AMD Beema/Mullins Launch Thread

[AtenRa]

What the heck are you talking about ARM licensing for. Pay attention. I AM TALKING ABOUT THE OPERATING SYSTEM. Windows 8.1 versus android/iOS.

Windows 8.1 is a very poor touch driven mobile operating system. Period. Android apps are designed for tablet form factors while Windows 8.1 apps have horrible DPI scaling and are designed for everything. Being designed for everything is great for compatibility but GARBAGE for touch driven mobile specific apps. Using the ipad for a long period of time drove this home for me - using Win 8.1 on a tablet in a touch driven fashion is downright PAINFUL compared to using android or iOS. Heck using the gmail or youtube apps on the iPad are great to use. I dare say better than desktop. But using that crap in a browser window on an 8 inch windows tablet with no mouse/kb is just LOL compared to the ipad. Win 8.1, great for desktop. Terrible as a touch driven mobile OS, mainly because Win 8.1 apps aren't mobile specific.

Now Windows 8.1 is great for an ultrabook with a touchpad or a desktop with a keyboard mouse. When you're actually, you know, MOBILE, the touch driven aspect of windows 8.1 is garbage. So that is why I said AMD needs android compatibility, they do not have it. ARM licensing has ABSOLUTELY NOTHING TO DO with being android compatible. Android is free. Android costs nothing. AMD has to implement it to get into that market. And they don't need ARM SOCs to do so.

11.6" Windows Tablets are going to be the next Netbooks. You will have a docking station at Home and use the Tablet as an ordinary Desktop. When you will leave the House/Work you will use it as a Tablet. So AMD right now doesnt care about Android compatibility that much. If they want to enter the Android 7-8" market they can do it with an ARM based SoC design, not x86.

Ohh one more thing,
You can run the Bluestacks software on AMD devices and run Android applications within Windows. So, you can have Win 8.1 Tablet and run Android apps today.

- Enables PCs to run both Windows and Android applications at the same time
- Share files between Windows and Android
- Eliminates the needs for OEMs to run separate driver test and support processes for the two OS’s

 

[blackened23]

Apparently you haven't heard, but bluestacks, being an emulator, is garbage. It is a far cry from real android when many applications do not work and/or run erratically and the performance for a simple mobile operating system is slow. Also requires a lot of configuration and using bluestacks own interface for installing apps (unless it changed). It doesn't even use the native android interface last I heard. Come on. Bluestacks isn't a proper android solution. Native compatibility is, and apparently AMD doesn't want to collaborate with google to get that done. Certainly an ARM SOC isn't' required to do so. If you want to say AMD should stick to x86, say that, but bluestacks is a joke for actual android use.

 

[Abwx]

Legitreviews review and article :

http://www.legitreviews.com/amd-beema-mullins-apu_139863

 

[AtenRa]

Apparently you haven't heard, but bluestacks, being an emulator, is garbage. It is a far cry from real android when many applications do not work and/or run erratically and the performance for a simple mobile operating system is slow. Also requires a lot of configuration and using bluestacks own interface for installing apps (unless it changed). It doesn't even use the native android interface last I heard. Come on. Bluestacks isn't a proper android solution. Native compatibility is, and apparently AMD doesn't want to collaborate with google to get that done. Certainly an ARM SOC isn't' required to do so. If you want to say AMD should stick to x86, say that, but bluestacks is a joke for actual android use.

Its all about resources, they dont have currently the resources to engage in to entering the Android market with x86 designs. Look how badly Intel is doing. Larger format Windows based Tablets is where they can make a difference and that is where they aiming.
But perhaps some OEMs could create a 8" Dual Core Mullins Windows tablet.

 

[tential]

He is making a very valid point in response to your post. You say that Mullins/Beema is not for tablets and that AMD needs to get into ~7" tablet form factors... He made the very logical point that they can simply use ARM to get into that form factor if x86 can't get them there. Perhaps before yelling at others for not listening you should listen to yourself

I personally agree with the sentiment that unless GCN 2.0 is a huge gain in efficiency and designed for mobile first like Maxwell then it probably won't be worth shoe-horning into an ARM SOC and trying to get it into products. I think AMD should just try to keep the Cat cores moving in the right direction as they are already close to ARM power levels and give us Windows compatibility.

