that is. The 3080 is a beast and to make matters worse (for AMD, not for us) it is priced reasonably for a change.
If the top-end Big Navi has roughly (i.e. within +/- 5%) the same performance as the 3070 and is priced 10 - 15% higher I believe the double memory space will be well worth it and thus would be a better buy. It would be $50 to $75 more expensive but you would get 16 GB of RAM instead of 8 GB, which would be unheard of for a consumer card at that price. If it is 20% more expensive or more I think I would choose the 3070.
p.s. I doubt the top-end Big Navi will cost less than $599 though. Even at that price, if it's really going to have 16 GB of RAM, AMD's margin should be very narrow. If AMD cannot compete in performance I expect their marketing team to spin their fatter memory in a variety of ways..
@andrewaggb - You wouldn't be won over by 10% cheaper for roughly the same performance? That's fair enough, but I absolutely would be. Same goes for 10% cheaper with 10% better performance, but I have no idea why they'd charge less for a superior card. Demanding that that they be faster, cheaper and more RAM before considering a purchase would strike me as an expression of brand loyalty more than anything else. Don't get me wrong - I acknowledge that's a big factor in the market at large - I just find it weird hearing it from "enthusiasts".
As for how they'd do that anyway - they have two potential assets here: 1) TSMC's 7nm process. It's more mature and has a higher density, so in theory, with the right chip they could get better yields than Nvidia and a higher proportion of chips that perform well, reducing the costs of their highest-binned GPU. 2) We already know Nvidia's driving their designs hard - early testing indicates that they appear to have burned much of their gains in performance per watt on getting *even more performance* out of them. AMD could (potentially) produce cards between the 3060 and 3080 performance levels based on smaller dies that require less power and have lower cooling requirements, much like they did back with the HD 4000 series. Adding up savings on the GPU itself, the board design and the cooler, that could come out quite nicely in their favour - *if* that's what they decided to do.
They don't need margins at this point, they need mindshare, and market share. Most people aren't brand agnostic and buy what's popular.
The big question is how will the 6 series perform? It needs to match perf & undercut Nvidia and sacrifice margins to do so if need be. ZEN can easily cover the loss. Otherwise they wont gain market share.
"That 3070 in particular looks to be solid 4k gaming card." Yes it does, but oddly the 3080 has even better bang (FP32 TFLOPs) for the buck, which I don't think has ever happened before. I concur that AMD's next-gen graphics cards will apparently* compete with the 3070 at best. The top Big Navi card will of course has twice the memory but it is unclear if it will be faster.
Some early leaks suggested that the top-end Big Navi will handily beat the 2080 Ti in performance, but according to the performance values of these leaks (which, I know, should be handled with a bucket of salt) that places it around the level of the 3070 in performance, at best. So, once again, the 3080 and later the 3080 Ti will be way beyond the reach of AMD, in a continuation of the status quo of the last several generations. So the only reason someone might pick AMD would be the much larger memory space at a far lower price than the 3090.
*Unless of course AMD surprises us all and pulls out a dual GPU chiplet based beast of a graphics card, which I strongly doubt will happen. Most likely both AMD and Nvidia will move to a chiplet scheme from the next generation (AMD from the next-next, i.e RDNA 3). Nvidia's Hopper will almost certainly be chiplet based; it makes little sense to fab huge reticle stretching dies with poorer yields, and besides Intel's first generation of graphics cards will be chiplet based anyway. If Nvidia and AMD do not move to chiplet based GPUs Intel might steal the performance crown from them simply by adding more chiplets.
Where's the information about the 3070-level performance coming from? The only things I've seen are from quite a long time ago - further back than would be useful for predictions now.
Based only on the known performance of Navi 10, the most likely configuration of a Big Navi (80 CUs), and the claimed improvements in performance per watt for RDNA 2, we're most likely to be looking at something that approaches 3080 performance levels in a similar power envelope and potentially a slightly smaller die area. Bear in mind that AMD don't have to match Nvidia's claimed FP32 performance to match the card's actual performance in games (there's clearly something unusual going on in that regard, which is why Nvidia are suddenly so keen on that metric).
AMD are extremely unlike to compete with the 3090 - but they don't really have to.
The really big question is how they do on RT performance - we could easily end up with a similar situation to the early days of DX10 tessellation here, where a different (and weaker) hardware implementation combined with games being designed for Nvidia's setup results in impaired performance on AMD hardware.
Looking forward to seeing what Zen 3 and RDNA2 can do. I'm quite worried for AMD that choosing not to have any comparisons out before the Ampere launch is going to cost them a measurable chunk of sales, though. People on the fence about buying Ampere might well decide to just do it and take Nvidia's proven performance to get a GPU 2 months sooner than AMD.
To my knowledge it is Samsung's 7nm node that has poor yields, which is probably why it was not chosen by Nvidia. Their (standard) 8nm node is just a souped up slightly denser 10nm node, which Samsung have long experience with so I don't see why there would be yield issues. The 8nm variant they fab Ampere with is probably not standard 8nm but 10nm optimized for high power GPUs.
I heard Ampere has a transistor density in the 50 to 55 million transistors per mm^2 (MTr/mm^2) range. Samsung's (mobile oriented) 10nm node has a transistor density of 52 MTr/mm^2, so the densities match. Samsung's mobile oriented 8nm has a transistor density of 61 MTr/mm^2. That's less than 2/3 the transistor density of their 7nm node (for mobile), which is 95 MTr/mm^2, while Ampere's 8nm variant is almost half as dense.
Transistor density is the primary factor of low yields at fabs, therefore I strongly doubt the 8nm variant of Ampere has a yield issue. Capacity, on the other hand, is another matter. Can Samsung fulfill the demand for Ampere? Judging from its reasonable price and spectacular performance increase over Turing I kind of doubt that.
edit about that "Transistor density is the primary factor of low yields at fabs" bit : The No1 factor that affects yields at fabs is arguably die size. Transistor density should be secondary to that (provided it is not unreasonably high).
Wasn't Samsung's 10nm node fairly underwhelming, though? I seem to recall SoCs based on it being more dense than the previous node, but without any improvements in power characteristics. That would explain why Nvidia are having to push Ampere past the edge of its voltage shoulder to get the performance they want (which is usually AMD's thing to do when they're playing catch-up).
i have a feeling thats thats the point. i'm thinking RDNA 2 will be competing with the rtx 3070 and not so much with the other 2 higher end nvidia GPUs. That still could be good news if the card in question ends up being priced less and with similar perf.
My thoughts? Unless you absolutely need a card immediately, you'd have to be an idiot to buy before both have had benchmarks come out from reliable sources.
How are they supposed to release benchmarks against an unreleased product? These cards are going to be selling for at least a year and probably two, it's better for them if they get the product right. Also, it's likely they wanted to avoid Nvidia just upping specs to beat them before release like happened with the Super series. Nvidia won't be able to just kill their own cards a month after they're released.
@mikeatx - I don't recall AMD ever having released benchmarks for an upcoming product before the product's launch, so I'm not sure you can make that judgement.
A lot of the people who'd buy Nvidia before AMD have even launched their competition would buy Nvidia anyway, though - so it's debatable just how much of an impact it will have. Can't argue it wouldn't be better if they'd been able to launch first, though.
You'd have to be an idiot not to buy *this time* before supply dries out. It's simply a no brainer if want to spend the money. Waiting a few weeks literally doesn't do you any good.
Only if you gotta have it at launch. NVidia will make more chips, the supply will stutter. It isn't like you won't be able to get a card ever again if you miss the launch window.
Obvious shill is obvious. When an account pops up and starts yammering about nebulous AMD driver issues "every day" and then says to buy a card before it's even been reviewed, you're not left with a lot of leeway in interpretation.
Waiting a few weeks literally doesn't do you any harm. Nobody buying these for games *needs it now*.
They tried this "just buy it" crap when Turing lunched. I didn't buy it then, I won't now.
Ampere is using Samsung 8nm node which has absolutely horrendous yields meaning Nvidia's 3080 and 3090 will have more restricted stock than the AMD's own past Radeon VII at launch. Especially when pushing the power envelopes like they are, one must question stability and longevity. Many RMA's to come. AMD is using TSMC's 7nm enhanced node which is yielding above 90% already. AMD, despite a later launch will have far more cards out by end of year than Nvidia and psooibly for the foreseeable future. As long as AMD doesn't shoot itself in the foot (or face, let's be honest) with bad drivers again, they have a good shot at effectively competing in market share.
I think the 3070 will be a far more dangerous comeptitor if they launch an additional sku with more video memory. That would be the sweet spot for PC gamers for sure. We will have to see what AMD's answer is. I hope it's good. Let's get some good hardware for good prices.
