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

The 2500U officially supports up to 2400MHz. Anything faster is a waste of money because it won't run at the higher speed.

 

According to AMD's official notes, you are correct that 2500u supported RAM speed is 2400MhZ...

https://www.amd.com/en/products/apu/amd-ryzen-5-2500u

Furthermore, the actual availability (not to mention price) of higher speed DDR4 SODIMM modules is a problem (especially here in UK - you'll be hard pressed finding 2600MhZ SODIMM, or 3000MHZ for that matter... most shops sell only 2400MhZ... and only Amazon might have 2600 and 3000 MhZ DDR4 for laptops - but again, the prices on those can be quite high and not worth the investment if it turns out they don't actually work - I was thinking on getting higher speed RAM for my GL702ZC eventually when prices of RAM go down, but alas, no one had a chance to test it yet... though another user who has that unit running Linux says that he slipstreamed Ryzen+ support into its BIOS, but he couldn't say whether higher speed RAM support was included).

Still, I find it odd that AMD would limit 2500u to mere 2400MhZ (considering Infinity Fabric's and iGP dependence on memory bandwidth).

 

I don't think that announcement of the RTX 20 series changes much as of yet. Until we know the actual performance of those GPU's, we won't know for certain how they might affect people's buying habits.
But its safe to say that they are not 7nm GPU's (at least not according to this FORBES article: https://www.forbes.com/sites/jasone...-should-jump-off-the-hype-train/#1c81ac623f8e) ... but rather 12nm (again high perf. TSMC process) with added raytracing capabilities (which is interesting considering that AMD already showcased an open-source version of this a while ago). Nvidia right now needs to convince devs to code for their raytracing (which in all probability is not open-source - and we already saw what that meant with devs adopting CUDA as opposed to OpenCL for example).

AMD might be in a pickle until they can get 7nm GPU's out the door for consumers. Vega and Polaris may have outsold previous gpu generations by a factor of 10, but we have yet to see wide-scale industry support for open-source features that Polaris and Vega have (in fact, most industry software is going for CUDA optimizations, whereas OpenCL and Vulkan optimizations are left on the sidelines or barely in the process of being used).

If AMD made a refresh of existing GPU's on 12nm LPP without raising performance (and specifically optimized power consumption further), it would certainly increase efficiency and serve as a stopgap measure until they get their 7nm consumer GPUs out, but that would likely incur additional costs for AMD that they might not see worthy of justification at this time (considering they are close with 7nm release and no plans were made to refresh GPU's on 12nm) because the 12nmLPP process is just as limited as the 14nm process was. If they try raising clocks, they probably won't gain anywhere near enough performance vs Nvidia on 12nm because the TSMC 12nm process will allow higher clocks on far lower voltages (whereas AMD is more likely to reach the exponential voltage threshold with even a small clock increase... much like what happened with Ryzen+ ).

As for the color change... seems a bit pointless really.
The GPU performance comparison is interesting, but too much focus on gaming. Why hadn't they showcased the GPU's computational capabilities in rendering, use of professional software etc.?
I realize the margin in this reduced with Maxwell and Pascal's introduction, but still, it seems a bit... incomplete.

 

Out of curiosity, does NewEgg or Fry's have a UK presence or ship to the UK? I live in the rural, Midwest USA. Nearest Fry's is about 200 mi (~320 km) away, so everything I do is pretty much shipped. Now it's been a while since I built the Mighty Mini, but there was no problem getting high speed DDR4 SODIMMs for 2666MHz and higher... Again this was about 10 months ago, so I cannot comment on availability today.

 

Fry's does not appear to have UK presence, but NewEgg apparently does (however, everything seems to be shipped from USA - and of course, the prices shown in the selection screen with other items don't include VAT... so the final prices are about 20% higher.
And I just checked about Customs duties... it seems the final price does not reflect those (they are added/calculated just before final checkout).

So for example, if I were to use NewEgg, 32GB of 3200MhZ SODIMM DDR4 (2x16GB CL18) would cost £340 with express delivery... however, that doesn't include customs duties... which would likely increase the price by another 20% (or more).

Not exactly useful.

Amazon UK is now apparently offering that item too... and costs £404 in total.
Comes out to pretty much similar actually (but its likely the customs duties might slap higher import duties and all-together would cost more than buying the thing from Amazon UK itself).

 

AMD MxGPU Technology - The World’s First Hardware Virtualized Graphics Solution
AMD
Published on Aug 26, 2018
Pure. Datacenter. Graphics.
MxGPU is the purest form of GPU virtualization. A built-in hardware engine does the heavy lifting for VM scheduling and management. MxGPU utilizes the underlying SR-IOV protocol to create a robust, data-secure VDI solution that is well suited for a wide range of hypervisors and applications. With no additional licensing fees to virtualize the GPU, the solution reduces the total cost of ownership.
Discover more: http://http://www.amd.com/mxgpu

Introducing the AMD Radeon™ Pro V340 Graphics
AMD
Published on Aug 26, 2018
Pure. Datacenter. Graphics.
The AMD Radeon™ Pro V340 datacenter graphics card delivers an impressively smooth GPU experience from the cloud to virtually any device, anywhere. With MxGPU technology at its core, this hardware-based virtualized graphics solution provides high levels of predictable performance, enhanced security and support for up to 32 VMs per card.
IT managers will also like the fact that this MxGPU solution is easy to set up and manage. Unlike other GPUs, the V340 delivers GPU virtualization to end users without end user licenses, providing enterprises with a lower cost per user.
Discover more: http://www.amd.com/V340Builtforbuilders

 

Real-Time Ray Tracing with Radeon ProRender
Posted on March 20, 2018 by Dan Flatt
https://gpuopen.com/announcing-real-time-ray-tracing

AMD also announces support for Real-Time Ray Tracing
by Hilbert Hagedoorn on: 03/20/2018 07:02 PM
https://www.guru3d.com/news-story/amd-also-announces-support-for-real-time-ray-tracing.html

"Nvidia and Microsoft yesterday made an announcement on new real-time raytracing support. AMD was a bit quiet about that, today the company has made an announcement.

AMD is announcing Radeon ProRender support for real-time GPU acceleration of ray tracing techniques mixed with traditional rasterization-based rendering.

This new process fuses the speed of rasterization with the physically-based realism that users of Radeon ProRender expect for their workflows.