He's saying that AMD needs to get android compatibility with their chips so they can also compete in the Android market. Not sure why this was so hard to understand but I hope that clarifies it for you.

 

[tential]

Its all about resources, they dont have currently the resources to engage in to entering the Android market with x86 designs. Look how badly Intel is doing. Larget format Windows based Tablets is where they can make a difference and that is where they aiming.
But perhaps some OEMs could create a 8" Dual Core Mullins Windows tablet.

http://blogs.computerworld.com/tabl...betting-our-tablet-future-android-not-windows

"He said that between 80% and 90% of the 5 million tablet processors Intel shipped this year are for Android tablets, with the rest for Windows tablets. He said that Intel hopes for big growth in tablet processors, and has a goal of selling processors for 40 million tablets -- and 80% to 90% of them would be Android."

So it seems intel was able to sell their tablet processors there. If AMD processors are that much better than intel's offerings, then all AMD needs to do is collaborate with google, get android support and they'd be able to easily snatch this market from intel.

If I see free cake, I'm gonna take it. Why won't AMD?

blackened has a valid point. If AMD has a better offering than intel and better pricing, and there is clearly a market for their product on Android, why not try to take that market from intel?

 

[blackened23]

Its all about resources, they dont have currently the resources to engage in to entering the Android market with x86 designs. Look how badly Intel is doing. Larget format Windows based Tablets is where they can make a difference and that is where they aiming.
But perhaps some OEMs could create a 8" Dual Core Mullins Windows tablet.

They aren't selling poorly. They're selling very well with several models being consistent top 10 sellers at amazon. I literally have no idea what you're talking about here because the Dell Venue 8 inch sells out nearly instantly anytime it goes on sale and it has sold ridiculously well. Many BT tablets have sold incredibly well. Don't try to twist the facts here.

I think you're really confused here. Yes, intel is selling BT cheap because ARM SOCs are cheap. They are not making much money off of Bay Trail. However, the bulk of their loss from their financials was from R+D expenditure. As i'm sure you're aware, investing in future silicon technologies requires spending money on R+D. If you don't spend money you cannot make money. As it turns out, the bulk of the loss was just that. Investing in the future of mobile. That requires spending money. Obviously AMD doesn't have the warchest that Intel has in this respect. Intel has tons and tons and tons of cash on hand so frankly, they can do what they do no problem. So to compare intel's situation to AMD's? They're far different. They're investing heavily into R+D and pushing hard to compete against ARM SOCs. The last piece of the puzzle was android x64 compatibility which was completed last month. And they have full chrome OS compatibility.

Now i'm not sure what this has to do with AMD having android compatibility. That was and is my point so i'm not sure why you're trying to talk about intel's BT situation here. Intel's situation isnt' AMD's situation. Intel has tons and tons of cash, AMD doesn't. Not that this requires anything except AMD's software engineers to collaborate with google: that certainly doesn't require millions in R+D. Besides which, Android is the bigger market for mobile. Android is better for mobile tablet form factors. Period. AMD should be collaborating with google to get that done just as intel did, IMO.

It's fairly simple here, let me spell it out for you. While the 15W TDP parts will obviously be better suited for Windows 8.1 devices and that's all great, AMD obviously wants to be on low power tablets as well. At a 4.5W TDP, that is a prime candidate for 7-8 inch tablets the TDP lives up to claims. And in a 7 inch mobile tablet form factor, Windows 8.1 is not a good touch driven mobile operating system. Android and iOS are better as touch driven operating systems. So my point is, if AMD wants to be on these devices, they should get real android compatibility. Understand? Android is free, so I really don't see what the big deal is. It's not like google is going to charge AMD money to collaborate for compatibility. There's also the fact that in some countries, low cost android devices sell by the millions. Wouldn't AMD want a piece of that pie? Low cost chips running android? I dunno. Sounds like the obvious path for AMD, but I have no idea if they're planning on real android compatibility or not.

