Windows tablets

well intel is claiming that a 7.8w haswell is as fast as a 17watt ivybridge. If thats true then we could see some truely great windows 8 pro tablets in early 2013.

http://fudzilla.com/home/item/28703-haswell-at-7w-does-job-of-ivy-at-17w


I'm still hopefully that surface pro will ship in febuary with one of these beauties in it. It could really put the tablet market on its head.

8watt x86/64 bit system capable of decent gaming and running all windows apps. That be nice.


Even better would be able to adjust the TDP of the part to 17watts when you need the performance

Holy sh*t, that would be awesome! And I think will be my next pc :)
 
I'm still hopefully that surface pro will ship in febuary with one of these beauties in it. It could really put the tablet market on its head.

8watt x86/64 bit system capable of decent gaming and running all windows apps. That be nice.

I know I said this before, but I really don't believe Surface Pro will get Haswell, but rather a <10W TDP variant of Ivy Bridge. Surface Pro is coming 3 months after October 26th, which puts it at late January. The <10W variant of Ivy Bridge is coming early 2013, some say its in January as well. The new lower TDP chip won't provide better battery life in most scenarios, but it enables sleeker form factors, which could have been attributed to why MS is waiting extra 3 months.

The reason is that in Tablets, even the 17W chips throttle(meaning it goes below base frequency, nevermind Turbo). Like in the Samsung Slate 7 device, the Sandy Bridge 17W there gets slightly better than 800MHz performance. <10W would mean it may have similar performance, but at least the chip will run cooler and/or cooling system smaller.
 
I know I said this before, but I really don't believe Surface Pro will get Haswell, but rather a <10W TDP variant of Ivy Bridge. Surface Pro is coming 3 months after October 26th, which puts it at late January. The <10W variant of Ivy Bridge is coming early 2013, some say its in January as well. The new lower TDP chip won't provide better battery life in most scenarios, but it enables sleeker form factors, which could have been attributed to why MS is waiting extra 3 months.

The reason is that in Tablets, even the 17W chips throttle(meaning it goes below base frequency, nevermind Turbo). Like in the Samsung Slate 7 device, the Sandy Bridge 17W there gets slightly better than 800MHz performance. <10W would mean it may have similar performance, but at least the chip will run cooler and/or cooling system smaller.

They said its coming about 3 months after october 26th (they actually used days to measure not months) The thing is late January can easily become late febuary or even march. Remember the original annoucnement was already what a month and a half ago ? So things can slip.

The 10w variant of ivy bridge is interesting but it seems like haswell not only has much better performance per watt but is also better at power gating itself .

Intel needs a huge hit in the mobile section and the 8watt haswell could be a great kick off to showing x86 tablets can compete with arm tablets. The surface is going to be one of the biggest tablets of 2013 and so it would make sense that intel supplys haswell for it. I don't think they'd need a huge amount up front anyway , january / febuary is after the holiday sales bump , so they would need a few million at most and the supply curve can follow the ramp of haswell


I think we might find out more info tommorow anyway , the surface cakes said 9:14 as the time . I don't think its far off to believe we can hear an announcement with intel for the surface tommorow during idf
 
no chance that there will be surface variant with Amd apu? or that hasyell already have good 3d performance for game?
 
no chance that there will be surface variant with Amd apu? or that hasyell already have good 3d performance for game?

Kaveri should have quite a bit better 3d performance than Haswell, unless Haswell's huge amount of stacked DRAM provides a very significant advantage.
512 GCN cores is nothing to sneeze at, unless the ULV versions come severely castrated.


That said, it always feels like Intel will win no matter what. They always get all the relevant design wins so a Win8 tablet with Kaveri is highly unlikely IMHO.
 
Haven't they always so far? The lower TDP IGPs from AMD have always had smaller GPUs, not just lower clocks..

Depends on your notion of "severely castrated".

The fastest 17W Trinity has about 2/3 the number of shaders/TMU and 2/3 the clockspeed of the full-fledged 35W models, which results in about 50% of the overal GPU speed (memory bandwidth bottlenecks usually makes the performance difference a lot shorter, though).

Even if there's a 10W Kaveri with 50% the speed of a 512 GCN @ 900MHz iGPU, it should still be spectacular for a 11" tablet.
 
