To be fair on AMD and GPU energy efficiency: the desktop Radeon cards are by no means a fair benchmark of GCN's efficiency, especially if you scale it down to a mobile SoC. What these cards' comparatively poor performance per watt (versus Nvidia for example) demonstrates, is poor performance per watt *scaling* at high performance levels. Ugh did I word this correctly?
I mean that, those cards are pushes so far out of the architecture efficiency curve, because they didn't scale up to that performance efficiently. I would point out that at much lower performance levels and power optimised, GCN cards can be just as efficient as Pascal and potentially Turing: just at a much lower overall performance tier.
The mobile Ryzen APUs can demonstrate how efficient GCN can be.
I think they should have BOUGHT Imagination Technologies. They have some impressive designs, and with Samsung's cash they could do great things again.
But yes, GCN can scale down and becomes far more efficient at lower power envelopes. You've already got laptop APUs on 14nm (and soon 12nm Picasso) that can TDP-down to 12W, and they include 4 full Zen cores and 10 CUs of last-gen GCN. That's just with old tech... take the next-gen GPU designs, customize them, and pair them with power-sipping ARM cores and more power efficient process tech.
They might very well buy Imagination Tech anyways. The way Samsung goes about business is to wage battle on multiple fronts and later determine which was the valid path forward. It's one of the biggest reasons why they're murdering the competition and probably approaching world-wide monopolistic scale in the tech world.
There is a massive divide between 200w and 2w of power though. Yes, GCN can scale down in powerlevels really well, Vega 64 for instance was actually extremely efficient when you dropped the clocks and voltages... But instead AMD blew those two aspects out to get as much performance as possible to remain semi-competitive in the mid-high-end.
If you consider existing Mobile Ryzen APU@15W, which already delivers up to 4 times the performance of Snapdragon 82x@5W, on the same process node, it's not that hard to believe even without substantial modification, they can deliver something at least equal to Adreno by the end of the year, just in time for 2021 Exynos.
AMD is far behind in efficiency, in both CPUs and GPUs. While APUs can be considered efficient, based on their TDPs, they can't offer the battery life that Intel processors can offer. That's why AMD is pushing in manufacturing by rushing to adopt 7nm.
They are NOT AT ALL. for GPU depending on the specific generation, they are "just as good" as the other ( Nvidia ) and more often than not it was the Radeons that used less power / ran cooler / gave a better final end result (image quality or various other features) that is FACT.
HOWEVER, the generations that Nv was ahead in this regard, often were noticeably so, CPU wise.....Sorry to burst bubble, but Intel and AMD for the most part use the SAME or roughly the same wattage "over time" and truth be tole it is AMD actually with CPU and GPU that are more oft than not actually giving a very close to 100% true rating on their TDP/Board power etc without relying on fancy BS to achieve such.
in other words, Radeons (at least for the most part) are like junk yard Mac wreckers, sure they are louder, they sometimes belch some cobwebs, but darn it, they get any task done you need to do, and if you really want, take that wonking engine and put into a strip down race body and she just flies.
Nv has over 7 going on to 8 generations cut things away, added "dual" compute (or however want to call) to "trick" apps etc into using it like it is the full fat version when it is not, so they need fancy other chips.circuits to allow this, so the modern Nv is more like a Ferrari with all the fancy gauges that temperatures, heatsinks etc are as "cheap as can be, but quick acting" so instead of beefing up things such as better solder, better VRM and the like Intel and Nvidia whom you likely reference have cut things away, went to lower quality sub components (i.e paste instead of solder, solder instead of eutectic, 85/95 sometime 105c spec capacitor for VRM instead of at least 105/125 or 115/125c ...
So, while in aspect AMD did not do all that well below the 15w level, neither did Intel vs smartphone others and Nv is not the "golden child" folks keep pimping them up to be, they both ( Intel - Nv) do and still do A LOT of crap to mainly trick folks into continuing to use them regardless of what they are doing to make it so, as well, many folk sare BLIND BLIND BLIND to the point where their brand loyalty is the only thing that makes an iota of sense to their peanut of a brain.
Depending on product generation as well as how is used etc, Intel was "slightly" (last 20mins more for a bit more performance in a laptop as an example) but, at the same time the Radeons and AMD CPU, MOST PEOPLE who use them side by side to competing they all generally walk away "it feels the same, but I can actually have far more tabs open in my browser and it does not hitch as well I can leave system running for days-weeks-months and not really have to reboot fully ever with AMD/Radeon but with Intel/Nvidia I had to do a full shutdown once every 2 weeks like clock work or performance would just tank like crazy.....
I not trust numbers any of the "big whales" put out, as when a chip with a 105w TDP label is able to and consistently blows past this number (Intel 7-8-9 generations most recent) but yet the AMD ones you "claim" are SO FAR BEHIND rarely, truthfully rise above to much beyond 10% or so (which is spec to about 15% +/- rating not 30%+ BS STOCK...........
why are you even bothering. Their new CPUs look to have caught up to Intel efficiency wise, but that'll be the first time since Conroe and everyone knows it. And nVidia? They've been lagging them on efficiency since Maxwell. https://www.techpowerup.com/reviews/EVGA/GTX_1650_... FFS. How do you look at that, and with a straight face vomit out some BS about them being toe to toe.
The best I can say for your post was that it was so long I couldn't be bothered to read the whole thing so I probably missed a bunch more dumb stuff
"Higher power efficiency" does not mean "more battery life" in the absolute sense. It simply means "higher performance per watt" (or "per hour of battery life", though that's technically *energy* efficiency). So, for instance : an APU with 2 TFLOPs of iGPU performance resulting in 6 hours of battery life has double the power efficiency (when only the iGPU is taken into account) of an APU with 500 GFLOPs of iGPU performance resulting in 12 hours of battery life.
An APU with 1 TFLOP of iGPU performance resulting in 6 hours of battery life, on the other hand, has exactly the same power efficiency as an APU with 500 GFLOPs of iGPU performance and 12 hours of battery life. Battery life alone is meaningless. The point is what performance you can milk from that battery life.
put down the crud pipe.... Radeon, GCN or no are and have always been just as good, if not more often than not better in regards to power efficiency/usage, a generation or two back and now they are/were also on at least if not better performance as well.
It is amazing how much power can be saved via a simple voltage adjustment etc, with Nv not so much, whatever amount of power could have been saved essentially they did for you, mainly by cheaping out with chips instead of higher grade components (beyond those)
There is a reason why Nvidia is only in the Switch (for price it is vs how little actually get hardware wise vs the 2 "uncontested" that is Ps4 and XB1, there are only so many BS tricks they can pull to get performance high (no matter the game/application) while keeping power use down, temperatures as minimal as possible(therefore also very quiet)
base example, I take my 7870, I can "lock it" to just under factory speed (1040/1280 core/memory @ 1.167v 0% power slide...idle 367/600 @ 0.856v) vs factory of 1050/1200 @ 1219 and 300/150 @ .836v in all tests, when I "lock it" it transitions more quickly between P0-P3 (low 2d, low 3d, high 3d basically)
if watch youtube etc, unless locked will jump to 450/1200 @0.900v and burn about 20w more than when I set the way I do so, also, factory wise, maxes out at 175w (spec) mine never above 150w (even with heavy OC) my 826/1235 0.969v gets me 75% of way to "stock", I easily OC mine and still save power via voltage tuning and the like.
Long story short, in a "closed environment" they have access to the best silicon and anything they need to do "under the hood" is right there.
They are saying Radeon 5000....they should say something else, mainly due to it was not that long ago there was HD 5000 (potent asf cards they were as well, another generation of Radeons that were beasts (power hungry but could back it up, big time...take the time to do a quick voltage tune and dropped power use significantly enough.
to take to 2022+ before this can happen.....I call utter BS on that, no way, no how, they have 3 generations of Ryzens to go by, countless generations of Radeons etc....the combined might of 2 of the "leaders" especially Samsung and AMD...not buying going to take that long, especially with Sammy, they not "drop" what they are doing, what potentially 3-4 generations to make a key spec change before it is done, especially not announce 2+ years before it "might" become tangible.
