AMD "Renoir" APU iGPU Configuration and Platform Spread Detailed

[Tomorrow]

10900KF sounds stupid but actually follows their established naming model. 10 - 10th gen, 900 - higher end, K - unlocked multiplier, F - without/disabled iGPU.

Well then there's mobile. The reason i mentioned 10900 was because i could actually remember that. I could not remeber the 10nm/14nm naming scheme mess they currently have on the mobile side.

 

[notb]

Well then there's mobile. The reason i mentioned 10900 was because i could actually remember that. I could not remeber the 10nm/14nm naming scheme mess they currently have on the mobile side.

And that mess is what exactly?
The new Gx naming? It's supposed to be for chips with better GPU. Isn't it intuitive?

 

[AnarchoPrimitiv]

If the memory bandwidth isn't there for more CUs, why not add 1-4GB of embedded HBM2(e)?

Personally, I would LOVE it if AMD stopped screwing around and made a powerhouse APU for desktop that would have enough graphical horsepower to basically gut the low end dGPU market. It would need 4GB of embedded HBM2, but personally I wouldn't care if the entire APU package was the size of threadripper and required 200W (as long as it has 8c/16T at least) and cost $500.

I've been following patents for about 5 years now, and from what I've seen, AMD has the IP to do this... So perhaps I'm the only one that wants an APU with the power of the new gaming consoles?

 

[Flanker]

I really hope AMD doesn't change the iGPU naming scheme again. Current Vega 3/8/11 is short and concise. If it's a mix of Navi and Vega like the leak claims just name it Vevi or better yet Naga.

I swear AMD is the fucking worst when it comes to naming their products. They ditched Radeon HD 8000, 9000 and went straight to Rn 200. Then after RX 500, they leaped back to RX 5000. Jesus fuck! And just forget about whatever they named their mobile and desktop APUs in the meantime. People make fun of Intel's 10k series naming scheme but at least they and Nvidia to some extent follow the numbering scheme.

Not to mention the fury's and radeon VII, what the hell? Even when there used to be some sort of consistency, we had 2900 XT being the flag ship, and then 3870 is...? uh, some more power efficient card with similar performance to 2900XT, then there is the brilliant 4780.

 

[Vya Domus]

the memory performance Is there, AMD refuses to use it. Look at the surface book 3. The AMD model has dual channel DDR4 2400, the intel model has dual channell 3733 LPDDR4X. That is nothing short of a MASSIVE jump in bandwidth. And it's not because MS is anti AMD, its because AMD only supports 2400MHz memory on their NEWEST APU!

How many LPDDR4X modules are out there in laptops ? How many laptops have more than 2400Mhz rated DDR4 modules ? Can you see the problem there ? It's a waste of effort and time for AMD to fit different memory controllers and change their architecture for a fringe memory type and speeds that aren't widespread. The only reason Intel did it is because it coincided with their newly developed architecture, 1065G7 is nothing short of a PR product showing supposedly that they are better than AMD. Go and tell me how many laptops with 1065G7 in them them that have 3733mhz rated LPDDR4X are there, I am waiting, then come back with a better theory.

LPDDR4X and higher speed DDR4 have no future with DDR5 around the corner which will make it worthwhile to change their APUs. Also, no, the memory performance isn't there, even with something like 3733 LPDDR4X these things are still memory bandwidth starved.

AMD refuses to employ a better memory controller to take advantage of higher memory speeds that could feed a newer, better iGPU. AMD's refual to embrace higher memory speeds, LPDDR4X, refusal to use the latest transistor processes, refual to use the newerst CPU arch, and refusal to use the latest GPU arch merely shows AMD doesn't take mobile seriously, or doesn't care about mobile users.

I cringe reading this, on the other hand it seems that you take this too seriously. I explained why, it's not worth it, as with any other company they care only about what's going to yield results.

It is curious though that you bring this up, are you aware that there have been practically zero innovations in the last decade or so with anything that was GPU and mobile related from Intel ? When AMD trashed them in this area in terms of performance Intel was suddenly able to come back with a newly developed architecture with LPDDR4X and all that. If Intel was so concerned with the well being of their mobile users why have they sold them gloried display adapters all this time ? It seems like AMD raising the bar tremendously was just them not caring about mobile. These are some interesting theories indeed.

If the memory bandwidth isn't there for more CUs, why not add 1-4GB of embedded HBM2(e)?

