AMD really does have good designs limited by slow fab. The market would be a different place if Richland had been 22nm. Imagine how good this would be if AMD was preparing to launch 14nm parts in 2014 like Intel.
I'm sure that if AMD was on the same node as Intel, their processors would be a lot better, but Intel has the most advanced fabs on the world, even Samsung can't match Intel on high performance silicon. On the other hand, AMD has always been after Intel on fab nodes, but they could pull clever and efficient designs like the Athlon and Athlon 64. And now the company seems to be finishing the integration between AMD and ATI, a company they bought 7 years ago. Once all of their processes share a common base, the designs should improve big time, and we'll start to see that on 2014, expanding to 2017.
I was just thinking that Intel shareholders must be content with basically just owning a Fab. The way the entire semiconductor industry is going, Intel is so far behind in everything except single core performance it is just bizarre (I love my OC'd i7 950 tyvm and plan to keep it for 3 more years until DDR4/etc...). Intel does have the best fabs but and despite the claims of silly TSMC and Sammy claims that they're catching up, it will still likely get harder for Intel to maintain their advantage with so much money pouring into the industry from the mobile boom. I predict that short some amazing design wins, we may see Intel turn into Fab. They could probably make more money being a Fab now anyways considering the idle capacity they undoubtedly have at 45, 32, and my guess is even at 22. Mantle is exciting innovation, so is the ARM x86 marriage in the form that it's in here. It all boils down to whether or not their movement to the process node before everyone else will continue to carry them. I suspect that it might, but not at the margins they have enjoyed in the past due to the difficulties of moving to smaller nodes and the bottom falling out of the PC industry. Microsoft failing so incredibly with 8 seriously damaged Intel's strategy.
I want to know where else Intel is behind? They have the best chips from 2W to 150W. Atom is better than anything ARM makes and is slightly better than S800. Intel undoubtly owns the. Laptop desktop and server markets as well.
Really? The newest atom is only on par with current ARM chips and only in terms of CPU performance, while the GPU is much slower, and ARM V8 is just around the corner. And what is the market share of intel in tablets/phones?
Intel is so far behind in mobile market share it's not funny. Mobile is where the market has hurtled since '07 and the shift is showing no signs of slowing down. Go look up international smartphone growth vs. international PC growth over the past 3 years to see my point. (Also remember, I was referencing this from a shareholder's POV.) Desktop growth may return in a few years but it will be incremental at best. Microsoft, Intel's most critical partner, has failed enormously with Windows 8. The OS continues to experience lackluster sales and poor enterprise adoption. I work for a fortune 500 company and our IT department refuses to adopt Win 8 because it is far too different and awkward for users who were accustomed to the previous versions. Meanwhile, HTML5/JS based system agnostic productivity solutions (e.g. Google Drive, Lucidchart) continue to march forward and look more attractive to cost sensitive corporations. Longer PC update cycles, and a shift towards VDI as a replacement for expensive desktops invariably erodes support for one of Intel's highest earning markets. Granted, Intel might have some *potentially* promising architectures moving forward but so far Intel is being carried solely based on their process node lead. I question if Intel is spending aggressively enough on catching up in their architecture to rip away a significant segment of the Mobile market. Sure Intel is ahead in single core performance per watt, and will likely lead in that category for years, but it isn't as important as other categories. Most mobile apps do not necessarily require strong CPU cores to fulfill the majority of their use case. Intel is training behind in GPU, GPU compute and heterogeneous compute. Crystalwell was a big step in the right direction but IMHO, Intel needs to invest big time in catching up. On the other hand, their Fabs are top notch. Intel has 3D transistors, and will be the first to scale production at 14 nm. With the entire world coalescing around the flexible, open ARM based solutions, what is an x86 architecture to do? Finally, don't forget the very real threat of low cost, low power ARM based micro servers to Intel's server market share. Bottom line is if there isn't a remarkable turnaround in Intel's mobile fortunes, we will be seeing all of the architecture growth headed to Intel's giant, hungry army of competitors. That leaves Intel's best and nearly guaranteed growth prospects with their Fabs.
