Re: GPD Win 2 handheld gaming PC details and benchmarks released
Quote:
Originally Posted by
LordRetroGamer
Lets not forget HP which had a chance to use AMD's Mullins chips for a tablet product (such as the ElitePad series), but instead ended up in low end notebooks like HP Stream 14 series (which used A4 Micro-6400T).
I was thinking more of the old 40nm products: http://www.cpu-world.com/CPUs/Bobcat...es%20Z-03.html
I still use my 11" HP laptop with an E350 in it as it is the right size, those later HP laptops all seemed to be 15.6" panels so of no interest to me.
Re: GPD Win 2 handheld gaming PC details and benchmarks released
Quote:
Originally Posted by
LordRetroGamer
As far as I know, those AMD engineering samples were actually Stoney Ridge and Merlin Falcon which were still using AMD's older Excavator cores. The lowest TDP for Stoney Ridge is 6W (for AMD E2-9000e only) and for Merlin Falcon is 12W. These are still in production and no replacements have been forthcoming yet. ...
Then let me enlighten you :)
https://videocardz.com/69428/amd-sno...eat-horned-owl
That's slides from a Feb 2016 presentation about AMD's industrial/embedded roadmap. The last but one slide is for "Banded Kestrel", an APU with 2 Zen cores and 3 'GFX9' (i.e. Vega) CUs. As I said, been on the embedded roadmap a long time. It may not make it to consumer, but the silicon is planned.
https://forums.hexus.net/cpus/371038...ml#post3873746
where hexus' own leak-sniffer CAT-THE-FIFTH posted a link to this tweet:
https://twitter.com/VideoCardz/statu...08001536307200
which details a dual core APU ES with Vega 3 graphics.
Note that I never said we'd see it soon, or indeed that we'd see it at all in mainstream consumer devices. That's kind of irrelevant to this thread. A handheld gaming console is exactly the kind of device that you'd expect an embedded or semi-custom chip to end up in, which means it'd be a perfect target for a 2C/3CU part that's not on the consumer roadmap but is already in production as ES. AMD are already making that chip, so if someone wanted to put it in a device it *would* be that simple.
Re: GPD Win 2 handheld gaming PC details and benchmarks released
Quote:
Originally Posted by
scaryjim
There aren't any on the mobile roadmap for notebooks. As for the embedded ones, I saw the news long ago. Aren't those (both Horned Owl and Banded Kestrel) supposed to be launched somewhere in 2H 2017 (as the chart shows)? In few days time, already near the end of 2017. Like I've said earlier, "no replacements have been forthcoming yet."
Quote:
Originally Posted by
scaryjim
I would not call that a leak nor an engineering sample at all, especially when looking at the only source (that twitter post) with no indications of where it came from. Only valid leaks always happens at Geekbench, SiSoft Sandra and GFXBench databases. Other valid leaks happens in far east places like China and Taiwan, where major ODMs resides. Wake me up when that happens, otherwise this is a non-leak for me.
Quote:
Originally Posted by
scaryjim
Note that I never said we'd see it soon, or indeed that we'd see it at all in mainstream consumer devices. That's kind of irrelevant to this thread. A handheld gaming console is exactly the kind of device that you'd expect an embedded or semi-custom chip to end up in, which means it'd be a perfect target for a 2C/3CU part that's not on the consumer roadmap but is already in production as ES. AMD are already making that chip, so if someone wanted to put it in a device it *would* be that simple.
Well, Intel's Core m3-7Y30 is not considered an embedded part and neither was Intel's Atom x7-Z8700 used in the original GPD Win. Others like Vastking G800 and LinX Vision 8 gaming tablets also used Intel's mobile SoCs. In other words, does not need to be "embedded" type. Should be noted we hardly see any of AMD's embedded chips in devices like tablets and handhelds (that includes AMD's current generation embedded SoCs). As for chip manufacture, unlikely to start yet until the slow ramp gets sorted out at Global Foundries (which is perhaps why both Horned Owl and Banded Kestrel never made it in 2017 at all).
Re: GPD Win 2 handheld gaming PC details and benchmarks released
Quote:
Originally Posted by
LordRetroGamer
There aren't any on the mobile roadmap for notebooks. As for the embedded ones, I saw the news long ago. Aren't those (both Horned Owl and Banded Kestrel) supposed to be launched somewhere in 2H 2017 (as the chart shows)? In few days time, already near the end of 2017. Like I've said earlier, "no replacements have been forthcoming yet." ...
You're not actually listening to what I'm saying, are you.
The whole question at the start of this was whether it would be possible to use a Ryzen APU in a 4.5W TDP. That's the conversation from posts 2, 3, 4 and 7. Not whether you could stick an existing Ryzen APU in there, but just generally whether a low-end Ryzen was possible and would be a better fit.
I think there's enough beef behind the rumours I've linked above (videocardz are actually pretty damn reliable when it comes to this kind of thing - they've leaked plenty of of ES details which were confirmed later) to believe that AMD already have working silicon of a 2C/3CU APU. And we can deduce - given that it has half the CPU cores and less than 1/3 the GPU cores - that it will fit in TDPs that are half or less of the launched Raven Ridge APUs.
That means that it would, in fact, be simple for AMD to release a suitable Ryzen APU in a 4.5W TDP. Just because something would be simple doesn't mean it happens. Ryzen 3 Mobile hasn't released yet, although that would be simple. It would've been simple for all the Ryzen CPUs to launch at once, since they're all based on the same silicon, but again AMD chose to release them in a staggered fashion. And I can't actually remember Threadripper being on the consumer roadmap ahead of launch, despite being a real product that launched.