I'll like to add that AMD as Lisa Su has stated...likes to package performance in smaller packages. I'm quite sure that AMD can get equal performance out of smaller dies. The 5700XT was a small chip and packed as much performance as 2070 Super/2080. I think you need to stop underestimating AMD.
That comparison's not really valid, though, because they're manufactured on very different processes. A die-shrunk TU106 would likely be smaller than Navi 10, while the TU104 that the 2070 Super/2080 uses has a higher performance ceiling than Navi 10.
I'm not really convinced that Samsung 8nm is *that* bad - but I guess we'll see, when the time comes.
I'm definitely prepared to believe that AMD will be going for a more conservative design that's guaranteed to yield well. The good competition should happen in the 3070 / 2080 range. Fingers crossed we finally see some significant price adjustments.
I don't think that's how any reasonable company would think about at its customers.
That said, I very much doubt AMD's goal is to "feed the fans" either. They need to win market share, not just maintain it. In theory that process also ought to please their current customers, but that's not guaranteed.
They are focused on Zen3. I think rightfully so. I'm not saying RDNA2 isn't going to be good just they are more worried about Zen3 landing this early fall than RDNA2. This is the right priorities.
Not actually! If you look pc hardware sold, it is very stable around the year! The is very small tip around holidayseason, but very small. So it does not matter when you release your product!
Just what I've been waiting for. I've been waiting on buying a new CPU to see what Zen 3 brings. I don't know if I'll end up buying a Zen 3 CPU or a heavily discounted Zen 2 CPU but it's nice to have the opinion.
If you want to be future proof, to the degree that it's possible, then at this point you may as well wait for the release of AM5/Zen4/DDR5 in Q4 2021 (2022?). Zen 3 is the last AM4-compatible chip, so you have nowhere to go except further up the 4000-series stack. That's not a terrible place to be, but five years from now it might be - I don't know how frequently you like to upgrade.
Rumor has it that there will be Zen 3(+) "Warhol" desktop CPUs in 2021, which could be the introduction to the AM5 socket. If that's true, it would be a way to future proof without having to wait until 2022.
Zen 3 will be even a bigger jump from Zen 2 compared to Zen 2 vs Zen+. IPC should be ~20% better and frequencies also a bit higher, so a hypothetical 4900x with 12 cores should be 25% faster than a 3900x which very close to 3950x.
I was hoping to hold out until Zen 3, but my Phenom II system bit the bucket a month or so ago and forced my hand.
Ended up with a B550, Ryzen 5 3600, and 32 GB of RAM on sale (using PSU, GPU, drives, etc from the existing system). Took awhile for the parts to arrive, and I haven't had a chance to put it together yet.
Sounds like what I'm planning. I had to do an emergency purchase to replace a failed system and ended up going for a Ryzen 3 3100 and second-hand RX580 to hold me over until I'm ready to build a proper new system - likely some time early next year.
Whatever I end up with - Zen 2 or 3, RDNA 1 or 2, or Ampere - it's going to kick the arse out of the Skylake / Maxwell system I've been running up until now. Hell, this last-minute build is already a noticeable improvement 😅
I'm waiting until both Ampere and Big Navi are released before I upgrade my GPU. I'm still rocking a 4770k & R9 285. The 4770k will still be enough until DDR5 Zen 4 comes out. I don't want to be on the end of a RAM cycle again so I'm not going Zen 3.
i have yet to even hear the fan on my x570 based board, i even put a piece of paper in it to see if it was even running after playing a game or encoding a vid via handbrake, not a sound.
Because it produces heat and the AMD thought it was a good idea. The quality of fans are much higher quality now than back in the day. And you don't really have a choice either go B550 or go intel.
Or get a X570 and rip the fan off. Enough people have done this, putting on a proper silent heatsink, to prove at this point that X570 doesnt generate enough heat to warrant a fan, and often the fans do a worse job cooling the chipset then a silent hunk of aluminum does.
X570 Aorus Extreme is the only option but it's batshit expensive at $699, ridiculous. I'm just waiting for Ryzen 4000 vs 10K benchmarks and then decide if Intel's Z590 is going to beat or not, along with X670 if it can beat then money is on Z590 Dark or ASUS Z590 Maximus Apex else will see what options I have.
People needing more lanes or PCIe 4 doesn't give a flying pig if the chipset has a fan or not. If they need they'll buy it. MTFB for the chipset fan is not a concern. I won't comment about "noise" (wich is none).
But the thing is, all hardware can fail so you need backups and to plan for potential downtime. If you're a company with 1000 laptops maybe it makes sense to talk about a failure rate, but for my personal computer it's always a possibility that it can fail today. Whether that risk is 1/1000 (one in three years) or 1/5000 (one in fifteen years) I can't really do anything differently. Even if it's less likely I'll need my backup plan, exercising it is such a small part of it.
I rather suspect that with most people concerned about PCIe 4.0 lanes it isn't about need but rather want. Or the age old "future-proofing" thing again. Wait for DDR5 to be "future-proof" only find that for the first 2 years of its life or so DDR5 is barely faster than DDR4 but costs 50% or more for the same capacity.
In the enterprise world is the only way to use PCIe 4 based on the workload. On typical gaming platform you don't NEED that much storage or networking bandwidth, and the GPU are mostly doing graphics work instead of compute, which means PCIe bandwidth isn't the bottleneck (compare to memory or inside the GPU).
There aren't enough physical space on motherboard to put in too many GPU, 25-40G ethernet, NVMe SSD, etc in consumer space yet. The only way to absolutely have to go PCIe 4 is when you already occupied all these lanes in PCIe 3 and you still need more.
You fan-is-no-big-deal people fail to understand that the fan is never the disease, it's a symptom. That the MB needs a fan means there are excess heat which means the circuits and components are under more duress which means there are more potential points of failure.
And what you fail to understand is that the fan on x570 boards is pure precautionary. It's not in use for 90% or more of regular use cases on most boards. It's only needed if you have several PCIe 4 devices, ie. more than just a graphics card and one/two NMVe. On top of that there's fail guards and temperature sensors in practically all modern ICs.
Then why don't you remove that little tiny fan if it's not a big deal? No? Then stop belittling people who prefer a chipset that doesn't need to be actively cooled.
considering i have never heard my x570 chipset fan at all, i doubt it has even spun for more then a few seconds, even while there is any load on the system, so no need to remove it :-)
Deciding you don't want a fan is a personal choice - an opinion - and that's all well and good. Hand-waving a bunch of specious nonsense to justify that opinion, though - well that's just silly, because now we know you're full of it.
I think AMD intentionally put their Radeon launch on Oct 28th. This way NVIDIA will be with all cards on deck and benchmarked. AMD will decide in the last minute probably the price and last core/vram adjustments.
Also the absolute lack of hype regarding Zen and Radeon is strange. It's either the best autumn (I don't want to say fall :-) ) in AMD's history (Zen 3, Radeon 600, Xbox SX/SS + PS5) or nothing to write home about. I'm rooting for them to amaze us, of course. Competition is good.
This is why Nvidia has RTX 3060 in the bag - 1060 and 1080p is the most widely GPU/resolution as per steam surveys (agree not the whole universe of gamers) - but no compromise ray-tracing 1080p gaming at $299 would be awesome if Nvidia chooses to launch RTX 3060 at that price - though knowing them would be definitely be priced higher
Just build a mid-range gaming PC for a friends kid with him last week out of an Asus B550, a Ryzen 3 3100 and a 5600XT. Total cost of the whole thing (and a nice 75Hz LG FreeSync monitor) was $900.
I was thoroughly impressed with how well it went together compared to the nightmare of an AMD system I built years ago for someone on a strict budget, an AM3 system with an Athlon 860K. There were like 3 BIOS updates that came out over the course of a year to address stability issues particularly with the 860K - the CPU we happened to use. And it wasn't just the ASUS board, a lot of other OEM's had trouble maintaining stability with that chip...really left a sour taste and reminded me why I'd been strictly Intel since Nehalem.
ROFL. No kidding, I was expecting benchmarks or something tangible data wise. Something big is coming tomorrow...uh, we're announcing something big is coming in october...So wait....WTF? So today you are announcing, um...NOTHING. Right, back to bed, I got up early for what again? I'm not sure I'll be interested on the 8th now.
I'm pretty sure I will be building my next system, the last of my working professional career, with these two offerings.. Zen3 / RDNA2... in very early 2021
Navi 2x series drivers for Linux have been steadily maturing the past 18 months. I guarantee Windows will be further along and nothing like the RDNA launch. TSMC yields are at or over 90%. AMD has a large pre-order that includes PS5/XBox X Zen2/RDNA 2.0 SoCs, never mind the large preorder for Ryzen/EPYC and more.
We also have the upcoming CDNA 1.0 release Compute Cards yet to be announced.
