AMD's Ryzen CPUs (Ryzen/TR/Epyc) & Vega/Polaris/Navi GPUs

Due the results we now have seen... Maybe a reason AMD won't put out the replacement for 1800x (R 2800x). Or they maybe wait for the 8core from Intel. I don't know, but this doesn't smell well. But we need to see more examples before final judgment falls. Edit (19. April 2018) From the release. Confirmed... The overclockability is bad!!

 

I've allowed some limited BGA convo to stay as it is pertinent too the topic of AMD and non-BGA, keep OEM's out of it though. Also lets get back and stay on topic of AMD CPU's and Gen-2.

 

On this, most of the time, they could. According to Prema, they did change, and need to change, some electrical stuff for Coffeelake and the upcoming 8-core. Also, when they did FIVR and then dropped it. By moving the Voltage Regulators onto the package, you then need a motherboard redesign for the power delivery, which required a MB redesign. You also had the updating of the chipset for some features, needing to do m.2 when that was on the scene (although older board owners could have just used an adapter in a 4 lane slot), etc. So, the fact we got most compatibility (but not full) between the z170 and Kaby was nice. Also, because BW was such a ****stain, it was compatible with the later chipset that showed up with the HW refresh 4790K.

Bull. You are saying it MUST be like Intel, and even if it has a similar jump in performance, or more than what Intel does between gens, that you bash them in a different way. Yes, they clock it closer to the edge and don't leave room for going further, but you wouldn't even like it if they put out a 5.5GHz chip, but you could only hit 5.45GHz on all cores. So, you need to make sure you measure the rants. Then, after that, you bring up an issue that has two components: 1) their drivers (legit to rip), and 2) software optimizations that are beyond their control, which you and others blame them for the actions of other companies regularly. They take it because they have to to maintain their relationships, but you see Nvidia and Intel often try to shift that blame to others, sometimes correctly, sometimes falsely.

Then comes your allegations on what can be done or not, and ripping them. I will deal with this in part below at your doom's day prediction for them, but you are being hyperbolic. Now, they were supposed to collapse in 2018 if nothing happened. Guess what? Zen happened. They bounced back with a 56% improvement over their prior gen. That was while nearly dying, reducing their staffing significantly, etc. They, because of their financial situation, designed one die, but did so targeting the server market, WHICH HAS NO OVERCLOCKING. That is what they could do with the money left, and they hit a ****ing homerun with what they had to work with. They even went into HEDT, something they telegraphed they would not do, instead suggesting they were relegating themselves to the low end because of their finances. It was over-hyped and a year late, just like Vega, those are fair points, but they did get back in the game. I'll address the rest in the other upcoming comment. Also, this isn't status quo. They have lost a little market share in GPU, but they have grown in market share of CPUs.

Yeah, that is because you are not looking at the large picture: miniaturization. Pure and simple, everyone hit a wall at 10nm. Intel keeps trying to pursue it, but there is a reason that everyone else jumped that node (except for like Samsung, which did the iterate node, but is also using odd nomenclature for 8nm and 6nm instead of 7nm and 5nm). Everyone reporting on Intel 10nm says it is crap, and Intel even said 14nm++ is better than 10nm. Now, if you are looking at what they did, they cannibalized their 10nm, instead trying to go to a 10nm+, which is shown by doing mostly full 14nm++ lines, which there is a question whether the 10nm+ will be much of an improvement over 14nm++. Simply said, they cannibalized future profits due to the competition that AMD slapped down on them. And that is before dealing with the fact covfefelake (8700K) likely acted as a violation or dumping of a defective product on the market. When examining the other companies and what is happening on the miniaturization front, they are waiting on EUV lithography, which has been a pain to develop and is easily 3-4 years late once deployed next year. Without the new lithography, things have sat and the boosts have had to come from a tweak here, a clock there, etc. Boy will you be upset when they are trying to do 3nm and 2nm in the next 5-8 years! We have to use different materials with those. I'm just hoping they figure out and choose to switch to graphene with the mag band gap solution. They also recently discovered how to use standard silicon CPU lines to create entangled Q-dots, which may jump up the adoption and number of quantum computers, while we still don't even know how to fully use such things and half the time the conversion of input from traditional to quantum takes more time or equal to the quantum computer computing the solution. So, it would be more fair if you mentioned the walls the industry has hit, as that is a huge part of your complaint.

Further, that voting with your wallet crap of capitalism is crap. Pure and simple, if people are not economically free, meaning having enough to buy whatever they would want, then they do not have a true vote with their cash. Around 50% in the US have income of $30K or less. That is poverty levels. As such, they don't get to vote which products they want, instead having to accept lower tier products, which is the true reason for the proliferation of BGA and lower end chips, while the real research is subsidized by corps buying the high end and server products. Simply, if so many did not make so little, then you would see the purchase of better products, which would reverse certain trends that you hate. McDonalds, for example, made over $90K in revenue per employee. If you look at their pay, if they make federal minimum wage, that is $15K or less on their labor. If they make $15/hr, that is closer to $30K, the current median income in the US (using IRS and Social Security numbers, which use a larger sample from raw data, unlike BLS which claims it is around $50K, but uses a smaller sample and throws those who are unemployed or have their phone cut off out of the survey, or if you make $25 in a week, you are not counted as unemployed). I digress.

Bull. Intel has locked skus that cannot really go above the 3GHz range. All chips unlocked means you could do better against those lower skus. Not all sales are only the top chip from Intel.

Do you remember the 880m? That was an overheating piece of ****. Do you remember BW? I'm sorry, you have a very selective memory that seems short at time, or at least when you are ripping on AMD.

