AMD's Future Chips & SoC's: News, Info & Rumours.

[adamsleath]

relative to early improvements pre-one decade ago maybe. but they still have managed to climb over your "wall" :lol:

better hope coders update their program code to be more parallel....multithreaded.

 

[juanrga]

relative to early improvements pre-one decade ago maybe. but they still have managed to climb over your "wall" :lol:

better hope coders update their program code to be more parallel....multithreaded.

The empirical graphs clearly show where the IPC and frequency walls are placed and how we are approaching asymptotically to both walls.

As mentioned before multithreading is useless for serial algorithms. Also multithreading have its own overheads. Sometimes adding more threads gives less performance than using less threads.

 

[adamsleath]

VLIW EPIC. various tricks.

scheduling...bla bla bla. if its a wall, then just drop the ipc concept as an issue. the 8086 can just drop dead in a corner when its replaced by more sophisticated cpu's with better compilers...etc.

some code is not suited to multi-threads, added latencies aswell.

 

[-Fran-]

What has to do compiler regression with hardware performance? If a new version of a compiler has some problem and reduces the performance on the same hardware, then one avoids that specific compiler version or the flags generating the problem until it is solved.

The scores given by Kanter are official SPEC submission rather than in house non-validated benches. The scores for the Xeon system are from Huawei, not from Intel. The scores for EPYC are from AMD itself. And they show that Skylake core is much faster than Zen core for integer stuff.

The overclock3d link has Dhrystone benches for the i9-7960X and the i9-7900X. The Sisoftware link has Dhrystone benches for i9-7900X and 1950X. As mentioned in a former post:

For instance in Dhrystone the 16C SKylake does about 662 whereas the 16C ThreadRipper does less than 500.

So this is a second bench where Skylake-X core is much faster in integer code than Zen core.

SIMD units in Intel cores don't "emulate integer operations". There are vector ALUs that execute integer instructions. This is an old diagram for Haswell, relating ports and execution units

AVX512 is something more than a "marketing keyword". AVX512 is a standard ISA in HPC and it starts to become a standard in servers as well

https://www.hpcwire.com/2017/06/29/reinders-avx-512-may-hidden-gem-intel-xeon-scalable-processors/

https://cloudplatform.googleblog.com/2017/02/Google-Cloud-Platform-is-the-first-cloud-provider-to-offer-Intel-Skylake.html

Er... You're using a lot of fancy words there that are mixing a lot of things.

What I mean by "emulating" is you'll no longer be calling the same instruction to operand on an int and will need to invoke the "Vasdfasdfdfsaf" new instruction forcing you to recompile your code, and if you're low level enough in C (or ASM), then you'll have to re-write your code for vectorial operations (good ol' packed stuff with a new lipstick color). Just like on any bleeding edge piece of hardware, the software you have in it will have to be validated again and that is a lot of fun if you want to make full use of AVX512.

For the rest of us plebeians, we'll need Windows 11, so the OS is fully aware of AVX512, just like we were able to use AVX with Win7 (I would imagine Win10 supports up to AVX2?). So, yes, it is a marketing term for even enthusiasts when Intel is selling it in the HEDT platform. We're not going to re-compile our games, software or even 3rd party specialized software that *could* use AVX512. For your point of "standard in servers as well". Care to elaborate on that? I'm pretty damn sure no one in the planet is going to port from their PHP+Apache+Linux/Windows installation just because it now you upgraded to a CPU that supports AVX512. The world you're talking about when referring to "servers" is very damn specific and small. I'd even call it niche. Google has the money to burn to just pop data centers due to the stupid amount of data and projects they have, but that is just one singular example. Amazon and other "cloud" providers will offer AVX512 for their stuff, but that won't make them best sellers unless people actually uses it and knows how to use it. Hence, it's just a fancy marketing term. There's a plethora of start ups making use of Amazon and their stack of software is as standard as the day. How do I know this? Because it's the safest way to migrate from one provider to the other.