Your conclusions are to extreme. Microsoft still making cash. BP still making cash. Wal-Mart still making cash. Intel will be around for years even making worse chips than everyone (they don't). P4/Thunderbird history alone should be your lesson of the status quo
Calm down :). Amd tested only shitty PcMark 8 home that do not gives any information about how good is a cpu running real world software, the cpus are not under stress with this horrible suite. Obviously nobody can archieve a 2X in performance/watt at the same node, sure not Amd in these bad days. 35% faster (on a very crap bench) of a slow 1/1.4Ghz soc is enough to be competitive with BayTrail, Amd need to wait a full node shrink. Anyway a good review will gives the real landscape of this so so Soc.
I doubt it. After the bulldozer disaster AMD didn't launch anything new in the CPU department. Just small tweaks, increasing frequency, releasing some wanna be mobile SOCs, etc. They are so far behind Intel, not to mention what happens in mobile space, that's it's not even funny.
This things don't sell. If not for the consoles, by now AMD would have prepared to close the doors, with GPUs and all.
It's bewildering me how is possible for Apple to come out with a truly magnificent SOC in the second iteration while a renown CPU manufacturer is wandering without direction.
AMD's Kaveri alone was interesting tech with Steamroller cores, advanced GCN HSA, Trueaudio and scaling from embedded to laptops, desktops and servers.
With Beema and Mullins for low and ultra low power mobile devices, with 100% perf per watt over Kabini/Temash, AMD has the potential to be the dominant player in all sectors from gaming consoles to tablets to laptops to desktops to servers.
Finally, if AMD Mantle is as good as it sounds, it will allow AMD to upend the computing market like a thermonuclear explosion.
The AMD Temash Jaguar APU has been designed for both tablets and light laptops (as well as winning designs with around 5 million preorders for Sony PS4 and Microsoft Xbox One).
One design is the Acer V5-122P with touchscreen, 4GB to 8GB DDR3, backlit keyboard, high quality IPS (versus low quality Baytrail displays), dockport and expandable battery capability.
Rated at 84% by Notebook check and sold everywhere from Amazon, Tiger Direct, Best Buy etc for around $375 to $475 (depending on memory).
There are CURRENTLY NO commercially available TEMASH/KABINI tablets. I've been looking since the Temash release in June. In fact there was a lot of talk, A LOT, about AMD's mobile APU lead since Baytrail was over 3-4 months away.
Anandtech had multiple articles with the phrase just good enough. The ACER V5-122P is a great value, @$350 for a 11.6", backlit, IPS (or high qual) 13x7 panel, quad core, DX 11 GPU, etc. Battery life is rated at 3.5 hours.
So, The Vizio tablet is Hondo (32 nm??). Fujitsu also has a AMD Hondo tablet. The MSI is the only Temash design win I've seen/read. It's also dual core and locked at 1Ghz (i.e. no turbo).
The are good things about either the Vizio Hondo or MSI W20 3M: 1080P screens, SATA SSD, full DX support. They are both 11.6 screens which I can assume is due to the battery/chassis size required due to TDP and high power draw.
To be blunt, I've been waiting almost 6 months now and nothing. Not one single Temash TABLET device. The worrisome thing is Intel is catching up from a platform perspective. And let's not start to compare the Intel Pentium/Celeron Haswell SKUs. These provide better GPU performance and native SATA, etc.
It's hard to continue waiting because my use case for an AMD win8 tablet is very similar to Intel devices. In fact, for a mobile device, the integration of CPU, GPU, MODEM, wire-less radio, Sensors (e.g. GPS, accelerometers, etc), I think the Qualcomm SnapDragron 800 is a great solution for the price point (i.e. Nokia 2520).
I think you're missing an important point here: AMD was able to lower power usage and increase performance on the same process node (28 nm) as many of the last gen chips. This means that there is still a lot of potential for efficiency increases on the existing node and that AMD is probably waiting for newer process nodes to mature. This means that when they do move to a smaller node, the process will be much more refined. For AMD this translates to lower leakage currents and fewer faulty chips. For the consumer, this translates to lower power usage and lower processor costs.
After all, no sense moving to a newer process node when Intel is having trouble getting Broadwell out the door. Not every company can absorb the financial loss of having to delay an entire product line because they jumped at a new process before it was really ready.
It means the process isn't ready for them yet. They are still at a disadvantage compared to Intel in terms of performance and power use. Have no doubt if they could shrink to a smaller process they would do so.
"Could" is kind of a loaded word. Yes they could sink revenue into moving their manufacturing to a lower node. But that would mean extra costs at a stage when AMD is just starting to prove to investors that they can be profitable again. You also need to take into account that TSMC's 20nm has not reached good yields yet and that GloFo is still at 28nm. Given that these are AMD's major fabs, it makes sense that they aren't pushing lower.