Radeon ProRender engine, used in Autodesk, Blender and such will get support for real-time ray tracing. This will possibly be mixed with traditional rasterization.

Radeon GPU Profiler gets support for ray tracing in version 1.2. With this tool, the performance of different GPU's can be accurately analyzed.

--- AMD ---

Announcing Real-Time Ray Tracing
The level of visual detail required of CAD models for the automotive industry or the most advanced film VFX requires a level of visual accuracy that can only be achieved with advanced ray tracing techniques.

Ray tracing maps the path of light rays from each pixel of the rendered object back to the camera, realistically simulating light interactions like shadows, reflections and transparency.

Ray tracing is very processor intensive resulting in long render times and low frame rates. Alternatively, rasterization (or scanline) rendering uses algorithms to approximate the appearance of 3D objects, surfaces and textures.

Rasterization can be accelerated by GPU hardware resulting in real-time rendering and high frame rates making it an ideal fit for gaming. Of course the trade-off with rasterization is that the approximated renders are nowhere close to the photorealism of their ray traced counterparts.

These trade-offs between rasterization and ray tracing traditionally impose corresponding constraints on 3D designers who would have to choose between realistic illumination and real-time speeds depending on the priorities of their project.

The differences between rasterization and ray tracing at a computational level have made the combination of these processes a desirable but difficult problem discussed more often in academic papers than practical application.

AMD is announcing Radeon™ ProRender support for real-time GPU acceleration of ray tracing techniques mixed with traditional rasterization based rendering.

This new process fuses the speed of rasterization with the physically-based realism that users of Radeon ProRender expect for their workflows.

Spoiler

At a high level, the process achieves these results by using rasterization to draw basic structures and surfaces before a ray tracing process is used to compute advanced light-effects like reflections, shadows and transparency.

The flexibility of the process allows users to decide when these advanced light effects are actually necessary and add noticeable new dimensions of realism to their renders.

So how could real-time ray tracing actually change the way 3D professionals work?

Professional users of this new process will be able to achieve far faster render times for their projects saving valuable time and system resources for 3D projects of any scale.

Game developers will also be able to achieve a greater degree of photorealism without sacrificing the high frame rates required for smooth real-time onscreen animation.

(We’ll talk more about how the new rendering technique can enhance gaming graphics in future blog posts.)

Like many of the other open source tools available from GPUOpen, real-time ray tracing will give users more control over the development process unlocking even more performance from your GPU."

Real-Time Ray Tracing with Radeon ProRender
AMD
Published on Mar 19, 2018
AMD is announcing Radeon™ ProRender support for real-time GPU acceleration of ray tracing techniques mixed with traditional rasterization based rendering. This new process fuses the speed of rasterization with the physically-based realism that users of Radeon™ ProRender expect for their workflows.
To learn more visit: gpuopen.com and subscribe to AMD Developer News at https://experience.amd.com/developer-...

Radeon™ ProRender: Real Time Ray Tracing Explained
AMD
Published on Mar 28, 2018
AMD is announcing Radeon™ ProRender support for real-time GPU acceleration of ray tracing techniques mixed with traditional rasterization based rendering. In this video, learn about the difference between rasterization and ray tracing and how the new process combines their best features.
To learn more visit: gpuopen.com and subscribe to AMD Developer News at https://experience.amd.com/developer-...

GPUOpen: Radeon Rays
AMD
Published on Jun 21, 2016
AMD FireRays is specialized software that accelerates and enhances ray tracing operations by processing the data on the GPU, asynchronously. Open-source and part of the GPUOpen initiative, FireRays enables developers to create and render photorealistic images and content that will make tomorrow’s ultra-immersive virtual reality experiences possible.
For more please visit: http://www.gpuopen.com

AMD Jumps on Ray Tracing! DX12 vs Vulkan: Lines Being Drawn in the Sand
The Good Old Gamer
Published on Mar 22, 2018
At GDC Nvidia revealed their RTX technology in conjunction with Microsoft's DXR (DirectX 12 extension.) AMD has followed up with Radeon Rays using Vulkan. It seems the battle of the low level APIs is just heating up.

Spoiler: Sources

Nvidia RTX Video: https://youtu.be/cPzIbzcrABo

Sources: TechPowerup:
https://www.techpowerup.com/242615/am...

https://www.techpowerup.com/242603/4a...

https://www.techpowerup.com/242504/mi...

Futuremark:
https://www.futuremark.com/pressrelea...

Other Tech Info:
Nvidia Tensor Cores: https://devblogs.nvidia.com/programmi...

AMD's Rapid Packed Math: https://www.anandtech.com/show/11717/...

Welcome to the Radeon ProRender Forum
https://community.amd.com/community/prorender

 

Strange Brigade: Extreme 52 GPU Benchmark, Vega 56 / 64 vs. New & Old Graphics Cards
Hardware Unboxed
Published on Aug 27, 2018

 

AMD accelerates 7nm process adoption
https://www.hexus.net/tech/news/industry/121508-amd-accelerates-7nm-process-adoption/

"AMD had originally planned to release the first 7nm Vega GPU in 2019 but, thanks to "immense focus" has been able to move the date forward to later this year. Thus we will see the next-generation Radeon Vega Instinct GPU, demonstrated at Computex with 32GB HBM2, launched before the year is out."

Isn't this older news though?
We already knew that 7nm Vega Instinct will be demoed late this year... actual consumer products for 7nm will still come in 2019.

 

One of the interesting things he mentions during the video is that most AMD sponsored titles seem to run well on both AMD and Nvidia hardware (Whereas NV optimized games seem to run great on NV but otherwise cripple AMD in turn).

This leads me to think that if games are 'optimized for AMD' they seem to be just better optimized for PC and GPU's across the board.
Both companies integrate similar functionality after all, but NV has a lot of proprietary garb.... ahem... software, whereas AMD uses open source approach which is available to both companies.

I think we witnessed something similar before with AMD's TressFX did we not?
That ran really good on both AMD and NV gpu's and didn't tax the hardware too much, but when NV insisted on using Gameworks instead, it crippled performance on both GPU's...and ran better on NV (more or less - depending on how complex you set the settings for GameWorks - because I think they modified the game to set those setting to ridiculously high levels on AMD gpu's - could be wrong though as its been a while).

 

Yeah, that seems to be referring to some kind of workstation card, so I'm guessing that means consumer gaming GPUs will be out sometime in (early) 2019.