I could keep on writing my views on your "AVX512 holy grail", but I'll stop here. Still, just don't say "it's a standard", because there's an organization for that. You can say it's "famous" or "popular", but not "standard". The implications are very different.

In regards to the benches, I'm still not convinced. It's good Intel was able to pack AVX2 to make use of "AVX512", since they'd need Win11 to display some of the goodness of it, but we're still not seeing it all. I'll read more on that, since I haven't found much information, other than the instructions look pretty much the same, but fatter arguments.

PS: http://www.officedaytime.com/simd512e/simd.html

 

[jaymc]

"GlobalFoundries will detail a 0.0269µm2 SRAM cell made in its 7nm FinFET process, capable of delivering 2.8x better routed logic density and more than 40 per cent more performance or 55 per cent lower power, compared to its 14nm products."
An Graphene chips on the horizon.."25GHz graphene circuits printed on paper"
http://hexus.net/business/news/components/111173-samsung-8nm-lpp-now-ready-production/

"AMD’s new notebook processor will put Intel on notice"
http://www.marketwatch.com/story/amds-new-notebook-processor-will-put-intel-on-notice-2017-10-23

 

[aldaia]

"GlobalFoundries will detail a 0.0269µm2 SRAM cell made in its 7nm FinFET process

I remember a lenghty discussion where some of us claimed exactly 0.0269µm2 SRAM cell for 7nm GloFo.

 

[8350rocks]

Spoiler

What has to do compiler regression with hardware performance? If a new version of a compiler has some problem and reduces the performance on the same hardware, then one avoids that specific compiler version or the flags generating the problem until it is solved.

The scores given by Kanter are official SPEC submission rather than in house non-validated benches. The scores for the Xeon system are from Huawei, not from Intel. The scores for EPYC are from AMD itself. And they show that Skylake core is much faster than Zen core for integer stuff.

The overclock3d link has Dhrystone benches for the i9-7960X and the i9-7900X. The Sisoftware link has Dhrystone benches for i9-7900X and 1950X. As mentioned in a former post:

For instance in Dhrystone the 16C SKylake does about 662 whereas the 16C ThreadRipper does less than 500.

So this is a second bench where Skylake-X core is much faster in integer code than Zen core.

SIMD units in Intel cores don't "emulate integer operations". There are vector ALUs that execute integer instructions. This is an old diagram for Haswell, relating ports and execution units

AVX512 is something more than a "marketing keyword". AVX512 is a standard ISA in HPC and it starts to become a standard in servers as well

https://www.hpcwire.com/2017/06/29/reinders-avx-512-may-hidden-gem-intel-xeon-scalable-processors/

https://cloudplatform.googleblog.com/2017/02/Google-Cloud-Platform-is-the-first-cloud-provider-to-offer-Intel-Skylake.html

Er... You're using a lot of fancy words there that are mixing a lot of things.

What I mean by "emulating" is you'll no longer be calling the same instruction to operand on an int and will need to invoke the "Vasdfasdfdfsaf" new instruction forcing you to recompile your code, and if you're low level enough in C (or ASM), then you'll have to re-write your code for vectorial operations (good ol' packed stuff with a new lipstick color). Just like on any bleeding edge piece of hardware, the software you have in it will have to be validated again and that is a lot of fun if you want to make full use of AVX512.