We are seeing a lot of interesting changes in the way that AMD does business and it really should be no surprise that they stayed at the current node. AMD switched from using their CPU layout tools for CPU design to their GPU tools and were able to decrease overall layout size by > 20%. AMD's newest chips will be competitive in the mobile market, especially having much faster graphics performance than anything Intel can offer at the same power envelope and cost.
If you really want to be concerned about a threat to AMD (or x86 in general), the focus should be on what ARM-based designs are going to be able to do once 64-bit chips hit the microserver and PC arena. Again, AMD has had the foresight to start design and testing of 64-bit ARM designs that are likely to be paired with the graphics superior GCN cores. This should help them get the jump on Intel in quite a few developing markets, and should make them competitive with ARM in other arenas.
Are it? Nobody is at a lower node on TSMC. We have been using 28nm TSMC for GPU's since what? 1H of 2012. The industry isn't slowing down, but it won't really follow Moore's law after 16/14 nm and shrinking becomes harder. 40 nm was basically around for 3 generations of GPU's btw, or 2009-2012.
As process becomes exponentially more complex, they also take much longer to mature. Obviously, Intel has a key advantage here: at 14nm they have 4X area savings over AMD's chips, however, 28nm process in 2014 is quite a different thing than 28nm process in 2012. Not to mention that designing properly/efficiently in such a small node is not straightforward. So, lacking Intel's Fab power, I think that exploiting 28nm is a good bet. The best bet, infact, in my opinion. Let's hope it's enough.
This has more to do with three other problems: caching latency, RAM bandwidth, and non-uniform memory addressing for CPU and GPU.
AMD has had significantly higher memory latencies on all levels than Intel for a little while now, and we are hoping to see this improve in newer designs. This has the greatest impact on performance for CPU and GPU as it causes the processor to become "data starved" and have to wait for data to be retrieved.
RAM bandwidth is another factor, as highly parallel workloads end up consuming and generating significant amounts of data. While some can be done by developers to alleviate these issues, the root problem is the limited throughput of DDR3 over GGR5 for these kinds of workloads (that's why both XB1 and PS4 have unified GDDR5).
Having a uniform memory space is important because it eliminates the need for copying between different memory pools, as well as pointer conversions. This alone can have a pronounced affect on high utilization scenarios.
The problem with eDRAM is the die space and power usage. Intel got away with eDRAM in their processors because they are running at much higher power and thermal envelopes than mobile parts. They also dodged the die space issue by having the eDRAM as a separate die in the same package. AMD is trying to unify their product lines so that they can focus more on developing new architectures. Adding something special to a few different models won't help them in the long run. This was really more of a marketing gimmick for Intel as they try to become recognized as a workstation level graphics solution.
Thanks! I sort of remembered that one, but should have checked to be sure. DDR3 + eDRAM vs GDDR5 on the PS4. It would be interesting to see a performance comparison, but I tend to side with GDDR5 or HMC as better memory solutions.
Using eDram will remove most of the ram latencies but it comes with great cost in die-space and power usage. You need x-MB of space and a controller logic to interface the bus. The Xbox 360 cpu had 10MB of eDram which solves its problems. I can predict that 12MB or 16MB of eDram for AMD can do so as well. The eDram of the Iris Pro-e takes 20 watts for 128MB. So 16MB will be less than 3 watts additional. That might be the way.
Is it possible that AMD switched to TSMC's 28HPM process? It would be the easiest way to increase speed without sacrificing power. Which process was AMD using for Temash/Kabini? 28HP or 28LP?
dockport sounds cool. that's not AMD exclusive is it?
thunderbolt has licensing fees that make it FAR too expensive to be practical. So I'd like to see EVERY laptop made have this dockport feature. As desktops fade away the need for a cheap, standardized docking technology is ever increasing.
If they can actually get an OEM to put Mullins in a tablet, it sounds fantastic. But I'm afraid it's going to be Temash all over again, with zero tablet wins and only a couple of mediocre netbook-ish design wins.
If Vizio updated their tablet with Mullins and priced it at around $500, I'd be all over that.
agreed!!! the vizio tablet would be a good execution, but you would need to add 802.11 A/C, 3G/4G/LTE, accelerometer, gyroscope, GPS. The Dell Venue 11 Pro's have similar equipment/features, and can range from BayTrail-Celeron-Pentium-Core. As I understand the Celeron, Pentium, Core and are all Haswell architectures. No availability/pricing on the Haswell yet. The dell venue pro also have replaceable batteries. The MSI W20 3M is another alternative with Temash that can be upgraded.
