| 42 posts found | |
|---|---|
|
10/07/12 6:33:14 PM#21
Originally posted by Quizzical
Not quoting this, too long(did not read it all i think too confusing) and buzzwords are killing meh... edit: bleh some one commented had to quote... You're view or Moores law is way too text book, die shrinks do not work as simply as that, since SiO replacements methods do no transfer immediately to smaller scale fabs, also many aspects of the die cannot be minimized, control asics, cache, and other elements do not scale as much as the transistors, also when you see a 32NM fab process it does not means the smallest gate transistor is 32NM
Also the "NM" does not matter, the number of nano meters in the fabrication process actually has nothing to do with the size of the transistors ;) The XX nano meters you read on the press release is the average distance between identical discrete SRAM elements, that's it, In Intel's 65NM process for example, the average gate size of the MOFSET transistor about 30-35NM depending on point of time in the cycle, while the overall length of the transistor was around 160-180nm. High school computer science was quite long time ago but power consumption of the processor is more dependent on the design of the processor it self(e.g. implementation of sleeping gates), and the chemical process of the silica it When any semi conductor manufacturer switches to it does it to increase the yield per wafer, in many cases actually have the same power consumption, or even higher one due to leakage, and not being to implement some chemical process into the higher resolution lithography. If you look at mil-spec chips some of them are still made at the 300nm range, with most of them today made around the 120-180 line of processes, yet they achieve much better power consumption, noise / EM internecine resistance, and even higher logical density in some cases than commercial IC's, when you care less about the cost per unit, and can afford to stick with a single process and perfect it you get much much better results. Intel has no reason to do so, since if can shrink your die size from 10mm to 5, you get 4 times more cores per wafer(based on a 300mm one). The N2800 has a 6.5W and the N2600 has a 3.5W TDP, also TDP is not power consumption so not sure why you mentioned it, the N2800 has a power consumption of about 8-10W when its under high load, the N2600 has about 5-7W, thats the one they released for tablets, and ultra thin netbooks, the D series was meant for low power desktops, and cheaper stuff, and has 12-14W power consumption. And again honestly you're view of the whole market is way different than what most consumer want, or need, i do not need a tablet i can play Crysis on, even if i had one i would not do it, if i have enough time to play Crysis ill do it on my PC, Laptop(Gaming one), or Console. Tablets, Smartphones, Laptops/Desktops, have their own role, and there is no need for any more overlap than the basic ability to access my data, and my basic workspace.
|
|
|
10/07/12 6:36:55 PM#22
AMD has a prototype tablet using a trinity A6 APU running windows 8. As of right now nothing really looks to come close to matching it in performance. I believe it was built by Compal if you want to check it out.
|
|
Originally posted by Wizardry Wait until you see what the next generation of tablet chips can do. It's going to be a huge jump when you have chips actually meant for tablets, rather than just using a cell phone chip (not enough performance) or a laptop chip (too much power consumption). As for operating systems, I expect Apple to eventually be reduced to a niche player, just as happened with desktops and laptops and is far along the way in happening with cell phones. Apple brings some important innovations, but then within a few years, competitors are able to match them with cheaper prices and more open platforms. I somewhat expect Google Android to end up as the dominant tablet OS, for basically the same reasons that it has it has become the dominant cell phone OS. Windows 8 tablets will also occupy a niche, with their main selling point that they can run all of your normal Windows hardware. You can use it like a tablet for web browsing, then plug in a keyboard and use it like a laptop and run all of your normal laptop software. As some people keep insisting, that's not really what people want in a tablet, and to a considerable degree, that's true. That's why Windows 8 tablets will only occupy a niche (for the people who want a tablet that is also a functional laptop), and the vast amount of software won't make Windows 8 the dominant tablet OS. Wandering off in a different direction, I think Nvidia has ambitions about Tegra-based laptops, eventually with discrete GeForce cards. They haven't announced it, and it's really just speculation on my part. But they're a graphics company, and need to survive in a world in which integrated graphics has already eaten up the low end of graphics and is moving toward eating up the mid-range. And now that integrated graphics are built into the same chip as the CPU, you can't do integrated graphics if you don't do the processor. |
|
Originally posted by GrayGhost79 ULV Ivy Bridge crushes it in CPU performance, though it loses badly in GPU performance. But they both suffer from the same problems: 17 W is really too much for a tablet. And they both have a bunch of bits burning power without doing anything productive in a tablet. |