 

[cbn]

Now can we please get a two socket motherboard with igpu crossfire support and a modified dual or quad channel memory controller.

For these? What would be the point? These are low cost chips.

Assuming gaming companies transition to well threaded code to work with the playstation and xbox, 8 low powered cores coupled with crossfire igpus and decent memory should be capable of running the next generation of games, at least those designed for xbox and playstation and ported to PC. It would be great for a home server/storage HTPC solution. Look at the 8core atom motherboard on anandtechs frontpage.

A two socket motherboard? How much extra complexity and cost would this add and how well would it even run on Linux or BSD?

At least with a two socket Beema or Mullins there would be four native SATA ports right? (two from each SOC)

Now as far as people saying things like "just add a PCI-E based two SATA port controller card" to the various two SATA mobile processors (like Beema, Mullins, Kabini, Bay Trail-D, etc.) there are numerous reports of problems with add-on controllers (in general) when using Linux.

Here is an example of third party SATA controller problems from the Anandtech c2750 review --> http://www.anandtech.com/show/7970/asrock-rack-c2750d4i-review-a-storage-motherboard-with-management

The only issue that surrounds the C2750D4I since its release is the use of Marvell controllers. Users have been reporting that in Linux and FreeBSD, high intensity read/write workloads cause the controller to reset and elements to any software array are lost.

Then there is the issue of whether or not the third part IC is even supported in Linux or BSD. I've even read cases where the aftermarket controller cards are even defective and cause silent data corruption:

http://lime-technology.com/wiki/index.php/Hardware_Compatibility#PCI_SATA_Controllers

Beware!!! There are numerous providers of SiL3132-based addon cards, and a few of them (unknown how many) are known to be faulty, causing unseen (SILENT) data corruption

Based on that I would much rather have a native four Sata port situation than a two native sata port plus two SATA port controller card.

If your dual socket idea was actually workable and cost effective enough......then I will take a pair of the lowest end dual core SKUs on my board. Even dual Kabini E1-2100s would probably work for me.

 

[blackened23]

http://blogs.computerworld.com/tabl...betting-our-tablet-future-android-not-windows
If AMD has a better offering than intel and better pricing, and there is clearly a market for their product on Android, why not try to take that market from intel?

Yes, precisely. Intel is going for premium devices with their upcoming 14nm successor to Bay Trail. If AMD has the cheaper product with android compatibility, well it seems to me that AMD would have the potential to sell millions of chips for that purpose: low cost 7 inch tablets with a 4.5W TDP chip.

I dunno. I just think that AMD is trying to get into 7 inch tablets with their 4.5W TDP parts. I can assure you that Android/IOS is a better *mobile* touch driven OS than Windows 8.1 is, and there's an absolutely huge market for low cost android chips and low cost android devices in China. As i've said several times....Windows 8.1 is great. I love using it. But it's only great on a desktop or ultrabook with a touchpad keyboard. As soon as Windows 8.1 goes mobile as a touch driven OS, forget it. The apps aren't designed specifically for that (well 98% of them aren't) and the other factors such as DPI scaling make it a mess for touch driven operation. Anyway, I think AMD would want a piece of that Android pie since they're not necessarily aiming for high cost/premium products as intel is. There's the potential to sell millions of chips, but they just need real android compatibility.

 

[Enigmoid]

Legitreviews review and article :

http://www.legitreviews.com/amd-beema-mullins-apu_139863

The CPU's seem pretty much the same. Baytrail has stronger memory performance. Mullins has slightly faster ST performance but it looks like it reduces clocks for MT (3dmark physics). Mullins has stronger GPU perf. The overall score there can't be directly compared as the storage on the discovery platform is ~2-3x faster than the HP tablet.

In the Dirt gaming test the two were pretty much identical.

Mullins has a performance edge but no clue on battery like.