I agree...512 gcn cores will rock...however I do think bandwidth is going to be a big factor next time round....we really would he spoiled if 17w kaveri could play most games at dx11 medium settings @ 1080p!...very very unlikely but we can dream :)
 
Surprisingly, clove trail has sgx545 graphics @533mhz, and not 544 as most had assumed. Having seen some videos of clover trail running it seems that it's very smooth and not suffering the same issues that cedar trail had.

http://www.anandtech.com/show/6340/intel-details-atom-z2760-clovertrail-for-windows-8-tablets

these would be amazing at $500 for tablets. But it look like companys are going to try and get $700 to $800.


Rumors put the Surface Pro @ 400-500 to make and $700-$800 at retail. I'd take i5 performance over atom performance any day of the week at the same price point.
 
Surprisingly, clove trail has sgx545 graphics @533mhz, and not 544 as most had assumed.
AFAIU, given that 5?5 devices (unlike 5?4) are for Windows markets, it seems like a strange assumption to make.
 
AFAIU, given that 5?5 devices (unlike 5?4) are for Windows markets, it seems like a strange assumption to make.


What ?, TI omap4470 with sgx544 is NOT for windows market. Omap5 with dual core sgx544 is not for windows market ? Someone would need to tell Microsoft, as TI was one of the first 3 hardware platforms confirmed for windows 8 RT, along with Qualcomm and nvidia.
 
I don't market these things. <shrug>

However, IMG clearly does market SGX544 for windows system, in contrary to your earlier statement. Hence my surprise at clovertrail not using 544 which we know is marketed at windows, and as clovertrail seems to be launching with dx9 only drivers, similar to cedartrail, it would have seem more logical to go with the smaller dx9 compliant product as per TI. One wonders if there is persistent problems in getting dx10 drivers running on SGX545.

Additionally given that "clovertrail+"' aka Z2580 the smartphone variant, is rumoured to have dual core 544, one would have thought it more logical to keep the same core across the same soc family.

But then again this is Intel and they seem to make strange decisions regarding gpu selections, ie sgx540 in medfield when everyone else moved on from 540 in new designs long ago.
 
However, IMG clearly does market SGX544 for windows system, in contrary to your earlier statement.
Oh. The "SGX Series5XT Factsheet" seems to suggest otherwise. <shrug>

As I said, I don't have anything to do with such matters, so feel free to ignore my comments.
 
Oh. The "SGX Series5XT Factsheet" seems to suggest otherwise. <shrug>

As I said, I don't have anything to do with such matters, so feel free to ignore my comments.

You obviously don't have anything to do with such matters, albeit I'd prefer engineers would have more of a saying than usually in marketing decisions.

In any case SGX545 seems like an expection for windows devices with IMG IP, since apart from Intel I doubt anyone else has lisenced it. SGX544 will be used in windows devices by several other of IMG's partners and on top of that Tegra3 and Qualcomm S4 SoCs have DX9.0L3 GPUs for windows devices and I'm not sure to count in also Vivante GPU IP (while it's DX9L3 I'm not sure if any of their partners will use it for a windows device).

Frankly if I only had the dilemma between a DX10.1 SGX545@533MHz and a DX9.0L3 SGX544MP2@533MHz I'd take the latter single handed despite it probably needing somewhat more die area for quite a bit more performance of course. How many DX10 games with full in game settings at decent framerates is a 545 really going to master in a typical native tablet resolution even if it would have WHQL'ed DX10.1 drivers?

With the price those Covertrail based tablets are going to cost I don't see all too many incentives for a potential buyer to not go for a similar priced iOS tablet (with much higher performance) or a much cheaper Android tablet for which both the user has a wealth of tablet tailored cheap 3D games to pick from.

In hindsight it's not what IMG's marketing is thinking for its crop of GPU IP cores, but rather Intel's marketing department with design decisions like that. For what it's worth they may plaster as many "Intel inside" warnings on device covers if that's all that's supposed to drive sales for those products.

If now Clovertrail shouldn't come along with WHQL DX10.1 drivers for a second SoC in a row it's rather IMG's driver team that deserves to get shot.
 
AMD recently launched their "AppZone" for Android apps. Apparently based off BlueStacks' emulation tech wrapped as a proprietary AMD "app player". Which is a bit weird, since I don't think I've seen a single Windows 8 tablet announced with an AMD apu in it.

Loads of coverage available via Google.

Edit: Or maybe I'm just not paying attention. I guess if Intel has their own app store, AMD should have one too. :)
 
Last edited by a moderator:
Back
Top