|
|
10/07/12 7:39:55 PM#25
Originally posted by Quizzical The battery life was actually pretty good. The GPU performance is why nothing comes close and the GPU is the biggest limiting factor with tablets. I'm not saying there isn't a lot of room for improvement but I truly think the Llano/Trinity route is the best hope for viable tablet CPU's or rather the APU route in general. |
|
|
10/07/12 9:10:32 PM#26
I have an original iPad. I have a not-insignificant list of gripes about it, but really, "Not fast enough" is not one of those gripes. I don't think the "next generation of tablet chips" is what the tablet revolution is waiting for to take off. The tablet revolution has already happened. It wasn't CPU horsepower that made it happen - it was a shift in thinking about the purpose of a tablet. It had failed for years because everyone just assumed that tablets would replace a laptop and take over many of the same functions. Apple may or may not be a significant player in the future - they don't really have a great track record of maintaining a lead in anything, but they have an outstanding record of innovating and capitalizing on that innovation (oddly enough, that also was tied directly to SPJ, and he's no longer there to drive that, so time will tell if anyone else can pick that up). That doesn't really have anything to do with the future of the tablet - it has everything to do with the fact that Apple single-handedly defined the tablet, and what consumers want to do with one (although it took them two tries, the Newton didn't quite have the technology yet). A smart phone processor proved to be more than enough horsepower, and a smartphone OS more than capable enough (from both Apple and Google) - what we are seeing now are just refinements and iterations. Microsoft may be able to catch up in the same manner they typically do - by throwing enough money and resources at it that they eventually get on the boat (Office and Xbox both were years behind their counterparts, but were able to become market leaders with this philosophy) - but that hasn't always worked for them. If they continue to try to push "Replacement for the PC" - which they may be doing with their general purpose OS - they will continue to lose, and lose badly no matter how much capitol or resources they throw behind it (ala Bob or Zune). |
|
|
Robokapp
Elite Member
Joined: 11/15/09
The only luck I had today was to have you as my opponent. |
10/07/12 9:23:14 PM#27
Originally posted by Ridelynn I finally foun a line of text where I can call someone wrong on.
Horsepower?
TEAM SUBSCRIPTION. P2P > P2W. |
|
10/07/12 10:37:03 PM#28
FLOPS, MIPS, IPC, clock speed, benchmarks - pick a metric relating to general processing performance. Horsepower. And if you really want to get technical, there is a direct conversion of watts to horsepower as a unit of power (1hp = 745.7W), and you can calculate, based on the Performance per Watt (an important metric in today's standards) to actually derive a relevant horsepower statistic for a piece of silicon, rather than just a straight conversion of power consumption or TDP. However, I am using it in a general sense to convey the amount of processing power available. |
|
|
10/07/12 11:11:15 PM#29
CPU power is so far from holding back tablets and smartphones this is laughable. Do you realize even the majority of the most advanced games on iOS still run on the iPhone 3GS and the first iPad will run them even better. They will be scaled back a bit, but the point is nothing will test the current A6 until well after it has already been replaced. Remember, developers can't instantly produce boundry breaking software moments after a new device is released. |
|
Originally posted by Fendel84M And the most advanced games for NES still run well on my very old NES. The "most advanced games on iOS" aren't anything remotely resembling advanced, as neither the hardware support nor the software support is there. In constrast, the "most advanced games for tablets" will likely be getting a lot more advanced soon. Would you argue that there was no need for the SNES or Sega Genesis (or any subsequent consoles) on the basis that NES games ran fine on an NES? Unlike iOS, Windows 8 has a lot of advanced games that will run. For example, nearly all of the games listed on this site, though many will be unplayable unless you plug in a keyboard, mouse, and/or gamepad. And it's not just games, either; do you think Apple's war on Flash is because Flash would run just great on various iDevices if only Apple would enable it? Slower software makes a need for faster hardware far more obvious. |
|
|
10/08/12 5:07:50 PM#31
None of these games can run on Windows 8 RT which is destined for tablets :P The tablets that will actually run Windows which is not windows 8 RT will cost much more money than what is reasonable to spend on a tablet, the prices will probably be at least around the ultra book price range(900-1100US+). Windows 8 RT is great for tablets for people that need a decent office app, and satisfied with running metro or .NET micro edition apps only(the latter needs a "hack" to actually run). Also since you cannot run any unsigned applications on Windows 8 RT, and the fact that it will be impossible to change the OS due to the mandatory UEFI secure boot option i doubt any one that owns an android based tablet will want to switch to it other than to get a better office suite.