 

Vega Radeon Instinct 7nm is an AI oriented GPU...

I was under the impression the consumer gaming GPU's (based on Navi 7nm) will be out by mid 2019.
Unless this new schedule accelerated things.

Zen 2 is likely to ship in the first quarter of 2019 though... not sure about GPU's.

 

Yeah, AMD is leaking like a sieve. Yet another of their major employes does not seem to work with Lisa Su.

Exclusive: AMD Senior Vice President Jim Anderson Resigns – Moving On To Become CEO-Wccftech.com
I have just heard something pretty massive: Jim Anderson, AMD’s Senior Vice President and General Manager has resigned and AMD should be announcing this fact very very soon. The reasons surrounding the sudden departure are unknown but I have heard it had something to do with “tensions” related to Lisa Su.


AMD loses another key executive, Jim Anderson, as it shifts manufacturing to TSMC-Pcworld.com
AMD said Monday that it lost a key executive in its processor group, Jim Anderson, and that it would migrate its leading-edge manufacturing to TSMC.

------------------------------------------------

AMD Is All In On 7nm CPUs For Strategic Advantage On Intel’s 10nm Stumble-Hothardware.com
"In the CPU space, AMD is firing on all cylinders currently, having reasserted itself into the performance sector with its mainstream Ryzen processors and high-end desktop (HEDT) Threadripper chips. Both are now in their second generation.

While it may sound like a gamble for AMD to put all of its eggs into a 7nm basket, it might be more out of necessity than choice. Intel contents that its 10nm process is more advanced than the competition, as it is packing more transistors and is doing a better job with scaling. In fact, Intel has even stated a desire to introduce a new metric to more accurately reflect where chip design technologies are really at...

 

AMD Moves All 7nm CPU, GPU Production to TSMC
https://www.extremetech.com/computing/276169-amd-moves-all-7nm-cpu-gpu-production-to-tsmc

It looks like AMD made the right call by making an agreement with TSMC... otherwise, Glofo would have seriously impacted their plans for Zen 2 and Vega/Navi if TSMC was not in play.

What do you guys make of this?

 

GlobalFoundries Puts its 7 nm Program on Hold Indefinitely
Techpowerup.com Tuesday, August 28th 2018 00:42

GLOBALFOUNDRIES today announced an important step in its transformation, continuing the trajectory launched with the appointment of Tom Caulfield as CEO earlier this year. In line with the strategic direction Caulfield has articulated, GF is reshaping its technology portfolio to intensify its focus on delivering truly differentiated offerings for clients in high-growth markets.

"Lifting the burden of investing at the leading edge will allow GF to make more targeted investments in technologies that really matter to the majority of chip designers in fast-growing markets such as RF, IoT, 5G, industrial and automotive," said Samuel Wang, research vice president at Gartner. "While the leading edge gets most of the headlines, fewer customers can afford the transition to 7nm and finer geometries. 14nm and above technologies will continue to be the important demand driver for the foundry business for many years to come. There is significant room for innovation on these nodes to fuel the next wave of technology."

 

Wow, that sounds like Intel (or?) could have bought a "significant" portion of production for 14nm and by that gesture bought themselves space in the queue for future 7nm production. Too late now? Or is GF drawing the line that invites Intel in?

Intel, by offering to help fund the 7nm gen 2 and 7nm gen 3 work, perhaps with collaboration from their successful portions of 10nm fab, could be a good partnership with GF.

 

If Intel continue on 14nm++++ or whatever they will call the next and same time beat AMD clock for clock, have higher clocks speed + overclock a lot higher exactly the same as today... Just let them continue with whatever they want

 

Yeah, that's the same poor thinking and planning that got Intel into the jam they are in now.

Hopefully there is enough TSMC 7nm production for AMD - and everyone else that was scheduled for GF 7nm production - to release on time, or early as it seems AMD was(is?) on track to do sooner than originally planned.

GlobalFoundries Stops All 7nm Development: Opts To Focus on Specialized Processeses
https://www.reddit.com/r/Amd/comments/9asamn/globalfoundries_stops_all_7nm_development_opts_to/

"...Keep in mind that AMD’s first 7 nm product was designed for TSMC’s CLN7FF from the beginning, so the company did not bet on GF’s 7LP in late 2018 anyway, and no rush with the manufacturing technology was needed for GF’s key customer."

"...Meanwhile, having spent well over $20 billion on GlobalFoundries over the last 10 years, Mubadala, the owner of the company, is not inclined to lose more money or invest tens of billions in the hopes of becoming profitable one day. The investor wants GlobalFoundries to stop bleeding and start generating profits."

AMD accelerates 7nm process adoption
https://www.reddit.com/r/Amd/comments/9ao94r/amd_accelerates_7nm_process_adoption/

"... We knew 7nm would be a big challenge, so we made the bet, we shifted our resources onto the new node,"Papermaster said in an exclusive interview with CRN. "We didn't just dip our toe in the water. We went all in."

AMD had originally planned to release the first 7nm Vega GPU in 2019 but, thanks to "immense focus" has been able to move the date forward to later this year. Thus we will see the next-generation Radeon Vega Instinct GPU, demonstrated at Computex with 32GB HBM2, launched before the year is out."

 

I hope there is enough 7nm Capacity for both AMD processors and graphics. Or maybe I talk too much... Of course it's enough for botht.... Less sales due Intel and Nvidia will help them out of the mess. One of those two scenarios will probably fit Yeah, I'm clever regarding backup, LOOL

 

That EPYC moment when AMD stock surpasses the $25 mark, the first time since the Athlon 64 years circa 2006...
https://www.reddit.com/r/Amd/comments/9ap2xc/that_epyc_moment_when_amd_stock_surpasses_the_25/


$27
https://www.reddit.com/r/AMD_Stock/comments/9ar133/27/

 

AMD Slashes Prices of First-Gen Ryzen Threadripper CPUs: Starting at $300
by Anton Shilov on August 27, 2018 10:00 AM EST
https://www.anandtech.com/show/13273/amd-slashes-prices-of-first-gen-ryzen-threadripper-cpus

"...As a result of the price cut, the eight-core Ryzen Threadripper 1900X will cost around $300, the twelve-core Ryzen Threadripper 1920X will be priced at circa $400, whereas the former flagship Ryzen Threadripper 1950X will be available for $799.

A quick check with Amazon indicates that retail prices of the said CPUs are very close to their SRPs, or even below them.