 

[Abwx]

The CPU's seem pretty much the same. Baytrail has stronger memory performance. Mullins has slightly faster ST performance but it looks like it reduces clocks for MT (3dmark physics). Mullins has stronger GPU perf. The overall score there can't be directly compared as the storage on the discovery platform is ~2-3x faster than the HP tablet.

In the Dirt gaming test the two were pretty much identical.

Mullins has a performance edge but no clue on battery like.

The higher memory bandwith dosnt appear to give an advantage, i guess that the 3795 was released to confront Mullins more closely given it s 1.6 base frequency....and its naming.

The two configuration have different memory sizes, Mullins has 2GB while the HP based 3795 has 4GB, dont know to how extent this has an influence in the benchs used.

Edit : Battery life should be comparable , no way the 3795 does better
than a 3740, on global tests PC mark yield nearby scores , sub scores
give a slight hedge to Mullins in the computing department, GPU being
on another category.

 

[Erenhardt]

If I see free cake, I'm gonna take it. Why won't AMD?

If AMD has a better offering than intel and better pricing, and there is clearly a market for their product on Android, why not try to take that market from intel?

The cake was real, "free" was a lie.

Intel have all the waiters collecting Tips at their table while AMD is waiting by the empty table

 

[TrulyUncouth]

He's saying that AMD needs to get android compatibility with their chips so they can also compete in the Android market. Not sure why this was so hard to understand but I hope that clarifies it for you.

You clearly didn't read all of blackened and atenra's posts or you are intentionally misunderstanding. Read it again and you will see why pointing that out was necessary.

 

[TrulyUncouth]

Its all about resources, they dont have currently the resources to engage in to entering the Android market with x86 designs. Look how badly Intel is doing. Larger format Windows based Tablets is where they can make a difference and that is where they aiming.
But perhaps some OEMs could create a 8" Dual Core Mullins Windows tablet.

They aren't selling poorly. They're selling very well with several models being consistent top 10 sellers at amazon.They aren't selling poorly. They're selling very well with several models being consistent top 10 sellers at amazon.

http://blogs.computerworld.com/tabl...betting-our-tablet-future-android-not-windows

"He said that between 80% and 90% of the 5 million tablet processors Intel shipped this year are for Android tablets, with the rest for Windows tablets. He said that Intel hopes for big growth in tablet processors, and has a goal of selling processors for 40 million tablets -- and 80% to 90% of them would be Android."

So it seems intel was able to sell their tablet processors there. If AMD processors are that much better than intel's offerings, then all AMD needs to do is collaborate with google, get android support and they'd be able to easily snatch this market from intel.

If I see free cake, I'm gonna take it. Why won't AMD?

blackened has a valid point. If AMD has a better offering than intel and better pricing, and there is clearly a market for their product on Android, why not try to take that market from intel?

Aren't a lot of intel's sales due to giving away the chip and also offsetting some of the cost through their contra-revenue program? Not that I am knocking it as it is definitely a valid way to enter a market they can't currently compete in at a profit. I am merely pointing out that AMD would also have to lose money to enter the market.

Hell, even if AMD did have the cash to give away the chips and some money to offset platform costs there would be no end game for them. They aren't coming out with a wireless chipset that can compete with Qualcomm or Intel's offerings in the coming years when even more is integrated- what would be the point of throwing away cash to enter a market they can't truly compete in?

So don't be obtuse, even if AMD's cpu/gpu performed 20% better for the same power usage they would be dead in the water for the small tablet market just because they don't have the money or integrated parts to compete.

 

[Phynaz]

Just to remind you and the rest of the forum, unlike Intel, AMD has an ARM license.

What in the world makes you think Intel doesn't have an ARM license? They used to produce ARM chips. Even if they didn't have a license do you think ARM would refuse to sell them one?

This advantage you think AMD holds over Intel doesn't exist.

 

[blackened23]

More importantly, having an ARM license means absolutely nothing for android compatibility. AtenRa suggested that an ARM license or ARM SOC is preferable for android operation. That, of course, is a completely stupid and absurd notion. Android in NO WAY SHAPE OR FORM requires an ARM SOC. If AMD wanted to pursue it, they could. Google doesn't charge money for such things, and android is free. But if they want to not be involved in a market that can bring them money, whatever.