LOL.........
Not count out whatever magic sauce them engineer folks can do, don't look at what Nv is doing and use that to gauge what AMD can or cannot do.
if you do that, that is not being fair to be honest, and, makes you act the putz
Your claim about how great the GCN architecture is, kinda goes against every benchmark and GPU test out there.
And as for why it’s AMD and not Nvidia who are getting console contracts, the answer is rather simple and has little to do with AMDs offerings being better or worse than Nvidia’s.
Nvidia just does t care to compete for the consoles. Why go into a bidding war with AMD for a contract worth pennies compared to what Nvidia can make with their 500$ graphic cards, that have a profit margin of 30-40%.
Except this licensing deal doesn't cover GCN. RDNA is significantly different from GCN in terms of how work is executed and how hardware is arranged.
GCN operated over 4 cycles using groups of SIMD16 (4x16 for 64 threads), so 16 threads could be finished in 1/4 of a cycle (0.25 IPC). RDNA does SIMD32 (2x over SIMD16) in 1 cycle via Wave32 (32 thread wavefront) and SIMD64 in 2 cycles (vs 4) via Wave64. That's a huge IPC shift right there.
And that's excluding the new cache hierarchies.
So, GCN's efficiency in lower wattage parts doesn't even remotely compare to RDNA.
"this this currently isn't possible with, say Mali designs, as a licensee isn't able" Extra "this" "this currently isn't possible with, say Mali designs, as a licensee isn't able"
Thanks for the writeup guys, I was also really curious about this subject!
And Navi isn't even 100% RDNA, it's somewhere between that and GCN (so the sources that used to claim Navi was still GCN weren't as wrong as they looked when RDNA was announced).
Navi uses the GCN instruction set -- and there is nothing wrong with that. The GCN instruction set is fine. Once AMD moved away from Terascale and it's vector ISA to GCN's scalar ISA there is no point in majorly changing the ISA again.
However, the physical implementation definitely had some weak points. Geometry performance (all fixed function hardware), the max limit of 4 shader engines, etc. These are things that can be completely re-arched while still implementing the same ISA. It's much like AMD and Intel can both build CPUs that work on the same ISA but use completely different physical implementations.
This is good because they finally decided to kill the 'technical debt' that had built up in GCN, but yet they don't need a 100% new driver stack. Even though they said it was 100% new (and Lisa Su DID say it was new from the ground up, and I don't think she would straight up lie, plus she comes from a technical background so I am sure she knows the implications of the words she said) that still doesn't mean they HAVE to get rid of the entire old architecture. There were plenty of parts of GCN that work fine, what's the point in getting rid of those? The big difference between this arch and previous ones is that they decided to actually fix some of the lower level roadblocks rather than just work around them because "fixing them would be too much work" -- it's going to be GCN derived, sure, it would honestly be stupid not to be, but it's going to ALSO be free from the chains that have dragged GCN down. Honestly I'm pretty fuckin excited, can't wait to see more details on it. This arch will almost for sure be VERY graphics focused, and have plenty of power in the graphics areas that GCN was typically weak (geometry, etc) because they are still KEEPING GCN for compute focused products. So they must be different enough to warrant doing that.
I'd like to draw upon the similarities of many archs we've seen in the past.
AMD, Hypertransport and so many design elements from Phenom II can be seen in the Zen architecture. IT's all just the same idea, brought forward to Zen. shame there wasn't as much love for the Phenom II to describe more of it's inner workings but it in essence looks very similar.. just didn't have the core to bring it forward. yet everything around it wasn't dumb, it was smart. Parts of it carried over to bulldozer and is it a bad thing that 1 inferior cpu and 1 very underperforming\ineffecient cpu has probably 50% of it's design ideas in the new cpu? No, it's what makes zen better than Intel's architecture from a Cpu vendors perspective.
So I am also not scared of things being based on previous inferior architectures and Intel and AMD have proven this in the cpu space - Not so good on gpu's, but probably there too :)
it reminds to be seen how big those changes are... They said many times, that chip was completely revamped, they said it about Polaris and then about Vega as well. But when you look at the CU unit with 64SP and L1 cache and how it´s being arranged inside of chip, you can clearly say, the are (almost) identical.
The descripction of RDNA and NAVI says it : Stremlined pipeline, new compute unit design, multilevel cache hierchy.
One SP still consists of 64 SP
*Steamlined pipeline could point to Vega minus all not game necessary stuff.Well we ´ll see. *New compute unit design sound more promising. *Multilevel cache hierarchy is very vague term, because it could mean literally anything
No, I don't think you understand microsoft's new strategy. It's new ceo is brilliant.. Microsoft's profit is shooting through the roof and market is more open than ever, they are biggest contributor to Linux, DirectX is loosing, you can now game on Linux thanks to many of Microsofts contributions.
Windows will likely not be based on a windows kernel for too long, cause what is the point as Microsoft doesn't earn money on our Windows licenses but rather our SQL, O365, powerbi etc etc. Azure servers running Linux from Microsoft and they are earning shit tons by embracing Linux instead of fighting it. Would not surprise me if we saw a Linux distribution from Microsoft....
What if we could build a tiny pc with a tiny heatsink that can use at most 7 watts... That would obviously be better than a bigger pc with a bigger heatsink that can use 500 watts when needed.
I am curious in mist of things, how ARM's own Mali GPU plays into this picture.
From what it looks like Samsung was looking to developed their own GPU to compete with Apple iPhone/iPad GPU's. So it sounds like Samsung was having trouble on their own so Samsung will get a major benefit of this arrangement with getting technology from AMD.
Maybe AMD could be even struggling to sell off their struggling GPU business. With a company as large as Samsung, one possibility in the future is that Samsung would buy out AMD. That would be easiest way that Samsung could gain the GPU technology at minimal cost.
oh look.. MORE AMD bashing and false info with NO proof from HStewart. amd's gpu side is not struggling... but you know who IS struggling?? your beloved intel... struggling for FOUR YEARS now to get 10nm working... and now.. struggling with the aspect of losing the performance crown....
Did I say anything about Intel in this message - some people are so blind sided and only think of one area of computers and that is gaming desktops.
This is a realistic question I propose here, what is going on behinds the scenes of this deal. Samsung is a huge company - and they want to create their own GPU also.. so a buy out of AMD would be logical option for this - it may not be a complete buy out - but NVidia without a doubt is the leader in discrete GPUs.
HStewart, you don't have to.. your pro-intel bias shows in the vast majority of your posts. you call some people blind sided.. but yet, YOU do the SAME thing when it comes to intel, losing the performance crown, is just not in games, as YOU assume it to be.. but it COULD also be in the server space too. Samsung was creating their own GPU, but have decided to licence the IP instead.. I think this was stated in this article, or the other one about this that AnandTech posted.
Phynaz what part of korguz's post was false info ? they HAVE been struggleing to get 10nm to market, it has ONLY been 4 years, or has it been 5 ?? that intel has been feeding everyone its lies about 10mn...
"Given what little we know about Navi (even though it’s launching next month) we won’t go into speculation about just what “Next Gen” entails, but regardless it’s going to be the basis of Samsung’s first generation of Radeon RDNA-derived GPUs."
That's where patent reading comes into the picture.
This article is written from the perspective of 'what is Samsung getting from AMD?' and that is fair because Samsung is indeed getting a lot. However, collaboration is a two way street so the raises the question of 'what could AMD get from Samsung?'.
The first thing that comes to mind is memory. In particular HBM as future products on AMD are to evolve past interposers and move toward direct die stacking. The level of integration between stacked memory and the core logic chip requires significant engineering to pull off, something they would have to collaborate together on. This also benefits Samsung as such die stacking can be beneficial in mobile as well but Samsung would needs more additional expertise in GPU and CPU areas, enter AMD. What else can AMD gain: integrated SSD controllers. Instead of relaying external NVMe controllers via PCIe, why not simplify an AMD SoC mobile board design by including a high speed ONFI interface directly on an AMD mobile chip? The same can be said about LCD and other display controllers currently handled by external drivers. Half a step in the other direction would be Samsung leveraging Infinity Fabric and/or AMD's own on-die interconnect. The benefit here for AMD is the increase in peripherals on their proprietary bus, like say ultra high speed SSDs for servers.