AMD isn't dropping any balls. Nvidia had to go first or risk obscurity with both Zen and RDNA 2.0 announced a few weeks apart from each other.
Intel had to go before them all because they have the least to show and the need for mind share, however short lived, was to stem the bleeding.
AMD has a huge stead stream of product lines far greater than Nvidia to keep developing and they are hitting their strides. Increases in the R&D budgets are paying off.
"Nvidia had to go first or risk obscurity with both Zen and RDNA 2.0 announced a few weeks apart from each other. Intel had to go before them all because they have the least to show and the need for mind share, however short lived, was to stem the bleeding."
Really not sure that's true... not even close. In terms of mind-share outside of enthusiasts, AMD doesn't come close to either of its competitors. Intel are indeed flailing for press amongst enthusiasts, but Nvidia quite clearly play to their own schedule at this point.
Just checking, isnt this the same nvidia that hit 80% marketshare despite their competitor owning both consoles and having a "huge advantage" in porting?
Hey, stop that fact crap. No need to make real points here...Those are shunned. 80%? Shouting facts, you freaking racist, etc etc . :) Nvidia sucks... /s
AMD is executing like a dream finally come true. Zen 2 stole ever bit of Intel's (INTeL?) thunder. Zen 3 at even 10% to 20% better is going to eat so much market share that they can afford keep the innovation train rolling. RDNA2 is already proven as a design. XboxX/PS5/Unreal5 show that it's got chops. Performance per watt and per dollar will beat Ampere BUT Nvidia's dies are so big that I am convinced big navi ain't big enough to beat 3080. Nvidia priced 3080 at a premium for that size and will do so again with a 16 or 20GB Super version to compete with Navi 16GB. I'd buy a 3080 fearlessly. But if I was waiting for 3070, I'd wait until AMD opens the kimono.
The only thing that keeps the innovation train rolling is R&D. That requires CASH, or if you prefer, NET INCOME in AMD's case, which they keep forgetting to make. Screw share, I'd rather be apple with higher margins. IF you make nothing on share, share is pointless as the poor man. As the rich man, share blocks the other guy from income you don't need (now), until you are better/stronger/faster whatever, than their parts and can charge a mint again. Again, this doesn't work from the poor side without a huge cash pile for R&D. AMD has no cash pile like apple etc.
They need to start charging appropriately before Intel is selling TSMC chips that are on the same process (or better, Intel can pay to leap ahead if a contract doesn't stop them), and with better designs. If intel had 7nm out now, we wouldn't be having SHARE conversations. That said, Intel will have 7nm (or better) shortly from TSMC...Then it will be Intel vs. AMD on the same fab. That won't work out well for AMD with no money to out R&D Intel. If it's my IP vs. your IP and I'm making 23B vs. your ~600mil, you're DEAD on the same process. If not at round 1, then certainly at some point, or worse when I get back in to my fabs with cheaper costs again to bludgeon you with forever. I mean surely Intel won't make the same 20B mobile pissed away mistake again right? It will go to fabs this time over 4-5yrs instead of pissing 4.1B+ a year away on mobile share buying. Do you people see how it fails? Share means nothing if you costs you to get it as the poor man (heck even as Intel - 20B a 5yrs to figure out you have to EARN it, not BUY it by giving away chips).
FYI macrotrends.net 2006-2020 data for any stock. AMD had a 1B+ NET INCOME Q in Q4 2009 IIRC. They had ASSETS then too (you know, fabs, land, leasing property etc). You can see 1/2 their assets lost in last 15yrs. The dream would be 20% share=20% of NET INCOME from x86 cpus. But for AMD 20% x86 share (Intel makes 23.6B NET INCOME on the other 80% share), merely means barely breaking even at ~600mil or less last decade (mostly losses BTW). IF it was "excuting like a dream finally come true", margins would be 62-64%, and NET INCOME ~6B TTM. Meaning exactly what 20% should equal based on INTEL NET INCOME of the other 80% share of the same crap. If you can't crack 1B a quarter, you're failing IMHO on all fronts but perf. Great chips, but what good are they if you fail to make NET INCOME? How many years before they make 6B/yr on that 20%, so the stock IS worth $80-90? It's worth $4 at 600mil. See value when they had MUCH MORE assets, and higher income 2009. Pick a number, it is surely not 80+...LOL.
If they'd just raise prices or at least launch ABOVE last gen or something, they'd make some cash. They refuse to earn, and at this point it is almost as if on purpose. I mean WTF? Charge more money. Don't get me wrong, I love a cheap price, but I make money on stocks, so chips are free (why are you not doing this?). You are already reading about what you should be buying. I basically only buy tech now, and it isn't hard to see who's way oversold or overbought with a few basic checks ruling out most work. 3 seconds on AMD and I know I'd RUN. Net income is nowhere near what the stock says (not even in the galaxy) it should be. DONE. You need a 1B+ Net income Q or I ask why are you worth more than 2009 with 1B+ Q and ASSETS.
Oh yeah, and your shares outstanding has doubled since 2009 too (check NV/Intel, not so). So anyone with shares from 2009 are worth half now from dilution right? Again, seems like simple checks show a stock price below 2009 Q4. IE, Intel shares 2009 5645 (millions) vs. 2020 4284. Meaning less shares than 2009 for Intel now by quite a lot. AMD is 100% the wrong way from 2009. BTW I'm using 2009 due to that 1B+ NET INCOME Q amd had. It was their best Q in time from what I can tell, so we compare against that and all that years data as a point of ref. Check all the stats back then vs. today and again, please explain 100B mktcap and $80-94 share price. Holy cow man who is backing this sh!te math? Forget balls of steel, if you own this stock now, you have balls made of stuff we haven't identified yet ;) I could keep going and destroy AMD here with data, but do your own homework. 16yrs of data goes a long way to proving what the future should look like for company X. A few checks elsewhere to verify data (your stock site for example), and you're likely almost done. You only need to research a few a night to have enough to make some good choices all year. The process isn't rocket science, just time. I'm excited about AMD products, just wish they'd make money on them instead of buying share on shareholders (doubled outstanding shares) dimes. Intel shares have dropped every year since 2010 (outstanding shares that is, they buy back making mine worth more!). Yes, I own Intel (for now). Until they make ~15B NET (not 23B), the stock is worth far more and setting Q records while taking a $27 hit. Intel took a hit like they just lost 1/3 of their NET INCOME. But they didn't have a 1/3 net income hit, they hit a record Q2. So if you thought it was worth $70 a month ago...I say Intel Q1 $70 again as people realize Net income isn't dropping 35%, actually going up by then probably (think this Q is a wash or slight drop until TSMC crap kicks in probably next Q). I'll also say $100 by Q1 2022 and probably $26-27B NET INCOME then (above 23.6B now). Again, EARNING that stock price as they grow INCOME, assets, etc. Revenue isn't earnings just ask tesla. 27B revenue, can't hit 500mil NET INCOME and a 350B Mktcap vs. Intel 200B and making 23.6B NET yearly...ROFLMAO. Do the math people. Look at their stats vs. all other car companies...LOL. Ridiculous. Their price is almost as if all other car companies just quit and don't exist now AND it's illegal to buy any car but Tesla. It is amazing shorts keep losing that war (I see them scratching their heads...ROFL). Of course talking semis, you should be hitting 58%+ margins or good luck making big bank (Intel NV both above 60), not the same with cars (just saying TSLA is crazy too). Tesla wouldn't be worth this much if we had an Obama type president handing out free money to batteries. You are worth more than Intel by a LOT and haven't made a single billion yet. This may be the first positive year in fact (none from 2008-2019 800mil lost 2019). AMAZING. Wait....Stupid.
OH, PE ratio of 958 for TSLA...Just saying. Someone once said don't buy a PE over 20 (maybe it's written somewhere in a basics of investing book or something). There are a lot of variables to consider, but I'd say that person is pretty correct still today for most stuff (again, do your homework!). S&P 500 PE is about 21 today. Do you get it? Math hasn't changed, people just got dumb or lazy so we get PE's of 950's and mkaps of 350B on no net income. Can't blame Elon for taking advantage of millions of fools.
They already followed your "charge more money" "advice" with Zen 2 and RDNA, dingus. They still have to get design wins, which Nvidia and Intel are still bullying them out of...
That's basically all I could extract of any relevance from that obscene wall of text, and it was useless information.
I am excited about the new Zen 3 processor architecture.
As for RDNA2, AMD silicon is good by every few generations I said to myself, "Surely AMD has hired a software team by now." I bought an ATI or AMD card and every time its astonishingly incompetent software prompted me to return or repurpose it. My most recent is a Dell Precision laptop an AMD FirePro W5170M. Flaky, crashy, glitchy, all the things that define AMD video.