For the rest of us plebeians, we'll need Windows 11, so the OS is fully aware of AVX512, just like we were able to use AVX with Win7 (I would imagine Win10 supports up to AVX2?). So, yes, it is a marketing term for even enthusiasts when Intel is selling it in the HEDT platform. We're not going to re-compile our games, software or even 3rd party specialized software that *could* use AVX512. For your point of "standard in servers as well". Care to elaborate on that? I'm pretty damn sure no one in the planet is going to port from their PHP+Apache+Linux/Windows installation just because it now you upgraded to a CPU that supports AVX512. The world you're talking about when referring to "servers" is very damn specific and small. I'd even call it niche. Google has the money to burn to just pop data centers due to the stupid amount of data and projects they have, but that is just one singular example. Amazon and other "cloud" providers will offer AVX512 for their stuff, but that won't make them best sellers unless people actually uses it and knows how to use it. Hence, it's just a fancy marketing term. There's a plethora of start ups making use of Amazon and their stack of software is as standard as the day. How do I know this? Because it's the safest way to migrate from one provider to the other.

I could keep on writing my views on your "AVX512 holy grail", but I'll stop here. Still, just don't say "it's a standard", because there's an organization for that. You can say it's "famous" or "popular", but not "standard". The implications are very different.

In regards to the benches, I'm still not convinced. It's good Intel was able to pack AVX2 to make use of "AVX512", since they'd need Win11 to display some of the goodness of it, but we're still not seeing it all. I'll read more on that, since I haven't found much information, other than the instructions look pretty much the same, but fatter arguments.

PS: http://www.officedaytime.com/simd512e/simd.html

No, you are pretty much spot on.

AVX512 is theoretical performance for consumers outside a few edge cases (like someone who would run Gentoo, for example...and even how much they would get out of it would be debatable).

It also tends to cause intel chips to run very hot, very quickly with any prolonged usage...which is partly why the Xeons are rated to run so much slower than the HEDT parts. Efficiency is the smallest part of that, the thermals from a rack server are the elephant in the room that no one is discussing, and AVX512 feeds that elephant tons of peanuts.

 

[-Fran-]

In other, yet related, news:

http://techreport.com/news/32736/amd-stayed-100-in-the-black-in-q3-2017
https://www.anandtech.com/show/11961/amd-announces-q3-earnings-71m-in-the-black

AMD seems to be staying afloat, but the race is far from over. They still need to secure more clients, but at least their quarter shows they are moving inventory, and not back to their own warehouses, so that is very good. I hope they put most of the money they are making back into R&D and production; they'd be stupid not to!

Cheers!

 

[TMTOWTSAC]

In other, yet related, news:

http://techreport.com/news/32736/amd-stayed-100-in-the-black-in-q3-2017
https://www.anandtech.com/show/11961/amd-announces-q3-earnings-71m-in-the-black

AMD seems to be staying afloat, but the race is far from over. They still need to secure more clients, but at least their quarter shows they are moving inventory, and not back to their own warehouses, so that is very good. I hope they put most of the money they are making back into R&D and production; they'd be stupid not to!

Cheers!

Their creditors will probably have a say in that. About 200 million of their debt comes due in 2019, and they need to avoid leveraging themselves any more than they already have. Something like 1.4 billion at last count. But you can't deny they're in a better position now, and making better products.

 

[-Fran-]

Their creditors will probably have a say in that. About 200 million of their debt comes due in 2019, and they need to avoid leveraging themselves any more than they already have. Something like 1.4 billion at last count. But you can't deny they're in a better position now, and making better products.

If my accounting course in University was any good, then all the debt, as long as it has been accounted for (hence the term), is already factored in (earning referrals and devengated costs) and it should not be a problem even "in the red".

I would imagine that information is for everyone to see, so it should be easy to confirm or deny what I'm saying.

Cheers!

 

[juanrga]

What has to do compiler regression with hardware performance? If a new version of a compiler has some problem and reduces the performance on the same hardware, then one avoids that specific compiler version or the flags generating the problem until it is solved.

The scores given by Kanter are official SPEC submission rather than in house non-validated benches. The scores for the Xeon system are from Huawei, not from Intel. The scores for EPYC are from AMD itself. And they show that Skylake core is much faster than Zen core for integer stuff.