Same. I was reading the whole thing, thinking what's in it for AMD. Great point about HBM. I've thought HBM as the future when HBM2 came out where it doubled over HBM1, and saw how it can make GPU cards shorter. AMD might need Samsung to scale HBM into more products. APUs, GPUs, Consoles. It is great for integration, bandwidth, and power efficient
thats not proof.. thats just a list of samsungs assets...nice try... the same was said a few years ago when there were rumours that various companies were interesting in buying AMD then. how about, oh i dont know.. proof that samsung has any interest in buying AMD.. or mention of it...
One thing to add to this, is Samsung is looking to build there own GPU's and it logical they would grab AMD for this part of business. To Samsung buying AMD would be only a scratch in their pocket book.
I think the first sign of this happening is when and if we see Samsung Notebooks and PC Tablets going with AMD CPU/GPUs but then again Samsung would rather put their own Processors - so maybe there will be Windows for ARM compatible with Samsung processors coming out.
HStewart, did you even READ this article ?? YOUR own words : " I think the first sign of this happening is when and if we see Samsung Notebooks and PC Tablets " from the article that you OBVIOUSLY didn't read : " In short, the AMD/Samsung deal is structured so that Samsung is only allowed to use AMD GPU IP in segments that AMD doesn’t compete in. In other words, AMD’s GPU tech can only be used in smartphone and tablet SoCs. For anything else, Samsung is either prohibited from using the tech, or at best, has to get AMD’s permission first. " where do you come up with this BS ?? do you make it up as you go along ??? whats next.. when AMD announces their zen 2 based APUs are you going to say they upped their voltage to 12v and 15v like you tried to tell people in another article that intel upped the voltage on the 10nm cpus to 5v and 9volts ???
I doubt US Government would be too fond of that passing sensitive US tech to a non-American company would raise alarms... like how Broadcom tried to acquire Qualcomm.
Well US Government did not stop AMD to give technology to China which is worst. There is also always way that Samsung could get around those restrictions like putting offices on American soil and limited what technology goes off countries that America does not like. I have Toyota Tundra and Toyota did exactly that.
Just FYI, I have Samsung Note 8, Samsung Tab S3 and Samsung Tab Pro. So I am not bias as some would like to think I am. I used Apple Phone and iPad for years and to me honest there OS technology did not advance. I was once very bias against Android.
Hstewart hasn’t got a clue. If anyone buys AMD they would have to get Intel’s approval to keep their x86 license. You think Intel would allow Samsung to keep the x86 license? Dream on.
DannyH246, that's not true. AMD can get acquired and the x86 licensing deal will not be affected. Back in 2015: "Devinder Kumar, chief financial officer of Advanced Micro Devices, said in a statement last week that the company could enter into joint ventures, mergers or acquisitions (M&A) agreements without fearing of termination its cross-license pact with Intel Corp. Many industry observers believe that a bigger company cannot acquire AMD since this will terminate the deal with Intel and will leave AMD without an x86 license immediately."
Shlong, actually.. it looks like if EITHER amd OR intel get acquired by another company, their cross licensing deal, is terminated, which means.. NO x86 license for the company that buys AMD : https://www.kitguru.net/components/cpu/anton-shilo...
i dont know how, reliable, kit guru is.. but if that is true.. then that could be why no one has bought amd over the years.....
DannyH246 he really doesnt.. i read in another thread about the HP laptop with a leather cover, where he tried to claim that intel upped the voltage of their 10nm cpus to use 5volts and 9 volts.. turns out.. he was referring to TDP. he has stated he has been using comps for years, and knows alot about them, but doesnt know the difference between TDP and voltage ?? come on .. the sad part, he even linked the the article that talked about it...
Wrong the specs in chart indicated that Intel up Y series to 9V from 5V. I only stating what the table states. Now the question is why was that required and it likely because WIFI 6 and TB3 integrated in chip. But we really don't know until it officially release.
come on HStewart.. even you should know cpus DON'T take 5 OR 9 VOLTS... and what table says that ?? i just checked.. and no where does it say those cpus take 5 or 9 volts, face it.. you made a mistake.. then tried to accuse others of not reading properly...
Just because I don't care for AMD Junk, It does not mean I only use Intel Products. Give me a break. I am not that naïve to Intel. I just believe one should go one that created the products. Times do change - for most of my smart phone uses was Apple. iPhone 3 - iPhone 6, but I like The Galaxy Note 8 because of my experience with Galaxy Tab S3.
I personally have not had good experience with AMD products. At my first Job, I actually went to AMD office and got technical books to complement my collection of Intel - this was in days before the Internet.
maybe not.. but 90% of your posts.. clearly show your intel bias, and your pro intel views... as i have read many times.. any time an article or news post on here, put intel in the bad said, or amd on the good.. there you are trying to twist things around, to make intel look better.. like the cray super computer, AMD getting pcie for what looks like could be months before intel ( then you claim intel will skip pcie 4 and go straight to pcie 5 ) there are MANY other such comments from you that i have read on here like this....
if it wasnt for some of that " AMD junk " we may not have some of the things we have now, look how long intel stuck the industry with quad core cpus for the mainstream, making us go to the HEDT for 5 or more cores... or the so generous 10% or less performance increases year over year, zen comes out.. and all of a sudden.. more then quad core for the mainstream. you may be fine with intels lies, BS and cpu stagnation over the last few years, and im sick of it. i9 series, 9820x, $1210, 9980xe, $2800!!!!!!!!!!!!!! 9900k $700. but if i were to upgrade my comp now.. it very well could be zen 2 based... wont know for sure for another month, but more cores, same or better performance, for A LOT less then what intel charges.
" I am not that naïve to Intel " heh, yea right, sorry hstewart, but 90% of your previous posts, say other wise
It wasn't just JBL, the Harman Group which included professional names like AMX, SVSI, Crown, Lexicon, BSS and others. Samsung is pretty much running those respected name brands into the ground. I can get way Samsung would wind some of those down as they overlap with some of their own products but they are giving themselves a bad name while doing so and not leveraging the technologies those Harman Group companies developed.
Samsung acquiring AMD would be a horrible move. This collaborative partnership avoids the disaster that Samsung wrought upon the Harman Group by keep the focus on mutually beneficial technologies.
Move maybe bad for AMD fans - but for Samsung it likely fits a new, they are interesting make their own GPU architecture and AMD is a drop in the bucket for Samsung. These type of deals happen after earlier deals all the time. I bet even you look back something similar happen when AMD got ATI. But it appears AMD actually wanted NVidia but of course NVidia didn't do it
ahh so ANYTIME 2 tech companies agree to do a joint venture, or share IP, it means one will eventually be bought out by the other.... yea ok. sure... all be cause HStewart said so...
That does happen from time to time. See Intel and Altera as they had partnered for both providing fab space to Altera for FPGAs and Intel got a chip to put into a Xeon package for data centers. The partnership worked out so well that before either of those two things actually hit the market to purchase, Intel and Altera had merged.
It also doesn't always work out. See Intel and Micron which transpired during the same time period as the Intel + Altera example.
While more of a negative trait, one the things that keeps AMD out there as an independent company is the amount of debt that it incurred a few years back. Granted the worst is over and the company is healthy once more but the effects are still present today. Various anti-trust regulators would also like to have a look at the merger before approving it. The last settlement AMD had with Intel wrote out that the x86 license AMD has can be transfered with the blessing of Intel or Intel themselves would invoke the wraith of antitrust regulators too.
Samsung could have easily acquired Imagination a year or two back fi they wanted a competitive mobile GPU company and it was even cheaper than AMD. It would have also gotten Samsung into another position as a both a supplier and competitor to Apple. Alas, that what-if never happend.
Samsung has its sights on mobile with no real interest in the PC or data center space. Samsung would run of what AMD is doing today into the ground and walk away.