|
|
|
10/10/12 11:15:18 AM#32
Originally posted by Wizardry I myself am not a fan of eco-system that is apple, there are things I like about IOS but mostly the eco system and fan boys drive me nutz! |
|
Originally posted by DOGMA1138 http://www.newegg.com/Product/Product.aspx?Item=N82E16834152278 Sure doesn't look like ultrabook prices to me. Do you think Windows 8 will be $400+ more expensive than Windows 7? For that matter, you'd be able to buy that tablet and then install Windows 8 on it if so inclined. Certainly, if you want ultrabook hardware, then you'll pay ultrabook prices. But that includes $225+ for the CPU alone, as opposed to the $40 that AMD charges. But returning to the original premise of this thread, neither that APU nor ULV Ivy Bridge are really what you want in a tablet. Intel's Haswell will be the go-to chip if you want an ultrabook experience in a tablet form factor, while AMD's Temash will be the thing to get if you want something more budget-friendly, or perhaps if you want video drivers that actually work. Unlike any x86 chips on the market now, those will be architectures designed from the start with tablets in mind. We will get baby steps in that direction before the real solution comes, though. AMD launched their Z-60 APU (basically the same thing as linked above, except with USB 3.0 and a 4.5 W TDP) yesterday, and Intel is reportedly going to ship very low quantities of a 10 W bin of Ivy Bridge. |
|
|
10/10/12 5:08:37 PM#34
A tablet that does what I would use it for doesn't need to run Windows. For me that negates any value in a tablet ready Win8.
The only thing that could pull me towards using a tablet for more than basic web and apps would be a tablet + gamepad with console quality GPU and a $250-ish price point. |
|
|
10/10/12 7:23:04 PM#35
Hahaha ROFL thats a tablet? sorry when you're talking about 3.3lbs, that's not a tablet, that's heavier than my Ultrabook ;) Microsoft Surface, and Surface Pro will be good examples, the Samsung Slates although the current ones are still a bit heavy(still under 2 lbs) are good example too... Nothing stopping you from buying a "Tablet" today with 5-6hours battery life(while working/surfing w/ WLAN) and with much better performance than that MSI poj, heck the Samsung Series 7 slate will give you a Core I5(ULV ofc), a bigger screen, and under 2lbs all in one package. The problem is the price... And again with the name calling.. :P I don't want tablets to be heavy, i don't want tablets to have fans in them, i don't want tablets to be bigger than 7-8", i want cheap devices(300-400" or under), that i can turn them off, forget about them and they will still work.... AMD's does not have a real SOC, Intel does, and honestly even then i would still go for an ARM SOC any time of the day. Running windows on a Tablet is the lazy way out, i do not use a Tablet the same way i use a PC, i should get applications that allow me to take advantages of the usage form, form factor, mobility, and user interface instead of selling me the same old junk saying that i can always connect a keyboard to it. The app on a tablet needs to give you the same functionality as it does on PC while providing you with a complete new interaction experience, and allow the user to actually take advantage of using a tablet and not to minimize the "disadvantages" of using one. Any thing else is just lazy and both from a developer and consumer point of views is completely unacceptable.