The overclock3d link has Dhrystone benches for the i9-7960X and the i9-7900X. The Sisoftware link has Dhrystone benches for i9-7900X and 1950X. As mentioned in a former post:

For instance in Dhrystone the 16C SKylake does about 662 whereas the 16C ThreadRipper does less than 500.

So this is a second bench where Skylake-X core is much faster in integer code than Zen core.

SIMD units in Intel cores don't "emulate integer operations". There are vector ALUs that execute integer instructions. This is an old diagram for Haswell, relating ports and execution units

AVX512 is something more than a "marketing keyword". AVX512 is a standard ISA in HPC and it starts to become a standard in servers as well

https://www.hpcwire.com/2017/06/29/reinders-avx-512-may-hidden-gem-intel-xeon-scalable-processors/

https://cloudplatform.googleblog.com/2017/02/Google-Cloud-Platform-is-the-first-cloud-provider-to-offer-Intel-Skylake.html

Er... You're using a lot of fancy words there that are mixing a lot of things.

What I mean by "emulating" is you'll no longer be calling the same instruction to operand on an int and will need to invoke the "Vasdfasdfdfsaf" new instruction forcing you to recompile your code, and if you're low level enough in C (or ASM), then you'll have to re-write your code for vectorial operations (good ol' packed stuff with a new lipstick color). Just like on any bleeding edge piece of hardware, the software you have in it will have to be validated again and that is a lot of fun if you want to make full use of AVX512.

For the rest of us plebeians, we'll need Windows 11, so the OS is fully aware of AVX512, just like we were able to use AVX with Win7 (I would imagine Win10 supports up to AVX2?). So, yes, it is a marketing term for even enthusiasts when Intel is selling it in the HEDT platform. We're not going to re-compile our games, software or even 3rd party specialized software that *could* use AVX512. For your point of "standard in servers as well". Care to elaborate on that? I'm pretty damn sure no one in the planet is going to port from their PHP+Apache+Linux/Windows installation just because it now you upgraded to a CPU that supports AVX512. The world you're talking about when referring to "servers" is very damn specific and small. I'd even call it niche. Google has the money to burn to just pop data centers due to the stupid amount of data and projects they have, but that is just one singular example. Amazon and other "cloud" providers will offer AVX512 for their stuff, but that won't make them best sellers unless people actually uses it and knows how to use it. Hence, it's just a fancy marketing term. There's a plethora of start ups making use of Amazon and their stack of software is as standard as the day. How do I know this? Because it's the safest way to migrate from one provider to the other.

I could keep on writing my views on your "AVX512 holy grail", but I'll stop here. Still, just don't say "it's a standard", because there's an organization for that. You can say it's "famous" or "popular", but not "standard". The implications are very different.

In regards to the benches, I'm still not convinced. It's good Intel was able to pack AVX2 to make use of "AVX512", since they'd need Win11 to display some of the goodness of it, but we're still not seeing it all. I'll read more on that, since I haven't found much information, other than the instructions look pretty much the same, but fatter arguments.

PS: http://www.officedaytime.com/simd512e/simd.html

AVX is a SIMD approach. So you have to recompile code to use wider vectors. This is normal, precisely the goal of vector/SIMD approach is to avoid the overheads and inefficiencies generated in the front-end (fetching, decoding, renaming, and rest of SS/OOOE logic).

One doesn't need an OS "fully aware of AVX512" to use AVX software. One can run AVX512 software in current windows/linux OSes. That is what reviews did and what professional people is doing. Some developers have already recompiled software to use AVX512. And customers are already using AVX512 in servers

https://www.hpcwire.com/2017/06/29/reinders-avx-512-may-hidden-gem-intel-xeon-scalable-processors/

When I say that AVX512 is a "standard" in HPC, I mean it is a defacto standard. Both CUDA and AVX512 are standards in the HPC world.

No one said that AVX512 is a "holy grail", but it is important enough that AMD is considering adding AVX512 support in future Zen cores.

