Report: AMD Working on Radeon R9 280 Graphics Card

[renz496]

last thing i want is both nvidia and AMD cards get overpriced because of mining

 

[smeezekitty]

last thing i want is both nvidia and AMD cards get overpriced because of mining

Looks like it might be coming

 

[somebodyspecial]

FACT: Intel has lost 21% of market share in notebooks to ARM. The same will happen to desktops as we hit 64bit 8/16GB etc.

FACT: We are already at 8/16GB on desktop and x86 still rules

FACT: Most people don't have more than 8GB in their PC's.

I will bet that will change in the next 3 years or less

FACT: The vast majority of desktop software is written for X86 which makes changing architecture beyond stupid

You're missing my point. Once you have more than 4GB in ARM devices you will start to see some potent apps start to be made for them. Many people will hook these up to a monitor/keyboard/mouse at home, and for the poor people, they won't by a PC at all perhaps.

Once you have a 500w arm box there is nothing stopping them from slapping 32GB in them either. They are already using the same DDR3 1600 most people have now (on recent pc's, actually most probably have 1333 as haswell doesn't even use anything above that and faster only gets a few percent). The only things holding apps back on ARM right now is 32bit+<4GB ram, and no discrete cards. You can fix all of that with A57's. We will have a desktop ARM box next year (that is what Denver and it's competition A57's are for, see Qcom, Arm, NV's etc roadmaps they all say DESKTOP on them). Most people don't use Office or pro apps. Those people can switch on a dime, just like the 21% that left notebooks for chromebooks. The second they have a desktop version (with no windows lic fee, no Intel premium fee, free android, chrome, linux etc - Triboot? Quad with steamos? ALL FREE), meaning A57's, you can slap an NV card, 500w PSU and 32GB of memory if desired.

Switching is beyond stupid right now for a SMALL portion of the market who use PRO apps or Office etc. The rest of the people? The ones I described before, email, browse, casual games, twitter/facebook, movies and other miner stuff, they can switch right now as the 21% leaving notebooks shows. A fairly large portion of the population doesn't even own a PC and their first experience will be an a $100-200 tablet etc to get to the web and do basic tasks.

PC sales off ~11.5% and 21% of notebooks just left the market for android. What part of the situation are you not getting? Losing 21% of your market in notebooks in ~6-12 months is a VERY bad situation. PC sales dropped from 384mil units to 350mil. That's a lot of people! A large portion of the public is NOT tied to Wintel in any way they can't dump overnight. Wintel didn't respond to ARM's march for 2-3yrs which let android get to a Billion units using android. That will prove to be a very costly mistake as sheer numbers gets devs attention already and they will likely double soon. MSFT should have had WinRT out 4-5yrs ago, but they didn't recognize the new enemy (mobile, getting enough power that it could force its way into low-end then move to higher stuff as power/abilities increased). This is good for users though, as we'll get OS competition/innovation, better games/directx, maybe a windows/office price cut FINALLY etc. MSFT will have to give you the OS you really want instead of a win8 they'd like to force on you

20% of american has NO internet. Mostly because it costs too much to get on previously (it required an expensive PC, and expensive connections). But now they can get on with a phone they already own (or seemingly free on contract...LOL) or tablet. As old people die off, the only ones left without internet are younger and more willing to use technology. Many seniors who can afford it, have never used tech and hate it, but they won't be here forever. They are being replaced with a generation that grew up with ipads etc in their schools. With google (and others soon I suspect) offering free internet to poor people, they don't even have to pay for that bill in some cases. 46% of people with income less than $30,000 have no broadband. 30% with income between $30,000-50,000 have no broadband. Those people say it's too expensive or too difficult to use. They've dumbed down my dad's tablet so much anyone could use it and people are figuring this out. You can't customize much of anything in chrome/firefox etc on mobile (which is very irritating to me, but easy for newbs I guess). A full 20% have just opted to stay off the net for the same two reasons. That is changing and none of these people are tied to ANY PC. The PC market isn't growing (shrinking actually) and mobile will be 2B-2.5B units in 2 years if Gartner etc is correct.