Now, I do believe you would jump ship, but the question is what that will take. As has been detailed since last summer, 7nm is where AMD matches, approximately depending on how density is calculated, the logic density of Intel. TSMC and GloFo both will be approximately as dense on transistors at that point, meaning the game comes down to design. As GloFo put out, they are targeting 5GHz on SERVER AND HPC systems. Considering they are licensing the process from IBM rather than Samsung, this lends credence to the statement, as IBM has regularly had 5-5.5GHz chips in servers for awhile. Now, IBM doesn't publish their yields and are able to cull more silicon to bin for the speeds. AMD doesn't have that room. But, that means that if EPYC successor hits 4-4.5GHz, the less heat constrained single and double die chips for the AM4 and TR4 should be 5GHz+. Still have to look at implementation, and that brings me back to discussing AMD finances, talked about below.

I agree here. The rumor came from the Starship rumor/leak, with the newer rumor discussing Rome suggesting 16 Cores, as Starship was a 48C chip rumor and Rome to 64C chips. I've actually discussed the fear of this becoming reality late this year or early next year being why Intel is doing clearance on Xeon Phis right now. There are also rumors of AMD adding a CPU die controller chip to the dies, like a central processor. If this is true, you could potentially scale with more dies. So we don't know if there are planned changes to the CCX core count or the number of CCXs per die, or if they will add more dies or what. It all is speculation, which then runs into how they will do the memory controllers and the IF. I've touched on these to varying degrees before in this thread. Either way, the Ryzen 2 series (not Zen 2) was nothing more than a refinement of Ryzen. The real design changes come with the team working on the 7nm. So we must remember that as well.

This is hyperbolic. AMD has growth in the PC market, reaching historic levels, meaning they are back in the game, just not pulling the moves you want yet. The debts have been refinanced so 2020-2023 is the maturities you have to worry about. Considering they are on track with their current sales to claw back even more market share in the next couple years, the threat of folding is not there. In fact, if you reviewed the R&D expenditures, in addition to early payments on debts, paying extra on wafer supply agreements due to demand and need, at a time when the costs of wafers are going up, you actually see a company in growth mode, not at the verge of collapse that you are suggesting. You purposely ignore the steps taken already on both fronts. I admit, they need more in the GPU side of things, but on the CPU side, you are displacing your disappointment in Intel not doing better onto AMD, which is hilarious considering the above explanation of the miniaturization wall hit by the industry as a whole. The stagnation on the CPU side is something that neither could control, specifically the development of EUV, which Intel spent BILLIONS, as well as others in the industry, to try to get it up and running faster.

Everyone has had to redesign chips because they expected EUV. With the 7nm chips and Intel's 10nm+, they are designing a compatible uarch so that for the replacement for those chips, they don't have to design from scratch to adopt EUV. But, by that time, the entire industry will be caught up, including possibly VIA being caught up by 2019 or 2020.

The real question is how well the uarch plays with the 7nm node for Zen. If it is implemented well, then you will get that fight you want. AMD has been greatly increasing R&D throughout last year every quarter, meaning they are playing the game you want them to play, just you are not able to wait for the results of that spending, apparent in your comments. There is also a question if NAVI will finally get them back to fighting on the GPU side. Not much is known about it, but due to Nvidia already admitting that they are moving to multi-die at 7nm, I would suggest AMD plans to also. Now, if they do, AMD has a couple advantages in their corner. Specifically, they have more experience doing multi-die chips, even though it is on the CPU side, currently. They have played with putting HBM on their APUs, and the High Bandwidth Cache Controller makes a LOT more sense on a multi-die chip. They are making changes to caches on GPUs as well, and are doing 7nm Vega this summer, which is to do the shrink and learn what they need to change before implementation on Navi. Now, the delay in competition is likely why Nvidia seems to be waiting to release an 11XX series card. But the overall point to be made is that it is in the works.

Here is a video discussing yields by adoredTV:
There is some good info on there!

 

I see somebody got up on the wrong side of the bed today, LOL. Take a deep breath and you will feel better. Don't pop a vein just because not everyone is content with what is on offer. Sometimes it's just not good enough.

I will wait until they pull the moves I want, if they do. No reason for me to do anything different until they do. If I wanted something that sucks at overclocking there are plenty of options like that to choose from. Not going to spend my money on something that runs at more or less a fixed performance level for everyone. I can do that with a game console or a NUC for a fraction of the cost. I find no pleasure in owning it if it runs more or less the same for everyone. Utility, sure... Pleasure, no.

 

My point is you are ignoring all other data on what is coming down the pipeline. That includes the rumor that the 7nm Vega will be done at GloFo while the 7nm Epyc chips will be done at TSMC, which is discussed in that video, the controller chip and potentially moving everything to that chip except for the cores and maybe cache, which can reduce latency and effects for the memory speeds, while also allowing implementation of faster integrated fabrics, such as an upgraded IF, or any number of other faster data fabrics. Then there is the discussion of yields, densities, etc.

If all you are going to do is simplistically point at max OC, you are missing the whole picture on what is happening, hence the doom and gloom predictions. I point out AMD does need to improve, but I also note the steps taken from them. When I bought my TR, you heard me say the real fight is with 7nm vs 10nm+ from Intel. That was 9 months ago. I made many other predictions and have been fairly accurate in those. Sure, I got a couple things wrong, but got the overall picture correct. Here, your comments are not targeted to actually inspire or get them to do more, it is to discourage people from deciding to get a product because it fulfills their needs, instead trying to say get the highest end Intel always, even if that is not in their budget or what they need. Now, on an enthusiast front, maybe, but it is also throwing that at people that think they are enthusiasts but are not. Now, for gaming, Intel still is king there, but it should be noted that the 7700K can STILL beat the 8700K with gaming on some benches.

But the point is, review the info I gave in the long explanation and that video attached. It may change your mind on what AMD is doing to actually accomplish what you want from them.