 

[-Fran-]

I know AVX512 is the next step in the "evolution" of SIMD, but you're giving it too much credit when it's just an evolution of something already developed; given all your rants with IF and how "non-revolutionary and non-innovative", I find it weird you praise AVX512 so much. Plus, you sacrifice some things in order to make full use of its capabilities in the CPU. Running at half the speed just to run an specific instruction set sounds really weird to me at least. That makes it sound like the implementation in the circuitry is really abnormal and might even require special clock generators.

You could say I just don't like the current implementation of AVX512 and your particular way of praising it so much without looking beyond the marketing material from Intel.

Cheers!

 

[YoAndy]

AMD..Shares of the chipmaker tumbled almost 11% after-hours on Tuesday following the report, and were down 9% to $12.97 to start the day on Wednesday. AMD shares have risen over 118% since Oct. 2016, as the company has benefited from increasing market share and cryptocurrency miners, among other factors.
The bar was set high this quarter, with AMD needing to a deliver a "near perfect" print, said Susquehanna Financial Group analyst Christopher Rolland. Instead, the company signaled a slowdown in the fourth quarter due to seasonally weaker console processor demand when it guided for revenue to decline 15% sequentially, plus or minus 3%. Adding to those concerns, AMD forecast a gross margin of 35% for the fourth quarter, which was 50 basis points below Rolland's estimate.

"We were disappointed by the -15% sequential 4Q17 drop, worse than our -11% estimate, which may indicate a deceleration in the core business as well as a marked slowdown in crypto-related sales (we witnessed a slowing in total GPU compute in October which may weigh here)," said Rolland, who has a Neutral rating and $15 price target on AMD's stock.

Maybe Advanced Micro Devices is just having issues of its own. Even though the company is doing a lot of things right, it's also going head-to-head with two great American companies right now, in Nvidia and Intel Corporation (INTC - Get Report) .
https://www.thestreet.com/story/14358868/1/advanced-micro-devices-stock-is-plunging-due-to-own-issues.html

 

[gamerk316]

AMD..Shares of the chipmaker tumbled almost 11% after-hours on Tuesday following the report, and were down 9% to $12.97 to start the day on Wednesday. AMD shares have risen over 118% since Oct. 2016, as the company has benefited from increasing market share and cryptocurrency miners, among other factors.
The bar was set high this quarter, with AMD needing to a deliver a "near perfect" print, said Susquehanna Financial Group analyst Christopher Rolland. Instead, the company signaled a slowdown in the fourth quarter due to seasonally weaker console processor demand when it guided for revenue to decline 15% sequentially, plus or minus 3%. Adding to those concerns, AMD forecast a gross margin of 35% for the fourth quarter, which was 50 basis points below Rolland's estimate.

"We were disappointed by the -15% sequential 4Q17 drop, worse than our -11% estimate, which may indicate a deceleration in the core business as well as a marked slowdown in crypto-related sales (we witnessed a slowing in total GPU compute in October which may weigh here)," said Rolland, who has a Neutral rating and $15 price target on AMD's stock.

Maybe Advanced Micro Devices is just having issues of its own. Even though the company is doing a lot of things right, it's also going head-to-head with two great American companies right now, in Nvidia and Intel Corporation (INTC - Get Report) .
https://www.thestreet.com/story/14358868/1/advanced-micro-devices-stock-is-plunging-due-to-own-issues.html

Expected. AMD's been due for an investor sell off for a while now. Nothing structural, just a natural consequence of the stock going up.

More interesting will be to see how many people jump back on once the price stabilizes. Will investors believe AMD has some more room for growth, or will they (like me) think AMDs stock has peaked and stay away? The next few days will tell a lot about how investors few AMDs long term outlook.

If my accounting course in University was any good, then all the debt, as long as it has been accounted for (hence the term), is already factored in (earning referrals and devengated costs) and it should not be a problem even "in the red".

I would imagine that information is for everyone to see, so it should be easy to confirm or deny what I'm saying.