For example, the octa-core 1900X is now available for $308, whereas the 16-core 1950X can be purchased for $720.

The 1900X has more memory channels and PCIe lanes than any regular Ryzen processor, so the chip makes sense for those who need a lot of DRAM bandwidth and/or plans to use multiple high-end SSDs and/or graphics cards..."

"...It is unknown how many Ryzen Threadripper 1000-series CPUs are still in stocks of retailers and AMD as well as how long these stocks will last.

AMD’s Ryzen Threadripper 2950X is set to be released on August 31, whereas the model 2920X is due in October..."

 

AMD loses another key executive, Jim Anderson, as it shifts manufacturing to TSMC
Anderson's departure is a blow to a company that has already lost two key executives, just as it appears that AMD has regained its footing.
By Mark Hachman, Senior Editor, PCWorld | AUG 27, 2018 1:34 PM PT
https://www.pcworld.com/article/330...erson-as-it-shifts-manufacturing-to-tsmc.html

"As AMD moves into the next phases of its Threadripper and Vega chip architectures, it has made two key business changes as well: Jim Anderson, who oversaw AMD’s Computing and Graphics Group, has left the company. In addition, AMD has shifted its leading-edge manufacturing to TSMC, and away from GlobalFoundries.

Anderson, who appeared on PCWorld’s The Full Nerd podcast just two weeks ago to explain the ins and outs of AMD’s new 2nd Gen Threadripper chip, has left to become chief executive of Lattice Semiconductor, an FPGA maker. Anderson will be paid several million dollars in stock and incentives, according to a Lattice statement. An AMD spokesman said that it was always Anderson's career ambition to eventually become a chief executive..."

Lattice Semiconductor Appoints Jim Anderson as CEO
Published: Aug 27, 2018 4:00 p.m. ET
Industry Veteran with Proven Track Record to Focus on Accelerating the Growth of Lattice’s FPGAs and Smart Connectivity Solutions
https://www.marketwatch.com/press-r...uctor-appoints-jim-anderson-as-ceo-2018-08-27

"Lattice Semiconductor Corporation LSCC, -0.79% a leading provider of customizable smart connectivity solutions, announced the appointment of Jim Anderson as the Company’s President and Chief Executive Officer, and to the Company’s Board of Directors, effective September 4, 2018.

Mr. Anderson brings broad technology industry experience and a proven track record of leading and transforming businesses to drive sustained growth and profitability.

Mr. Anderson joins Lattice from Advanced Micro Devices (AMD) where he served as the General Manager and Senior Vice President of the Computing and Graphics Business Group..."

 

We haven't heard from this guy in a while, remember his early "unbelievable" OC'ing Ryzen 2700X to 4.3ghz+? It turned out he was for real, and now he is moving to Phase Change cooling and reaching for 5ghz.

See how noisy Phase Change is when running...

Extreme Overclocking: 2700X and the race for 5GHz (-93°C) in Cinebench R15
Hardware Numb3rs
Published on Aug 28, 2018
Are curious about how far a Ryzen 2700X can go ? This is the first of a series of video about Extreme Overclocking, using a Cascade Phase Change Cooling I can go down to -96°C and push the frequency where water cooling can't!

Spoiler: Details...

I tried to break my personal record in Cinebench R15 and to run the benchmarks at 5GHz or above.

The component used:

AMD Ryzen 2700X
ASUS Crosshair VI Hero X370
Gskill Flare X 3200MHz C14
ASUS Strix Vega 64
Corsair RM1000i

 

I genuinely don't think Intel will be able to that.
The 14nm process (Even the high performing one Intel uses) has its limits. Intel is more likely to release overclocked versions of their CPU's, but they will probably also run much hotter and consume more power than AMD's Zen 2 on same clocks and 7nm (that's assuming Intel manages to squeeze out same clock speeds like AMD will).
AMD is expecting over 5GhZ frequencies... with unknown IPC gains.
40% performance increase from the process improvement over Zen 1 would easily put Zen 2 to 5.5-5.7GhZ on base for most lower core CPU's and progressively less as the core count goes up (8 cores would likely reach 5-5.2 GhZ on base, with 16 c parts probably on the 4.8-5GhZ base).
That's of course if the 40% perf. jump is from frequencies only and IPC separate - but still, 5GhZ or over that on base is equally good across the board.

 

knowing math as well as I do the numbers can mean many different things. It could just be a 40% improvement per watt where as in the end the chip can only handle in all 75% of the wattage the 12nm's could. Without real silicon we have to wait and see. As far as Intel, who would have thought they would have gotten where they are with 14nm to begin with?

I still can not trust AMD without actual silicon since the initial Vega claims and subsequent let down.

 

In regards to AMD overhyping...

They were fairly consistent with ryzen and delivery as promised.
Besides the 7mm process is designed for high performance and efficiency, whereas 14 and 12nm from glofo was designed for low clocks and mobile parts (so tjia isn't a direct comparison and they were rather specific in saying 40% performance increase on 7nm - we just don't know if its from frequencies alone or a combination or frequencies and IPC).

As for the gpus... Correct me if I'm wrong, but I don't think vega was ever touted by amd that it would go up against 1080ti for example.
Also, AMD overpowered it with really high amount of stream processors (much higher than what NV put on Pascal in terms of cuda cores). Then there is overvolting and the issue of using a manuf. Process suited for mobile parts and low frequencies.

Mind you, vega did achieve much higher clocks than Fury, but it could have done that with lower amount of stream Processors (at which point it would likely consume less power and gaming performance would in all probability not suffer as compute doesn't translate readily to games).

I do recall that AMD might have said vega would be a great overclocker... But evidently that never happened.

In any case, they were inconsistent with the gpu claim(s?)…. Not cpu ones.

But I agree that we should really wait for the actual silicon before we can specifically say what's what in regards to performance of zen2 parts.

Still, right now, the info we have on 7nm TSMC points to high performance and efficiency.

 

Insane Deal: 12-core Threadripper For Only $420
Hardware Unboxed
Published on Aug 29, 2018

 

Your concern is warranted about 7nm capacity, because it seems that Apple gobbled up 75% -80% of the total capacity according to this:
https://www.eetasia.com/news/article/18082404-tsmc-to-continue-as-apple-sole-supplier

I'm not sure how much 7nm capacity is allocated to AMD.