 

No, that's the thing. It is not here yet. The road to hell is paved with good intentions. If it works out like planned and I can enjoy robust overclocking fun, awesome. I will probably buy it when the time comes. Since it is not here yet, none of that matters to me. I am the same way about stuff from Intel and the NVIDIOTS. I never get excited about rumors, speculation or roadmaps. I always take a wait and see approach. I do not trust any of them. We have been screwed around way too many times to burn any calories on rumors and concepts. But, that is just me and my approach. Wake me up when tomorrow comes and maybe there will be something to be excited about. I hope so. I love surprises. I hate disappointment, but that is an unavoidable part of life. I expect it based on the status quo.

 

Why do you think I said 4.3-4.4 is all to expect from this refresh? I said the extra core count was likely absurd at this refresh, that expecting crazy clocks was horseshit, and if you already own the first Ryzen chips, it likely wasn't worth upgrading. Kinda said that 4-9 months ago, depending on the part being discussed. I don't predict on things like that without pretty good idea that is all we can expect, and I solidified those arguments with data leaked in January that mostly bolstered my claims. I didn't push hype on this second round, nor do I push doom and gloom.

I also look at what AMD MADE Intel do as responses, including pushing out chips never planned (everything above 12 core) while accelerating the six-core on mainstream, all while not fully being ready for it, but needing to dump covfefe due to Meltdown. Now, all your complaints on their GPU side, I won't touch because those are real. We can hope what they are pushing on 7nm will be better, and they have the tech and experience for it to be good, but they have failed on execution on that side a fair amount. Considering there are 6 skus for Vega 7nm, we do not know what is targeted, if it will ALL be commercial cards with AI being the target, with maybe a frontier edition, or what. Do I have hope for a consumer sku? No, I don't. But they have generated enough fear from Nvidia that they may close the gap that they seem to be holding off release of the new 1100 series cards to tweak them, if need be, so they don't get hit with a top consumer sku off-guard. That does say something. Now, once Nvidia does multi-die 7nm (and it is likely their own development at that node that generated the fear), then, so long as they can get down the multi-chip issues and drivers, they can really stay on top. But with the advent of a crypto-miner dropping prices of GPUs back down, 7nm coming to market before them, and the lack of innovation on their part since Maxwell (not saying they have not made good incremental change, as they have), I'd say that explains some of the behavior of the companies, as well as why Nvidia feels it can sit on its laurels until getting something worthwhile from AMD to respond. You will see a lot more agreement with me on the GPU side. It is the CPU side that we disagree. All I'm using is publicly available information, ignoring the spin, and matching up timelines to look at convergence points on tech advancements. As I said, GloFo says 5GHz, but implementation of the node and density can effect speeds, so that doesn't guarantee hitting that. Further, with the rumor of flipping the CPU and GPU fabs, you have to then ignore the 5GHz number when looking at what may come from the CPU side. That would make the potential speed comments on what 7nm come into question. Also, there is a question if the consumer chips would remain at TSMC or not. Remember, there are two revisions on dies, one that goes in Epyc and one that goes into Ryzen and TR. So they could split them depending on how the transistor performance and yields turn out on Vega 7nm, if that rumor is true.

Meanwhile, I brought up TR lowest sku being compared to Intel's 8-core for a reason: they will both drop around the same time and with similar pricing if the rumors are true of the 8-core from Intel costing closer to $500.

But I get it. I am not saying get excited. I consider all the talk on what will come to be an exercise in futility, masturbation, if you will. I don't get really excited when performing such activities. But, if you can see what comes when with what timelines, it isn't a matter of getting excited, it is all just planning. The better intelligence you have on the situation, and the better conception on how it implements and executes, the better your predictive models. You don't need a surprise to get excited. But, there is way too much hype because of pent up desires by the consumer, which is why all sides seem to cause some form of disappointment these days. Personally, I'm just waiting for them to figure out how to use carbon nanotubes, graphene, or nano-vacuum tubes. At that point, you have it go from GHz to almost THz speeds. The last computer you would likely need to buy for a LONG WHILE. Earliest predictions on them figuring that out is around 2023, with 2025-28 being much more likely. Nano-vacuum tubes are more for space, quantum for cloud as well as classical graphene for cloud, then trickle down to consumers, lots more in wearables and the true IoT issues appear, reduction in energy costs which are offset by proliferation of integration, and solid state batteries advancing times for not having to be tethered. Just standard predictions. You also have optical, which moves data faster and will have light based ram by 2023-25, but there are ups and downs with the tech. Optical interposers were due for integration into server side this year as of five years ago. It will likely be delayed. Now, with optical, you have a humongous reduction in energy for the same processing power, but we still have a ways to go on devo. It will also be competition for the 2nm and below chips. But, that plays out through 2032. I cannot see beyond those fights as they are too far out and are more theoretical. Also, it depends on implementation of AIs in devo. But, once again, I digress.

 

that is some crazy knowledge bomb right there mate! quite the wall of text, but a very interesting read indeed same goes for the follow-up posts, naturally ^^

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk

 

As it is Intel is the overclocking crown and also pure performance. Few here care about the OC crown. A few more need the overall performance crown but this still is few and far between. Overall of those that need these features over and above AMD offerings is maybe 2% of the market. Until 7nm AMD just has to yield that part of the market and even then I am sure a fight will ensue.

Also where 5.0 GHz may be the target we have to accept at first 4.7 Ghz or so may be the only offerings. Epyc at 5.0 GHz I am sure is a joke. If the process is a hit I could see an Epyc 32c at maybe 4.0 GHz or there about.

 

It will be interesting to see if the owners with current gen MB will be screwed with trash overclock if they upgrade to 2nd gen Ryzen processors. Maybe they will be forced into upgrade to new MB if they want max overclock from newest gen Ryzen processors!!