Cheers!

It's factored into their financial reports, but AMD still needs the hard physical cash to pay said debts off. That being said, they pushed the majority of their debt out a few more years, but they still don't have a ton of cash on hand at any given point in time.

 

[jdwii]

If only VEGA was successful Amd should be thankful for mining!

 

[juanrga]

I know AVX512 is the next step in the "evolution" of SIMD, but you're giving it too much credit when it's just an evolution of something already developed; given all your rants with IF and how "non-revolutionary and non-innovative", I find it weird you praise AVX512 so much. Plus, you sacrifice some things in order to make full use of its capabilities in the CPU. Running at half the speed just to run an specific instruction set sounds really weird to me at least. That makes it sound like the implementation in the circuitry is really abnormal and might even require special clock generators.

You could say I just don't like the current implementation of AVX512 and your particular way of praising it so much without looking beyond the marketing material from Intel.

I didn't say that AVX512 is something revolutionary and innovative. Never!

Running at half the speed increases the efficiency doing that CPUs with AVX512 can start to compete with TCUs (accelerators, GPGPUs,...) in performance per watt. It is the reason why some recent supercomputers don't use GPGPUs or MIMD accelerators anymore, but only CPUs equipated with AVX512 support. It is also the reason why the first exascale supercomputer will use only CPUs equipped with SVE (which is a similar technology to AVX512). GPGPUs aren't needed anymore. Approaches as AMD's HSA are DOA. On the other hand you hear multiple rumors about AMD considering adding AVX512 to future Zen CPUs.

This doesn't have anything to do with "marketing material". It is physics. At RWT some of us demonstrated how a future CPU with AVX1024 would be better than any GPU even in 3D rendering. He already developed an AVX-based graphics renderer...

I will stop here.

 

[-Fran-]

It's factored into their financial reports, but AMD still needs the hard physical cash to pay said debts off. That being said, they pushed the majority of their debt out a few more years, but they still don't have a ton of cash on hand at any given point in time.

Uhm... Not necessarily... I remember that one of the fundamentals on big corporation management was to *not* have cash "at hand", but have liquidity (similar, but not the same). You have your assets and all that, but you always have credit to pay for stuff in the long term and a small amount of cash for the short term.

Since you're talking about ~2020, then it could be "long term" debt, so I would imagine they would handle that with credit instead. So, this is to say, they might be in trouble if and only if, they can't get any credit. As long as a company is able to get access to credit, they'll have a way forward. The amount of "cash at hand" becomes irrelevant, unless you mean "liquidity", which is fairly close, but not quite the same.

Cheers!

 

[juanrga]

As expected the AMD stock dropped by more than 13% immediately after the financial report.

https://finance.google.com/finance?chdnp=0&chdd=1&chds=1&chdv=1&chvs=maximized&chdeh=1&chfdeh=0&chdet=1509016469256&chddm=1955&chls=IntervalBasedLine&q=NASDAQ:AMD&ntsp=0&ei=TMPxWdiwAsyNUvGOvLAN

 

[-Fran-]

https://www.anandtech.com/show/11964/ryzen-mobile-is-launched-amd-apus-for-laptops-with-vega-and-updated-zen

All I can read there is "hot, steamy and very wild sex".

I have to admit I don't like they're aiming for ultra/super thin devices first, but I can see why they might be. All I can see in the shelves as of late from OEMs are systems that try to be sleek first and useful second. I hope AMD strikes a balance there Intel can't really get with their GPU. I can't wait for official reviews!

Also, Zen 1.5; man, if only someone would have guessed! Yes, I'm being sarcastic. EDIT: Ok, it's not the Zen core itself, but the power controls instead. So, strictly speaking, there is not a pure performance jump due to IPC. Bummer.

Cheers!

 

[goldstone77]

Advanced Micro Devices (AMD) beat expectations for Q3 with revenue of $1.64B and EPS of $0.10. Analyst consensus for revenues was $1.51B and EPS of $0.08.

