New Upcoming ATI/AMD GPU's Thread: Leaks, Hopes & Aftermarket GPU's

7nm? if Yes then ATI is on Track to release Vega2 for Gaming
Now it can be sooner than later

 

So far the confirmed order is 7nm Vega20 (The workstation card for deep learning and such.) is going to be late 2018 and then Ryzen on 7nm with probably Zen2 (CPU following Zen+ like the 2700) should be 2019
Late 2019 is where I'm expecting Navi to land but there's not too much info on it yet, it's using the primitive shader stage going by code commits for Vulkan to Linux kernels AMD has submitted although I expect it will revamped to either rely on extensions or be per-profile enabled since it didn't pan out with a global solution for Vega though there could be more to it than that and Vega also lost some other features in the end.

If Navi is still GCN I'm hoping they've revamped the architecture extensively to remove some of inherent weaknesses and scaling problems but we'll see. GCN 1.1 and later seems fairly similar although with various tweaks and improvements particularly the last variants for Polaris and Vega though scaling could be better still so I'm interested in hearing what Navi will be bringing once AMD starts talking more about it.
(Enthusiast class wise If the 2080Ti is about 20% - 25% faster than the 1080Ti which is already 20 - 25% faster than Vega 64 if not more then that's not going to be easy to close in on for AMD though price/performance could be one alternative.)
(Or competing with the 2080 and landing between that and the 2080Ti perhaps but it's too early and just down to speculation, price is going to be a factor though as is NVIDIA's head start with the 2000 series launching shortly.)


EDIT: The HardOCP article recently is a pretty good example, Vega has advantages to tessellation, geometry, memory and clock speed including overall memory bandwidth when compared to Fury. It still only reaches 20 - 30% faster performance. This is good but overall on the low side of things especially for a GPU almost 2 years newer and with several key improvements. For a the full 5 almost 6 years since the 200 series and the 290X specifically landing on a 95% - 98% gain is pretty low too but it is improving and again the GCN architecture seems to have remained fairly consistent probably limiting things a bit as well.

EDIT: Not a bad card at all or anything and it excels at compute work but it's going to be a lot of work closing in on NVIDIA's lead now and them having GPU's out for months in advance will see people upgrade to those and then not really be in a hurry to upgrade again for a while though we'll see how it works out, ideally the mining craze won't see price inflation quite as bad as last time too but it could flare up again as well.



EDIT: Hmm might sound a bit negative but I'm trying to keep realistic expectations and seeing the overall state of things or how to say, would be nice if AMD could compete more with NVIDIA and ultimately us regular customers would also benefit from this but it's not easy to compete with their market share lead, larger teams and overall difference in revenue and total budget although AMD is shaping up but it's going to be tough.

But it's also going to be interesting to hear about both GPU vendors plans for 2019 and beyond, next after the 2000 series for NVIDIA and then Navi and beyond for AMD.

 

nvm

 

ATI/AMD upcoming Vega 20 GPU should use up to 400W of power, but pack 20 TFLOP's of compute performance !

AMD will be unleashing its refreshed Vega 20 GPU in the coming weeks, made on the industry-leading 7nm node and it'll feature some great improvements over the current Vega 10 GPU made on 14nm.
Vega 20 isn't being aimed at gamers and rather the HPC/automotive/AI markets with its huge 32GB of HBM2 and PCIe 4.0 standard.
AMD is looking to aim Vega 20 at NVIDIA's current flagship Tesla V100 solution, which was recently bumped up to 32GB of HBM2 at GTC 2018 earlier this year.

The new details on Vega 20 see it coming in at just 360mm2 compared to Vega 10 at 510mm2, a huge 70% reduction in total die size thanks to the fresh 7nm node.
This is where a fork in the road happens: AMD can choose power savings with the die size savings, or they can ramp clock speeds up. Vega 20 could benefit from around 55% power savings over Vega 10, or GPU clocks can be cranked by up to 40%, which would also see the now 4-stack HBM2 drawing more power.

This would drive the Vega 20 card with 32GB of HBM2 to 400W, or more.