 

I can see they talk about “putting all the eggs in one basket” also for Apple. It’s a risk but “they haven’t other choices” as stated in the article about AMD I posted above. Aka no backup plan if something unexpected happens. This maybe doesn’t look good for next gen AMD graphics. Their processors have most likely first priority.

 

Well, GLOFO did end up suddenly discontinuing their 7nm production...
Had AMD stayed with them and not made a deal with TSMC, they'd end up with huge issues right now (or at the very least, be forced to use 12nmLPP for Zen 2 and gPu's).

I say that's not such a bad thing. Glofo was a problem for AMD for a while now because they were using unsuitable manuf. process for desktop hardware.
At the very least, TSMC seems to be more reliable in this regard, and they have more experience.

Granted, GLOFO was bound to use IBM's high performance 7nm process, but there were concerns about them being able to meed demands... its probable AMD got the wind of this early on and delegated to negotiate with TSMC.

On another hand, this apparently indicates that Nvidia won't get 7nm access soon, which is probably why they are going with 12nm TSMC process.

AMD seems confident enough capacity won't be an issue for TSMC as far as their products go.. in fact, they were able to push a demo of Vega on 7nm to late this year as opposed to next year... so at the very least, that gives us a possible positive outlook on things.

I was already under the impression that Zen 2 was indeed AMD's priority... with GPU's coming in afterwards.

I suspect that Zen 2 server parts along with Radeon Instinct will be coming in early 2019 (market availability that is)... with consumer Zen 2 following steady in the first quarter (or basically just a year after Ryzen+).
As for GPU's... the consumer variety (Navi) is likely coming mid next year or might be released with consumer Zen 2.
Don't know.

But I'm thinking TSMC meeting AMD demand won't be too much of a problem.. even with APPLE gobbling up most of the capacity.

You know... I can't help but think AMD could have also opted for 12nm TSMC much like NV did.
It's still a high perf. process designed for high clocks and efficiency... so it would offer a substantial performance/efficiency increase over Glofo manuf. process, bringing them effectively on par with NV in GPU's... and possibly Intel in the CPU arena (maybe not 5GhZ + frequencies, but well above 4GhZ baseline most likely).
Or delegate GPU's to 12nm, with Zen 2 on 7nm...

Still, it also depends on how ready they are for rollout. There might not be too much of a difference if 7nm is ready as is (and AMD certainly claims it is), so, sticking to 12nm might be counterproductive.

If problems arise with 7nm... we will have to wait and see what they are... though as of right now, AMD seems to be on track... and accelerating if anything (they are a much smaller company than APPLE, NV and Intel).

 

Yeah, it was one of the moves against AMD left, to slow down or stop their 7nm release due to fab capacity.

Not just GPU's, but consumer CPU's, console APU's, datacenter CPU's too.

Someone funding the GF fab's 7nm push either withdrew or offered to help fund if GF delayed. It will be interesting to see who fills up the 14nm / 12nm capacity increases...

It could also be GF holding out for more money... to motivate those originally scheduled @ 7m to call in and complain...GF -> "Hey, help us with 7nm capital investment or order 14nm/12nm production."

GlobalFoundries Puts 7nm ‘On-hold Indefinitely; Spins Off ASIC Division As Subsidiary
By Ramish Zafar, Aug 28 2018
https://wccftech.com/globalfoundrie...nitely-spins-off-asic-division-as-subsidiary/

 

Introducing AMD Radeon Pro™ V340
Redefining High Performance Datacenter Graphics
AMD is announcing the Radeon™ Pro V340 graphics card, a high-performance dual-GPU Virtual Desktop Infrastructure (VDI) solution purpose-built to accelerate and power the most demanding datacenter visualization workloads, including CAD, design, desktop-as-a service (DaaS) and rendering.

The AMD Radeon™ Pro V340 datacenter graphics card delivers an impressively smooth GPU experience from the cloud to virtually any device, anywhere. With MxGPU technology at its core, this hardware-based virtualized graphics solution provides high levels of predictable performance, enhanced security and support for up to 32 VMs per card.

IT managers will also like the fact that this MxGPU solution is easy to set up and manage. Unlike other GPUs, the V340 delivers GPU virtualization to end users without end user licenses, providing enterprises with a lower cost per user.

Learn more about the AMD Radeon™ Pro V340 and the value of MxGPU Technology

For more information on Radeon Pro, check out pro.radeon.com . For the latest updates on Radeon Pro, follow us on Twitter at @RadeonPro and on Facebook.com/Radeon .

 

The problem with AMD and Vega, for me, is the raw performance of the chip out did even the 1080TI and they were touting it was better but just needed game optimized drivers. This is when the pre consumer level released hardware was out. Well we all see where that ended up. At consumer level release they quietly told reviewers to compare them to the 1080 and not the 1080TI but neven came back saying they knew it was not just a driver issue.

This gave AMD a huge black eye in the promise to perform area, and now 12nm does this as well as the touted 10-15% performance increase was from tapping into overclocking headroom not delivered by the process. Sorry I need actual silicon at this point!

 

Wow, where did that come from? Did that thought timewarp in from the Vega 64 release? What triggered that outburst?

We aren't seeing any AMD GPU's in consumer-land for a while, you can take a rest on the pre-recorded outrage against AMD until next year at least.

The new 1080ti is about $650, Vega 64 $530, and Vega 56 $485, and 1080 $447 according to https://gpu.userbenchmark.com/

Those seem like fair prices if you can find them at or lower new, and even lower used.

No need to step into the twilight zone with the Nvidia RTX release...

 

2.5 Geeks Webcast 8/22/18: LIVE 2nd Gen Threadripper Q&A With James Of AMD!
HotHardware
Streamed live on Aug 22, 2018
WATCH Wednesday 8/22 at 5:30PM EST! - On the next episode of HotHardware’s Two And Half Geeks, we’re going to have a special guest from AMD on the show to answer all of your burning questions regarding the just-launched 2nd Generation Threadripper processors. AMD’s James Prior will be participating to chat 2nd Gen Threadripper, the X399 platform, Wraith Ripper, and more.

 

LoL, time warp, not even. 12 nm is this year and V64 is last year. So far about one bad reveal a year, is there another one due shortly? Agreed I have a long memory but that even goes back to the Athlon days and Intel's infamous vaporware releases (what is old, is new again).