It is little surprise that AMD’s core metrics (image below) showed solid improvement across the board.

The big disappointment was that the company’s guidance for Q4 is down 15% - which indicates that none of AMD’s high-profile CPU or GPU products will be ramping meaningfully in Q4.

While the company’s guidance is higher than consensus for Q4, it appears that several investors may have been expecting/hoping for a stronger Q4 based on a Ryzen Mobile launch and strong EPYC or Vega revenues.

Prognosis

From a quantitative view point, Advanced Micro Devices beat Q3 numbers comfortably and guided up compared to analyst forecasts.

The company is solidly profitable after many years and is now starting to consistently beat guidance and expectations. The level of revenues being witnessed by the company are levels that the company has not seen in half a decade. These are positives in terms of sentiment.

Despite the CEO’s vagueness, which is not helping the company’s cause, the company has a strong product cycle ahead of it and, looking at 2018, the company’s strong product story and prospects remains unchanged.

Considering the product strength, and Intel's relatively tame competitive position, AMD remains a solid multi-year investment at current levels.

We see a bright future for the company and believe the stock is likely to move up once the initial negative response is out of the way.

Our View: Strong Buy

https://seekingalpha.com/article/4116792-amd-remains-solid-investment-despite-slow-product-ramp

This article points out a few reasons why I advised people to sell before earnings report. One point that is not mentioned was that there is no response to Intel's 8th Gen processors until February, and the low expectations of 12nm Ryzen.

 

[goldstone77]

https://www.anandtech.com/show/11964/ryzen-mobile-is-launched-amd-apus-for-laptops-with-vega-and-updated-zen

All I can read there is "hot, steamy and very wild sex".

I have to admit I don't like they're aiming for ultra/super thin devices first, but I can see why they might be. All I can see in the shelves as of late from OEMs are systems that try to be sleek first and useful second. I hope AMD strikes a balance there Intel can't really get with their GPU. I can't wait for official reviews!

Also, Zen 1.5; man, if only someone would have guessed! Yes, I'm being sarcastic. EDIT: Ok, it's not the Zen core itself, but the power controls instead. So, strictly speaking, there is not a pure performance jump due to IPC. Bummer.

Cheers!

I think they have a much better product now, and they will do well as long as the price is right!

 

[-Fran-]

https://www.anandtech.com/show/11964/ryzen-mobile-is-launched-amd-apus-for-laptops-with-vega-and-updated-zen

All I can read there is "hot, steamy and very wild sex".

I have to admit I don't like they're aiming for ultra/super thin devices first, but I can see why they might be. All I can see in the shelves as of late from OEMs are systems that try to be sleek first and useful second. I hope AMD strikes a balance there Intel can't really get with their GPU. I can't wait for official reviews!

Also, Zen 1.5; man, if only someone would have guessed! Yes, I'm being sarcastic. EDIT: Ok, it's not the Zen core itself, but the power controls instead. So, strictly speaking, there is not a pure performance jump due to IPC. Bummer.

Cheers!

I think they have a much better product now, and they will do well as long as the price is right!

The numbers quoted by AMD itself are the ones that were posted a few months ago in the thread; or at least they're very in-line with them. So the performance of the APUs is decent, but not really breath taking. I still don't like the way the marketing teams present the data. The i7 used seemed to be really thermal limited in the notebook and i suspect it might have been limited with other factors, since the i5 was actually better in everything.

That being said, I do agree it's a good product and they need to make people realize it, so they can actually sell. I hope HP, Acer and Lenovo don't offer sucky laptops.

Cheers!

 

[jdwii]

https://www.anandtech.com/show/11964/ryzen-mobile-is-launched-amd-apus-for-laptops-with-vega-and-updated-zen

All I can read there is "hot, steamy and very wild sex".