After 5 tries and so much forgiveness, I am done with AMD video cards permanently.
AMD's FirePro cards are definitely not something I'd recommend outside of a few narrow circumstances, and especially not the laptop ones.
That is from 2015, though - you might want to rethink our "done with AMD cards permanently" approach. I just switched across to an RX 580 - not exactly the latest tech - and it's doing great; no driver issues whatsoever.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
143 Comments
Back to Article
FreckledTrout - Wednesday, September 9, 2020 - link
Exciting times.p1esk - Thursday, September 10, 2020 - link
What about next gen ThreadRipper?nandnandnand - Thursday, September 10, 2020 - link
It's called "Genesis Peak". Expect it in 2021.Chaitanya - Thursday, September 10, 2020 - link
Couple of months after Mainstream launch.Kevin G - Thursday, September 10, 2020 - link
This would have been the perfect for a CES 2021 launch but that trade show has been cancelled so who knows.webdoctors - Thursday, September 10, 2020 - link
Yes, an announcement of a product announcement presentation, haha. Very exciting!!!Hopefully the trend of bring quality cheap stuff continues, I got my 8 core AMD CPU that's 50% the price of an Intel one. Lets see if their GPUs also 50% cheaper.
Showtime - Thursday, September 10, 2020 - link
It won't be anywhere near half price going by current pricing. NVIDIA had their last gen RTX artificially high due to lack of competition, but AMD brought their current cards out to be barely competitive with those high prices.The next gen AMD cards have to compete with what looks to be much, much better NVIDIA cards. That 3070 in particular looks to be solid 4k gaming card. We'll be lucky if AMD has something with that much power, and even luckier if AMD can match, or top that price to performance ratio.
webdoctors - Thursday, September 10, 2020 - link
But would you buy AMD if it was only 10% cheaper? I'm not a charity, it would have to be at least 30% or more for me to consider it. The NVIDIA ecosystem is definitely worth more than whats' just the sales tax difference in most states depending on online store.zavalita2002 - Friday, September 11, 2020 - link
Oh man, buy what you want and that's thatSpunjji - Friday, September 11, 2020 - link
You're saying that AMD would have to sell their products at a level that wouldn't provide them with any profit margins in order for you to consider them. Just say you're not interested 🤷♂️andrewaggb - Saturday, September 12, 2020 - link
He's not wrong though. I don't think 10% cheaper will win many people over. I think if they want market share they'll need to be 10% cheaper and/or 10% faster and have 16gb of ram instead of 8 or 10. But I don't see how they can be cheaper or faster and have more ram and still have respectable margins.Santoval - Sunday, September 13, 2020 - link
The top-end Big Navi will almost certainly have 16 GB of RAM, according to reasonably trustworthy leaks. Only the top-end though. In performance it will probably be targeted against the 3070; *if* it can even compete with it, that is. The 3080 is a beast and to make matters worse (for AMD, not for us) it is priced reasonably for a change.If the top-end Big Navi has roughly (i.e. within +/- 5%) the same performance as the 3070 and is priced 10 - 15% higher I believe the double memory space will be well worth it and thus would be a better buy. It would be $50 to $75 more expensive but you would get 16 GB of RAM instead of 8 GB, which would be unheard of for a consumer card at that price. If it is 20% more expensive or more I think I would choose the 3070.
Santoval - Sunday, September 13, 2020 - link
p.s. I doubt the top-end Big Navi will cost less than $599 though. Even at that price, if it's really going to have 16 GB of RAM, AMD's margin should be very narrow. If AMD cannot compete in performance I expect their marketing team to spin their fatter memory in a variety of ways..Spunjji - Tuesday, September 15, 2020 - link
@andrewaggb - You wouldn't be won over by 10% cheaper for roughly the same performance? That's fair enough, but I absolutely would be. Same goes for 10% cheaper with 10% better performance, but I have no idea why they'd charge less for a superior card. Demanding that that they be faster, cheaper and more RAM before considering a purchase would strike me as an expression of brand loyalty more than anything else. Don't get me wrong - I acknowledge that's a big factor in the market at large - I just find it weird hearing it from "enthusiasts".As for how they'd do that anyway - they have two potential assets here:
1) TSMC's 7nm process. It's more mature and has a higher density, so in theory, with the right chip they could get better yields than Nvidia and a higher proportion of chips that perform well, reducing the costs of their highest-binned GPU.
2) We already know Nvidia's driving their designs hard - early testing indicates that they appear to have burned much of their gains in performance per watt on getting *even more performance* out of them. AMD could (potentially) produce cards between the 3060 and 3080 performance levels based on smaller dies that require less power and have lower cooling requirements, much like they did back with the HD 4000 series. Adding up savings on the GPU itself, the board design and the cooler, that could come out quite nicely in their favour - *if* that's what they decided to do.
Manch - Tuesday, September 15, 2020 - link
They don't need margins at this point, they need mindshare, and market share. Most people aren't brand agnostic and buy what's popular.The big question is how will the 6 series perform? It needs to match perf & undercut Nvidia and sacrifice margins to do so if need be. ZEN can easily cover the loss. Otherwise they wont gain market share.
Santoval - Sunday, September 13, 2020 - link
"That 3070 in particular looks to be solid 4k gaming card."Yes it does, but oddly the 3080 has even better bang (FP32 TFLOPs) for the buck, which I don't think has ever happened before. I concur that AMD's next-gen graphics cards will apparently* compete with the 3070 at best. The top Big Navi card will of course has twice the memory but it is unclear if it will be faster.
Some early leaks suggested that the top-end Big Navi will handily beat the 2080 Ti in performance, but according to the performance values of these leaks (which, I know, should be handled with a bucket of salt) that places it around the level of the 3070 in performance, at best. So, once again, the 3080 and later the 3080 Ti will be way beyond the reach of AMD, in a continuation of the status quo of the last several generations. So the only reason someone might pick AMD would be the much larger memory space at a far lower price than the 3090.
*Unless of course AMD surprises us all and pulls out a dual GPU chiplet based beast of a graphics card, which I strongly doubt will happen. Most likely both AMD and Nvidia will move to a chiplet scheme from the next generation (AMD from the next-next, i.e RDNA 3). Nvidia's Hopper will almost certainly be chiplet based; it makes little sense to fab huge reticle stretching dies with poorer yields, and besides Intel's first generation of graphics cards will be chiplet based anyway. If Nvidia and AMD do not move to chiplet based GPUs Intel might steal the performance crown from them simply by adding more chiplets.
Spunjji - Tuesday, September 15, 2020 - link
Where's the information about the 3070-level performance coming from? The only things I've seen are from quite a long time ago - further back than would be useful for predictions now.Based only on the known performance of Navi 10, the most likely configuration of a Big Navi (80 CUs), and the claimed improvements in performance per watt for RDNA 2, we're most likely to be looking at something that approaches 3080 performance levels in a similar power envelope and potentially a slightly smaller die area. Bear in mind that AMD don't have to match Nvidia's claimed FP32 performance to match the card's actual performance in games (there's clearly something unusual going on in that regard, which is why Nvidia are suddenly so keen on that metric).
AMD are extremely unlike to compete with the 3090 - but they don't really have to.
The really big question is how they do on RT performance - we could easily end up with a similar situation to the early days of DX10 tessellation here, where a different (and weaker) hardware implementation combined with games being designed for Nvidia's setup results in impaired performance on AMD hardware.
sonny73n - Friday, September 11, 2020 - link
“Exciting times.”Get a life!
Drumsticks - Wednesday, September 9, 2020 - link
Looking forward to seeing what Zen 3 and RDNA2 can do. I'm quite worried for AMD that choosing not to have any comparisons out before the Ampere launch is going to cost them a measurable chunk of sales, though. People on the fence about buying Ampere might well decide to just do it and take Nvidia's proven performance to get a GPU 2 months sooner than AMD.nandnandnand - Wednesday, September 9, 2020 - link
Even RTX 3070 could be out by then. Seems very late.Cellar Door - Wednesday, September 9, 2020 - link
Initial stock will be vapour and at inflated prices, 2nd batch for Christmas also at inflated prices. Realistic availability in February 2021 @ msrp.Throw in driver possible issues...