This is some of the best tech-journalism I’ve ever read. Nice work. I wonder if AMD sees this as a path towards being in more mobile devices than phones or tablets, like the arm-based VR headset market. It could force some game developers wanting to be on those platforms to gear code for AMD’s new GPU architectures, and maybe make it easier for those games to be tuned for desktop GPUs.
I am willing to bet we will see a AMD IP tech in a Samsung chip in 2021! Knowing Samsung, they will test run it with a mid-range Exynos, not flagship. Samsung are far, far behind QC, Huawie, let alone Apple. Exynos's performance per watt is a generation behind everyone else on almost any level, especially the CPU part.
"Samsung are far, far behind QC, Huawie, let alone Apple. Exynos's performance per watt is a generation behind everyone else on almost any level, especially the CPU part. " I'd suggest taking a look at the recent articles here on AT (the A77 one for example). Samsung isn't that far behind. 9820 made up a lot of ground on power consumption. They are definitely better than 845 and 970. So not a full generation by any measurement.
"Thankfully, AMD’s GPU public roadmap is short and simple. There’s only one architecture on the roadmap after Navi, and that is the enigmatic “Next Gen” architecture. Essentially nothing is known about this architecture..." It is no longer called merely "Next Gen". AMD already announced that it is going to be called "Arcturus", which is much less enigmatic and much more stellar (literally). That graph is just old. Arcturus is expected to be AMD's first post-GCN architecture, however it is not yet clear if it will be a 2020 release. It will most likely slip into 2021, perhaps Q1 2021. If the deal with Samsung is for 2022+ they will apparently license the original Arcturus and work with AMD around that μarch in the fashion they want.
Thanks Andrei and Ryan! Regardless of whether LSI will actually produce mobile graphics tech based on AMD IP (they probably will use at least some), I believe this agreement also means: - Samsung has effectively given AMD a non-compete agreement in the Windows/mobile computing space - this, in turn, could give AMD the peace of mind to turn to Samsung for future manufacturing deals, giving them both leverage when negotiating with TSMC and access to Samsung's 5 nm EUV processes.
I would be surprised not seeing some of AMD's upcoming chips being fabbed by Samsung. One of the (many) reasons why Intel's attempt to offer contract fabbing failed utterly was the understandable reluctance of potential competitors to hand other their designs to Intel and trust the firewalls inside Intel's to actually hold. That, and Intel's inability to actually get their fabs making 10 nm chips in quantity for several years.
VR usage is growing exponentialy. Mobile processors are obviously going to dominate untethered VR devices. Current untethered VR users have to make many concessions because of the low GPU capabilities in mobile devices. If AMD and Samsung can dominate this arena they will control what is bcoming a very large tech sector.
Sounds like Samsung dipping their toe in the water to see if they want to buy AMD (or at least RPG), but then the news that Ampere will be on Samsung 7nm kind of puts that to question unless Nvidia just got blindsided here.
What about smart-TV that's a huge market where Samsung is a major player with big volumes. I wouldn't be surprised if we see AMD-based graphics in Samsung smart-TV's. It wouldn't surprise me if AMD-Samsung has other deals as well since AMD tries to sell their portfolio where they can so if i guess we will see AMD product in other Samsung products as well and this is only the beginning. I don't think windows arm laptops with AMD will be any problem only another deal since AMD wants to be everywhere. About efficiency AMD gpu's scales down very well just look at mobile APU, this deal means scaling much more but i don't think it will be any problems since the new AMD architecture has been marked as even more scalable. AMD's new chromebook (A6-9220C, A4-9120C) APU's has an 6w TDP so we know AMD can scale down with that older 7th gen hardware.
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
94 Comments
Back to Article
AshlayW - Friday, June 7, 2019 - link
To be fair on AMD and GPU energy efficiency: the desktop Radeon cards are by no means a fair benchmark of GCN's efficiency, especially if you scale it down to a mobile SoC. What these cards' comparatively poor performance per watt (versus Nvidia for example) demonstrates, is poor performance per watt *scaling* at high performance levels. Ugh did I word this correctly?I mean that, those cards are pushes so far out of the architecture efficiency curve, because they didn't scale up to that performance efficiently. I would point out that at much lower performance levels and power optimised, GCN cards can be just as efficient as Pascal and potentially Turing: just at a much lower overall performance tier.
The mobile Ryzen APUs can demonstrate how efficient GCN can be.
webdoctors - Friday, June 7, 2019 - link
But can GCN scale down to the tablet/phone sub 5W envelope that well? Wouldn't Samsung partnering with Imagination make way more sense?Alexvrb - Friday, June 7, 2019 - link
I think they should have BOUGHT Imagination Technologies. They have some impressive designs, and with Samsung's cash they could do great things again.But yes, GCN can scale down and becomes far more efficient at lower power envelopes. You've already got laptop APUs on 14nm (and soon 12nm Picasso) that can TDP-down to 12W, and they include 4 full Zen cores and 10 CUs of last-gen GCN. That's just with old tech... take the next-gen GPU designs, customize them, and pair them with power-sipping ARM cores and more power efficient process tech.
niva - Monday, June 10, 2019 - link
They might very well buy Imagination Tech anyways. The way Samsung goes about business is to wage battle on multiple fronts and later determine which was the valid path forward. It's one of the biggest reasons why they're murdering the competition and probably approaching world-wide monopolistic scale in the tech world.StevoLincolnite - Friday, June 7, 2019 - link
There is a massive divide between 200w and 2w of power though.Yes, GCN can scale down in powerlevels really well, Vega 64 for instance was actually extremely efficient when you dropped the clocks and voltages... But instead AMD blew those two aspects out to get as much performance as possible to remain semi-competitive in the mid-high-end.
levizx - Monday, June 10, 2019 - link
If you consider existing Mobile Ryzen APU@15W, which already delivers up to 4 times the performance of Snapdragon 82x@5W, on the same process node, it's not that hard to believe even without substantial modification, they can deliver something at least equal to Adreno by the end of the year, just in time for 2021 Exynos.yannigr2 - Saturday, June 8, 2019 - link
AMD is far behind in efficiency, in both CPUs and GPUs. While APUs can be considered efficient, based on their TDPs, they can't offer the battery life that Intel processors can offer. That's why AMD is pushing in manufacturing by rushing to adopt 7nm.Dragonstongue - Saturday, June 8, 2019 - link
They are NOT AT ALL.for GPU depending on the specific generation, they are "just as good" as the other ( Nvidia ) and more often than not it was the Radeons that used less power / ran cooler / gave a better final end result (image quality or various other features) that is FACT.
HOWEVER, the generations that Nv was ahead in this regard, often were noticeably so, CPU wise.....Sorry to burst bubble, but Intel and AMD for the most part use the SAME or roughly the same wattage "over time" and truth be tole it is AMD actually with CPU and GPU that are more oft than not actually giving a very close to 100% true rating on their TDP/Board power etc without relying on fancy BS to achieve such.
in other words, Radeons (at least for the most part) are like junk yard Mac wreckers, sure they are louder, they sometimes belch some cobwebs, but darn it, they get any task done you need to do, and if you really want, take that wonking engine and put into a strip down race body and she just flies.
Nv has over 7 going on to 8 generations cut things away, added "dual" compute (or however want to call) to "trick" apps etc into using it like it is the full fat version when it is not, so they need fancy other chips.circuits to allow this, so the modern Nv is more like a Ferrari with all the fancy gauges that temperatures, heatsinks etc are as "cheap as can be, but quick acting" so instead of beefing up things such as better solder, better VRM and the like Intel and Nvidia whom you likely reference have cut things away, went to lower quality sub components (i.e paste instead of solder, solder instead of eutectic, 85/95 sometime 105c spec capacitor for VRM instead of at least 105/125 or 115/125c ...
So, while in aspect AMD did not do all that well below the 15w level, neither did Intel vs smartphone others and Nv is not the "golden child" folks keep pimping them up to be, they both ( Intel - Nv) do and still do A LOT of crap to mainly trick folks into continuing to use them regardless of what they are doing to make it so, as well, many folk sare BLIND BLIND BLIND to the point where their brand loyalty is the only thing that makes an iota of sense to their peanut of a brain.