AMD could use a moderate 20% gain in GPU clock speeds and reach 300-350W with a performance leap of 65% all while saving 30-40% on power consumption.
This would allow AMD's new Vega 20 GPU to beat NVIDIA's current Tesla V100 in compute at 300W with around 20.9 TFLOPs compared to the V100 with 15.7 TFLOPs.
To cap it off: Vega 20 is not a gamers GPU. Navi is the next step for AMD Radeon and gamers, and that won't be appearing until around Computex 2019 (11 months from now). We could expect AMD to unleash Vega 20 on 7nm to gamers as well in time



Read more:
https://www.tweaktown.com/news/62491/amd-vega-20-7nm-32gb-hbm2-up-tflops-400w-tdp/index.html

 

AMD Is Negotiating A 7th Amendment To The WSA (Wafer Supply Agreement)

Sources familiar with the matter have confirmed that AMD is in talks right now to renegotiate the Wafer Supply Agreement in light of GlobalFoundries decision to drop out of the race for 7nm.
The last negotiation gave AMD quite a lot of freedom (that came at a cost) and was known as the 6th amendment to the WSA.
This iteration would be known as the 7th amendment and would probably offer even more freedom for AMD.

The 7th amendment to the WSA (Wafer Supply Agreement) could give AMD even more freedom to tap TSMC and Samsung for fabrication

This is a very big deal. To understand how big of a deal this is, we first need to look at the 6th amendment to the WSA and the conditions it had set for AMD.
To those who don’t know, the WSA (or the Wafer Supply Agreement) is the document that dictates the design-manufacturer relationship between GlobalFoundries and AMD.
As some of you may know, GloFo used to be part of AMD before the latter spun it off into a separate entity.

This was done due to the fact that the foundry business was starting to hurt in terms to an increasingly cash-strapped AMD and it did not make sense at the time to retain such an R&D intensive branch. Ironically, even though the company was spun off,
AMD was still bound to purchase a ‘quota’ of wafers from the company every year and could not reach out to other pure-play foundries without heavy penalties.

Lisa Su entered the scene and renegotiated the terms of the WSA with GloFo to introduce the 6th amendment to the scene.
This gave AMD massive amounts of freedom to outsource its GPU and CPU ambitions to other pureplay foundries and is one of the reasons Zen has flourished.
But all of this came at a cost. It is here that I will start to get into specifics.

Assume AMD wants to fabricate a single wafer. It will have the following two choices:

Go with GloFo:

• Simply pay the cost of the lone wafer
• 1 Wafer deducted from annual wafer quota (This is good for AMD)

Go with TSMC/Samsung:

• Pay the cost of the lone wafer
• Pay a relatively small penalty to GloFo for each wafer produced at TSMC/Samsung. This will be included in the annual wafer quota calculations.
• If the annual wafer quota is not met, then pay a penalty on a fraction of the difference too.

So as you can see, fabricating a wafer at any other foundry than GloFo is basically double jeopardy for AMD.
Simply put, these were/are shackles that still bind the company to its former whole. All of that, however, might change with GlobalFoundries officially dropping out of the race and materially changing the nature of their relationship with AMD.
See, the understanding between both companies was that GlobalFoundries will continue to churn out new process nodes and AMD will keep buying – with the former no longer happening the latter can now be, well, amended.

I have been told that AMD is certain they will agree to the new terms that are “mutually beneficial” to both parties.
Keep in mind, my source did not actually state that the terms would benefit AMD, just that they would be ‘mutually beneficial’ and they are fairly confident AMD will agree to them.
This implies that we are probably looking at increased leniency (in the annual wafer quota) and even better terms for the company.

More cash freed up from the WSA means more money it can put into R&D or just about anywhere else.
Not only this, but its also a reminder that the WSA has a limited lifespan and it will start to wind up in 2020. This is the first sign of AMD transitioning to a truly fabless entity and probably the only silver lining to come out of GloFo dropping out of the nm race.
It goes without saying that there is also the very slim (but non-zero) chance that GloFo manages to increase AMD’s burden with the 7th amendment instead of alleviating it, but I would think that unlikely.