Android already has 60% of devs making games for it, while PC's only have a little less than 40% (see GDC 2013 survey). Game makers have already spoken and will probably get worse (greater than 60% for mobile, lower than 40% perhaps on PC) soon as we hit GDC 2014 next month and they update their survey with this years results. Apps will follow once you get the box that's just like a PC, just cheaper without WINTEL in it. You talk like the war is over this month. It's just getting started and will go for years to come (this is a LONG game). Everyone making ARM socs is coming for x86's $52B in revenue and MSFT revenue also (~$83B), as surely android will cost $5 or something at some point (x 1B-2B units is $5B-10B in revenue YEARLY). Not doing this is really is "stupid", which is why they are ALL doing it. Low-end fruit was notebooks, but the next stop is desktops/servers. I never said apps will be here next week to replace windows...LOL. But each year the other side will keep getting better as games already show. Hardware first (A57/64bit/memory matching desktops/servers, big PSU's, discrete cards), then you get the REAL software apps. You can't really write software that is leaps and bounds above the hardware, you have to do the reverse so someone wants to write for the higher end hardware once you have a lot of it in the market running. Not many are writing for consoles until they hit 10mil (always the case). Same story happens on the app side. You don't write tons of software now for 32GB on android when nobody is really running more than 3GB. See the point? Don't forget phones/tablets get updated yearly or every 2yrs for many unlike a PC which might last 4-7yrs. It's much easier for ARM to move to 64bit than windows was and they will move much quicker because of the short buying cycle.

I'm sure you're right about desktops going up from 8GB (many have less today), but once you have 64bit on ARM (xmas this year or earlier as everyone races to catch apple's buzzword) there is nothing stopping them from matching your "upgraded" PC memory in the next 3yrs on desktops. They can run the same amount as you shortly. You can outrun them now, but only because they have no reason to go above 4GB currently (32bit still, except apple). That is over in ~10 months. How much market share do you think they take from x86 then? They have taken 21% of notebooks already with no 64bit, no discrete, crap software, and just getting to unreal 3 engine etc games. They will continue to take more IMHO until Intel figures out how to slow arms march into it's space. IMHO the only way to do that is to buy NV. Which is probably why Street.com just wrote they are a takeover target. I agree and have been saying it for a while. If they want to make that purchase they had better do it before AMD gets much weaker in GPU or the FTC might freak (also NV is getting more expensive, now it would cost them ~22-24B or so, a month ago probably ~17-18B). I don't like that happening in the end (prices would go up on GPU as Intel die shrinks on NV gpus kills AMD and fends off arm with good soc GPU's at 14/10nm too), but it would take a little while for Intel to fend off ARM and during that time we'd get some 14/10nm GPU's, so I'd like this for at least 5yrs But then we'd probably get screwed

 

[somebodyspecial]

http://www.bloomberg.com/news/2014-02-23/microsoft-targets-cheaper-phones-to-tap-global-internet-growth.html?cmpid=msnmoney

"At the Mobile World Congress, other companies are also announcing lower-end devices. Mozilla Corp., developer of the Firefox browser, unveiled a $25 phone prototype.

“The story of Internet-for-everyone will be written at MWC this year,” Maitre said. "

So already realizing the poor need to be on tech in order to expand the market (and to get their countries developed further, if they ignore tech they'll never develop). Unfortunately for MS most people don't want windows phone as the 96% on ios/android shows. Again they should have made this announcement 4-5yrs ago to stop android from getting to 1B+ units.

 

[InvalidError]

You're missing my point. Once you have more than 4GB in ARM devices you will start to see some potent apps start to be made for them.

Android developers who want to make their games and apps as broadly available as they possibly can still target 512MB RAM because there are still some new low-end devices shipping with only 512MB RAM and about half the devices launched in 2013 still had only 1GB RAM. It does not look like 2GB is going to become the lowest common denominator in 2014 either so we aren't going to see many mobile developers targeting 3-4GB for another 3+ years.

If you forgo unnecessary frou-frou and feature bloat, you can do a fair amount of stuff with only 1GB RAM.

 

[somebodyspecial]

You're missing my point. Once you have more than 4GB in ARM devices you will start to see some potent apps start to be made for them.

Android developers who want to make their games and apps as broadly available as they possibly can still target 512MB RAM because there are still some new low-end devices shipping with only 512MB RAM and about half the devices launched in 2013 still had only 1GB RAM. It does not look like 2GB is going to become the lowest common denominator in 2014 either so we aren't going to see many mobile developers targeting 3-4GB for another 3+ years.

If you forgo unnecessary frou-frou and feature bloat, you can do a fair amount of stuff with only 1GB RAM.

Agree with your comment they CAN do that, but I'm talking the platform evolving to desktop etc. You're thinking this is a phone/tablet market only. The point of A57/64bit is to get to a laptop/desktop/server box replacement with 16-32GB etc of the same memory we are using in your PC (DDR3L 1600 already in phones etc, DDR4 in there at some point). There is nothing saying ARM has to stay in a 4-10w envelope with no discrete card and I can't wait for the A57 jump to see what they do with basically a PC they can run at 3.5-4ghz in a PC style box with discrete. That should at least cause major fear in Wintel right? Maybe they get win9 right this time? To get Intel's market share you have to go into a 100-500watt box etc with discrete options (or even the lower stuff with APU, maxwell A57 sounds great). Intel is marching into mobile, while ARM comes for x86 on the top end.