AMD dropping the ball, yet again.
mazz7 - Wednesday, September 9, 2020 - link
Who are you? fortune teller??Kigerone - Thursday, September 10, 2020 - link
inflated prices? ProbablyRealistic availability 2021? Nvidia yes, AMD no
Driver issues? For AMD? It's definitely possible
Dropping the ball? 🤔 Not sure where you're getting that from.
nandnandnand - Thursday, September 10, 2020 - link
Nvidia could have worse availability problems. Bad yields at Samsung. Nvidia can also have "possible driver issues".dotjaz - Saturday, September 12, 2020 - link
NVIDIA can have noticeable driver issues. AMD have those on a daily basis.Qasar - Saturday, September 12, 2020 - link
tell that to my 7970 which has been running fine since day one :-) even the phenom X4 is still running along with no issues.Spunjji - Tuesday, September 15, 2020 - link
They never have evidence.Santoval - Sunday, September 13, 2020 - link
To my knowledge it is Samsung's 7nm node that has poor yields, which is probably why it was not chosen by Nvidia. Their (standard) 8nm node is just a souped up slightly denser 10nm node, which Samsung have long experience with so I don't see why there would be yield issues. The 8nm variant they fab Ampere with is probably not standard 8nm but 10nm optimized for high power GPUs.I heard Ampere has a transistor density in the 50 to 55 million transistors per mm^2 (MTr/mm^2) range. Samsung's (mobile oriented) 10nm node has a transistor density of 52 MTr/mm^2, so the densities match. Samsung's mobile oriented 8nm has a transistor density of 61 MTr/mm^2. That's less than 2/3 the transistor density of their 7nm node (for mobile), which is 95 MTr/mm^2, while Ampere's 8nm variant is almost half as dense.
Transistor density is the primary factor of low yields at fabs, therefore I strongly doubt the 8nm variant of Ampere has a yield issue. Capacity, on the other hand, is another matter. Can Samsung fulfill the demand for Ampere? Judging from its reasonable price and spectacular performance increase over Turing I kind of doubt that.
Santoval - Sunday, September 13, 2020 - link
edit about that "Transistor density is the primary factor of low yields at fabs" bit :The No1 factor that affects yields at fabs is arguably die size. Transistor density should be secondary to that (provided it is not unreasonably high).
Spunjji - Tuesday, September 15, 2020 - link
Wasn't Samsung's 10nm node fairly underwhelming, though? I seem to recall SoCs based on it being more dense than the previous node, but without any improvements in power characteristics. That would explain why Nvidia are having to push Ampere past the edge of its voltage shoulder to get the performance they want (which is usually AMD's thing to do when they're playing catch-up).Hxx - Wednesday, September 9, 2020 - link
i have a feeling thats thats the point. i'm thinking RDNA 2 will be competing with the rtx 3070 and not so much with the other 2 higher end nvidia GPUs. That still could be good news if the card in question ends up being priced less and with similar perf.Kigerone - Thursday, September 10, 2020 - link
My thoughts? Unless you absolutely need a card immediately, you'd have to be an idiot to buy before both have had benchmarks come out from reliable sources.nandnandnand - Thursday, September 10, 2020 - link
Well there are plenty of idiots out there, but those benchmarks for Nvidia 30-series might be out before any AMD RDNA2 card is available. Lost salez.Qasar - Thursday, September 10, 2020 - link
the smart people will wait, those that cant, dont want to wait, or only buy nvidia, will buy asapmikeatx - Thursday, September 10, 2020 - link
also tbh I might just buy to make sure I get that 3080 (or 3090) and don't have to deal with msrp markups or backordersI mean, also I just don't think AMD has the goods, either, or I feel like they would release benchmarks before Nvidia releases.
San Pedro - Thursday, September 10, 2020 - link
How are they supposed to release benchmarks against an unreleased product? These cards are going to be selling for at least a year and probably two, it's better for them if they get the product right. Also, it's likely they wanted to avoid Nvidia just upping specs to beat them before release like happened with the Super series. Nvidia won't be able to just kill their own cards a month after they're released.Spunjji - Friday, September 11, 2020 - link
@mikeatx - I don't recall AMD ever having released benchmarks for an upcoming product before the product's launch, so I'm not sure you can make that judgement.dromoxen - Sunday, September 13, 2020 - link
These cards will be much cheaper in two years Nv will win out on DLSS and ray tracing rather than sheer power/price.Spunjji - Friday, September 11, 2020 - link
You're both right.A lot of the people who'd buy Nvidia before AMD have even launched their competition would buy Nvidia anyway, though - so it's debatable just how much of an impact it will have. Can't argue it wouldn't be better if they'd been able to launch first, though.
FreckledTrout - Thursday, September 10, 2020 - link
Yeah agree.dotjaz - Saturday, September 12, 2020 - link
You'd have to be an idiot not to buy *this time* before supply dries out. It's simply a no brainer if want to spend the money. Waiting a few weeks literally doesn't do you any good.Lord of the Bored - Sunday, September 13, 2020 - link
Only if you gotta have it at launch. NVidia will make more chips, the supply will stutter. It isn't like you won't be able to get a card ever again if you miss the launch window.Spunjji - Tuesday, September 15, 2020 - link
Obvious shill is obvious. When an account pops up and starts yammering about nebulous AMD driver issues "every day" and then says to buy a card before it's even been reviewed, you're not left with a lot of leeway in interpretation.Waiting a few weeks literally doesn't do you any harm. Nobody buying these for games *needs it now*.
They tried this "just buy it" crap when Turing lunched. I didn't buy it then, I won't now.
Kigerone - Thursday, September 10, 2020 - link
Ampere is using Samsung 8nm node which has absolutely horrendous yields meaning Nvidia's 3080 and 3090 will have more restricted stock than the AMD's own past Radeon VII at launch. Especially when pushing the power envelopes like they are, one must question stability and longevity. Many RMA's to come.AMD is using TSMC's 7nm enhanced node which is yielding above 90% already. AMD, despite a later launch will have far more cards out by end of year than Nvidia and psooibly for the foreseeable future. As long as AMD doesn't shoot itself in the foot (or face, let's be honest) with bad drivers again, they have a good shot at effectively competing in market share.
I think the 3070 will be a far more dangerous comeptitor if they launch an additional sku with more video memory. That would be the sweet spot for PC gamers for sure. We will have to see what AMD's answer is. I hope it's good. Let's get some good hardware for good prices.
lmcd - Thursday, September 10, 2020 - link
That 90% number is apples to oranges because the 90% yield is NOT for GPU-sized chips.NeuralNexus - Thursday, September 10, 2020 - link
I'll like to add that AMD as Lisa Su has stated...likes to package performance in smaller packages. I'm quite sure that AMD can get equal performance out of smaller dies. The 5700XT was a small chip and packed as much performance as 2070 Super/2080. I think you need to stop underestimating AMD.Spunjji - Friday, September 11, 2020 - link
That comparison's not really valid, though, because they're manufactured on very different processes. A die-shrunk TU106 would likely be smaller than Navi 10, while the TU104 that the 2070 Super/2080 uses has a higher performance ceiling than Navi 10.Spunjji - Friday, September 11, 2020 - link
I'm not really convinced that Samsung 8nm is *that* bad - but I guess we'll see, when the time comes.I'm definitely prepared to believe that AMD will be going for a more conservative design that's guaranteed to yield well. The good competition should happen in the 3070 / 2080 range. Fingers crossed we finally see some significant price adjustments.
TEAMSWITCHER - Thursday, September 10, 2020 - link
Stringing AMD fans along... I know many are about to jump ship and buy an RTX 3000 series card.Qasar - Thursday, September 10, 2020 - link
based on the prices rtx 3000 could be in canada, 3070 and for sure 3080 and higher, could be out of reach for some people.schujj07 - Thursday, September 10, 2020 - link
Who remembers when the top of the line GPUs cost all of $400. Now it costs $400 for a highend mid-tier card.Spunjji - Friday, September 11, 2020 - link
Pepperidge farm remembers.Also me. I've been whining about this on every GPU launch since Pascal 😬
Tams80 - Thursday, September 10, 2020 - link
If someone can't wait a month or so, then they're obviously a flakey customer and aren't that important.Spunjji - Friday, September 11, 2020 - link
I don't think that's how any reasonable company would think about at its customers.That said, I very much doubt AMD's goal is to "feed the fans" either. They need to win market share, not just maintain it. In theory that process also ought to please their current customers, but that's not guaranteed.
prophet001 - Thursday, September 10, 2020 - link
The RDNA2 is probably not going to hold up to the Ampere cards. Maybe in a niche performance per dollar category but not overall performance.FreckledTrout - Thursday, September 10, 2020 - link
They are focused on Zen3. I think rightfully so. I'm not saying RDNA2 isn't going to be good just they are more worried about Zen3 landing this early fall than RDNA2. This is the right priorities.San Pedro - Thursday, September 10, 2020 - link
They can't release any comparisons to Ampere yet, it is not possible as those cards haven't been released.haukionkannel - Friday, September 11, 2020 - link
Not actually! If you look pc hardware sold, it is very stable around the year! The is very small tip around holidayseason, but very small.So it does not matter when you release your product!