 

This one, unlike the others (which Keller shouldn't be included in when bringing up employees), varies in key ways. He worked his way up at Intel, then got notoriety for his taking the idea of the engineers at AMD for an HEDT product, then delivering on it in a truncated time span so condensed, it is amazing it even exists. He green-lit the project and threadripper was born. Him leaving now, after the refresh came out, but while still having huge acclaim on his role for that, makes sense to get CEO experience. It is rare to pick a CEO that has not been a CEO at some other company before. He is using his project delivery to now work for a great ASIC company doing good things. Considering the condition Intel is currently in, he may be trying to pad his resume to try to make a run at getting back on with Intel in a couple years, but this time directing the ship. So, I would disagree with the articles and say this likely does not mean what they think it means.

Raja, with Su's direction being opposed to what he wanted to do with the Radeon Technologies group, leaving makes a lot of sense. His close friend heading marketing heading out afterwords also makes sense. Funny how the ads for Intel's GPU resemble the same type of campaigns we saw previously from AMD, wouldn't you say?

There are multiple layers that I found to this after looking into it. It ranges from Intel and GF working on 12nm FD-SOI together to the largest shareholder in GF trying to sell a substantial number of shares in AMD the day of the Ferrari event to SOI being a candidate for implementation of Nanowire and Nanosheet Gate All Around, with 12nm being the most dense form of FD-SOI currently, IIRC, and that potentially coming into play on 3nm with EUV, which GF STILL likely has the EUV machines being installed from February of this year. Considering GF was looking at going from 7nm to 3nm GAA, they may have, due to costs, decided not to try to play 7nm finFET with DUV and just wait a little longer for the re-tool at 3nm, while AMD continues to use them for the 12nm and 14nm nodes.

Look through recent information on 12FDX being rumored to be Intel/GF. I just reloaded windows fresh and haven't moved over my browser profile yet, so I cannot find the image that showed this, but think I saw something about it recently, like hotchips time frame. So, there is that side of it. Also, there is the SOI and nanosheet switch I mentioned. Here is some info on it:

https://www.researchgate.net/public...Development_from_Single_Gate_to_Multiple_Gate
https://www.cell.com/joule/pdf/S2542-4351(17)30138-1.pdf
https://irds.ieee.org/images/files/pdf/2017/2017IRDS_MET.pdf

For these reasons, after some reflection, I'm not so sure it isn't to do a collaboration on FDSOI to get certain techniques ready for Intel and GF to deliver on nanosheet GAA down the road at 2021+ (Intel's 7nm, GF 3nm).

They reached their limit with 14nm++. That is what Coffee, the Coffee refresh, Cascade, Cooper (Cascade refresh) all use. Then, it is due or die time on 10nm, which is broken and had to be revised for less density and is why some have called it 12nm. I've given you those articles already, at this point, it seems you refuse to review content.

So, the second article, looking at the timelines, is just repeating what AMD said in their post that is being quoted for the TSMC move, but still fits with months old info, although some are now saying a full EPYC release before mainstream, which would shake up timelines. As to TSMC, AMD would not have purchased the fab time without some confirmation of having, at minimum, what they think the market needs. So, that is less of a fear for me sense they have an idea on the desktop and HEDT demand, and know that the 20% market share will be, in part, due to limited supplies.

Talked about above. Pretty sure he wants to get this to try to get back to running big blue.

Is this chart just for this year, or all scheduled 7nm fab time? Reason I say that is Apple is ramping up a full product line this year, so limited supply on Vega is expected, Qualcomm ramps up their Snapdragon and 5G products in Q4 for a likely Q1 launch, which Samsung has signed on for their chips and Samsung phones use a lot of Snapdragons, so they will eventually take up more of that production, and volume on AMD is scheduled for mainstream and server in the H1 next year, with Navi being H2, and TR being binned off mainstream, etc. So, even though true for 2018, the fear is a little absent due to looking at product timelines and that chart saying 2018, meaning the full year.

This is a BS talking point that came about the time of fabless chip design companies and when AMD spun off GF. It is the fear mongering that without ownership, you won't get production volume. That is why Wafer Supply Agreements and contracts are put in place, so that you can get damages for the FAB not delivering on their promises. And, of course the CPU takes precedent, as Su said. They project moving to 4-5% by the end of the year, and Intel's own statements said they are hoping to hold AMD to 15-20% market share in the server side next year. Now, whether or not Intel is leveraging their work with GF on 12nm FDSOI, I'm not going to speculate on, and if this is an attempt to limit fab time at TSMC, or if AMD got the assurances necessary from TSMC (and the 7nm GPU they tried at GF sucked) so they gave TSMC all of the 7nm work, since the WSA was filled with enough other work needed under 12/14nm products, while running out the clock otherwise.

Now, the owners of GF tried to sell at an AMD event in July, the Ferrari event. That was like a spit in the face, and is likely the reason that AMD NDA'd everyone in attendance for a week or so, which was a counterstrike to the investor liquidating a substantial amount in stock on what could have been inside information. This could be a second part of that retaliation. Lot's of moving parts.


Edit: Need to address a mistake I made. Evidently, all plans beyond 7nm are scrapped. This is in part because the new CEO appointed in March or so decided to take the company in a different direction. It will license IP for the newly spun-off ASIC division to provide 7nm production for that, but the main company will be with the 12nm finFET, 12nm FD-SOI, and larger. https://semiengineering.com/gf-puts-7nm-on-hold/

Still, looking at the executive switchup, the owner of GF selling off AMD at a time when the price should have popped, etc., it looks like you have only TSMC as an option in 2018 and part of 2019, followed by Samsung and TSMC in 2020 for EUV. Intel will have EUV up and running either in H2 2019 to sometime in 2020. But, with losing Apple and Qualcomm, among others, their fab may start to bleed badly as they lose market share to AMD on server chips and possibly low end low power to ARM. Would be interesting if Intel went fabless (and I believe good for them long term).

But, had to clarify, even though the planer FD-SOI can have some overlap with NW and NS GAA, looks like that is not the goal, at least not stated as a goal (although, with Intel working on it with GF)...

 

AMD Ryzen™ Threadripper™ X Series CPUs – The Ultimate Processor for Enthusiasts and Gamers
AMD
Published on Aug 31, 2018
The AMD Ryzen™ Threadripper™ X Series CPUs – more for gamers and creators.


eGPU Multitasking with AMD Radeon Pro Graphics on macOS
AMD
Published on Aug 30, 2018
High-end Pro eGPU hits macOS for heavy GPU multitasking.