 

AMD is "working with both Sony and Microsoft on consoles" and their "secret sauce"

In a recent interview with CNBC's Jim Cramer, AMD's CEO, Dr Lisa Su, stated that the company is "working with both Sony and Microsoft on consoles" and their "secret sauce", seemingly confirming that they are working on both Sony's PS5 console and Microsoft's Xbox Project Scarlet.

Su's wording suggests that AMD's work with both console manufacturers is ongoing, making it likely that next-generation consoles from both Microsoft and Sony will be powered by AMD hardware.

The use of the term "secret sauce" also implies that both next-generation consoles will contain more bespoke hardware customisations than their predecessors, making their hardware differences larger than they have ever been before.
Comparing the relative power of next-generation consoles may be more than a simple TFLOPS-TFLOPS comparison.
Could this custom silicon/"Secret Sauce" be related to next-generation checkerboarding/upscaling techniques, or perhaps a move into hybrid ray-tracing like Nvidia's latest RTX GPU offerings?

Rumour has it that AMD's Navi graphics architecture was heavily influenced by Sony's planned PS5 console, with AMD's Zen CPU architecture being the hardware platform that is likely to drive both next-generation console offerings,
assuming that both Sony and Microsoft plan to continue using the x86 instruction set architecture.
AMD's Zen processors will offer a transformative performance improvement over the Jaguar cores used in today's consoles, as Jaguar was initially designed with low power consumption levels in mind, not cutting-edge performance.

AMD believes that there can be two winners within the graphics market, with AMD securing a tight hold onto the console market thanks to the combination of high-end CPU and graphics IP, as well as their ability to create bespoke silicon for the specific needs of their clients.
Right now AMD powers both the PS4/PS4 Pro, Xbox One/Xbox One X and Zhongshan Subor's custom Ryzen+Vega console/PC hybrid system, which is exclusive to the Chinese market.


AMD's Semi-Custom division appears to have a strong future in the console market, though at this time it remains unclear whether or not these console developments will ultimately benefit AMD in the PC market.

 

Secret-sauce... Render 2 viewports for VR without double rendering cost. Per-Pixel-Precision control for shaders (tiny objects on screen with shader code optimization). Subpixel-AA.

 

Asynchronous Radeon Rays (Yup RT Pass is made simultaneously with GFX)
A.I. Controlled Primitive Shader

December Big driver update (also packs ~100 bug fixes)

 

That requires extra hardware.

If the rumor about the two die PS5 is correct, things will be interesting. I can't see hbm in a console, and neither any double die stuff.
I expect a Ryzen with eight cores and something like a Vega 56, with 16GB of GDDR6, on a single APU, at 7nm.

 

PrMinisterGR yo you're back

 

HH, this guy has once in a lifetime opportunity to make exciting video... and he blows it.

To this computer. It is lovely. If it was available prior me building 2400G, I think I would sacrifice 600MHz on CPU for that iGPU. And going down form 16GB to 8GB RAM.
+ one never knows if there is way to OC it...

 

Next-generation Geometry for Vega is history

https://www.hardwareluxx.de/index.php/news/hardware/grafikkarten/47231-next-generation-geometry-fuer-vega-offenbar-geschichte.html

 

AMD Ryzen Embedded V1000 Processor to Power SMACH Z Handheld Gaming PC

Remember the SMACH Z? Me neither. It was a long attempt in the making, with multiple Kickstarter campaigns that finally resulted in an actual product coming to market later this year.
The SMACH Z is a handheld gaming PC with an x86 processor, that was originally supposed to use a 5" 720p screen powered off an AMD G-series SoC (System on a Chip).
It was then updated to use an
AMD Merlin Falcon (Excavator) R-Series RX-421BD clocked at 2.1 GHz SoC with an integrated Radeon R7 series GPU at 800 MHz to run a 6" 1080p display,
that was in turn embedded in a handheld gaming console, effectively making it a portable gaming PC.
Perhaps it was a good thing that the product took its time, since they have now decided to use updated internals for the final product- once more from AMD- in the form of the Ryzen Embedded V1605B SoC with AMD Radeon Vega 8 graphics.

eBOM reports that AMD has announced it will be showcasing the SMACH Z in their booth presence at the Tokyo Game Show which commences later this week. This marks yet another gaming platform that AMD will be powering, although it is less powerful than the other ones. The SMACH Z is available for pre-order in three different variants (Z, Z Pro, and Z Ultra) which have different starting configurations, but can be customized as with most pre-built PCs.