SaturnusDK - Friday, September 11, 2020 - link
Isn't most of the excess sales during the Holiday season from Black Friday sales anyway so not relevant to newly released products.Flunk - Wednesday, September 9, 2020 - link
Just what I've been waiting for. I've been waiting on buying a new CPU to see what Zen 3 brings. I don't know if I'll end up buying a Zen 3 CPU or a heavily discounted Zen 2 CPU but it's nice to have the opinion.sorten - Wednesday, September 9, 2020 - link
Same here. Either Zen 2 or 3 will be a massive jump for me, but I'm looking for the best value plus some future proofing.nathanddrews - Wednesday, September 9, 2020 - link
If you want to be future proof, to the degree that it's possible, then at this point you may as well wait for the release of AM5/Zen4/DDR5 in Q4 2021 (2022?). Zen 3 is the last AM4-compatible chip, so you have nowhere to go except further up the 4000-series stack. That's not a terrible place to be, but five years from now it might be - I don't know how frequently you like to upgrade.nandnandnand - Thursday, September 10, 2020 - link
Rumor has it that there will be Zen 3(+) "Warhol" desktop CPUs in 2021, which could be the introduction to the AM5 socket. If that's true, it would be a way to future proof without having to wait until 2022.mdriftmeyer - Saturday, September 12, 2020 - link
Good luck. Zen 4 is the introduction to AM5 Socket.nandnandnand - Sunday, September 13, 2020 - link
You don't know that.sorten - Thursday, September 10, 2020 - link
Argh. Fair points, but it's taken all of my self control to not buy a Zen 2 system :)lilkwarrior - Friday, September 11, 2020 - link
Agree, it makes far more sense to skip initial Zen 3 chips until the one that will use the AM5 Socket comes out.yeeeeman - Thursday, September 10, 2020 - link
Zen 3 will be even a bigger jump from Zen 2 compared to Zen 2 vs Zen+. IPC should be ~20% better and frequencies also a bit higher, so a hypothetical 4900x with 12 cores should be 25% faster than a 3900x which very close to 3950x.phoenix_rizzen - Thursday, September 10, 2020 - link
I was hoping to hold out until Zen 3, but my Phenom II system bit the bucket a month or so ago and forced my hand.Ended up with a B550, Ryzen 5 3600, and 32 GB of RAM on sale (using PSU, GPU, drives, etc from the existing system). Took awhile for the parts to arrive, and I haven't had a chance to put it together yet.
Here's hoping it lasts as long as the Phenom did!
Spunjji - Friday, September 11, 2020 - link
That must feel like a mighty improvement!Spunjji - Friday, September 11, 2020 - link
Sounds like what I'm planning. I had to do an emergency purchase to replace a failed system and ended up going for a Ryzen 3 3100 and second-hand RX580 to hold me over until I'm ready to build a proper new system - likely some time early next year.Whatever I end up with - Zen 2 or 3, RDNA 1 or 2, or Ampere - it's going to kick the arse out of the Skylake / Maxwell system I've been running up until now. Hell, this last-minute build is already a noticeable improvement 😅
schujj07 - Friday, September 11, 2020 - link
I'm waiting until both Ampere and Big Navi are released before I upgrade my GPU. I'm still rocking a 4770k & R9 285. The 4770k will still be enough until DDR5 Zen 4 comes out. I don't want to be on the end of a RAM cycle again so I'm not going Zen 3.Spunjji - Tuesday, September 15, 2020 - link
That's fair! I prefer to land somewhere near the middle - after the initial price hikes have faded, but before the platform gets binned entirely.austinsguitar - Wednesday, September 9, 2020 - link
grabs popcornsilencer12 - Wednesday, September 9, 2020 - link
Grab me a pumpkin beer would you.coburn_c - Wednesday, September 9, 2020 - link
Wake Me Up When September EndsMrVibrato - Thursday, September 10, 2020 - link
...and so he kept slumbering to eternity.wr3zzz - Wednesday, September 9, 2020 - link
Nothing about new chipset? I have been waiting for a successor to x570 not needing active cooling to begin my new build.shabby - Wednesday, September 9, 2020 - link
Go with the b550 then, i doubt newer x570 chipsets will have passive cooling.inighthawki - Wednesday, September 9, 2020 - link
I think they were hoping for an x670 that was perhaps built on a smaller node that didn't require active cooling.B550 may not suffice for many users who are expecting to use more gen 4 lanes.
shabby - Wednesday, September 9, 2020 - link
That's understandable, but not buying a mobo just because of a fan is less understandable.inighthawki - Wednesday, September 9, 2020 - link
Eh, back in the day motherboards with active cooling fans were known to be not so reliable. I can understand if they had similar worries.Qasar - Wednesday, September 9, 2020 - link
i have yet to even hear the fan on my x570 based board, i even put a piece of paper in it to see if it was even running after playing a game or encoding a vid via handbrake, not a sound.inighthawki - Wednesday, September 9, 2020 - link
The concern here would be reliability, not sound.Qasar - Thursday, September 10, 2020 - link
i have some old A64 boards still in use, that have a chipset fan on them, still spinning nice an quiet.lmcd - Thursday, September 10, 2020 - link
Not the same age but my X79 extreme4-m doesn't have any issues ~10 years later with its mobo fan.wr3zzz - Wednesday, September 9, 2020 - link
You better start backing up everything if you can hear that tiny chipset fan. It's usually the last gasp just before the death of MB.Makaveli - Thursday, September 10, 2020 - link
Were not in the 90's anymore!wr3zzz - Thursday, September 10, 2020 - link
Yeah, it's not the 90s then why is a there still a fan on my chipset!?Makaveli - Thursday, September 10, 2020 - link
Because it produces heat and the AMD thought it was a good idea. The quality of fans are much higher quality now than back in the day. And you don't really have a choice either go B550 or go intel.TheinsanegamerN - Friday, September 11, 2020 - link
Or get a X570 and rip the fan off. Enough people have done this, putting on a proper silent heatsink, to prove at this point that X570 doesnt generate enough heat to warrant a fan, and often the fans do a worse job cooling the chipset then a silent hunk of aluminum does.Quantumz0d - Thursday, September 10, 2020 - link
X570 Aorus Extreme is the only option but it's batshit expensive at $699, ridiculous. I'm just waiting for Ryzen 4000 vs 10K benchmarks and then decide if Intel's Z590 is going to beat or not, along with X670 if it can beat then money is on Z590 Dark or ASUS Z590 Maximus Apex else will see what options I have.End of story for DDR4 and easy purchase choice.
Icehawk - Friday, September 11, 2020 - link
My fan failed after 2 weeks, seems fine with it unplugged - gets warm but not hot. /shrugCaparroz - Wednesday, September 9, 2020 - link
People needing more lanes or PCIe 4 doesn't give a flying pig if the chipset has a fan or not. If they need they'll buy it. MTFB for the chipset fan is not a concern. I won't comment about "noise" (wich is none).wr3zzz - Wednesday, September 9, 2020 - link
"MTFB for the chipset fan is not a concern." Famous last words of people who have never experienced HW failure.Kjella - Thursday, September 10, 2020 - link
But the thing is, all hardware can fail so you need backups and to plan for potential downtime. If you're a company with 1000 laptops maybe it makes sense to talk about a failure rate, but for my personal computer it's always a possibility that it can fail today. Whether that risk is 1/1000 (one in three years) or 1/5000 (one in fifteen years) I can't really do anything differently. Even if it's less likely I'll need my backup plan, exercising it is such a small part of it.TheinsanegamerN - Friday, September 11, 2020 - link
Having good backups doesnt justify buying hardware with likely faulty components. It's still a black mark against X570.KompuKare - Thursday, September 10, 2020 - link
I rather suspect that with most people concerned about PCIe 4.0 lanes it isn't about need but rather want.Or the age old "future-proofing" thing again.
Wait for DDR5 to be "future-proof" only find that for the first 2 years of its life or so DDR5 is barely faster than DDR4 but costs 50% or more for the same capacity.
PandaBear - Thursday, September 10, 2020 - link
In the enterprise world is the only way to use PCIe 4 based on the workload. On typical gaming platform you don't NEED that much storage or networking bandwidth, and the GPU are mostly doing graphics work instead of compute, which means PCIe bandwidth isn't the bottleneck (compare to memory or inside the GPU).There aren't enough physical space on motherboard to put in too many GPU, 25-40G ethernet, NVMe SSD, etc in consumer space yet. The only way to absolutely have to go PCIe 4 is when you already occupied all these lanes in PCIe 3 and you still need more.