AMD Ryzen 7 2700X, OCed to 4.2GHz & Benchmarked in Battlefield 1

AMD Ryzen 7 2700X, OCed to 4.225GHz & Benchmarked in Battlefield 1
The latest one comes via a user that has managed to snag a Ryzen 7 2700X and overclock it to 4.225GHz, then proceed to benchmark it while playing Battlefield 1. The user has also benchmarked the chip at stock clocks, and the 2700X, surprisingly, sustained a 4.0GHz clock speed on all 8 cores throughout the testing right out of the box.

​

By comparison, the 1700X would only Turbo to 4.0GHz in single threaded workloads on one core, and run at a maximum clock speed of 3.6GHz on all 8 cores with XFR enabled. This indicates that the Ryzen 2000 series chips have a significantly more aggressive Turbo algorithm than their predecessors.

Overall, it looks like the average OC on a 2700X paired with an X370 motherboard so far is between 4.2-4.3GHz. It will be interesting to see how much of a difference the new X470 motherboards will make for overclocking.

 

Hopefully, the new X470 motherboards will make for better overclocking. If it does help, I guess we can scratch multi-generation CPU upgrades off the AMD advantage list. That's kind of a bummer, but there is not much point in being able to add a newer CPU to an older motherboard if it provides no practical benefit for the overclocking enthusiast. Might as well just keep using the old CPU until you can afford a new motherboard to go with the CPU. Let's see how it plays out.

 

Multi-generation CPU upgrades would be useful for those who don't want to overclock, but instead use the CPU's at their stock speeds and lower voltages.
However, in that case, skipping the refresh is probably the better idea (unless it offers much better performance improvements - the Zen+ though doesn't appear to do this).

 

Perhaps it's a bit early to be passing judgement, considering the CPU's are not released yet

It might take a few revs of BIOS and some learning to get the most out of the new CPU's on x370 and x470 (etc) motherboards, and tuning them might provide more options to explore as well.

IDK what anyone was expecting, but traditionally it's always been about a 10% improvement for 18 months to 18 months updates, and usually more power is used as well - Intel has been doing this for many years, AMD had a large gap but multiplying out the performance bump, it was about the same.

It is fun to swap CPU's along the way, and even 10% with additional fixes and features in the CPU can be worth it, and it's good practice for the follow-on updates as well, practice makes perfect.

 

I think more than a 10% increase in stock versus stock would be an unrealistic expectation and probably not a financially intelligent upgrade for those that do not overclock. I was only hoping for some awesomely wicked overclocking, which could work out to a LOT MORE than 10% with a nicely binned CPU. Let's see what happens after they are officially released and there is a BIOS update or two. It might not make them overclock any better, but we never really know what's going to happen until the fat lady sings.

 

The issue is Ryzen was originally blamed for only getting to 4.0 GHz on 14nm because it was said the process "LPP" was for slower silicon compared to Intels 14nm++. It has been touted for some time that when Ryzen went to the supposed much more efficient "LP" process it would be faster. Then even faster yet as now it was 12nm instead of 14nm. It seems this idea is a dead end, it seems there is just 5% better overclocking headroom. That is severely disappointing.

I think at stock speeds the x470 chipset will have its advantages. Other than that overclocking will be the same other than possible advantages in power control on the boards yielding slightly better overclocks. That is the nature of the beast sometimes with newer boards.

It seems unfortunately 12nm is a lackluster release for Ryzen. This will not help AMD much if at all. The fact is because of doubt it will place on the new 7nm it could for a while even hurt AMD. What could have been a save grace is Vega and it maybe now being able to compete against the 1080TI. My doubts are we will ever see that. The inability to get a lot of HBM easily and cheaply will make it where AMD does not want Vega being in high demand.

On that note too, you have to place some suspect on AMD itself. Let us say they could have tweaked Ryzen where with 12nm 4.8 GHz were obtainable. Then yet another 5% for the process itself in speed enhancement. This 25% would put the CPU's well within competing against and even beating out Intel offerings. Now could AMD get a hold of the foundries needed to compete in a market place where 50% or more of the CPU's might be theirs. Let alone all the extra CCX's needed for Epyc processors. AMD would not benefit having a market it can not supply.

Again disclaimer, so far every thing above is pure speculation..........

 

AMD today has officially released specs and the listing of graphics cards that are being rebranded to the OEM-only RX 500X series.

What were before expected reports have now been rendered true: these are nothing more than an OEM-specific rebrand of AMD's RX 500 graphics cards. They're just direct rebadges - not a single MHz was increased across the entire portfolio, except for one lonely graphics card: the RX 550X has apparently seen a bump in clockspeeds, from the RX 550's "up to 1183 MHz" to the RX 550X's "up to 1287 MHz). Aside from that, folks, move along: there's nothing to see here.

AMD Rebrands Polaris GPUs As The Radeon RX 500X For OEMs – Same Specs Retained Across The Lineup

I mean I have seen similar before. And Nvidia can continue as usual... Unlike the Radeon RX 500 series which brought slightly improved performance and better power efficiency to the table, the RX 500X lineup brings nothing new. There are no specs changed but it will not matter to the DIY audience as these parts are built solely for OEMs. Simply put, these are the same graphics cards that launched last year but for pre-built PCs and mobility platforms.

 

RX 400 -> RX 500 -> RX 500X

Rebrandeon amirite

 

It seems AMD don't want be in front. And Intel and Nvidia can continue as before. This won't give people real choices. AMD threw away all opportunities to catch more buyers as it is now. They push people into Intel/Nvidia as they now keep on.

Edit. Btw.
Challenges With 7 nm, 5 nm EUV Technologies Could Lead to Delays In Process' TTM

 

Not sure what that "X" line is supposed to accomplish, I heard it was OEM branding - maybe to counter the Nvidia GPP carnage done to the AMD GPU Vendor branding?

Maybe it's something to give the AMD GPU integrators a rallying name to use to unify them against the loss of their individual Gamer branding for AMD cards included in builds?

