french toast
Veteran
Thanks for breakdown.
Hopefully that is not the foundation for antutu cpu benchmark then..it is odd that they dont specify just what their benchmark measures. .its been round long enough now.
Intel's smartphone platforms
- Thread starter tangey
- Start date
Thanks for breakdown.
Hopefully that is not the foundation for antutu cpu benchmark then..it is odd that they dont specify just what their benchmark measures. .its been round long enough now.
The perceived wisdom has said that although next get tablet SOCs from Intel (Bay trail) are using Intel in-house graphics, that it was thought highly unlikely that intel would be able to practically include its own graphics IP in the next gen smartphone SOC (merrifield).
Does the above need a rethink given the roadmap chart below, or is the chart just plain wrong ?
http://pc.watch.impress.co.jp/video/pcw/docs/569/575/p2.pdf
I note the chart says the tablet clovertrail SOC has SGX544MP2, whilst my understanding is that it has single core 544.
Does Intel really have Graphics IP ready at for inclusion on a 22nm SOC that has decent performance within the smartphone power envelope ?
Does the above need a rethink given the roadmap chart below, or is the chart just plain wrong ?
http://pc.watch.impress.co.jp/video/pcw/docs/569/575/p2.pdf
I note the chart says the tablet clovertrail SOC has SGX544MP2, whilst my understanding is that it has single core 544.
Does Intel really have Graphics IP ready at for inclusion on a 22nm SOC that has decent performance within the smartphone power envelope ?
D
Deleted member 13524
Guest
If that graph is correct, they have to.
Nonetheless, looking at the cross-platform results we're seeing in the latest 3dmark and gl/dxbenchmark, the performance/watt of ivybridge seems to be in the same order of magnitude as the powervr solutions, at least the series 5. So why not try to use their own solutions everywhere?
BTW, I suggested this exact approach for Intel mobile GPUs about a year ago (dropping PowerVR alltogether and only use their in-house solutions).
Needless to say, my question was gangbanged with a synchronous "NO! Never gonna happen!", as with most "what if" questions posed in the forum.
Nonetheless, looking at the cross-platform results we're seeing in the latest 3dmark and gl/dxbenchmark, the performance/watt of ivybridge seems to be in the same order of magnitude as the powervr solutions, at least the series 5. So why not try to use their own solutions everywhere?
BTW, I suggested this exact approach for Intel mobile GPUs about a year ago (dropping PowerVR alltogether and only use their in-house solutions).
Needless to say, my question was gangbanged with a synchronous "NO! Never gonna happen!", as with most "what if" questions posed in the forum.
Isn't Clovertrail using 545? At least it is according to Anandtech.
That casts some doubt on the diagram I'm afraid.
For tablets yes; going though into the <1W-2W TDP range would need a from ground up GPU design.
It must have been way longer than a year. IMG is not all out obviously. In any case since you've hit 50% of Intel's so far minority report market share we'll nominate you for this year's [glitter]TOLD YOU SOOO[/glitter] award and you'll get half the statue
I don't know, but refreshing to see someone wait a while after registering before posting. Perhaps 8 years might be being a bit too cautious ?
This site pretends it's PowerVR SGX 544MP: http://neogadgetreview.com/merrifield-intel-and-the-first-benchmarks.html
Reported also here: http://www.guru3d.com/news_story/intel_merrifield_soc_benchmark_scores_online.html
I guess they are all wrong and what they call Merrifield is in fact Z2580
Reported also here: http://www.guru3d.com/news_story/intel_merrifield_soc_benchmark_scores_online.html
I guess they are all wrong and what they call Merrifield is in fact Z2580
Intel has a rogue license as I remember, will be interesting if merrifiled has rogue.
I get the longtime lurker award I suppose!
I get the longtime lurker award I suppose!
In the past its been suggested by many including me that Intel holds a Rogue licence, however such a licence has never been announced by either Intel or IMG. Which doesn't in itself mean much as the last Intel related announcement we had from IMG was 6-7 years ago as I recall.
I don't know what Merrifield has in terms of graphics. I have been working under the assumption that Merrifield will retain IMG graphics regardless that Bay Trail is going with Intel Gen graphics. Only Intel knows if they have a capable power/performance/area graphics solution for mobile phone format with their own gen graphics on 22nm.
Asus ME302C 1080p tablet with Intel Clovertrail+:
http://gfxbench.com/compare.jsp?D1=Samsung+GT-I9500+Galaxy+S4&D2=Asus+ME302C&cols=2
http://gfxbench.com/compare.jsp?D1=Samsung+GT-I9500+Galaxy+S4&D2=Asus+ME302C&cols=2
Allright is relative when a tablet barely gets today's mainstream smartphone GPU performance. Of course might it get more "interesting" with their next generation, but my gut feeling doesn't tell me it's going to be anything that's going to catch the competition with its pants down; rather the opposite.
It's not the fastest Clover Trail+ device though. It's using a Z2560 which clocks the SGX544MP2 at 400MHz, compared to the 533MHz on the Z2580. There's also the 300MHz clocked Z2520.
An important distinction to make, Cedar Trail/Clover Trail supports Windows, but Clover Trail+ is only for Android, just like Medfield.
Interesting; I wouldn't suggest that any kind of MP2 at 533MHz would be a problem in terms of power consumption even for a smartphone plaform nowadays.
While it makes sense under the light that Intel will use soon it's own GenX GPUs for tablet designs, it forms a nice oxymoron considering how much faster the Clovertrail+ GPU actually is compared to the other two.