Depending on product generation as well as how is used etc, Intel was "slightly" (last 20mins more for a bit more performance in a laptop as an example) but, at the same time the Radeons and AMD CPU, MOST PEOPLE who use them side by side to competing they all generally walk away "it feels the same, but I can actually have far more tabs open in my browser and it does not hitch as well I can leave system running for days-weeks-months and not really have to reboot fully ever with AMD/Radeon but with Intel/Nvidia I had to do a full shutdown once every 2 weeks like clock work or performance would just tank like crazy.....
I not trust numbers any of the "big whales" put out, as when a chip with a 105w TDP label is able to and consistently blows past this number (Intel 7-8-9 generations most recent) but yet the AMD ones you "claim" are SO FAR BEHIND rarely, truthfully rise above to much beyond 10% or so (which is spec to about 15% +/- rating not 30%+ BS STOCK...........
Phynaz - Saturday, June 8, 2019 - link
AMD fanboys are weirdMamiyaOtaru - Saturday, June 8, 2019 - link
why are you even bothering. Their new CPUs look to have caught up to Intel efficiency wise, but that'll be the first time since Conroe and everyone knows it. And nVidia? They've been lagging them on efficiency since Maxwell. https://www.techpowerup.com/reviews/EVGA/GTX_1650_... FFS. How do you look at that, and with a straight face vomit out some BS about them being toe to toe.The best I can say for your post was that it was so long I couldn't be bothered to read the whole thing so I probably missed a bunch more dumb stuff
jhon1236 - Thursday, June 13, 2019 - link
<a href="http://projectigigame.com/download/igi-3-game-free...
http://projectigigame.com/download/igi-3-game-free...
gdansk - Saturday, June 8, 2019 - link
Their CPUs are competitive performance per watt, excepting Xeon D which is not scalable anyway.Santoval - Saturday, June 8, 2019 - link
"Higher power efficiency" does not mean "more battery life" in the absolute sense. It simply means "higher performance per watt" (or "per hour of battery life", though that's technically *energy* efficiency).So, for instance : an APU with 2 TFLOPs of iGPU performance resulting in 6 hours of battery life has double the power efficiency (when only the iGPU is taken into account) of an APU with 500 GFLOPs of iGPU performance resulting in 12 hours of battery life.
An APU with 1 TFLOP of iGPU performance resulting in 6 hours of battery life, on the other hand, has exactly the same power efficiency as an APU with 500 GFLOPs of iGPU performance and 12 hours of battery life. Battery life alone is meaningless. The point is what performance you can milk from that battery life.
Dragonstongue - Saturday, June 8, 2019 - link
put down the crud pipe.... Radeon, GCN or no are and have always been just as good, if not more often than not better in regards to power efficiency/usage, a generation or two back and now they are/were also on at least if not better performance as well.It is amazing how much power can be saved via a simple voltage adjustment etc, with Nv not so much, whatever amount of power could have been saved essentially they did for you, mainly by cheaping out with chips instead of higher grade components (beyond those)
There is a reason why Nvidia is only in the Switch (for price it is vs how little actually get hardware wise vs the 2 "uncontested" that is Ps4 and XB1, there are only so many BS tricks they can pull to get performance high (no matter the game/application) while keeping power use down, temperatures as minimal as possible(therefore also very quiet)
base example, I take my 7870, I can "lock it" to just under factory speed (1040/1280 core/memory @ 1.167v 0% power slide...idle 367/600 @ 0.856v) vs factory of 1050/1200 @ 1219 and 300/150 @ .836v in all tests, when I "lock it" it transitions more quickly between P0-P3 (low 2d, low 3d, high 3d basically)
if watch youtube etc, unless locked will jump to 450/1200 @0.900v and burn about 20w more than when I set the way I do so, also, factory wise, maxes out at 175w (spec) mine never above 150w (even with heavy OC) my 826/1235 0.969v gets me 75% of way to "stock", I easily OC mine and still save power via voltage tuning and the like.
Long story short, in a "closed environment" they have access to the best silicon and anything they need to do "under the hood" is right there.
They are saying Radeon 5000....they should say something else, mainly due to it was not that long ago there was HD 5000 (potent asf cards they were as well, another generation of Radeons that were beasts (power hungry but could back it up, big time...take the time to do a quick voltage tune and dropped power use significantly enough.
to take to 2022+ before this can happen.....I call utter BS on that, no way, no how, they have 3 generations of Ryzens to go by, countless generations of Radeons etc....the combined might of 2 of the "leaders" especially Samsung and AMD...not buying going to take that long, especially with Sammy, they not "drop" what they are doing, what potentially 3-4 generations to make a key spec change before it is done, especially not announce 2+ years before it "might" become tangible.
LOL.........
Not count out whatever magic sauce them engineer folks can do, don't look at what Nv is doing and use that to gauge what AMD can or cannot do.
if you do that, that is not being fair to be honest, and, makes you act the putz
Phynaz - Saturday, June 8, 2019 - link
Let me restate thatAMD fanboys are a bunch of lying shills
Korguz - Saturday, June 8, 2019 - link
Phynaz,and the intel fan boys arent as well ??
FreckledTrout - Saturday, June 8, 2019 - link
At least you are concise and unbiased.V900 - Sunday, June 23, 2019 - link
Your claim about how great the GCN architecture is, kinda goes against every benchmark and GPU test out there.And as for why it’s AMD and not Nvidia who are getting console contracts, the answer is rather simple and has little to do with AMDs offerings being better or worse than Nvidia’s.
Nvidia just does t care to compete for the consoles. Why go into a bidding war with AMD for a contract worth pennies compared to what Nvidia can make with their 500$ graphic cards, that have a profit margin of 30-40%.
JasonMZW20 - Thursday, June 13, 2019 - link
Except this licensing deal doesn't cover GCN. RDNA is significantly different from GCN in terms of how work is executed and how hardware is arranged.GCN operated over 4 cycles using groups of SIMD16 (4x16 for 64 threads), so 16 threads could be finished in 1/4 of a cycle (0.25 IPC). RDNA does SIMD32 (2x over SIMD16) in 1 cycle via Wave32 (32 thread wavefront) and SIMD64 in 2 cycles (vs 4) via Wave64. That's a huge IPC shift right there.
And that's excluding the new cache hierarchies.
So, GCN's efficiency in lower wattage parts doesn't even remotely compare to RDNA.
ballsystemlord - Friday, June 7, 2019 - link
1 grammar mistake:"this this currently isn't possible with, say Mali designs, as a licensee isn't able"
Extra "this"
"this currently isn't possible with, say Mali designs, as a licensee isn't able"
Thanks for the writeup guys, I was also really curious about this subject!
Ryan Smith - Friday, June 7, 2019 - link
Thanks!ballsystemlord - Friday, June 7, 2019 - link
OT: Wasn't AMD's Zen 2 supposed to launch today?futrtrubl - Friday, June 7, 2019 - link
7th July, not June.ballsystemlord - Friday, June 7, 2019 - link
Ah, my mistake, zen 2 launches in 1 month from today. :embarrassed:futrtrubl - Friday, June 7, 2019 - link
Not so sure this is not Navi based. They specifically said RDNA architecture.ajp_anton - Friday, June 7, 2019 - link
And Navi isn't even 100% RDNA, it's somewhere between that and GCN (so the sources that used to claim Navi was still GCN weren't as wrong as they looked when RDNA was announced).What comes after Navi will actually be RDNA.
extide - Friday, June 7, 2019 - link
Navi uses the GCN instruction set -- and there is nothing wrong with that. The GCN instruction set is fine. Once AMD moved away from Terascale and it's vector ISA to GCN's scalar ISA there is no point in majorly changing the ISA again.However, the physical implementation definitely had some weak points. Geometry performance (all fixed function hardware), the max limit of 4 shader engines, etc. These are things that can be completely re-arched while still implementing the same ISA. It's much like AMD and Intel can both build CPUs that work on the same ISA but use completely different physical implementations.