It must be something like that as there are no performance boosts, the RX550X gets a little boost, but that could have already been in progress when the Nvidia GPP hit.

 

They must have lost their enthusiast mojo after retreating into the "one size fits all" game console market to form their own monopoly in that niche. It would be a fatal error for them to assume that PC enthusiasts are OK with the Procrustean approach to things, with or without overclocking prowess. Console gamers and mainstream consumers do not care, and they don't know what they don't know. The real losers here are all of us that do know better and expect more... destined to continue on the path of decline with the status quo; stuck in a rut with the Intel and NVIDIA monopoly. Really want to see AMD open a can of whoop ass on them. And, they still might. But, it's not really looking that way to me so far.

 

Ryzen™ Processors: One Year Later
Published on Apr 9, 2018
One year after the launch of the first Ryzen™ Processors, we checked back in with John Taylor (Chief Marketing Officer), Suzanne Plummer (CVP Radeon Technologies Group), Mike Clark (Sr. Fellow Design Engineering), James Prior (Product Manager) and Christina Iron (Director, Global Campaigns) to get their first-hand account of launching Ryzen™ Processors. Find out what a difference a year makes and what’s next!
Learn more: www.amd.com/ryzen


AMD Chief Technology Officer Mark Papermaster’s Keynote at Embedded World 2018
Published on Apr 4, 2018
Mark Papermaster presents “Evolving Embedded Systems in a Self-Directed World” at Embedded World 2018!
Learn more: www.amd.com/embedded

 

We can always count on you to put the negative spin of "disappointment" on AMD's good progress and good news.

 

[Rumor] AMD Navi Mainstream GPU to Have GTX 1080 Class Performance, Nextgen Architecture is The “Zen” of GPUs
AMD Navi Mainstream GPU to Have GTX 1080 Class Performance, Succeed RX 580
Navi is Radeon’s last Graphics Core Next based architecture and is expected to be the world’s first high performance GPU built on 7nm process technology. Little is known about Navi in the techspere to date. One detail that we seldomly see reported is that AMD is working on two Navi GPUs, we’ll call them Navi 10 and Navi 11 for the time being. One is designed for the desktop market and the other for the mobile market.

According to this report from Fudzilla, Navi will not be a large high-end GPU. Although the report doesn’t specify which Navi GPU is being talked about, we’re going to assume that it’s the Navi 10 desktop part. The report further states that this Navi part will be a high performance, low power chip with the performance of today’s high-end GPUs and the power consumption of mainstream parts.

This makes sense from several perspectives. From a manufacturing point of view, it’s not feasible to produce a large GPU on a brand new cutting edge process like 7nm early in the node’s life-cycle. The yields and wafer costs make this prohibitively expensive. This is why NVIDIA and AMD were only able to introduce the GTX 1080 Ti and Vega in 2017, rather than 2016

 

I do not count a 5% increase in overclock as good news when we were led to believe going from LPP to LP and the new process of 12nm would yield to much better results. I am looking from 4.0 GHz max to 4.2 GHz max. That is pathetic not good news, TBH. So I do apologize for holding AMD to a higher standard, the truth.

 

It's not released, these are leaks from who know's what non-production units or old motherboards without BIOS updates, and even those numbers go much higher than 5%, so please stop rating it before it's even out.

It would be best to wait until the inevitable BIOS updates have gone their cycle, and the learning curve on how to get the best out of the new CPU's passes as well. The reviewers need time and the board vendors do too before getting the best results.

Your truth is your imagination and your desire to condemn AMD before it's time is the problem. Instead of rushing to be the first to condemn AMD, let's wait for the production release results from reputable sources, and again it's better to wait a few weeks to let the new hardware get settled in for the reviewers to do their best work.

 

Why should I stop rating it before it is out when all those posts about LPP vs. LP and the better 12nm were posted before it was out too? you were not so fast to go after those posts were you? And they were way before this point too! So you can not have it both ways, sorry.

 

It's the way you go on and on about how disappointed you are over nothing - every time anything comes up for AMD you find something to kvetch about, and how disappointed you are.

Again, what are you doing here anyway if you don't like what AMD is doing and continually find nothing but negative things to say?

Really man, you clearly need help over this AMD negativity bend you are on.

Do you need some positive spin on the nothings you are complaining about, are you looking for an argument over those imaginary points?

So far the rumors don't seen outlandish, and the numbers seen so far are what I expected, there's "nothing" that I see I'd complain about.

Where did you get that 5% anyway, I measure 13%.

 

Yes I need help, help from AMD themselves would be best too. I just get disappointed on over hype and vent it accordingly. AMD essentially tapped into the overclocking headroom to get their boots. Not a supposedly better LP nor 12nm advantage.

As far as before or other issues it is the same, "do what you say" and I will be quiet. BTW as far as AMD I like them but I hold them accountable.

5% is just the OC, 4.0 to 4.2. TR's are known for 4.1 or 4.2 so we will see where that goes. I run 4.0 but I like a low Vcore. (Edit; I can run 4.1 but do not like 1.450 vcore)

 

And if people complain about lackluster performance from mentioned new AMD chips.... They can't post or be here?

 

This isn't a ThreadRipper discussion right now, it's Ryzen 1.0 vs 2.0.

You can't compare your TR 1.0 results against Ryzen 2.0 results - ThreadRipper 1.0 OC'd better than Ryzen 1.0 - there is no valid comparison between TR 1.0 and Ryzen 2.0.

The news is about Ryzen 1.0 vs 2.0, which only did all core OC 3.8-3.9ghz, it's possible though rare to get higher to 4.0ghz, especially from day one release of Ryzen 1.0.

If the new Ryzen 2.0 hits 4.3ghz-4.4ghz that's 13-16% improvement from 3.8-3.9ghz. Even 10% is 2x what you are suggesting.