However, it sounds like at best Beema will be a fair amount ahead of Bay Trail in performance. Mullins will be about on par with Bay Trail in terms of power consumption.
Oh, but wait, Beema generally IS NOT in the same market as Bay Trail, unless you are looking at the laptop/desktop Pentium and Celeron parts...in which case the faster Pentium Bay Trail parts are likely as fast or a little faster than Beema will be...with a lot lower power consumption.
Mullins is likely to still lay Bay Trail a lot in CPU performance, but consume roughly the same power.
It moves the markers for AMD a lot, but it doesn't move them much in comparison to Intel or ARM.
Keep in mind, unless Intel changes its mind, these won't even be going up against Haswell ULT (massively faster) or Bay Trail-T, D or N. It'll be going up again Broadwell ULT and Cherry Trail-T, N and D based on AMD's release dates.
Uh, Saber...one point. Most/all similar power envelope Haswell chips blow AMD mobile chips out of the water in graphical performance. Desktop is a little different, but even there Intel isn't that far behind a lot of the AMD APUs.
Mobile though, especially low power Mobile, Intel has around a 2-4x CPU performance advantage and a 1.5-2x GPU advantage (Kabini versus Haswell ULT).
AMD pretty much only wins in graphics on the desktop and only if you don't care about power consumption and much lower performing CPUs going with those GPUs.
Wow, DockPort, if priced right could be a deciding factor for me between an Intel and AMD laptop.
Also, a standard in terms of physical port placement would go a long way to have independent docking stations that hold the device (either with lid open or with lid closed and external monitor only.
Although, even more I'd like a (software) solution that ports into a desktop, so I can use the resources (files and programs) on the laptop integrated into my desktop seamlessly. Ideally combined with a cloud (really ideally private cloud) solution, making the files just available faster, then over the network.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
47 Comments
Back to Article
Drumsticks - Wednesday, November 13, 2013 - link
AMD really does have good designs limited by slow fab. The market would be a different place if Richland had been 22nm. Imagine how good this would be if AMD was preparing to launch 14nm parts in 2014 like Intel.Filiprino - Wednesday, November 13, 2013 - link
I'm sure that if AMD was on the same node as Intel, their processors would be a lot better, but Intel has the most advanced fabs on the world, even Samsung can't match Intel on high performance silicon.On the other hand, AMD has always been after Intel on fab nodes, but they could pull clever and efficient designs like the Athlon and Athlon 64. And now the company seems to be finishing the integration between AMD and ATI, a company they bought 7 years ago. Once all of their processes share a common base, the designs should improve big time, and we'll start to see that on 2014, expanding to 2017.
djscrew - Thursday, November 14, 2013 - link
I was just thinking that Intel shareholders must be content with basically just owning a Fab. The way the entire semiconductor industry is going, Intel is so far behind in everything except single core performance it is just bizarre (I love my OC'd i7 950 tyvm and plan to keep it for 3 more years until DDR4/etc...). Intel does have the best fabs but and despite the claims of silly TSMC and Sammy claims that they're catching up, it will still likely get harder for Intel to maintain their advantage with so much money pouring into the industry from the mobile boom. I predict that short some amazing design wins, we may see Intel turn into Fab. They could probably make more money being a Fab now anyways considering the idle capacity they undoubtedly have at 45, 32, and my guess is even at 22. Mantle is exciting innovation, so is the ARM x86 marriage in the form that it's in here. It all boils down to whether or not their movement to the process node before everyone else will continue to carry them. I suspect that it might, but not at the margins they have enjoyed in the past due to the difficulties of moving to smaller nodes and the bottom falling out of the PC industry. Microsoft failing so incredibly with 8 seriously damaged Intel's strategy.dylan522p - Thursday, November 14, 2013 - link
I want to know where else Intel is behind? They have the best chips from 2W to 150W. Atom is better than anything ARM makes and is slightly better than S800. Intel undoubtly owns the. Laptop desktop and server markets as well.ddriver - Thursday, November 14, 2013 - link
Really? The newest atom is only on par with current ARM chips and only in terms of CPU performance, while the GPU is much slower, and ARM V8 is just around the corner. And what is the market share of intel in tablets/phones?Anato - Thursday, November 14, 2013 - link