 

AMD's response to GF's departure from the running for 7nm FAB work:

Expanding our High-Performance Leadership with Focused 7nm Development
Posted by mark.papermaster in AMD Corporate on Aug 27, 2018 3:00:08 PM
https://community.amd.com/community...mance-leadership-with-focused-7nm-development

"AMD is unique in the world of computing as the only company with both high-performance CPU and GPU products. For the past several years we have been executing our multi-generational leadership product and architectural roadmap.

Just in the last 18 months, we successfully introduced and ramped our strongest set of products in more than a decade and our business has grown dramatically as we gained market share across the PC, gaming and datacenter markets.

The industry is at a significant inflection point as the pace of Moore’s Law slows while the demand for computing and graphics performance continues to grow.

This trend is fueling significant shifts throughout the industry and creating new opportunities for companies that can successfully bring together architectural, packaging, system and software innovations with leading-edge process technologies.

That is why at AMD we have invested heavily in our architecture and product roadmaps, while also making the strategic decision to bet big on the 7nm process node.

While it is still too early to provide more details on the architectural and product advances we have in store with our next wave of products, it is the right time to provide more detail on the flexible foundry sourcing strategy we put in place several years ago.

AMD’s next major milestone is the introduction of our upcoming 7nm product portfolio, including the initial products with our second generation “Zen 2” CPU core and our new “Navi” GPU architecture.

We have already taped out multiple 7nm products at TSMC, including our first 7nm GPU planned to launch later this year and our first 7nm server CPU that we plan to launch in 2019.

Our work with TSMC on their 7nm node has gone very well and we have seen excellent results from early silicon.

To streamline our development and align our investments closely with each of our foundry partner’s investments, today we are announcing we intend to focus the breadth of our 7nm product portfolio on TSMC’s industry-leading 7nm process.

We also continue to have a broad partnership with GLOBALFOUNDRIES spanning multiple process nodes and technologies. We will leverage the additional investments GLOBALFOUNDRIES is making in their robust 14nm and 12nm technologies at their New York fab to support the ongoing ramp of our AMD Ryzen, AMD Radeon and AMD EPYC processors. We do not expect any changes to our product roadmaps as a result of these changes.

We are proud of the long-standing and successful relationships we have built with our multiple foundry partners, and we will continue to strengthen these relationships to enable the manufacturing capacity required to support our product roadmaps.

I look forward to providing more details on those innovations as we prepare to introduce the industry’s first 7nm GPU later this year and our first 7nm CPUs next year."

 

AMD keeps the lead over Intel on boosted Ryzen 2XXX sales in August @ mindfactory.de imgur.com
Submitted 7 hours ago by ingebor
https://www.reddit.com/r/Amd/comments/9bxdy4/amd_keeps_the_lead_over_intel_on_boosted_ryzen/

 

The GlobalFoundries and AMD Story
AdoredTV
Published on Sep 1, 2018
A marriage made in hell.

 

2nd Generation AMD Ryzen Threadripper 2950X Now Available
by Hilbert Hagedoorn on: 08/31/2018 03:12 PM | source: | 8 comment(s)
https://www.guru3d.com/news-story/2nd-generation-amd-ryzen-threadripper-2950x-now-available.html

"AMD announced that the Gen2 AMD Ryzen Threadripper 2950X, with 16-cores and 32-threads, is now available for purchase.

At $899, the Ryzen Threadripper 2950X delivers up to 41 percent more multi-threaded performance than the $999 Core i9-7900X.

AMD Ryzen Threadripper “X” series is designed for prosumers, enthusiasts, and multitasking gamers, and now with 2nd Gen AMD Ryzen Threadripper 2950X, AMD ups the ante with significant technology advancements and higher clockspeeds than ever before.

Not only that, but the incredible X399 platform – the only solution for users who need the most I/O and exclusive ECC memory support on desktop – gets even better with StoreMI storage acceleration technology."

"The four models will be released gradually over the coming months, today, however, we kick off with the 16-core version, which will become available at the end of the month for 899 USD. So what will AMD be releasing?

• Ryzen Threadripper 2920X
  This is a 12-core part with 24 threads. The maximum turbo frequency will be 4.3 GHz, and it has a base frequency of 3.5 GHz. This processor is to be priced at 649 USD. This processor under full stress has a 180 Watt TDP.
• Ryzen Threadripper 2950X
  The second gen 16-core part with 32 threads will run a turbo frequency of 4.4 GHz(!) with a base clock of 3.5 GHz, this will go for 899 USD. This processor under full stress has a 180 Watt TDP.
• Ryzen Threadripper 2970WX
  Advancing in cores we now reach a 24-core part with 48 threads. Here the maximum turbo frequency will be 4.2 GHz with a base clock of 3.0 GHz, this is a 1299 USD SKU. This processor under full stress has a 250 Watt TDP.
• Ryzen Threadripper 2990WX
  You have reached your final destination - the 2990WX has no less than 32 cores and 64 threads. The proc has a 4.2 GHz turbo allowance and runs a 3.0 GHz base clock. This massive manycore product will be sold for 1799 USD, which is 56 USD per core. This processor under full stress has a 250 Watt TDP.

AMD Ryzen Threadripper 2950X review
by Hilbert Hagedoorn on: 08/13/2018 03:00 PM
https://www.guru3d.com/articles-pages/amd-ryzen-threadripper-2950x-review,1.html

"You probably just read our 32-core 2990WX review and noticed my recommendation. As a more 'normal' enthusiast end-user, you really should not be looking at 32 cores man. Eight is fine, sixteen is sweet.

However, with the Ryzen Threadripper 2950X, AMD brings something special to the table. Sixteen cores and thirty-two threads are not something you'll be using anytime soon unless you are crunching data in extensive workloads. However, over time the CPU ecosystem will evolve, and this is why I think 16-core processors deserve a place in the HEDT PC enthusiast and gaming space.

AMD did well with the 2950X, honestly, it is 100 bucks cheaper compared to the original 1950X and, thanks to the Pinnacle Ridge ZEN+ dies, on some threads this puppy will manage 4.4 GHz.

I'll immediately say though, that is at XFR2 frequency, mandatory cooling is required, but any LCS kit will do just fine.