We don't know what typical OC will be, and for XFR 2.0 with no all core OC, the improvement could be the same or more on a x470 or for x370's with upgraded BIOS with all features (if AMD can do that for x370).

I don't see where 5% comes in. Maybe in games FPS that would make sense, as CPU improvement might be unnoticeable in some GPU intensive games.

But, again, you are drawing out a discussion from "nothing" solid, these are all pre-release numbers and who knows what the production x470 + Ryzen 2.0 + 2-3 BIOS updates will provide.

Certainly there will be far more than a 5% straight CPU improvement, that's silly to believe that's all there is, all the leak numbers are much higher.

 

I would almost put down money that this Navi 10 is the GPU going inside next-gen PS5 in 2019 or 2020. GTX 1080 class performance will suffice for next-gen consoles targeting native 4K gaming at 30 FPS. On the AMD side, GTX 1080 class performance currently means Vega 64, a 13 TFLOPS GPU. 13 TFLOPS, compared to base PS4 GPU's 1.84 TFLOPS, is an increase by a factor of 7, which pretty close to order of magnitude increase expected from a full console cycle refresh (not mid-gen upgrade a la PS4 Pro/XB1X). But Vega 64 is 300W--totally unfeasible for a console. If Navi on 7nm can double perf/W and trim that down to 150W while keeping same performance, it would be the perfect candidate. Maybe asking too much, though.

 

If Sony can wait long enough for AMD to release it.

I'd think it's 2 years out, after the consumer release of Navi / Ryzen 3.0, 7nm on both.

Hopefully the Sony / AMD PS5 project won't come first and delay the AMD consumer GPU / CPU releases on 7nm.

The current rumors about PS5 2018 wouldn't be Navi / Ryzen 3.0 on 7nm... way too soon for the date and technology to match up, besides if PS5 was coming out this Christmas we would already have more on it's internals - maybe we will hear about it more this year, but Xmas 2019 would now be the soonest.

SemiAcccurate gets some Playstation 5/Next details
Exlcusive: Once again first with real console news
Apr 3, 2018 by Charlie Demerjian
https://semiaccurate.com/2018/04/03/semiacccurate-gets-playstation-5next-details/

PS5 Releasing in 2018? First Hardware Specs Rumours | PlayStation 5

Published on Apr 4, 2018
The first real PS5 rumour has arrived courtesy of SemiAccurate, with the tech site claiming that Sony’s next console is set to release this year – in 2018. The site claims that the system will be powered by AMD’s new Navi and Vega technology, and will have PlayStation VR features baked into the silicon. Even more importantly, devkits are reportedly in the hands of publishers right now.

Original source: https://semiaccurate.com/2018/04/03/semiacccurate-gets-playstation-5next-details/

 

Barrowch Whale Shark Threadripper TR4 Water Block Startup and Flow


Barrowch Whale Shark Threadripper Water Block Unboxing and Breakdown

 

I am not comparing the TR, many as top 2800x hit overclocked at 4.0 GHz, the 2700x is purportedly the new high end chip. If I were comparing the TR CCX there are 4.2 GHz TR's out there already, albeit few and far between.

Agreed limited cores higher and this is all over stock, as I said they went into the overclocking headroom to achieve the boost. Since we are now using the LP vs, LPP and Gen2 is 12nm where is the increase to overclocking headroom as well? the increase is from what people are showing is to 4.225, pathetic if you ask me. Now at least overclocking too 4.4 easily or better yet 4.5 to 4.6 at safe vcore would be more expected to see from 10% for the processes and then maybe another 5% performance from the architecture.

 

You literally said TR when referring to the increase in clocks you used to come up with your 5% increase.

"5% is just the OC, 4.0 to 4.2. TR's are known for 4.1 or 4.2 so we will see where that goes. I run 4.0 but I like a low Vcore."

At least we know where you got your numbers.

Ryzen 2.0 release isn't that far off now, April 19th we should see lots of reviews released, and get some better idea of the performance in benchmarks, applications, games, IO, and OC - maybe even some undervolt info.

Perhaps ThreadRipper 2.0 won't be too far off as well.

Edit: It occurs to me that the OC improvement from Ryzen 1.0 to ThreadRipper 1.0 was due to process improvements - things learned making the Ryzen 1.0, which gave ThreadRipper 1.0 more OC headroom.

This time, it's possible that ThreadRipper 2.0 might only have the same process improvements that Ryzen 2.0 will have. That means a comparison of Ryzen 2.0 vs ThreadRipper 1.0 might be telling for what OC headroom Threadipper 2.0 will have when it comes out.

I hope ThreadRipper 2.0 gets the same extra bump in OC headroom from Ryzen 2.0 to ThreadRipper 2.0 as before, but it is something to consider.

 

I am not saying an improvement to TR, I only use it as it is my reference but I have seen up to 4.1 with Gen1 Ryzen, again albeit far and few between. 4.0 is the more common Gen1 overclock.

Now what may be interesting is where the clocks lie on Epyc at 12nm as to keep in the 180w TDP the cocks are a bit lower to begin with over the Ryzen cores. At these lower clocks maybe the 12nm will shine. So maybe there is still some hope out there.

 

Please refresh and re-read my last post, I added an "edit" about TR 2.0.

Epyc may need an increase in power as well even at the lower clock rates, time will tell.

 

How to play FORTNITE on a budget
Ryzen 2400G Radeon RX Vega 11

Published on Apr 5, 2018
Fortnite is a pretty lenient game when it comes to system requirements... so how does it play when we remove the video card all together and play on an AMD APU? Let's find out!