TheinsanegamerN - Friday, September 11, 2020 - link
X570 doesnt really require it either. Cooling 5-10W of power draw in a case with airflow shouldnt be this hard.Falloutboy - Saturday, September 12, 2020 - link
X570 will prolly be the last high-end chipset for this platform.shaddixboggs - Wednesday, September 9, 2020 - link
well dude, can't you just set an alarm if it stops spinning and replace it?wr3zzz - Wednesday, September 9, 2020 - link
You fan-is-no-big-deal people fail to understand that the fan is never the disease, it's a symptom. That the MB needs a fan means there are excess heat which means the circuits and components are under more duress which means there are more potential points of failure.SaturnusDK - Thursday, September 10, 2020 - link
And what you fail to understand is that the fan on x570 boards is pure precautionary. It's not in use for 90% or more of regular use cases on most boards. It's only needed if you have several PCIe 4 devices, ie. more than just a graphics card and one/two NMVe.On top of that there's fail guards and temperature sensors in practically all modern ICs.
wr3zzz - Thursday, September 10, 2020 - link
Then why don't you remove that little tiny fan if it's not a big deal? No? Then stop belittling people who prefer a chipset that doesn't need to be actively cooled.Dizoja86 - Thursday, September 10, 2020 - link
wr3zzz is feeling oppressed over what's pretty much the most first-world problem in existence.shabby - Thursday, September 10, 2020 - link
Bet you $5 that Intel's pcie4 chipset will also need one... whenever that comes out 🤷🏼♂️Qasar - Thursday, September 10, 2020 - link
considering i have never heard my x570 chipset fan at all, i doubt it has even spun for more then a few seconds, even while there is any load on the system, so no need to remove it :-)Spunjji - Friday, September 11, 2020 - link
Invert your logic: why remove it when it does no harm to be there, and ensures system stability in edge-cases?You're transparently trying to work backwards from a preference to a "rational" justification for it, which is the opposite of rationality.
Spunjji - Friday, September 11, 2020 - link
Deciding you don't want a fan is a personal choice - an opinion - and that's all well and good. Hand-waving a bunch of specious nonsense to justify that opinion, though - well that's just silly, because now we know you're full of it.If you were talking the difference between 50W and 100W to dissipate, you might have a point. 4-10W, not so much:
https://hexus.net/tech/news/mainboard/132515-der8a...
TheinsanegamerN - Friday, September 11, 2020 - link
Assuming you can find one, these are not standard size fans that are meant to be user replaceable.mazz7 - Wednesday, September 9, 2020 - link
Let us see if AMD will be up to the competition or not? do not make false assumption, competition always good for us ;)liquid_c - Thursday, September 10, 2020 - link
An announcement for a presentation for a (potential) release. This ain’t even a paper launch.Spunjji - Friday, September 11, 2020 - link
Yes, how dare they announce a date when they're going to announce a launch. Nobody ever does that! /s(Nvidia just did that, but you know, doing it first makes it okay or something)
Mugur - Thursday, September 10, 2020 - link
I think AMD intentionally put their Radeon launch on Oct 28th. This way NVIDIA will be with all cards on deck and benchmarked. AMD will decide in the last minute probably the price and last core/vram adjustments.Also the absolute lack of hype regarding Zen and Radeon is strange. It's either the best autumn (I don't want to say fall :-) ) in AMD's history (Zen 3, Radeon 600, Xbox SX/SS + PS5) or nothing to write home about. I'm rooting for them to amaze us, of course. Competition is good.
nandnandnand - Thursday, September 10, 2020 - link
"I think AMD intentionally put their Radeon launch on Oct 28th."AMD intends to do the things they intentionally doing? Say it ain't so!
"This way NVIDIA will be with all cards on deck and benchmarked."
They already got enough information on September 1st. Careful analysis of Nvidia marketing reveals the non-inflated gains.
"AMD will decide in the last minute probably the price and last core/vram adjustments."
Too late to adjust core/VRAM.
"Also the absolute lack of hype regarding Zen and Radeon is strange."
There is some hype. There will be a lot more hype if Zen 3 can beat Intel in single-threaded and gaming, and if RDNA 2 can beat the RTX 3080 20 GB.
edzieba - Thursday, September 10, 2020 - link
>Too late to adjust core/VRAM.AMD did exactly that before for the 5600XT: https://www.anandtech.com/show/15422/the-amd-radeo...
PandaBear - Thursday, September 10, 2020 - link
They can always have all the adjustment and decide which one to NOT release.Spunjji - Friday, September 11, 2020 - link
They can most definitely fine-tune clocks and voltages at this stage, which is likely what they'll do.I doubt that's *why* they're announcing at this point, though. They're more than likely doing this all as quickly as they can manage.
Gigaplex - Thursday, September 10, 2020 - link
"It's either the best autumn (I don't want to say fall :-)"Well, they're both "wrong". It's spring in the southern hemisphere.
Makaveli - Thursday, September 10, 2020 - link
Except not everyone lives in the Southern Hemisphere......Dribble - Thursday, September 10, 2020 - link
It tells me the priority is Zen 3 not RDNA 2.Sherlock - Friday, September 11, 2020 - link
This is why Nvidia has RTX 3060 in the bag - 1060 and 1080p is the most widely GPU/resolution as per steam surveys (agree not the whole universe of gamers) - but no compromise ray-tracing 1080p gaming at $299 would be awesome if Nvidia chooses to launch RTX 3060 at that price - though knowing them would be definitely be priced higherSamus - Thursday, September 10, 2020 - link
Just build a mid-range gaming PC for a friends kid with him last week out of an Asus B550, a Ryzen 3 3100 and a 5600XT. Total cost of the whole thing (and a nice 75Hz LG FreeSync monitor) was $900.I was thoroughly impressed with how well it went together compared to the nightmare of an AMD system I built years ago for someone on a strict budget, an AM3 system with an Athlon 860K. There were like 3 BIOS updates that came out over the course of a year to address stability issues particularly with the 860K - the CPU we happened to use. And it wasn't just the ASUS board, a lot of other OEM's had trouble maintaining stability with that chip...really left a sour taste and reminded me why I'd been strictly Intel since Nehalem.
Not anymore.
Peskarik - Thursday, September 10, 2020 - link
Can't wait! I got my new case, just waiting for AMD release and then I build a new system. :-)TheJian - Thursday, September 10, 2020 - link
ROFL. No kidding, I was expecting benchmarks or something tangible data wise. Something big is coming tomorrow...uh, we're announcing something big is coming in october...So wait....WTF? So today you are announcing, um...NOTHING. Right, back to bed, I got up early for what again? I'm not sure I'll be interested on the 8th now.Spunjji - Friday, September 11, 2020 - link
I'm reminded of the South Park episode with Cartman and the Nintendo Wii...LedHed - Thursday, September 10, 2020 - link
Dear AMD's Frank Azor,This is not the sun, this is barely a sunrise...
How about some details, even just a tiny bit?
HardwareDufus - Thursday, September 10, 2020 - link
I'm pretty sure I will be building my next system, the last of my working professional career, with these two offerings.. Zen3 / RDNA2... in very early 2021mdriftmeyer - Thursday, September 10, 2020 - link
Navi 2x series drivers for Linux have been steadily maturing the past 18 months. I guarantee Windows will be further along and nothing like the RDNA launch. TSMC yields are at or over 90%. AMD has a large pre-order that includes PS5/XBox X Zen2/RDNA 2.0 SoCs, never mind the large preorder for Ryzen/EPYC and more.We also have the upcoming CDNA 1.0 release Compute Cards yet to be announced.
AMD isn't dropping any balls. Nvidia had to go first or risk obscurity with both Zen and RDNA 2.0 announced a few weeks apart from each other.
Intel had to go before them all because they have the least to show and the need for mind share, however short lived, was to stem the bleeding.
AMD has a huge stead stream of product lines far greater than Nvidia to keep developing and they are hitting their strides. Increases in the R&D budgets are paying off.
Spunjji - Friday, September 11, 2020 - link
"Nvidia had to go first or risk obscurity with both Zen and RDNA 2.0 announced a few weeks apart from each other. Intel had to go before them all because they have the least to show and the need for mind share, however short lived, was to stem the bleeding."Really not sure that's true... not even close. In terms of mind-share outside of enthusiasts, AMD doesn't come close to either of its competitors. Intel are indeed flailing for press amongst enthusiasts, but Nvidia quite clearly play to their own schedule at this point.
TheinsanegamerN - Friday, September 11, 2020 - link
Nvidia? Risk obscurity?Just checking, isnt this the same nvidia that hit 80% marketshare despite their competitor owning both consoles and having a "huge advantage" in porting?