This is good because they finally decided to kill the 'technical debt' that had built up in GCN, but yet they don't need a 100% new driver stack. Even though they said it was 100% new (and Lisa Su DID say it was new from the ground up, and I don't think she would straight up lie, plus she comes from a technical background so I am sure she knows the implications of the words she said) that still doesn't mean they HAVE to get rid of the entire old architecture. There were plenty of parts of GCN that work fine, what's the point in getting rid of those? The big difference between this arch and previous ones is that they decided to actually fix some of the lower level roadblocks rather than just work around them because "fixing them would be too much work" -- it's going to be GCN derived, sure, it would honestly be stupid not to be, but it's going to ALSO be free from the chains that have dragged GCN down. Honestly I'm pretty fuckin excited, can't wait to see more details on it. This arch will almost for sure be VERY graphics focused, and have plenty of power in the graphics areas that GCN was typically weak (geometry, etc) because they are still KEEPING GCN for compute focused products. So they must be different enough to warrant doing that.
dr.denton - Saturday, June 8, 2019 - link
Thank you. Everyone who thinks GCN is some kind of deal breaker for Navi should read your post.oleyska - Saturday, June 8, 2019 - link
I'd like to draw upon the similarities of many archs we've seen in the past.AMD, Hypertransport and so many design elements from Phenom II can be seen in the Zen architecture.
IT's all just the same idea, brought forward to Zen.
shame there wasn't as much love for the Phenom II to describe more of it's inner workings but it in essence looks very similar.. just didn't have the core to bring it forward.
yet everything around it wasn't dumb, it was smart.
Parts of it carried over to bulldozer and is it a bad thing that 1 inferior cpu and 1 very underperforming\ineffecient cpu has probably 50% of it's design ideas in the new cpu?
No, it's what makes zen better than Intel's architecture from a Cpu vendors perspective.
So I am also not scared of things being based on previous inferior architectures and Intel and AMD have proven this in the cpu space - Not so good on gpu's, but probably there too :)
del42sa - Saturday, June 8, 2019 - link
it reminds to be seen how big those changes are... They said many times, that chip was completely revamped, they said it about Polaris and then about Vega as well. But when you look at the CU unit with 64SP and L1 cache and how it´s being arranged inside of chip, you can clearly say, the are (almost) identical.The descripction of RDNA and NAVI says it : Stremlined pipeline, new compute unit design, multilevel cache hierchy.
One SP still consists of 64 SP
*Steamlined pipeline could point to Vega minus all not game necessary stuff.Well we ´ll see.
*New compute unit design sound more promising.
*Multilevel cache hierarchy is very vague term, because it could mean literally anything
del42sa - Monday, June 10, 2019 - link
it remains to be seen....Streamlined..
Sorry for mistakes in grammar
TristanSDX - Friday, June 7, 2019 - link
next gen PC will be smartphone. Expect CPU, GPU and Windows installed on phone, docked to keyboard, mouse and monitor via wireless connectionsHStewart - Friday, June 7, 2019 - link
Windows Phone failed and even Windows for ARM can not save it.Best thing Microsoft can do it make a new kind of device that in between Phone and Tablet.
oleyska - Saturday, June 8, 2019 - link
No, I don't think you understand microsoft's new strategy.It's new ceo is brilliant..
Microsoft's profit is shooting through the roof and market is more open than ever, they are biggest contributor to Linux, DirectX is loosing, you can now game on Linux thanks to many of Microsofts contributions.
Windows will likely not be based on a windows kernel for too long, cause what is the point as Microsoft doesn't earn money on our Windows licenses but rather our SQL, O365, powerbi etc etc.
Azure servers running Linux from Microsoft and they are earning shit tons by embracing Linux instead of fighting it.
Would not surprise me if we saw a Linux distribution from Microsoft....
Fergy - Saturday, June 8, 2019 - link
What if we could build a tiny pc with a tiny heatsink that can use at most 7 watts... That would obviously be better than a bigger pc with a bigger heatsink that can use 500 watts when needed.HStewart - Friday, June 7, 2019 - link
I am curious in mist of things, how ARM's own Mali GPU plays into this picture.From what it looks like Samsung was looking to developed their own GPU to compete with Apple iPhone/iPad GPU's. So it sounds like Samsung was having trouble on their own so Samsung will get a major benefit of this arrangement with getting technology from AMD.
https://www.androidauthority.com/samsung-exynos-gp...
Maybe AMD could be even struggling to sell off their struggling GPU business. With a company as large as Samsung, one possibility in the future is that Samsung would buy out AMD. That would be easiest way that Samsung could gain the GPU technology at minimal cost.
Threska - Friday, June 7, 2019 - link
Struggling? Seems the GPU business is recovering nicely.Korguz - Friday, June 7, 2019 - link
oh look.. MORE AMD bashing and false info with NO proof from HStewart. amd's gpu side is not struggling... but you know who IS struggling?? your beloved intel... struggling for FOUR YEARS now to get 10nm working... and now.. struggling with the aspect of losing the performance crown....HStewart - Saturday, June 8, 2019 - link
Did I say anything about Intel in this message - some people are so blind sided and only think of one area of computers and that is gaming desktops.This is a realistic question I propose here, what is going on behinds the scenes of this deal. Samsung is a huge company - and they want to create their own GPU also.. so a buy out of AMD would be logical option for this - it may not be a complete buy out - but NVidia without a doubt is the leader in discrete GPUs.
Korguz - Saturday, June 8, 2019 - link
HStewart, you don't have to.. your pro-intel bias shows in the vast majority of your posts. you call some people blind sided.. but yet, YOU do the SAME thing when it comes to intel, losing the performance crown, is just not in games, as YOU assume it to be.. but it COULD also be in the server space too. Samsung was creating their own GPU, but have decided to licence the IP instead.. I think this was stated in this article, or the other one about this that AnandTech posted.Phynaz - Saturday, June 8, 2019 - link
Oh look...MORE AMD shilling and false info. AMD consistently loses money in GPUs because nobody but poor fanboys want them.Korguz - Saturday, June 8, 2019 - link
Phynaz um yea sure... if you say so...Qasar - Sunday, June 9, 2019 - link
Phynaz what part of korguz's post was false info ? they HAVE been struggleing to get 10nm to market, it has ONLY been 4 years, or has it been 5 ?? that intel has been feeding everyone its lies about 10mn...Threska - Friday, June 7, 2019 - link
"Given what little we know about Navi (even though it’s launching next month) we won’t go into speculation about just what “Next Gen” entails, but regardless it’s going to be the basis of Samsung’s first generation of Radeon RDNA-derived GPUs."That's where patent reading comes into the picture.
Kevin G - Friday, June 7, 2019 - link
This article is written from the perspective of 'what is Samsung getting from AMD?' and that is fair because Samsung is indeed getting a lot. However, collaboration is a two way street so the raises the question of 'what could AMD get from Samsung?'.The first thing that comes to mind is memory. In particular HBM as future products on AMD are to evolve past interposers and move toward direct die stacking. The level of integration between stacked memory and the core logic chip requires significant engineering to pull off, something they would have to collaborate together on. This also benefits Samsung as such die stacking can be beneficial in mobile as well but Samsung would needs more additional expertise in GPU and CPU areas, enter AMD. What else can AMD gain: integrated SSD controllers. Instead of relaying external NVMe controllers via PCIe, why not simplify an AMD SoC mobile board design by including a high speed ONFI interface directly on an AMD mobile chip? The same can be said about LCD and other display controllers currently handled by external drivers. Half a step in the other direction would be Samsung leveraging Infinity Fabric and/or AMD's own on-die interconnect. The benefit here for AMD is the increase in peripherals on their proprietary bus, like say ultra high speed SSDs for servers.
zodiacfml - Friday, June 7, 2019 - link
Same. I was reading the whole thing, thinking what's in it for AMD.Great point about HBM. I've thought HBM as the future when HBM2 came out where it doubled over HBM1, and saw how it can make GPU cards shorter.