The question isn't "Is there a launched AMD product that matches this description?". It's "Would it be possible, or even simple, for AMD to launch a product that meets this description, if they wanted to?". And given what we know, I'd say the answer to that second question is a resounding yes.
Re: GPD Win 2 handheld gaming PC details and benchmarks released
Quote:
Originally Posted by
scaryjim
You're not actually listening to what I'm saying, are you.
The whole question at the start of this was whether it would be possible to use a Ryzen APU in a 4.5W TDP. That's the conversation from posts 2, 3, 4 and 7. Not whether you could stick an existing Ryzen APU in there, but just generally whether a low-end Ryzen was possible and would be a better fit.
You have to look at the lowest TDP of AMD's current Raven Ridge APUs (which is nowhere near as low as 4.5W). Any dual core AMD Raven Ridge mobile chip would be harvested from AMD's current quad core Raven Ridge chips. Similar pattern as AMD's Bristol Ridge and Stoney Ridge. Basically AMD's dual core Stoney Ridge is the same chip as AMD's quad core Bristol Ridge, only half the number of cores enabled and less graphical cores to reduce power.
Quote:
Originally Posted by
scaryjim
I think there's enough beef behind the rumours I've linked above (videocardz are actually pretty damn reliable when it comes to this kind of thing - they've leaked plenty of of ES details which were confirmed later) to believe that AMD already have working silicon of a 2C/3CU APU. And we can deduce - given that it has half the CPU cores and less than 1/3 the GPU cores - that it will fit in TDPs that are half or less of the launched Raven Ridge APUs.
Again, you have not shown any real leaks (no "beef" there). Here is an example of a real leak at Passmark database https://www.passmark.com/baselines/V...id=94512522319 Real leak means there is real silicon being tested, not some model number on a random Twitter post not by any Taiwanese nor Chinese.
Quote:
Originally Posted by
scaryjim
That means that it would, in fact, be simple for AMD to release a suitable Ryzen APU in a 4.5W TDP. Just because something would be simple doesn't mean it happens. Ryzen 3 Mobile hasn't released yet, although that would be simple. It would've been simple for all the Ryzen CPUs to launch at once, since they're all based on the same silicon, but again AMD chose to release them in a staggered fashion. And I can't actually remember Threadripper being on the consumer roadmap ahead of launch, despite being a real product that launched.
The question isn't "Is there a launched AMD product that matches this description?". It's "Would it be possible, or even simple, for AMD to launch a product that meets this description, if they wanted to?". And given what we know, I'd say the answer to that second question is a resounding yes.
If it were that simple (as I have mentioned earlier) then AMD would have released them simultaneously with Raven Ridge or heard about them in AMD's future official roadmaps (around the same time as mobile Ryzen launch). Also AMD's Threadripper is basically the same as Epyc (since the existing MCM package contains the same 4 dies), thus easier to release as a HEDT product line. Thats is similar to Intel's HEDT line which were basically Intel's Xeon chips rebranded (which is why Intel's Xeons can be used in the same HEDT motherboard).
Re: GPD Win 2 handheld gaming PC details and benchmarks released
Quote:
Originally Posted by
LordRetroGamer
You have to look at the lowest TDP of AMD's current Raven Ridge APUs (which is nowhere near as low as 4.5W). Any dual core AMD Raven Ridge mobile chip would be harvested from AMD's current quad core Raven Ridge chips. Similar pattern as AMD's Bristol Ridge and Stoney Ridge. Basically AMD's dual core Stoney Ridge is the same chip as AMD's quad core Bristol Ridge, only half the number of cores enabled and less graphical cores to reduce power. ...
Huh. I thought you'd let this one go, but apparently not.
Your entire premise is wrong, on two factors.
First up, I thought Stoney Ridge and its embedded counter-parts were separate silicon, but even if they're not, they drop as low as 6W TDP for the embedded versions (Prairie Falcon). The quad-core versions of those chips only drop as low as 12W (Merlin Falcon). Whether it's a cut down version or a separate die, the 2 core versions can work at as low as half the TDP of the 4 core versions. 4 core Raven Ridge is configurable down to a 9W TDP, so whether a 2C APU was a salvaged part or a separate die, it's reasonable to assume it could have half the TDP - which gets us down to the 4.5W TDP we're discussing. Whether it's a separate die or not is, I admit, a little moot.
Secondly, your claim that if AMD could make a 4.5W APU they'd have released it already is - to be blunt - utter nonsense. They didn't even release the Ryzen 3 mobile APUs at the same time as the 5 and 7, despite them being cut straight from the same die, so I don't see how you can justify that claim.
Bottom line - it's reasonable to think AMD could produce a Zen/Vega APU in a 4.5W TDP, and it's reasonable to think they might not have released it even if they could. It apparently doesn't even need a separate 2C die to get there - perhaps that was just wishful thinking on my part. So really, I don't get what you're arguing about?
Re: GPD Win 2 handheld gaming PC details and benchmarks released
Quote:
Originally Posted by
scaryjim
It apparently doesn't even need a separate 2C die to get there
In a world where even smart TVs are advertised as "quad core" let alone £100 Atom based tablets, dual core would seem an increasingly hard sell.
Re: GPD Win 2 handheld gaming PC details and benchmarks released
Quote:
Originally Posted by
DanceswithUnix
In a world where even smart TVs are advertised as "quad core" let alone £100 Atom based tablets, dual core would seem an increasingly hard sell.
Current top-end intel competition is a 3.6/1.3 GHz 2c4t part with an HD 615 for graphics - a pretty low bar for AMD to match, and it goes it some very pricey laptops (+£1k for the acer swift 7)
Re: GPD Win 2 handheld gaming PC details and benchmarks released
Re: GPD Win 2 handheld gaming PC details and benchmarks released
Has anyone bought one of these, yet?