All variants can run off the Linux-based SMACH OS or Windows 10, retain the 6" 1080p touch screen, support up to 16 GB of dual channel DDR4 memory and 256 GB of solid state storage, and have an optional camera as well.
Prices begin at $699 and go up to $1199, with a 10% discount for pre-order for the console expected to be available December, 2018. For those interested, there are some preliminary benchmarks released by SMACH from earlier this year.

 

I do actually remember it. I Considered it and then came to conclusion that in that cooling capacity, there is no good gameplay.
Updated... good, enough... no. Decent gameplay at 720p, but still "portable gaming PC/console" is adequate name as much as "Portable Workstation notebook". Both really require one to be plugged in.
8GB version is apparently necessity even with low/medium detail textures as it is still eaten from system memory. And then for $899 one can get pretty portable 13,3'' laptop.
(One thing which can make it much more desirable is LTE modem. But that's for 7nm times.)

7nm is edge of good device like this. 7nm+ is where fun starts.
@pharma : I do not even know how many times I said that those are dead weight transistors in Vega 14nm. I wonder if they kept them in Vega 7nm or removed them.

 

That's a good question. I just don't know but not sure whether they can be removed. Likely will not impact performance gained at 7nm.

 

Philips Stunning New E Series Narrow-Border Monitors with FreeSync:

Today EPI, the brand license partner for Philips monitors and leading technology company, announces the availability of the Philips E Series. The E Series includes three monitors - a 27-inch (276E9QDSB), 24-inch (246E9QDSB) and 22-inch (226E9QDSB).
This family of professional monitors provides vivid color in a stylish, minimal design. With an IPS panel with wide viewing angles and incredibly narrow side bezels, the E series is especially suited for a multi-display or tiling setup for graphic design or professional applications. The E line is available at Newegg starting at $109.99.

The Philips E Line features stylish design with extraordinary picture performance. The display is a Full HD IPS panel with a wide range of colors that provides better viewing angles and high color accuracy for color-intensive tasks like photo editing or graphic design.
Users can expect to see more realistic colors and a true-to-life picture.These new Philips displays are designed with ultra-narrow borders which allow for minimal distraction and maximum viewing size when used together. The ultra-narrow border provides an experience similar to the feeling of using a single large display. Paired with the IPS panel, the E line is ideal not only for professional applications, but also for photos, movies, web browsing and gaming. For console gaming, the E line also uses AMD FreeSync technology for artifact-free performance and smooth video playback.

Philips has also packed the E line with a suite of Philips exclusive features including its LowBlue mode and Flicker-Free technology. Studies have shown that just as ultra-violet rays can cause eye damage, shortwave length blue light rays from LED displays can cause eye damage and affect vision over time. Philips LowBlue mode was developed for wellbeing and uses a smart software technology to reduce harmful blue light. Due to how brightness is controlled on LED screens, many users experience flicker on their screen which causes eye fatigue.
Philips Flicker-Free technology regulates brightness and reduces flicker for a more comfortable viewing experience.

The Philips E series is now available at Newegg for $179.99 for the 27-inch model, $139.99 for the 24-inch and $109.99 for the 22-inch.

 

Yup
399-499€ for 30% more than Vega 64
199-249€ for ~Vega 64 LC Performance

Thanks to this more ppl will have 1440p HDR capable affordable GPU and market can & will go on.
We need this more than 4k capable gaming, small steps are better.
People will love Gaming with FreeSync2 HDR at 1440p 100-144Hz for ~800€