AMD might need Samsung to scale HBM into more products. APUs, GPUs, Consoles. It is great for integration, bandwidth, and power efficient
Threska - Saturday, June 8, 2019 - link
Have a Vega 56. About same length as the old card. Only change is power on the side instead of ends.Vivaxeman - Friday, June 7, 2019 - link
In short, like JBL, AMD will be gobbled up by Samsung in 2 years. Intel & Nvidia can kiss their respectove a$$es goodbyeKorguz - Friday, June 7, 2019 - link
um yea.. ok sure Vivaxeman, you have poof of this ?Vivaxeman - Friday, June 7, 2019 - link
Samsung - Total equityUS$188.9 billion Revenue
US$210.9 billion - Products & Services -
Apparel, automotive, chemicals, consumer electronics, electronic components, medical equipment, semiconductors, solid state drives, DRAM, ships, telecommunications equipment, home appliances[2]
Services
Advertising, construction, entertainment, financial services, hospitality, information and communications technology, medical and health care services, retail, shipbuilding || AMD Total equity
US$1.266 billion Revenue
US$6.48 billion
Korguz - Saturday, June 8, 2019 - link
thats not proof.. thats just a list of samsungs assets...nice try... the same was said a few years ago when there were rumours that various companies were interesting in buying AMD then. how about, oh i dont know.. proof that samsung has any interest in buying AMD.. or mention of it...Vivaxeman - Saturday, June 8, 2019 - link
Well, now they defenitely have interest in their IP, makes their chips & have loads of cash.HStewart - Saturday, June 8, 2019 - link
One thing to add to this, is Samsung is looking to build there own GPU's and it logical they would grab AMD for this part of business. To Samsung buying AMD would be only a scratch in their pocket book.https://www.androidauthority.com/samsung-exynos-gp...
I think the first sign of this happening is when and if we see Samsung Notebooks and PC Tablets going with AMD CPU/GPUs but then again Samsung would rather put their own Processors - so maybe there will be Windows for ARM compatible with Samsung processors coming out.
Korguz - Saturday, June 8, 2019 - link
HStewart, did you even READ this article ?? YOUR own words : " I think the first sign of this happening is when and if we see Samsung Notebooks and PC Tablets " from the article that you OBVIOUSLY didn't read : " In short, the AMD/Samsung deal is structured so that Samsung is only allowed to use AMD GPU IP in segments that AMD doesn’t compete in. In other words, AMD’s GPU tech can only be used in smartphone and tablet SoCs. For anything else, Samsung is either prohibited from using the tech, or at best, has to get AMD’s permission first. " where do you come up with this BS ?? do you make it up as you go along ??? whats next.. when AMD announces their zen 2 based APUs are you going to say they upped their voltage to 12v and 15v like you tried to tell people in another article that intel upped the voltage on the 10nm cpus to 5v and 9volts ???visvim - Saturday, June 8, 2019 - link
I doubt US Government would be too fond of that passing sensitive US tech to a non-American company would raise alarms... like how Broadcom tried to acquire Qualcomm.HStewart - Saturday, June 8, 2019 - link
Well US Government did not stop AMD to give technology to China which is worst. There is also always way that Samsung could get around those restrictions like putting offices on American soil and limited what technology goes off countries that America does not like. I have Toyota Tundra and Toyota did exactly that.Just FYI, I have Samsung Note 8, Samsung Tab S3 and Samsung Tab Pro. So I am not bias as some would like to think I am. I used Apple Phone and iPad for years and to me honest there OS technology did not advance. I was once very bias against Android.
Korguz - Saturday, June 8, 2019 - link
oh but you ARE as bias as some know you to be... Mr Pro intel fanboyPhynaz - Saturday, June 8, 2019 - link
Hello, Pot?DannyH246 - Saturday, June 8, 2019 - link
Hstewart hasn’t got a clue. If anyone buys AMD they would have to get Intel’s approval to keep their x86 license. You think Intel would allow Samsung to keep the x86 license? Dream on.Shlong - Saturday, June 8, 2019 - link
DannyH246, that's not true. AMD can get acquired and the x86 licensing deal will not be affected. Back in 2015: "Devinder Kumar, chief financial officer of Advanced Micro Devices, said in a statement last week that the company could enter into joint ventures, mergers or acquisitions (M&A) agreements without fearing of termination its cross-license pact with Intel Corp. Many industry observers believe that a bigger company cannot acquire AMD since this will terminate the deal with Intel and will leave AMD without an x86 license immediately."Qasar - Sunday, June 9, 2019 - link
Shlong, actually.. it looks like if EITHER amd OR intel get acquired by another company, their cross licensing deal, is terminated, which means.. NO x86 license for the company that buys AMD :https://www.kitguru.net/components/cpu/anton-shilo...
i dont know how, reliable, kit guru is.. but if that is true.. then that could be why no one has bought amd over the years.....
Qasar - Saturday, June 8, 2019 - link
DannyH246 he really doesnt.. i read in another thread about the HP laptop with a leather cover, where he tried to claim that intel upped the voltage of their 10nm cpus to use 5volts and 9 volts.. turns out.. he was referring to TDP. he has stated he has been using comps for years, and knows alot about them, but doesnt know the difference between TDP and voltage ?? come on .. the sad part, he even linked the the article that talked about it...HStewart - Sunday, June 9, 2019 - link
Wrong the specs in chart indicated that Intel up Y series to 9V from 5V. I only stating what the table states. Now the question is why was that required and it likely because WIFI 6 and TB3 integrated in chip. But we really don't know until it officially release.Qasar - Sunday, June 9, 2019 - link
come on HStewart.. even you should know cpus DON'T take 5 OR 9 VOLTS... and what table says that ?? i just checked.. and no where does it say those cpus take 5 or 9 volts, face it.. you made a mistake.. then tried to accuse others of not reading properly...Korguz - Saturday, June 8, 2019 - link
hello, HStewart ?gadrew - Thursday, July 11, 2019 - link
what does pot mean ?HStewart - Sunday, June 9, 2019 - link
Just because I don't care for AMD Junk, It does not mean I only use Intel Products. Give me a break. I am not that naïve to Intel. I just believe one should go one that created the products. Times do change - for most of my smart phone uses was Apple. iPhone 3 - iPhone 6, but I like The Galaxy Note 8 because of my experience with Galaxy Tab S3.I personally have not had good experience with AMD products. At my first Job, I actually went to AMD office and got technical books to complement my collection of Intel - this was in days before the Internet.
Qasar - Sunday, June 9, 2019 - link
maybe not.. but 90% of your posts.. clearly show your intel bias, and your pro intel views... as i have read many times.. any time an article or news post on here, put intel in the bad said, or amd on the good.. there you are trying to twist things around, to make intel look better.. like the cray super computer, AMD getting pcie for what looks like could be months before intel ( then you claim intel will skip pcie 4 and go straight to pcie 5 ) there are MANY other such comments from you that i have read on here like this....Korguz - Monday, June 10, 2019 - link
if it wasnt for some of that " AMD junk " we may not have some of the things we have now, look how long intel stuck the industry with quad core cpus for the mainstream, making us go to the HEDT for 5 or more cores... or the so generous 10% or less performance increases year over year, zen comes out.. and all of a sudden.. more then quad core for the mainstream. you may be fine with intels lies, BS and cpu stagnation over the last few years, and im sick of it. i9 series, 9820x, $1210, 9980xe, $2800!!!!!!!!!!!!!! 9900k $700. but if i were to upgrade my comp now.. it very well could be zen 2 based... wont know for sure for another month, but more cores, same or better performance, for A LOT less then what intel charges." I am not that naïve to Intel " heh, yea right, sorry hstewart, but 90% of your previous posts, say other wise
Kevin G - Saturday, June 8, 2019 - link
It wasn't just JBL, the Harman Group which included professional names like AMX, SVSI, Crown, Lexicon, BSS and others. Samsung is pretty much running those respected name brands into the ground. I can get way Samsung would wind some of those down as they overlap with some of their own products but they are giving themselves a bad name while doing so and not leveraging the technologies those Harman Group companies developed.Samsung acquiring AMD would be a horrible move. This collaborative partnership avoids the disaster that Samsung wrought upon the Harman Group by keep the focus on mutually beneficial technologies.