Too costly and complicated for now, if AMD ever gets some considerable market share I suspect they will pull this off eventually.

 

[Tomorrow]

And that mess is what exactly?
The new Gx naming? It's supposed to be for chips with better GPU. Isn't it intuitive?

 

[R0H1T]

LPDDR4X and higher speed DDR4 have no future with DDR5 around the corner which will make it worthwhile to change their APUs. Also, no, the memory performance isn't there, even with something like 3733 LPDDR4X these things are still memory bandwidth starved.

A ddr5 APU is at least 2 years away, I don't see any mass market APU supporting ddr5 till late 2021. The key reason why Intel & AMD have moved to LPDDR4x is power savings & I'm pretty sure ddr5 at launch will consume more power than LPDDR4x ~ so till the time LPDDR5 becomes mainstream LPDDR4x isn't going anywhere. Secondly it's used in most premium & mid range devices (smartphones) these days & hence the cost/bit will favor it massively over the foreseeable.

Yes & this is why we'll need HBM in there somewhere, ddr5 or ddr6 will not solve that problem.

 

[notb]

This may seem weird to you, but not everyone likes to watch these lengthy videos.

Is there anything objectively interesting or is this just an extreme opinion that you identify with?

 

[Vya Domus]

A ddr5 APU is at least 2 years away, I don't see any mass market APU supporting ddr5 till late 2021. The key reason why Intel & AMD have moved to LPDDR4x is power savings & I'm pretty sure ddr5 at launch will consume more power than LPDDR4x ~ so till the time LPDDR5 becomes mainstream LPDDR4x isn't going anywhere. Secondly it's used in most premium & mid range devices (smartphones) these days & hence the cost/bit will favor it massively over the foreseeable.

Yes & this is why we'll need HBM in there somewhere, ddr5 or ddr6 will not solve that problem.

In a laptop the power saved is negligible you can't really compare the two.

 

[notb]

A ddr5 APU is at least 2 years away, I don't see any mass market APU supporting ddr5 till late 2021. The key reason why Intel & AMD have moved to LPDDR4x is power savings & I'm pretty sure ddr5 at launch will consume more power than LPDDR4x ~ so till the time LPDDR5 becomes mainstream LPDDR4x isn't going anywhere. Secondly it's used in most premium & mid range devices (smartphones) these days & hence the cost/bit will favor it massively over the foreseeable.

LPDDR5 will arrive around a year after DDR5, so at least 2 generations of mobile APU (Zen 2, Zen 3) could have used LPDDR4X.
Actually, this is where weird AMD launch strategy makes sense (by coincidence; likely for the first time). Mobile Zen4 will come a year after desktop Zen4, so both just in time for appropriate DDR5.

In a world dominated by mobile devices, LPDDR4(X) will soon be the most popular RAM (maybe it already is). And it's really fast as well.

 

[Apocalypsee]

AMD better coupled the B12 variant that carries 12-13CU with fast LPDDR4X RAM or there is no point having more CU in mobile platform but using DDR4 that usually at 2400MHz or 2666MHz and they better be strict about it like setting minimum speed for RAM and dual channel. Some vendor like HP usually cripples AMD hardware by employing single channel RAM (sometimes without the ability to upgrade), limiting TDP and/or putting inadequate cooler while at the same time putting crappy dedicated GPU that is either on par or lower than the iGPU itself.

 

[gamefoo21]

It's really interesting because these GPU cores are basically from what I read as the V20, but AMD disabled the enhanced Navi functionality. Vega is massively better at OpenCL than Navi, for pro purposes you want GCN still.

What I think is annoying though from those product leaks if true, the desktop tops out at 10cu, while the notebook parts get 12cu. That makes little sense to me, give me power budget and all the cus! 12 CU and a 15W power budget means a couple hundred MHz if the CPU cores are doing anything.

I still think it's sad that the best IGP that AMD has made in recent years was for Intel... 4GB of HBM that could be used as cache for the CPU... or that monster in the One X... Beautiful...

 

[INSTG8R]

still think it's sad that the best IGP that AMD has made in recent years was for Intel... 4GB of HBM that could be used as cache for the CPU... Beautiful...

And they literally ignored it. I‘m guessing maybe 6 months or so now they finally let AMD give it some actual support and updated and functional drivers for it, and they continue support it, where Intel literally all but left it useless never giving it a chance to shine.