I mean with their 15W TDP SOCs they aren't going to come close to Haswell U and Y SKUs which are obviously better performers. Despite that AMD will have a decently priced alternative that performs relatively well. At 4.5W TDP? That's the area for 7 to 8 inch tablets if AMD really meets that TDP with decent performance, then they should really get some android compatibility going for those parts. 4.5W TDP SKUs would be perfect for android; Windows 8.1 is a poor mobile OS for 7 to 8 inch touch driven devices while Android and iOS aren't. So if AMD wants to be on a 7 inch tablet, obvious conclusion is, android compatibility. But I suppose AMD doesn't want to pursue it. I'd say AMD being tied to Windows 8.1 with no alternative would be the bad business decision with the horrible DPI situation in Win 8.1 and very poor mobile touch driven development.

That said, Windows 8.1 is the obvious choice for SKUs past the 10W TDP area, although AMD would be competing with Haswell there. But at 4.5W TDP, that's tablet area, and Windows 8.1 isn't the best choice for a 7 to 8 inch completely touch driven tablet. I mean, i've tried to use Win 8.1 in that fashion and it really is a joke. Yeah it's great on a desktop or ultrabook with a trackpad/keyboard, but as soon as it goes mobile touch, no thanks. Hopefully AMD will pursue this, because I think it really would be the best business decision for their sub 5W TDP SKUs.

 

[piesquared]

The CPU's seem pretty much the same. Baytrail has stronger memory performance. Mullins has slightly faster ST performance but it looks like it reduces clocks for MT (3dmark physics). Mullins has stronger GPU perf. The overall score there can't be directly compared as the storage on the discovery platform is ~2-3x faster than the HP tablet.

In the Dirt gaming test the two were pretty much identical.

Mullins has a performance edge but no clue on battery like.

"performance edge" and "stronger GPU perf". Those are substantial understatements. Mullins crushes bay trail by at least 2x on average, and beats it in CPU tasks by a substantial margin as well. Memory performance is evidently meaningless, as Mullins does this beat down regardless, with a single channel. That's quite impressive and shows the superiority of the Mullins/Beema SoC design.

 

[Abwx]

Actualy Dirt Showdown seems to be a favourable case for BT, surely due to memory bandwith, otherwise the difference can be hefty.

http://techreport.com/review/26377/a-first-look-at-amd-mullins-mobile-apu/4

 

[Shivansps]

Actualy Dirt Showdown seems to be a favourable case for BT, surely due to memory bandwith, otherwise the difference can be hefty.

http://techreport.com/review/26377/a-first-look-at-amd-mullins-mobile-apu/4

Actually that looks like a severe CPU clock problem for the 6700T, it should not be performing way below the A4-5000. It should not be performing below the A4-5000 at all, that just mean turbos are not working while gaming.

 

[Abwx]

Actually that looks like a severe CPU clock problem for the 6700T, it should not be performing way below the A4-5000. It should not be performing below the A4-5000 at all, that just mean turbos are not working while gaming.

In dirt showdown the A4 5000 benefit from 1600MHz memory instead of 1333 as well as from Mullins being limited to 2GB in this test, that has nothing to do with turbo as shown by the benchs, besides Kabini is not TDP limited up to 15W.

 

[Enigmoid]

"performance edge" and "stronger GPU perf". Those are substantial understatements. Mullins crushes bay trail by at least 2x on average, and beats it in CPU tasks by a substantial margin as well. Memory performance is evidently meaningless, as Mullins does this beat down regardless, with a single channel. That's quite impressive and shows the superiority of the Mullins/Beema SoC design.

Yes. Bandwidth isn't really having an impact but intel has already implemented it for scale-ability to future designs. AMD Mullins is in "no-mans land" currently (similar to kaveri). Its easily fast enough for any android titles (if it gets android support) but not fast enough for any real gaming. Techreport tests showed unplayable in Bioshock or TR, barely playable in dirt. Basic tablet tasks BT is fast enough so i'm not sure where this puts mullins (and how much an advantage this really is).