Spoiler: Parts List

Parts List for this video:

CPU: AMD Ryzen 5 2400G - https://www.amazon.com/AMD-Ryzen-Processor-Radeon-Graphics/dp/B079D8FD28
Cooler: Stock
Motherboard: MSI B350M Gaming Pro - https://www.amazon.com/MSI-Motherboard-B350M-GAMING-PRO/dp/B06X3Y7KB5
Memory: 16GB G.Skill Flare X DDR4 - https://www.amazon.com/G-SKILL-288-Pin-Desktop-Memory-F4-2400C15D-16GFX/dp/B06ZZ7QS6L
PSU: Cooler Master 500w - https://www.amazon.com/Cooler-Master-Bronze-certified-supply/dp/B074ZDLGM3
SSD: Samsung 950 EVO - https://www.amazon.com/s/ref=as_li_ss_tl?url=search-alias=electronics&field-keywords=samsung+850+evo
(Save money and get an HDD if you are on a very tight budget)

 

ASRock AMD 400 Series Motherboards : TAICHI REVOLUTION
Published on Apr 13, 2018

 

Here they come, a slew of new Ryzen 2.0 pre-release preview videos...and it is Pre-Order time!!

Unboxing Boxes: The Ryzen 7 2700X & Ryzen 5 2600X Special!
Hardware Unboxed
Published on Apr 13, 2018

Pre-order 2nd-gen Ryzen now:
Ryzen 7 2700X: https://www.amazon.com/AMD-Ryzen-Processor-Wraith-Cooler/dp/B07B428M7F
Ryzen 7 2700: https://www.amazon.com/AMD-Ryzen-Processor-Wraith-Cooler/dp/B07B41717Z
Ryzen 5 2600X: https://www.amazon.com/AMD-Ryzen-Processor-Wraith-Cooler/dp/B07B428V2L
Ryzen 5 2600: https://www.amazon.com/AMD-Processor-Wraith-Stealth-Cooler/dp/B07B41WS48
X470 Motherboards: https://www.amazon.com/s/ref=as_li_ss_tl?url=search-alias=aps&field-keywords=X470+Motherboard
ASRock X470 Taichi Ultimate: https://www.amazon.com/s/ref=as_li_...ps&field-keywords=ASRock+X470+Taichi+Ultimate
Aorus X470 Gaming 7 WiFi: https://www.amazon.com/s/ref=as_li_...s=aps&field-keywords=Aorus+X470+Gaming+7+WiFi
MSI X470 Gaming M7 AC: https://www.amazon.com/MSI-X470-GAMING-M7-Motherboard/dp/B07BZ424WV

Click Titles to watch Videos...

Ryzen 2 Pre-Test: X370 & X470 Memory Scaling on R7 2700X, 1700X
Gamers Nexus
Published on Apr 13, 2018
Before getting to our AMD Ryzen 2 review, including the R7 2700X, R7 2700, R5 2600X, and R5 2600 CPU reviews, we first have to lay down the basics.

2nd Gen Ryzen 2600X and 2700X review kit unboxing
PCWorldVideos
Published on Apr 13, 2018
AMD's 2nd Gen Ryzen (aka Ryzen 2) CPUs are here - promising some nice clock speed boosts. But not only that, the 2700X gets an RGB wraith cooler! We unbox both CPUs and two X470 motherboards: MSI's Gaming M7AC and Asus's ROG Crosshair VII Hero (Wi-fi). This is all part of the review kit provided by AMD, so keep your eyes peeled for our official test results!

2nd Gen AMD Ryzen Reviewer's Kit Unboxing
Tech ARP
Published on Apr 13, 2018
AMD sent us a gray box, as well as the large black box. Let's take a look inside! Read https://www.techarp.com/articles/amd-ryzen-gen-2-reviewers-kit/

RYZEN 2nd GEN - WE ARE EXCITED!
HardwareCanucks
Published on Apr 13, 2018
AMD Ryzen 2nd generation processors are finally revealed The new line up promises significant performance boosts with higher clock speeds and useful X470 platform upgrades. Eber goes into the full details. Official launch is April 19th, 2018.

Unboxing Embargoes are Stupid
Science Studio
Published on Apr 13, 2018

AMD Ryzen Gen 2 Unboxing
TechteamGB
Published on Apr 13, 2018
AMD's Ryzen Gen 2 Lineup is here for a quick unboxing and showcase - along with a couple of new X470 motherboards from AORUS and ASRock! Let's take a look.

A Quick Ryzen 2700x/2600x Unboxing Video
Joker Productions
Published on Apr 13, 2018

 

Asus ROG X470 Crosshair VII Hero - Ryzen 2 2700X 2600X Preview
OC3D TV
Published on Apr 13, 2018
https://overclock3d.net/reviews/cpu_mainboard/asus_rog_x470_crosshair_vii_hero_preview/1


Click Titles to watch Video...

Gigabyte Aorus X470 Gaming 7 - Ryzen 2 2700X 2600X Preview
OC3D TV
Published on Apr 13, 2018

Asus X470 Prime Pro Preview Ryzen 2 2600X 2700X
OC3D RushKit
Published on Apr 13, 2018
https://overclock3d.net/reviews/cpu_mainboard/asus_prime_x470-pro_preview/1

Asus ROG Strix X470 H Preview Ryzen 2 2600X 2700X
OC3D RushKit
Published on Apr 13, 2018
https://overclock3d.net/reviews/cpu_mainboard/asus_prime_x470-pro_preview/1

MSI X470 Gaming M7 Preview
OC3D TV
Published on Apr 13, 2018
https://overclock3d.net/reviews/cpu_mainboard/msi_x470_gaming_m7_ac_preview/1

What you need to know about X470 GAMING PRO CARBON | Gaming Motherboard | MSI
MSI Gaming
Published on Apr 13, 2018
More about MSI X470 GAMING PRO CARBON: http://msi.gm/2HhF8SE

ASUS PRIME X470-PRO Motherboard Overview
ASUS North America
Published on Apr 13, 2018

 

Should I get Core i7 8700 or Ryzen 2700?
https://www.reddit.com/r/intel/comments/8bykup/should_i_get_core_i7_8700_or_ryzen_2700/

transcendent-R7-1700 3 hours ago
"WAIT FOR THE REVIEW. Embargo lifts on April 19. Just go do something for a week then come back."