 

[Sybaris_Caesar]

I sometimes dream AMD would one day make such a powerful APU that it would all but dominate low end dGPUs. Then AMD would leave the sub-$200 GPU market to its APU and their discreet GPU lineup would start from $200 or $250.

BTW mobile APU having more CU makes sense I think. Nvidia tactic maybe. More cores but all clocked very low. And desktop APU having too much power doesn't make sense. While I'll dream on my little dream, it's an expensive upgrade path tbh. Instead of just upgrading CPU or GPU individually when needed you have to upgrade the APU that won't be cheap if it is a decent CPU as well as sub-$200 dGPU-slaying iGPU. Not to mention it'll be memory bandwidth starved on DDR5 or whatever DRAM tech is used.

 

[R0H1T]

LPDDR5 will arrive around a year after DDR5, so at least 2 generations of mobile APU (Zen 2, Zen 3) could have used LPDDR4X.
Actually, this is where weird AMD launch strategy makes sense (by coincidence; likely for the first time). Mobile Zen4 will come a year after desktop Zen4, so both just in time for appropriate DDR5.

In a world dominated by mobile devices, LPDDR4(X) will soon be the most popular RAM (maybe it already is). And it's really fast as well.

Your info is outdated SD865 launches with LPDDR5, so does Axx & likely Exynos 9xxx also I'm not sure where you originally got this from

The DDR5 launch will actually happen with the mobile world first. Unless your point was something else entirely?

Samsung Starts Production of LPDDR5-5500 Devices: 12 GB of DRAM in a Smartphone

www.anandtech.com

Some vendor like HP usually cripples AMD hardware by employing single channel RAM

Yeah unfortunately you can't (easily, if at all?) upgrade LPxxx as far as I know, so if the OEM's pulled the same trick you're stuck with that till the end of product's life.

 

[gamefoo21]

And they literally ignored it. I‘m guessing maybe 6 months or so now they finally let AMD give it some actual support and updated and functional drivers for it, and they continue support it, where Intel literally all but left it useless never giving it a chance to shine.

Yeah, great hardware is often crippled by horrible software. That's really kind of depressing that it took that long for driver updates to start rolling out.

Though considering how things have been with my Fury X, the drivers can be moody, since that 'Vega' is really more of a baby Fury, since it uses Polaris CUs with HBM attached. I think that's also why you are seeing the new APU getting the Vega compute units because OpenCL performance on the Fury branch was never great.

 

[Tomorrow]

This may seem weird to you, but not everyone likes to watch these lengthy videos.

Is there anything objectively interesting or is this just an extreme opinion that you identify with?

Watch from 6 minutes on. Takes a few minutes to watch. Is it so hard? Or did you not bother to watch it because it goes against your opinion?

 

[notb]

Your info is outdated SD865 launches with LPDDR5, so does Axx & likely Exynos 9xxx also I'm not sure where you originally got this from

I obviously meant laptops, not devices in general.
Smartphones are getting LPDDR5 next year.

LPDDR4's first appearance in smartphones: Samsung S6, early 2015.

Watch from 6 minutes on. Takes a few minutes to watch. Is it so hard? Or did you not bother to watch it because it goes against your opinion?

How can I know if it goes against my opinion, if I haven't seen it?

Seriously, some of us just don't watch vlogs.

 

[Turmania]

So what kind of perdormance increase can we expect say from ryzen 5 3400G?

 

[notb]

So what kind of perdormance increase can we expect say from ryzen 5 3400G?

What do you mean? 3400G exists.

 

[R0H1T]

I obviously meant laptops, not devices in general.

Yeah well there's rumors that Apple is launching Macbooks with Axx (5nm EUV, LPDDR5) next year, so there's that.

 

[notb]

Yeah well there's rumors that Apple is launching Macbooks with Axx (5nm EUV, LPDDR5) next year, so there's that.

Obviously, when ARM arrives in high power PCs, I'll have to change terminology to "x86 mobile devices". But that's the future and today is what we have today.

And the ARM-powered MacBooks will be in the Air lineup - probably with an x86 variant available as well.
MacBook Pro will remain on x86.

 

[Chrispy_]

the memory performance Is there, AMD refuses to use it. Look at the surface book 3. The AMD model has dual channel DDR4 2400, the intel model has dual channell 3733 LPDDR4X. That is nothing short of a MASSIVE jump in bandwidth. And it's not because MS is anti AMD, its because AMD only supports 2400MHz memory on their NEWEST APU!