Yes, my friends, Ryzen processors have been refreshed, now the turn goes to AMD Ryzen Threadripper processors receiving that Pinnacle Ridge upgrade. In this review, we check out the all-new 16-core / 32 thread update, the Ryzen Threadripper 2950X.
..."

https://www.pcper.com/news/Processors/AMDs-Ryzen-Threadripper-2950X-now-available-purchase-899

 

AMD 2950X vs 1950X - Threadripper 2 Comparison and Benchmarks
Jarrod'sTech
Published on Aug 13, 2018
Threadripper 2 is here, so what are the differences in performance between AMD’s 1950X and the new 2950X CPU? In this comparison I’ve run a number of CPU specific benchmarks as well as some games to show you the improvements and features present in the second generation.

 

https://www.reddit.com/r/AyyMD/comments/9bzvfz/petition_to_approve_this_newspaper/

Newspaper Headline

TrashcanPrint 121 points 1 day ago
"Headline says “(Lisa Su) resurrects a falling company and uppercuts Intel”."

 

Seems a bit excessive, I hope to see some reasonable builds at much lower cost...

The $7500 Threadripper 2950X ‘CoffeeRipper’ Build [Almost Done]
Hardware Unboxed
Published on Sep 2, 2018

Spoiler: Build Details...

I'll strip the affiliate url's at some point, be aware they are as they came from the channel post:

Corsair 1000D: https://amzn.to/2MI3Sah
MSI X399 Gaming Pro: https://amzn.to/2C9kCCf
AMD Threadripper 2950X: https://amzn.to/2NHkv1Z
32GB Vengeance RGB Pro: https://amzn.to/2N5n4y0
GeForce GTX 1080 FE: https://amzn.to/2wyJA8Z
Corsair RM850x: https://amzn.to/2N6jCmG
Samsung 970 PRO 1TB NVMe M.2: https://amzn.to/2PWHcRi

x16 Corsair ML120: https://amzn.to/2PThldj
x3 Corsair ML140: https://amzn.to/2wvTAjs

MSI Z370I Gaming Pro Carbon AC: https://amzn.to/2NE3dmF
Intel Core i7-8700K: https://amzn.to/2MGoy2r
16GB Vengeance RGB Pro: https://amzn.to/2C91Z1l
Radeon RX Vega 64: https://amzn.to/2CfUmWK
Corsair SF600: https://amzn.to/2N26Eqf
Samsung 970 PRO 1TB NVMe M.2: https://amzn.to/2PWHcRi

EK-Supremacy Edge Nickel: https://amzn.to/2PSlNc2
EK-FC Radeon Vega RGB: https://amzn.to/2ouIqXz
EK-CoolStream CE 420: https://amzn.to/2PY1UQY
EK-RES X3 400: https://amzn.to/2wvcOpa
EK-XTOP Revo D5 PWM: https://amzn.to/2N6GUci

EK-FB MSI X399 Gaming RGB Monoblock: https://amzn.to/2MITSgH
EK-FC GeForce GTX FE: https://amzn.to/2PYzaay
x2 EK-CoolStream XE 480: https://amzn.to/2PYnWTN
EK-RES X3 400: https://amzn.to/2wvcOpa
EK-XTOP Revo D5 PWM: https://amzn.to/2N6GUci

x4 EK-AF Extender Rotary M-M G1/4: https://amzn.to/2N9EcT5
x4 EK-AF Extender Rotary M-F G1/4: https://amzn.to/2CkFQNR
x6 EK-AF Angled 90° G1/4 Black Nickel: https://amzn.to/2N5nsMY
x6 EK-AF Angled 2×45° G1/4 Black Nickel: https://amzn.to/2Na7xNl
x22 EK-HDC Fitting 16mm G1/4 - Black Nickel: https://amzn.to/2PX72on
x3 EK-HD PETG Tube 12/16mm 500mm: https://amzn.to/2wCqiPg
x2 EK-CryoFuel Blood Red Premix 900 mL: https://amzn.to/2PYzhD0
x2 EK-CryoFuel Clear Premix 900 mL: https://amzn.to/2N6hGdX

Support us on Patreon https://www.patreon.com/hardwareunboxed

The $7500 Threadripper 2950X ‘CoffeeRipper’ Build [Almost Done]

 

So any idea when we’ll see next gen APU’s for laptops?
The 2700U is decent enough if laptop makers use dual channel RAM but it would be nice to see something better.

 

There is some recently published info, and these reports are all I have right now:

Info on new CPU's and slide presentation...

AMD Teases New Raven Ridge 2018 APUs
by Paul Alcorn August 22, 2018 at 11:01 AM
https://www.tomshardware.com/news/amd-announces-new-raven-ridge-2018-apus,37678.html

AMD lists Ryzen 2100/2300X/2500X and 2800U ready for ‘mass market’
The most recent update to AMD product portfolio has revealed few upcoming products.
Published: 28th Apr 2018, 11:53 GMT
https://videocardz.com/76064/amd-lists-ryzen-2100-2300x-2500x-and-2800u-ready-for-mass-market

Also of interest for APU's:

Four Years In, Our 25x20 Energy Efficiency Goal Report Card
Posted by mark.papermaster in AMD Corporate on Aug 20, 2018 2:30:21 PM
https://community.amd.com/community...-our-25x20-energy-efficiency-goal-report-card

25 x 20 Energy Efficiency Initiative
https://www.amd.com/en/technologies/25x20?utm_campaign=www.amd.com_25x20

 

Threadripper 2 Update! ECC Support, Linux, and More!
Level1Techs
Published on Sep 3, 2018

Spoiler: Show Index

#Level1Presents

ECC Memory Support on Threadripper?: 6:20
Newer Kernels wont boot because of SEV/PSP: 7:51
Process Lasso and NUMA: 10:50

ASRock Fatal1ty X399 Professional Gaming Review: https://youtu.be/fNr6m13jmlM

 

In the current market, I don't think the 2700u is worth it.
The OEM's TDP constraint the APU's to 15W usually, and 2700u needs all 25W (probably more under full load) to fully stretch its legs.
This (And relatively bad cooling for 2700u) is probably why the 2500u gives same or better performance than 2700u.

Other factors might be at play as to why 2500u gives same/better performance than 2700u, but if you want an AMD apu, just go with 2500u right now, or wait for Zen+ APU's to hit the market (though, I think Zen 2 APU's would be a better deal).

 

How memory bandwidth is killing AMD's 32-core Threadripper performance

https://www.pcworld.com/article/329...core-threadripper-performance.html#tk.rss_all