What are your criterions? Does it include market share or price?djscrew - Thursday, November 14, 2013 - link
Intel is so far behind in mobile market share it's not funny. Mobile is where the market has hurtled since '07 and the shift is showing no signs of slowing down. Go look up international smartphone growth vs. international PC growth over the past 3 years to see my point. (Also remember, I was referencing this from a shareholder's POV.) Desktop growth may return in a few years but it will be incremental at best. Microsoft, Intel's most critical partner, has failed enormously with Windows 8. The OS continues to experience lackluster sales and poor enterprise adoption. I work for a fortune 500 company and our IT department refuses to adopt Win 8 because it is far too different and awkward for users who were accustomed to the previous versions. Meanwhile, HTML5/JS based system agnostic productivity solutions (e.g. Google Drive, Lucidchart) continue to march forward and look more attractive to cost sensitive corporations. Longer PC update cycles, and a shift towards VDI as a replacement for expensive desktops invariably erodes support for one of Intel's highest earning markets. Granted, Intel might have some *potentially* promising architectures moving forward but so far Intel is being carried solely based on their process node lead. I question if Intel is spending aggressively enough on catching up in their architecture to rip away a significant segment of the Mobile market. Sure Intel is ahead in single core performance per watt, and will likely lead in that category for years, but it isn't as important as other categories. Most mobile apps do not necessarily require strong CPU cores to fulfill the majority of their use case. Intel is training behind in GPU, GPU compute and heterogeneous compute. Crystalwell was a big step in the right direction but IMHO, Intel needs to invest big time in catching up. On the other hand, their Fabs are top notch. Intel has 3D transistors, and will be the first to scale production at 14 nm. With the entire world coalescing around the flexible, open ARM based solutions, what is an x86 architecture to do? Finally, don't forget the very real threat of low cost, low power ARM based micro servers to Intel's server market share. Bottom line is if there isn't a remarkable turnaround in Intel's mobile fortunes, we will be seeing all of the architecture growth headed to Intel's giant, hungry army of competitors. That leaves Intel's best and nearly guaranteed growth prospects with their Fabs.medi02 - Wednesday, December 18, 2013 - link
You forget that ARM margins are laughable for Intel.Even if they had entire mobile market, it would be a joke income for them.
They are messing with it only because of ear, that mobile might eventually kill PC.
SetiroN - Thursday, November 14, 2013 - link
You just made a whole lot of nonsense.madgabz - Friday, November 15, 2013 - link
And You didn't?hero4hire - Sunday, November 17, 2013 - link
Your conclusions are to extreme. Microsoft still making cash. BP still making cash. Wal-Mart still making cash. Intel will be around for years even making worse chips than everyone (they don't). P4/Thunderbird history alone should be your lesson of the status quoseolinkresearch - Friday, December 13, 2013 - link
if you think like this all this word will bee a nonsense ! watch tubevids.orgGondalf - Friday, November 15, 2013 - link
Calm down :). Amd tested only shitty PcMark 8 home that do not gives any information about how good is a cpu running real world software, the cpus are not under stress with this horrible suite. Obviously nobody can archieve a 2X in performance/watt at the same node, sure not Amd in these bad days. 35% faster (on a very crap bench) of a slow 1/1.4Ghz soc is enough to be competitive with BayTrail, Amd need to wait a full node shrink. Anyway a good review will gives the real landscape of this so so Soc.etre - Wednesday, December 4, 2013 - link
I doubt it. After the bulldozer disaster AMD didn't launch anything new in the CPU department. Just small tweaks, increasing frequency, releasing some wanna be mobile SOCs, etc. They are so far behind Intel, not to mention what happens in mobile space, that's it's not even funny.This things don't sell. If not for the consoles, by now AMD would have prepared to close the doors, with GPUs and all.
It's bewildering me how is possible for Apple to come out with a truly magnificent SOC in the second iteration while a renown CPU manufacturer is wandering without direction.
axien86 - Wednesday, November 13, 2013 - link
AMD's Kaveri alone was interesting tech with Steamroller cores, advanced GCN HSA, Trueaudio and scaling from embedded to laptops, desktops and servers.
With Beema and Mullins for low and ultra low power mobile devices, with 100% perf per watt over Kabini/Temash, AMD has the potential to be the dominant player in all sectors from gaming consoles to tablets to laptops to desktops to servers.