HStewart - Sunday, June 9, 2019 - link
Move maybe bad for AMD fans - but for Samsung it likely fits a new, they are interesting make their own GPU architecture and AMD is a drop in the bucket for Samsung. These type of deals happen after earlier deals all the time. I bet even you look back something similar happen when AMD got ATI. But it appears AMD actually wanted NVidia but of course NVidia didn't do ithttps://www.forbes.com/sites/briancaulfield/2012/0...
Qasar - Sunday, June 9, 2019 - link
ahh so ANYTIME 2 tech companies agree to do a joint venture, or share IP, it means one will eventually be bought out by the other.... yea ok. sure... all be cause HStewart said so...Kevin G - Monday, June 10, 2019 - link
That does happen from time to time. See Intel and Altera as they had partnered for both providing fab space to Altera for FPGAs and Intel got a chip to put into a Xeon package for data centers. The partnership worked out so well that before either of those two things actually hit the market to purchase, Intel and Altera had merged.It also doesn't always work out. See Intel and Micron which transpired during the same time period as the Intel + Altera example.
While more of a negative trait, one the things that keeps AMD out there as an independent company is the amount of debt that it incurred a few years back. Granted the worst is over and the company is healthy once more but the effects are still present today. Various anti-trust regulators would also like to have a look at the merger before approving it. The last settlement AMD had with Intel wrote out that the x86 license AMD has can be transfered with the blessing of Intel or Intel themselves would invoke the wraith of antitrust regulators too.
Kevin G - Monday, June 10, 2019 - link
Samsung could have easily acquired Imagination a year or two back fi they wanted a competitive mobile GPU company and it was even cheaper than AMD. It would have also gotten Samsung into another position as a both a supplier and competitor to Apple. Alas, that what-if never happend.Samsung has its sights on mobile with no real interest in the PC or data center space. Samsung would run of what AMD is doing today into the ground and walk away.
RedGreenBlue - Friday, June 7, 2019 - link
This is some of the best tech-journalism I’ve ever read. Nice work.I wonder if AMD sees this as a path towards being in more mobile devices than phones or tablets, like the arm-based VR headset market. It could force some game developers wanting to be on those platforms to gear code for AMD’s new GPU architectures, and maybe make it easier for those games to be tuned for desktop GPUs.
RedGreenBlue - Friday, June 7, 2019 - link
Have to wonder if Sony and Microsoft are interested in ARM-based VR headsets in the future.Threska - Sunday, June 9, 2019 - link
AMD currently has momentum, and a lead. As history shows one can't sit on their hands milking a lead because eventually everyone else catches up.RealBeast - Saturday, June 8, 2019 - link
In phones and tablets -- sounds like a win/win and another great source of more cash for AMD to continue pushing the envelope in CPUs.spaceship9876 - Saturday, June 8, 2019 - link
Samsung should just buy Imagination Technologies (creators of PowerVR).tkSteveFOX - Saturday, June 8, 2019 - link
I am willing to bet we will see a AMD IP tech in a Samsung chip in 2021! Knowing Samsung, they will test run it with a mid-range Exynos, not flagship.Samsung are far, far behind QC, Huawie, let alone Apple. Exynos's performance per watt is a generation behind everyone else on almost any level, especially the CPU part.
Death666Angel - Saturday, June 8, 2019 - link
"Samsung are far, far behind QC, Huawie, let alone Apple. Exynos's performance per watt is a generation behind everyone else on almost any level, especially the CPU part. "I'd suggest taking a look at the recent articles here on AT (the A77 one for example). Samsung isn't that far behind. 9820 made up a lot of ground on power consumption. They are definitely better than 845 and 970. So not a full generation by any measurement.
Santoval - Saturday, June 8, 2019 - link
"Thankfully, AMD’s GPU public roadmap is short and simple. There’s only one architecture on the roadmap after Navi, and that is the enigmatic “Next Gen” architecture. Essentially nothing is known about this architecture..."It is no longer called merely "Next Gen". AMD already announced that it is going to be called "Arcturus", which is much less enigmatic and much more stellar (literally). That graph is just old. Arcturus is expected to be AMD's first post-GCN architecture, however it is not yet clear if it will be a 2020 release. It will most likely slip into 2021, perhaps Q1 2021.
If the deal with Samsung is for 2022+ they will apparently license the original Arcturus and work with AMD around that μarch in the fashion they want.
del42sa - Sunday, June 9, 2019 - link
No, they never say that ! Arcturus is not the name of the architecture, but name of a chip. No name announced yet, so it´s still "next gen" moniker.https://www.anandtech.com/comments/14412/amd-tease...
eastcoast_pete - Sunday, June 9, 2019 - link
Thanks Andrei and Ryan! Regardless of whether LSI will actually produce mobile graphics tech based on AMD IP (they probably will use at least some), I believe this agreement also means:- Samsung has effectively given AMD a non-compete agreement in the Windows/mobile computing space
- this, in turn, could give AMD the peace of mind to turn to Samsung for future manufacturing deals, giving them both leverage when negotiating with TSMC and access to Samsung's 5 nm EUV processes.
I would be surprised not seeing some of AMD's upcoming chips being fabbed by Samsung. One of the (many) reasons why Intel's attempt to offer contract fabbing failed utterly was the understandable reluctance of potential competitors to hand other their designs to Intel and trust the firewalls inside Intel's to actually hold. That, and Intel's inability to actually get their fabs making 10 nm chips in quantity for several years.
mrvco - Sunday, June 9, 2019 - link
Saw the headline and braced for "Taizen Gaming Console"MadManMark - Monday, June 10, 2019 - link
Based on that comment I can tell that you only saw the headline before posting, and didn't read the article (or at least read it closely).Try going back and reading the section with the bolded subtitle "Samsung Can’t Compete with AMD: Smartphones & Tablets Only"
YouInspireMe - Sunday, June 9, 2019 - link
VR usage is growing exponentialy. Mobile processors are obviously going to dominate untethered VR devices. Current untethered VR users have to make many concessions because of the low GPU capabilities in mobile devices. If AMD and Samsung can dominate this arena they will control what is bcoming a very large tech sector.FunBunny2 - Sunday, June 9, 2019 - link
"VR usage is growing exponentialy. "other than inforainment uses? those are just drains on the economy. not a good thing.
YouInspireMe - Monday, June 10, 2019 - link
https://www.roadtovr.com/monthly-connected-vr-head...carcakes - Sunday, June 9, 2019 - link
New SD Express standard brings PCIe NVMe speeds to regular SD cardshttps://m.gsmarena.com/new_sd_express_standard_bri...
croc - Monday, June 10, 2019 - link
Just wait until Trump gets the shits with S. Korea's trading practices I mean how dare they take money away from Apple....MadManMark - Monday, June 10, 2019 - link
"National Security Threat" for sure! I'm waiting for him to declare blue states still being in the union a "national security threat."Dragonrider - Monday, June 10, 2019 - link
Sounds like Samsung dipping their toe in the water to see if they want to buy AMD (or at least RPG), but then the news that Ampere will be on Samsung 7nm kind of puts that to question unless Nvidia just got blindsided here.tekniknord - Monday, June 10, 2019 - link
What about smart-TV that's a huge market where Samsung is a major player with big volumes.I wouldn't be surprised if we see AMD-based graphics in Samsung smart-TV's.
It wouldn't surprise me if AMD-Samsung has other deals as well since AMD tries to sell their portfolio where they can so if i guess we will see AMD product in other Samsung products as well and this is only the beginning.
I don't think windows arm laptops with AMD will be any problem only another deal since AMD wants to be everywhere.
About efficiency AMD gpu's scales down very well just look at mobile APU, this deal means scaling much more but i don't think it will be any problems since the new AMD architecture has been marked as even more scalable.
AMD's new chromebook (A6-9220C, A4-9120C) APU's has an 6w TDP so we know AMD can scale down with that older 7th gen hardware.