I understand why there's a delay between latest μArch launch and APUs with AMD, but it sure as hell hurts their competitiveness in the mobile market.
7nm and Zen2's improved memory speeds are absolutely perfect for mobile APUs but AMD isn't woke enough to get it onto their mobile line as a priority.

I'm super glad that Zen2 and Navi happened on desktop, since I'm a desktop user first and foremost. However, the mass market is mobile-first. Laptops outsell desktops 3-1. AMD should be focusing on Mobile products first. That's where the money is, that's where they need to gain marketshare. It's mind-boggling that AMD are willing to concede this vastly larger market to Intel by using prior-gen architecture and process nodes. At times, their mobile line is two generations behind their current offerings; If I was a laptop manufacturer, I wouldn't be interested.

 

[INSTG8R]

I understand why there's a delay between latest μArch launch and APUs with AMD, but it sure as hell hurts their competitiveness in the mobile market.
7nm and Zen2's improved memory speeds are absolutely perfect for mobile APUs but AMD isn't woke enough to get it onto their mobile line as a priority.

I'm super glad that Zen2 and Navi happened on desktop, since I'm a desktop user first and foremost. However, the mass market is mobile-first. Laptops outsell desktops 3-1. AMD should be focusing on Mobile products first. That's where the money is, that's where they need to gain marketshare. It's mind-boggling that AMD are willing to concede this vastly larger market to Intel by using prior-gen architecture and process nodes. At times, their mobile line is two generations behind their current offerings; If I was a laptop manufacturer, I wouldn't be interested.

Doesn’t the new. MS Surface have an AMD APU as an option or an upcoming option? I totally agree if the6 get this APU right they can be taken seriously as an option and or alternative to Intel “same same” choices. I mean my last laptop had a Turion X2 paired with an 8800M dGPU(imagine when they used to get along)
But if I’m in my thinking of a possible MS Surface thats a great way to break back in and prove themselves.in low power portables so the stronger models should sell themselves.

 

[Chrispy_]

Doesn’t the new. MS Surface have an AMD APU as an option or an upcoming option? I totally agree if the6 get this APU right they can be taken seriously as an option and or alternative to Intel “same same” choices. I mean my last laptop had a Turion X2 paired with an 8800M dGPU(imagine when they used to get along)
But if I’m in my thinking of a possible MS Surface thats a great way to break back in and prove themselves.in low power portables so the stronger models should sell themselves.

The Surface Laptop 3 comes with either a 3580U or a 3780U. These are, for all intents and purposes identical to the 3500U and 3800U, based of April 2018's Zen+ launch. There's one more Vega CU in each of them, but neither model can use them because 15W and DDR4 2400 are the bottleneck here, not the CU count. It's Samsung's LP 12nm process and Zen+ architecture and DDR4 speeds with old, 100% Vega IGP architecture.

Given that the Surface Laptop 3 is only pre-ordering still (some lucky few might get theirs before Christmas), it's going to be something that's two years out of date by the time it's available for mainstream buyers. That's pretty awful for something that a lot of people will replace every 2-3 years.

If MS waited and used a Renoir APU with 7nm power efficiency gains and presumably AMD recommended memory speeds (DDR4-3200/CL14 or DDR4-3600/CL16) then the AMD would likely spank the Intel in most, if not all, of the tests - and at a lower price.

As it is, it's AMD's 2-year-old architecture on a 2-year-old manufacturing process going up against Intel's single most significant and advanced offering in the last 5 years, in a market segment where architectural efficiency and process node are quite literally the deciding factor. Power efficiency is what determines the size/weight/thickness/performance/temperature/noise/battery life of a laptop and despite their 'launch date' the Ryzen 3580U and 3780U are old and tired; It's a miracle that there's even any comparison between them and the Intel Ice-Lake options at all!

Edit:
Now I think about it, even the "Vega IGP" in the AMD APUs is likely just a rebrand. GCN5 differed from GCN4 as used by Polaris in the 570/580/590 almost entirely by HBM, HBCC and rapid-packed maths - primarily for gaining traction in the server/compute market. Guess what the Vega IGPs are missing; Yep - all of things I just mentioned. They're Polaris cores, in all but name - two generations behind Navi/RDNA already.