Finally, if AMD Mantle is as good as it sounds, it will allow AMD to upend the computing market like a thermonuclear explosion.
Wade_Jensen - Wednesday, November 13, 2013 - link
Could someone point me towards these kabini/temash tablets, please? I've never heard a whisper of them in an actual retail product.TeeRex - Wednesday, November 13, 2013 - link
The AMD Temash Jaguar APU has been designed for both tablets and light laptops (as well as winning designs with around 5 million preorders for Sony PS4 and Microsoft Xbox One).One design is the Acer V5-122P with touchscreen, 4GB to 8GB DDR3, backlit keyboard, high quality IPS (versus low quality Baytrail displays), dockport and expandable battery capability.
Rated at 84% by Notebook check and sold everywhere from Amazon, Tiger Direct, Best Buy etc for around $375 to $475 (depending on memory).
http://www.amazon.com/Acer-Aspire-V5-122P-0637-11-...
Gigaplex - Thursday, November 14, 2013 - link
Wade_Jensen asked for tablet products. Not a laptop or gaming consoles.JFish222 - Wednesday, November 13, 2013 - link
Apparently Vizio (based on the example product on this page).http://www.amd.com/us/products/notebook/tablets/Pa...
Quick google search found this MSI Wind
http://www.msi.com/product/windpad/W20-3M.html
There aren't many, but they are out there.
kj_anandtech - Thursday, November 14, 2013 - link
There are CURRENTLY NO commercially available TEMASH/KABINI tablets. I've been looking since the Temash release in June. In fact there was a lot of talk, A LOT, about AMD's mobile APU lead since Baytrail was over 3-4 months away.Anandtech had multiple articles with the phrase just good enough. The ACER V5-122P is a great value, @$350 for a 11.6", backlit, IPS (or high qual) 13x7 panel, quad core, DX 11 GPU, etc. Battery life is rated at 3.5 hours.
So, The Vizio tablet is Hondo (32 nm??). Fujitsu also has a AMD Hondo tablet. The MSI is the only Temash design win I've seen/read. It's also dual core and locked at 1Ghz (i.e. no turbo).
The are good things about either the Vizio Hondo or MSI W20 3M: 1080P screens, SATA SSD, full DX support. They are both 11.6 screens which I can assume is due to the battery/chassis size required due to TDP and high power draw.
To be blunt, I've been waiting almost 6 months now and nothing. Not one single Temash TABLET device. The worrisome thing is Intel is catching up from a platform perspective. And let's not start to compare the Intel Pentium/Celeron Haswell SKUs. These provide better GPU performance and native SATA, etc.
It's hard to continue waiting because my use case for an AMD win8 tablet is very similar to Intel devices. In fact, for a mobile device, the integration of CPU, GPU, MODEM, wire-less radio, Sensors (e.g. GPS, accelerometers, etc), I think the Qualcomm SnapDragron 800 is a great solution for the price point (i.e. Nokia 2520).
monstercameron - Wednesday, November 13, 2013 - link
unfortunately, no killer systems/value have used amd temash but here is onehttp://www.bestbuy.com/site/toshiba-satellite-clic...
blanarahul - Wednesday, November 13, 2013 - link
We are still on 28 nm?! I heard the industry was slowing down. But this is unexpected!!SaberKOG91 - Wednesday, November 13, 2013 - link
I think you're missing an important point here: AMD was able to lower power usage and increase performance on the same process node (28 nm) as many of the last gen chips. This means that there is still a lot of potential for efficiency increases on the existing node and that AMD is probably waiting for newer process nodes to mature. This means that when they do move to a smaller node, the process will be much more refined. For AMD this translates to lower leakage currents and fewer faulty chips. For the consumer, this translates to lower power usage and lower processor costs.After all, no sense moving to a newer process node when Intel is having trouble getting Broadwell out the door. Not every company can absorb the financial loss of having to delay an entire product line because they jumped at a new process before it was really ready.