In dirt showdown the A4 5000 benefit from 1600MHz memory instead of 1333 as well as from Mullins being limited to 2GB in this test, that has nothing to do with turbo as shown by the benchs, besides Kabini is not TDP limited up to 15W.

EVERY single MT test pretty much every review site has posted has shown that the Discovery tablet is clocking down under 4C load (not max turbo like 1C load). CB puts it at 1.5 ghz (equal to the a4-5000). Other synthetic tests put it well below the a4-5000.

Near perfect scaling for BT and the a4-5000. Mullins is definitely clocking down. BT appears to run at max turbo under a lot of conditions (prime 95 + furmark will really knock in down though), Kabini runs a constant speed under all conditions. Mullins most definitely is dropping to base clocks under heavier loads that do not affect BT and Kabini. The TDP (or possibly temperature target/cooling) is forcing lower clocks in MT tests for Mullins. Nevertheless AMD has implemented an extremely effective turbo, especially as heavy MT loads are uncommon on tablets.

This behaviour can be seen even in real world tasks.

It looks like for typical MT tasks 1.5 ghz is possible (good improvement). Turn the GPU on and 1.2 ghz is all you are going to get.

 

[Shivansps]

SC mem from 1333 to 1600 its not gona yield a 10fps increase, it yields something like you see in the other 2 test, those look like mem making the difference.

On Dirt it goes below because of something else, and THAT seems consistent with CB MT numbers, CB MT shows that the best case escenario for CPU only 4 threads load is 1.5ghz, dont forget that.

I was expecting higher numbers on gaming considering the higher ST perf compared to kabini, its still way too close to BT, remember that CT is coming with x4 the EUs and a arch change in just a few months, and it has a DC controller to use it too.

Just remember what happened last year, AMD lead over Intel Atoms was way wider than this.

 

[Abwx]

Yes. Bandwidth isn't really having an impact but intel has already implemented it for scale-ability to future designs. AMD Mullins is in "no-mans land" currently (similar to kaveri). Its easily fast enough for any android titles (if it gets android support) but not fast enough for any real gaming. Techreport tests showed unplayable in Bioshock or TR, barely playable in dirt. Basic tablet tasks BT is fast enough so i'm not sure where this puts mullins (and how much an advantage this really is).



EVERY single MT test pretty much every review site has posted has shown that the Discovery tablet is clocking down under 4C load (not max turbo like 1C load). CB puts it at 1.5 ghz (equal to the a4-5000). Other synthetic tests put it well below the a4-5000.

Near perfect scaling for BT and the a4-5000. Mullins is definitely clocking down. BT appears to run at max turbo under a lot of conditions (prime 95 + furmark will really knock in down though), Kabini runs a constant speed under all conditions. Mullins most definitely is dropping to base clocks under heavier loads that do not affect BT and Kabini. The TDP (or possibly temperature target/cooling) is forcing lower clocks in MT tests for Mullins. Nevertheless AMD has implemented an extremely effective turbo, especially as heavy MT loads are uncommon on tablets.

In this test both Kabini and BT works at their base frequency, hence the perfect scaling and also overall scores, Mullins is at about 1.5.

 

[Enigmoid]

In this test both Kabini and BT works at their base frequency, hence the perfect scaling and also overall scores, Mullins is at about 1.5.

No.

http://www.notebookcheck.net/Review-Acer-Iconia-W4-820-2466-Tablet.114871.0.html

BT 3740 runs Prime 95 at max turbo (1.86 mhz) stable.

There is no downclocking to base clocks.

Luxmark will run at max turbo.

Under Prime and Furmark CPU drops to 1333 mhz

Note that some of the other BT tablets show massive throttling which appears to be temperature based, hitting over 70 degrees on the SOC; the iconia is around 50.

Needless to say these are extremely intensive tests.

 

[Shivansps]

yeah the turbos, both Intel and AMD are designed to run on normal escenarios, Prime+Furmark throws that out the window.