TheJian - Sunday, September 13, 2020 - link
Hey, stop that fact crap. No need to make real points here...Those are shunned. 80%? Shouting facts, you freaking racist, etc etc . :) Nvidia sucks... /sSpunjji - Tuesday, September 15, 2020 - link
What do facts have to do with racism?janolsen - Thursday, September 10, 2020 - link
Does this entail that current Ryzen 4000 series APU will remain only as OEM, as they now talk about Ryzen 3 ?Spunjji - Friday, September 11, 2020 - link
🤷♂️ Not sure we can infer anything about that from this.TheJian - Sunday, September 13, 2020 - link
We can infer nothing because that is exactly what they said. NOTHING. Wait, wake up on...Whatever. NOTHING.xba - Friday, September 11, 2020 - link
AMD is executing like a dream finally come true.Zen 2 stole ever bit of Intel's (INTeL?) thunder. Zen 3 at even 10% to 20% better is going to eat so much market share that they can afford keep the innovation train rolling.
RDNA2 is already proven as a design. XboxX/PS5/Unreal5 show that it's got chops. Performance per watt and per dollar will beat Ampere BUT Nvidia's dies are so big that I am convinced big navi ain't big enough to beat 3080. Nvidia priced 3080 at a premium for that size and will do so again with a 16 or 20GB Super version to compete with Navi 16GB. I'd buy a 3080 fearlessly. But if I was waiting for 3070, I'd wait until AMD opens the kimono.
TheJian - Sunday, September 13, 2020 - link
The only thing that keeps the innovation train rolling is R&D. That requires CASH, or if you prefer, NET INCOME in AMD's case, which they keep forgetting to make. Screw share, I'd rather be apple with higher margins. IF you make nothing on share, share is pointless as the poor man. As the rich man, share blocks the other guy from income you don't need (now), until you are better/stronger/faster whatever, than their parts and can charge a mint again. Again, this doesn't work from the poor side without a huge cash pile for R&D. AMD has no cash pile like apple etc.They need to start charging appropriately before Intel is selling TSMC chips that are on the same process (or better, Intel can pay to leap ahead if a contract doesn't stop them), and with better designs. If intel had 7nm out now, we wouldn't be having SHARE conversations. That said, Intel will have 7nm (or better) shortly from TSMC...Then it will be Intel vs. AMD on the same fab. That won't work out well for AMD with no money to out R&D Intel. If it's my IP vs. your IP and I'm making 23B vs. your ~600mil, you're DEAD on the same process. If not at round 1, then certainly at some point, or worse when I get back in to my fabs with cheaper costs again to bludgeon you with forever. I mean surely Intel won't make the same 20B mobile pissed away mistake again right? It will go to fabs this time over 4-5yrs instead of pissing 4.1B+ a year away on mobile share buying. Do you people see how it fails? Share means nothing if you costs you to get it as the poor man (heck even as Intel - 20B a 5yrs to figure out you have to EARN it, not BUY it by giving away chips).
FYI macrotrends.net 2006-2020 data for any stock. AMD had a 1B+ NET INCOME Q in Q4 2009 IIRC. They had ASSETS then too (you know, fabs, land, leasing property etc). You can see 1/2 their assets lost in last 15yrs. The dream would be 20% share=20% of NET INCOME from x86 cpus. But for AMD 20% x86 share (Intel makes 23.6B NET INCOME on the other 80% share), merely means barely breaking even at ~600mil or less last decade (mostly losses BTW). IF it was "excuting like a dream finally come true", margins would be 62-64%, and NET INCOME ~6B TTM. Meaning exactly what 20% should equal based on INTEL NET INCOME of the other 80% share of the same crap. If you can't crack 1B a quarter, you're failing IMHO on all fronts but perf. Great chips, but what good are they if you fail to make NET INCOME? How many years before they make 6B/yr on that 20%, so the stock IS worth $80-90? It's worth $4 at 600mil. See value when they had MUCH MORE assets, and higher income 2009. Pick a number, it is surely not 80+...LOL.
If they'd just raise prices or at least launch ABOVE last gen or something, they'd make some cash. They refuse to earn, and at this point it is almost as if on purpose. I mean WTF? Charge more money. Don't get me wrong, I love a cheap price, but I make money on stocks, so chips are free (why are you not doing this?). You are already reading about what you should be buying. I basically only buy tech now, and it isn't hard to see who's way oversold or overbought with a few basic checks ruling out most work. 3 seconds on AMD and I know I'd RUN. Net income is nowhere near what the stock says (not even in the galaxy) it should be. DONE. You need a 1B+ Net income Q or I ask why are you worth more than 2009 with 1B+ Q and ASSETS.
Oh yeah, and your shares outstanding has doubled since 2009 too (check NV/Intel, not so). So anyone with shares from 2009 are worth half now from dilution right? Again, seems like simple checks show a stock price below 2009 Q4. IE, Intel shares 2009 5645 (millions) vs. 2020 4284. Meaning less shares than 2009 for Intel now by quite a lot. AMD is 100% the wrong way from 2009. BTW I'm using 2009 due to that 1B+ NET INCOME Q amd had. It was their best Q in time from what I can tell, so we compare against that and all that years data as a point of ref. Check all the stats back then vs. today and again, please explain 100B mktcap and $80-94 share price. Holy cow man who is backing this sh!te math? Forget balls of steel, if you own this stock now, you have balls made of stuff we haven't identified yet ;) I could keep going and destroy AMD here with data, but do your own homework. 16yrs of data goes a long way to proving what the future should look like for company X. A few checks elsewhere to verify data (your stock site for example), and you're likely almost done. You only need to research a few a night to have enough to make some good choices all year. The process isn't rocket science, just time. I'm excited about AMD products, just wish they'd make money on them instead of buying share on shareholders (doubled outstanding shares) dimes. Intel shares have dropped every year since 2010 (outstanding shares that is, they buy back making mine worth more!). Yes, I own Intel (for now). Until they make ~15B NET (not 23B), the stock is worth far more and setting Q records while taking a $27 hit. Intel took a hit like they just lost 1/3 of their NET INCOME. But they didn't have a 1/3 net income hit, they hit a record Q2. So if you thought it was worth $70 a month ago...I say Intel Q1 $70 again as people realize Net income isn't dropping 35%, actually going up by then probably (think this Q is a wash or slight drop until TSMC crap kicks in probably next Q). I'll also say $100 by Q1 2022 and probably $26-27B NET INCOME then (above 23.6B now). Again, EARNING that stock price as they grow INCOME, assets, etc. Revenue isn't earnings just ask tesla. 27B revenue, can't hit 500mil NET INCOME and a 350B Mktcap vs. Intel 200B and making 23.6B NET yearly...ROFLMAO. Do the math people. Look at their stats vs. all other car companies...LOL. Ridiculous. Their price is almost as if all other car companies just quit and don't exist now AND it's illegal to buy any car but Tesla. It is amazing shorts keep losing that war (I see them scratching their heads...ROFL). Of course talking semis, you should be hitting 58%+ margins or good luck making big bank (Intel NV both above 60), not the same with cars (just saying TSLA is crazy too). Tesla wouldn't be worth this much if we had an Obama type president handing out free money to batteries. You are worth more than Intel by a LOT and haven't made a single billion yet. This may be the first positive year in fact (none from 2008-2019 800mil lost 2019). AMAZING. Wait....Stupid.
TheJian - Sunday, September 13, 2020 - link
OH, PE ratio of 958 for TSLA...Just saying. Someone once said don't buy a PE over 20 (maybe it's written somewhere in a basics of investing book or something). There are a lot of variables to consider, but I'd say that person is pretty correct still today for most stuff (again, do your homework!). S&P 500 PE is about 21 today. Do you get it? Math hasn't changed, people just got dumb or lazy so we get PE's of 950's and mkaps of 350B on no net income. Can't blame Elon for taking advantage of millions of fools.Spunjji - Tuesday, September 15, 2020 - link
They already followed your "charge more money" "advice" with Zen 2 and RDNA, dingus. They still have to get design wins, which Nvidia and Intel are still bullying them out of...That's basically all I could extract of any relevance from that obscene wall of text, and it was useless information.
Sivar - Friday, September 11, 2020 - link
I am excited about the new Zen 3 processor architecture.As for RDNA2, AMD silicon is good by every few generations I said to myself, "Surely AMD has hired a software team by now." I bought an ATI or AMD card and every time its astonishingly incompetent software prompted me to return or repurpose it.
My most recent is a Dell Precision laptop an AMD FirePro W5170M. Flaky, crashy, glitchy, all the things that define AMD video.
After 5 tries and so much forgiveness, I am done with AMD video cards permanently.
Spunjji - Tuesday, September 15, 2020 - link
AMD's FirePro cards are definitely not something I'd recommend outside of a few narrow circumstances, and especially not the laptop ones.That is from 2015, though - you might want to rethink our "done with AMD cards permanently" approach. I just switched across to an RX 580 - not exactly the latest tech - and it's doing great; no driver issues whatsoever.