Nagorak - Thursday, November 14, 2013 - link
It means the process isn't ready for them yet. They are still at a disadvantage compared to Intel in terms of performance and power use. Have no doubt if they could shrink to a smaller process they would do so.SaberKOG91 - Thursday, November 14, 2013 - link
"Could" is kind of a loaded word. Yes they could sink revenue into moving their manufacturing to a lower node. But that would mean extra costs at a stage when AMD is just starting to prove to investors that they can be profitable again. You also need to take into account that TSMC's 20nm has not reached good yields yet and that GloFo is still at 28nm. Given that these are AMD's major fabs, it makes sense that they aren't pushing lower.We are seeing a lot of interesting changes in the way that AMD does business and it really should be no surprise that they stayed at the current node. AMD switched from using their CPU layout tools for CPU design to their GPU tools and were able to decrease overall layout size by > 20%. AMD's newest chips will be competitive in the mobile market, especially having much faster graphics performance than anything Intel can offer at the same power envelope and cost.
If you really want to be concerned about a threat to AMD (or x86 in general), the focus should be on what ARM-based designs are going to be able to do once 64-bit chips hit the microserver and PC arena. Again, AMD has had the foresight to start design and testing of 64-bit ARM designs that are likely to be paired with the graphics superior GCN cores. This should help them get the jump on Intel in quite a few developing markets, and should make them competitive with ARM in other arenas.
Penti - Thursday, November 14, 2013 - link
Are it? Nobody is at a lower node on TSMC. We have been using 28nm TSMC for GPU's since what? 1H of 2012. The industry isn't slowing down, but it won't really follow Moore's law after 16/14 nm and shrinking becomes harder. 40 nm was basically around for 3 generations of GPU's btw, or 2009-2012.yankeeDDL - Thursday, November 14, 2013 - link
As process becomes exponentially more complex, they also take much longer to mature. Obviously, Intel has a key advantage here: at 14nm they have 4X area savings over AMD's chips, however, 28nm process in 2014 is quite a different thing than 28nm process in 2012. Not to mention that designing properly/efficiently in such a small node is not straightforward. So, lacking Intel's Fab power, I think that exploiting 28nm is a good bet. The best bet, infact, in my opinion. Let's hope it's enough.bleh0 - Wednesday, November 13, 2013 - link
I'm just waiting mainly on notebook Kaveri and whatever they can put Mullins into.tipoo - Wednesday, November 13, 2013 - link
I wonder if AMD could ever go the eDRAM route? Bandwidth seems to be the limitation on APUs. Lack of fab space for it maybe?SaberKOG91 - Thursday, November 14, 2013 - link
This has more to do with three other problems: caching latency, RAM bandwidth, and non-uniform memory addressing for CPU and GPU.AMD has had significantly higher memory latencies on all levels than Intel for a little while now, and we are hoping to see this improve in newer designs. This has the greatest impact on performance for CPU and GPU as it causes the processor to become "data starved" and have to wait for data to be retrieved.
RAM bandwidth is another factor, as highly parallel workloads end up consuming and generating significant amounts of data. While some can be done by developers to alleviate these issues, the root problem is the limited throughput of DDR3 over GGR5 for these kinds of workloads (that's why both XB1 and PS4 have unified GDDR5).
Having a uniform memory space is important because it eliminates the need for copying between different memory pools, as well as pointer conversions. This alone can have a pronounced affect on high utilization scenarios.
The problem with eDRAM is the die space and power usage. Intel got away with eDRAM in their processors because they are running at much higher power and thermal envelopes than mobile parts. They also dodged the die space issue by having the eDRAM as a separate die in the same package. AMD is trying to unify their product lines so that they can focus more on developing new architectures. Adding something special to a few different models won't help them in the long run. This was really more of a marketing gimmick for Intel as they try to become recognized as a workstation level graphics solution.
Mugur - Thursday, November 14, 2013 - link
Xbox One has DDR3.SaberKOG91 - Thursday, November 14, 2013 - link
Thanks! I sort of remembered that one, but should have checked to be sure. DDR3 + eDRAM vs GDDR5 on the PS4. It would be interesting to see a performance comparison, but I tend to side with GDDR5 or HMC as better memory solutions.fteoath64 - Saturday, November 16, 2013 - link
Using eDram will remove most of the ram latencies but it comes with great cost in die-space and power usage. You need x-MB of space and a controller logic to interface the bus. The Xbox 360 cpu had 10MB of eDram which solves its problems. I can predict that 12MB or 16MB of eDram for AMD can do so as well. The eDram of the Iris Pro-e takes 20 watts for 128MB. So 16MB will be less than 3 watts additional. That might be the way.iwod - Wednesday, November 13, 2013 - link
Why do they not make a network processor SoC with PUMA? Something for NAS and other similar devices.SaberKOG91 - Thursday, November 14, 2013 - link
This would be a good place for a G-series version of Mullins.Lolimaster - Wednesday, November 13, 2013 - link
For the massive gains in performance/watt it seems better IPC and more gpu core at lower clocks.192SP on the new low power APU's.