I have to admit I don't like they're aiming for ultra/super thin devices first, but I can see why they might be. All I can see in the shelves as of late from OEMs are systems that try to be sleek first and useful second. I hope AMD strikes a balance there Intel can't really get with their GPU. I can't wait for official reviews!

Also, Zen 1.5; man, if only someone would have guessed! Yes, I'm being sarcastic. EDIT: Ok, it's not the Zen core itself, but the power controls instead. So, strictly speaking, there is not a pure performance jump due to IPC. Bummer.

Cheers!

Man Yuka the 2700U scored 707 in R15 that's pretty darn impressive for a 15 watt CPU from Amd even more so since it has half the L3 cache per core compared to summit ridge in reference my 4790K at stock scored 805. If only Microsoft put ryzen in their new xbox one x then i'd be a bit more excited they could have fit a 8 core in it based on what i see here. To think man a stock Phenom II x6 1100T scored like 610-640.

Now we just have to wait to see what their higher TDP parts can do.

 

[8350rocks]

https://www.anandtech.com/show/11964/ryzen-mobile-is-launched-amd-apus-for-laptops-with-vega-and-updated-zen

All I can read there is "hot, steamy and very wild sex".

I have to admit I don't like they're aiming for ultra/super thin devices first, but I can see why they might be. All I can see in the shelves as of late from OEMs are systems that try to be sleek first and useful second. I hope AMD strikes a balance there Intel can't really get with their GPU. I can't wait for official reviews!

Also, Zen 1.5; man, if only someone would have guessed! Yes, I'm being sarcastic. EDIT: Ok, it's not the Zen core itself, but the power controls instead. So, strictly speaking, there is not a pure performance jump due to IPC. Bummer.

Cheers!

I think they have a much better product now, and they will do well as long as the price is right!

The numbers quoted by AMD itself are the ones that were posted a few months ago in the thread; or at least they're very in-line with them. So the performance of the APUs is decent, but not really breath taking. I still don't like the way the marketing teams present the data. The i7 used seemed to be really thermal limited in the notebook and i suspect it might have been limited with other factors, since the i5 was actually better in everything.

That being said, I do agree it's a good product and they need to make people realize it, so they can actually sell. I hope HP, Acer and Lenovo don't offer sucky laptops.

Cheers!

Also, the turbo core technology they have been working on since the discovery tablet is finally coming to market on tech that consumers can get.

Expect that in Zen+ and moving forward.

I would almost buy that lenovo as it stands, but single channel memory kills it for me.

2700U + 500GB SSD and about 8 GB RAM and that is all I need from a slick lightweight notebook.

 

[juanrga]

https://www.anandtech.com/show/11964/ryzen-mobile-is-launched-amd-apus-for-laptops-with-vega-and-updated-zen

All I can read there is "hot, steamy and very wild sex".

I have to admit I don't like they're aiming for ultra/super thin devices first, but I can see why they might be. All I can see in the shelves as of late from OEMs are systems that try to be sleek first and useful second. I hope AMD strikes a balance there Intel can't really get with their GPU. I can't wait for official reviews!

Also, Zen 1.5; man, if only someone would have guessed! Yes, I'm being sarcastic. EDIT: Ok, it's not the Zen core itself, but the power controls instead. So, strictly speaking, there is not a pure performance jump due to IPC. Bummer.

Cheers!

I think they have a much better product now, and they will do well as long as the price is right!

The numbers quoted by AMD itself are the ones that were posted a few months ago in the thread; or at least they're very in-line with them. So the performance of the APUs is decent, but not really breath taking. I still don't like the way the marketing teams present the data. The i7 used seemed to be really thermal limited in the notebook and i suspect it might have been limited with other factors, since the i5 was actually better in everything.

That being said, I do agree it's a good product and they need to make people realize it, so they can actually sell. I hope HP, Acer and Lenovo don't offer sucky laptops.

Like this?

Pay attention to the specs...