MobiusOne_ISAF an hour ago
"Seriously, why is everyone so darn impatient?"
"Tell me about this unreleased product no one has / is NDAed" - Reddit poster #52

AMG_Chopper 5 hours ago
"I believe the 2700 will outperform the 8700k in raw power potential. Considering the current strengths of the Ryzen 7 lineup getting reinforced with additions of better gaming performances than before, you're getting a great value with the 2700. Also don't forget the further compatibility with the future chips using an AM4 motherboard, it's better if you think you're gonna upgrade in a short time even after getting a 2700 (if you get it ofc). Plus you got a very good stock cooler.
On the other hand, the 8700, as everyone may have told you, has a good single core power, and I believe it will still keep that advantage for now.
Still, I consider going for Ryzen is the best bet."

 

It counts on what you want from the CPU. If the primary purpose is gaming the 8700K may be the best choice. If you are using production applications in the end the 2700x is the better choice. Just remember for Intel to have the major advantage in gaming you need at least a 1080 or 1080TI and be looking to not run @4K resolution.

 

I was one of the people making the theoretical claims that LPP was a limiting factor behind Ryzen 1's inability to clock reliably past 4Ghz (and you have to admit that it usually IS the process that's the problem in such situations.. not necessarily the architecture)... however, in my defense, I had good reason to make those claims. For one thing, we knew the 14nm LPP was designed for low clocks and low power, whereas 12nm LP was designed for high power and high clocks at lower operational voltages (per the technical specs, we were supposed to get 10% performance uplift for lower power requirements - this is not something I claimed alone, this much was stated in other articles that described Zen+ and its' manuf. process).

Now we see preliminary indications that the 12nm LP is not shaping up to the technical specs we read about... and we're also seeing a higher TDP to boot (on an unreleased product, yes, but still).

And I was also vocal in expressing the notion that AMD may not be able to achieve 4.5GhZ and above... simply that the technical specs of the 12nm LP process INDICATED we might be able to reach those speeds with independent overclocking (and now, we supposedly see AMD tapping into that overclocking headroom - or not, remains to be seen, but the increase in 10% clocks seems to have resulted resulted in 10% TPD increase... which is disconcerting, and doesn't mesh with what we read).
So, from my point of view, this situation is... peculiar.

The claims people made were more or less valid extrapolations based on the information we had at the time. I would definitely like to ask AMD what exactly happened with Zen+ that seemed to have skewed the numbers.

 

Agreed you were not alone in the thought process of LP and 12nm. There were a few here and elsewhere. I believe even rumors from AMD were 10% just from the process. as I said 4.4 as a minimum would have been nice, 4.5 or 4.6 would have been awesome.

 

As I said, we had very good reasons to think that the new process will allow Zen+ to reach those speeds due to what was stated in the technical specs of the process itself (so, it wasn't far fetched). To be honest, those speeds you mentioned might still be achieved from independent overclocking... and we don't know if Zen+ will encounter same type of 'wall' which will prevent it from clocking higher.

But you also have to admit that something is amiss. If the reported voltages and TDP are higher from just 10% clock increases (whereas they should probably be LOWER than Zen 1 or at least the same at 10% higher frequencies), then, what exactly happened to have skewed the numbers contrary to the manuf. process specs?

Is it using the older motherboard, lack of BIOS optimizations, or was the process used just the same as 14nm (aka, again, not designed for high clocks, etc) but just small shrink down to 12nm?
But then... why post technical specs for 12nm LP for which the paper says it works differently?
As I said, what we know of the technical process and what was leaked of Zen+ doesn't mesh well together at all (at least, not to me)... and I'd like to know if we're accounting for (or missing) some factors?

 

Preorder Your 2nd Gen AMD Ryzen™ Desktop Processor Today.
Improved performance and advanced features provide faster, smoother computing experiences than you thought possible.
https://www.amd.com/en/where-to-buy/2nd-gen-ryzen

*You will be directed to a third party site

• NORTH AMERICA
• LATIN AMERICA
• WESTERN EUROPE
• EASTERN EUROPE
• MIDDLE EAST & AFRICA
• GREATER CHINA
• ASIA-PACIFIC

 

Sounds like Navi replacing Polaris will be good IF the price is right. If the pricing for 1080 power is $250-350 and this is their low-end part (hoping that the asic money normalizes prices before release), then it really will be a grand slam. If they then do a high end card with multi-die as Nvidia plans, that could be a really good jump for the high-end as well, especially if you can crossfire and they fix crossfire. Imagine, if power is low enough on a low power 7nm process adding two or four of those on one card. Could be interesting. This is all speculation, though.

Sent from my SM-G900P using Tapatalk

 

So, a couple answers.

1) the process and fab targets are what they have gotten for a reference design they play with internally to tweak and refine the process. It is theoretical. Every architecture has its own nuance and special qualities, so once adapted to a process, it may not match. With Intel, they do both because of owning the fab, so those numbers are matching. I need to pull up 12nm at GF to compare target to what is produced. But if that is what you are referencing, that is a partial explanation. Where did you get 4.5 from?

2) the process or the silicon itself limited the clock speed. Now, some of the extra voltage and heat is going to the improved SOC. If you've overclocking that for ram, you've seen the voltage increase and the heat. The other part is that now it can be pushed further. What we don't know is whether the overclocks shown are heat constrained or what may be happening.

We'll know more soon, though.

Sent from my SM-G900P using Tapatalk

 

Going into pages such as f.eksample page 429 in this thread (with all the YouTube videos) is a nightmare when you do not have the best internet speed or surfing a smartphone And I'm quite sure other may feel the same although they may not say anything about it.