It remains to see if they finally added dual channel support which bottlenecks Kabini/Temash hard regarding 3D games.
aryonoco - Wednesday, November 13, 2013 - link
Me thinks that Mullins could be a perfect SoC for fanless Chromebook. Can AMD pick the phone up and call Google please?milli - Thursday, November 14, 2013 - link
Is it possible that AMD switched to TSMC's 28HPM process? It would be the easiest way to increase speed without sacrificing power. Which process was AMD using for Temash/Kabini? 28HP or 28LP?ddriver - Thursday, November 14, 2013 - link
Finally something that may be a reasonable choice for a tablet.Hrel - Thursday, November 14, 2013 - link
dockport sounds cool. that's not AMD exclusive is it?thunderbolt has licensing fees that make it FAR too expensive to be practical. So I'd like to see EVERY laptop made have this dockport feature. As desktops fade away the need for a cheap, standardized docking technology is ever increasing.
kyuu - Thursday, November 14, 2013 - link
If they can actually get an OEM to put Mullins in a tablet, it sounds fantastic. But I'm afraid it's going to be Temash all over again, with zero tablet wins and only a couple of mediocre netbook-ish design wins.If Vizio updated their tablet with Mullins and priced it at around $500, I'd be all over that.
kj_anandtech - Tuesday, November 19, 2013 - link
agreed!!! the vizio tablet would be a good execution, but you would need to add 802.11 A/C, 3G/4G/LTE, accelerometer, gyroscope, GPS. The Dell Venue 11 Pro's have similar equipment/features, and can range from BayTrail-Celeron-Pentium-Core. As I understand the Celeron, Pentium, Core and are all Haswell architectures. No availability/pricing on the Haswell yet. The dell venue pro also have replaceable batteries. The MSI W20 3M is another alternative with Temash that can be upgraded.azazel1024 - Thursday, November 14, 2013 - link
It'll be interesting to see how they perform.However, it sounds like at best Beema will be a fair amount ahead of Bay Trail in performance. Mullins will be about on par with Bay Trail in terms of power consumption.
Oh, but wait, Beema generally IS NOT in the same market as Bay Trail, unless you are looking at the laptop/desktop Pentium and Celeron parts...in which case the faster Pentium Bay Trail parts are likely as fast or a little faster than Beema will be...with a lot lower power consumption.
Mullins is likely to still lay Bay Trail a lot in CPU performance, but consume roughly the same power.
It moves the markers for AMD a lot, but it doesn't move them much in comparison to Intel or ARM.
Keep in mind, unless Intel changes its mind, these won't even be going up against Haswell ULT (massively faster) or Bay Trail-T, D or N. It'll be going up again Broadwell ULT and Cherry Trail-T, N and D based on AMD's release dates.
azazel1024 - Thursday, November 14, 2013 - link
Uh, Saber...one point. Most/all similar power envelope Haswell chips blow AMD mobile chips out of the water in graphical performance. Desktop is a little different, but even there Intel isn't that far behind a lot of the AMD APUs.Mobile though, especially low power Mobile, Intel has around a 2-4x CPU performance advantage and a 1.5-2x GPU advantage (Kabini versus Haswell ULT).
AMD pretty much only wins in graphics on the desktop and only if you don't care about power consumption and much lower performing CPUs going with those GPUs.
Conficio - Friday, November 15, 2013 - link
Wow, DockPort, if priced right could be a deciding factor for me between an Intel and AMD laptop.Also, a standard in terms of physical port placement would go a long way to have independent docking stations that hold the device (either with lid open or with lid closed and external monitor only.
Although, even more I'd like a (software) solution that ports into a desktop, so I can use the resources (files and programs) on the laptop integrated into my desktop seamlessly. Ideally combined with a cloud (really ideally private cloud) solution, making the files just available faster, then over the network.
AlB80 - Friday, November 15, 2013 - link
http://www.x.org/wiki/RadeonFeature/#index6h2Bonaire, Kabini and Kaveri has the same 3d core revision - GCN 1.1.
parkerm35 - Tuesday, November 19, 2013 - link
If AMD had access to intel 22nm, they would be well and truly licked.