|
|
|
10/10/12 7:32:19 PM#36
Originally posted by lizardbones If you have not. . you need to try the swype keyboard. Now that I am used to it I have at least doubled (if not more) my "typing" speed on my phone. Wa min God! Se æx on min heafod is! |
|
Originally posted by DOGMA1138 Well yes, that's the whole point of the thread. The hardware isn't ready. Hondo isn't an SoC because it's a laptop chip crammed into a tablet. And that's a problem, for a lot of reasons. And it's a problem that Temash will fix, except that it isn't ready. And, of course, Ivy Bridge isn't an SoC, either. Haswell will have an SoC variant, but that's not ready yet, either. Clover Trail Atom is an SoC, but with graphics performance that is absolutely dreadful. You can argue that you don't need that much graphics performance, but you do at least need a little. Silvermont Atom will massively overhaul the whole Atom architecture, but that's not ready, either. Plenty of ARM vendors offer an SoC, of course. But Windows 8 needs x86, not ARM. Windows RT is the ARM version, and that has problems of its own, such as a complete lack of software support. ----- For the benefit of others who may be confused, SoC = "system on a chip", meaning one chip with the CPU, GPU, memory controller, SATA ports, USB ports, and everything, rather than spreading those functions across multiple chips inside the device. You can do a lot more in a device if you can have a bunch of separate logic chips to handle different functions. That gives you a lot more versatility, as you can make independent choices of the CPU, GPU, SATA controller, USB controller, and so forth that you want. It also has production advantages, as the various chips can each be made on process nodes well-suited to them, and if one chip is defective, you don't have to throw out all of them as you would if you combined them all into one huge chip. So there's no real point in having an SoC for a desktop, when it's easy to scatter a bunch of chips across a 12" x 9" motherboard. But an SoC is very desirable for tablets or cell phones, as having everything in a single chip means you don't need a bunch of space for separate chips. It means you don't have to cool a bunch of separate chips. It means you don't have to burn a bunch of power letting separate chips communiate with each other. Burning several watts to allow various chips to communicate with each other doesn't matter much in a desktop, but it's a big problem in a tablet when you want the entire device to have a total system power consumption of only several watts. |
|
|
10/10/12 8:13:42 PM#38
The graphic performance of the Atom SoC's is on par with other tablets, i have no issues playing andorid games, flash, and Full HD videos on it at all. I don't need more than that, i'm not going to play Crysis on my tablet, and tablet games are looking good enough for what they provide. Take a look at the PSVita, or even the 3Ds they are bearly selling hardware or games even tho they support much better graphics than most mobile phones. For the type of games that people can play on the go thats sufficient, and as poor as the Atom SOC(Power VR 5XX) GPU is it is still much better than what AMD/Intel used to have for what you would need on a tablet, a 6250 is just a horrible GPU it can bearly play high bit rate HD films, no flash acc, and it's raw performance are not that far from a Power VR, i don't care if it supports SM3, 4, or 56 since it just can't backup it's on paper specs with real world performance. |
|
|
Ah, Cedar Trail/Clover Trail uses PowerVR graphics instead of Intel graphics. I hadn't caught that. Reviews found that it performs very poorly in general applications (the Radeon HD 6250 that you malign gives several times the performance), but presumably it has fixed function blocks to handle things like video decoding. I don't know if it will be able to handle the standard Windows desktop, which is something earlier versions of Atom graphics choked on. That presumably can't be done in fixed-function blocks like video decoding can.
|
|
|
10/10/12 9:14:06 PM#40
All "GPU"'s which Intel put out for the Atom platform were PowerVR SGX based, the newer ones are quite similar to what you get in a PSVita, or an Ipad(3nd/3rd gen) these days.. The GMA500/600 which is what most people had in their Netbook could run windows XP well, and run Linux even with desktop effects with more than reasonable performance. PowerVR 545 at 400/650Mhz which is what Intel clocks it to these days, will give you pretty much the same fill rate(2-3.5Gpixels/s) that a 6XXX will which is the only really important factor for desktop rendering.
|
|