I agree though, they can aim as low as they want. I'm just expecting them to move on shortly, much faster than Wintel did from 32bit to 64bit. Upgrade cycles on arm are much faster with many buying new devices yearly while a PC owner may hold their PC device for 7yrs (people still run 32bit today and will for a while on Wintel). We are just starting to see PC games require 64bit only. I don't know many people that have a phone over 3yrs old. Well, NONE but me...ROFL. But after 8hrs on a phone most days at work, if my cell rings I ignore it (or any phone...LOL). Unless it's work related I pretend phones don't exist after 5pm most days

I not interested much in what they can do today. The future won't be on a phone/tablet for long for ARM as sales are slowing now as the market gets pretty saturated. They want to sell you a $100-300 chip, not a $25 chip. They want to compete with Intel for everything they have and Intel wants to (has to do it?) do the same to ARM. It's just a question of who wins and opinions vary on that, but not a question of what each is doing or what their goals are. Intel wants to replace ARM (all mobile) and ARM wants to replace Intel & probably WINTEL really, as google would rather have android or something everywhere than windows. Mobile is worth less than $30B, while x86 is $52B (on Intel's side, another $89B on MS side). Samsung puts more on mobiles side, but they don't use much outside samsung stuff (just the soc, not much else as 67% of Galaxy S4 is internal). So I'm not really including their money here or apple, who again uses a lot of their own stuff and IOS.

Think BIGGER PICTURE and longer term. You can't take out a desktop without becoming one at least to some extent. You can make a game look like ultima 1 and it will run on everything including crap laptops etc, but how many want that game vs. BF4? To get Adobe etc ported or something to ARM we need an ARM PC/Laptop/Workstations just like PC's. 500w etc, 16-32GB, vid cards etc, monitor/mouse/keyboard. A57 makes this reality. You can hook up tablets/phones to a mon/key/mouse but not much point for power users/gamers without the power that A57/64bit large memory, 3.5ghz+, psu, vid card etc brings with it all. We won't get REAL apps until ARM gets to desktops with 64bit. Until then we'll keep getting the junk you're describing for 512mb. It won't get much better even with 4GB on a phone/tablet and 64bit really doesn't become worth it until over 4GB which can happen on a desktop overnight. We need ARM to escape the 10w envelope to get real work apps and far superior gaming compared to mobile stuff, so we might actually be able to totally dump WINTEL for a ton of people and not feel pain. This war won't be over in 3yrs and probably not even 10, but we'll have a good idea of how successful each side is in the battle by 5yrs probably. It will take a while to get the whole 64bit ecosystem going on ARM obviously.

 

[InvalidError]

Think BIGGER PICTURE and longer term. You can't take out a desktop without becoming one at least to some extent. (...) It will take a while to get the whole 64bit ecosystem going on ARM obviously.

It will not take any time at all for the market to move to ARMv8 (64bits) because most Android apps are Java-based so all you need to take advantage of extra and wider registers is an updated Java recompiler and JRE optimized for the new instruction set - that's why x86-based Android phones and tablets can run the majority of Google Play apps and games using the same APK install file even though the hardware is completely different.

Performance-wise, current ARM chips are about 1/10th as fast as desktop chips so even if you slightly more than double the power budget to double clocks to desktop levels, we are still only talking 1/5th as fast as desktop and as I said above, most Android apps are Java-based so they take an additional performance hit from the JRE on raw processing power compared to PCs running native code. You can double the core count if you want but most mainstream apps hardly make meaningful use of more than 2-3 concurrent threads so there would be very little benefit from that. Sure, ARM-based PCs may run some other Linux/GNU variant more favorable to native applications but I am not convinced those are going to be anywhere near as successful as Android and ChromeOS.

Over the past couple of years, SoC performance was roughly doubling every year and based on that, I was predicting that there were around three years left to x86's significant performance lead but based on the latest bunch of SoC press releases, it looks like ARM SoC progress will be down to ~50% this year, which may not be a good sign: when PC performance scaling started tapering off, it ended up falling off a cliff to the glacial pace we know today so the dip in ARM performance progress this year may indicate that ARM is starting to roll down the hill near the cliff. On the other hand, you have Intel aggressively pushing their SoC agenda and outperforming many ARM chips on performance-per-watt so they certainly have not lost the fight there.

As for ARM wanting to sell $300 chips, that is unlikely to happen in the mainstream since there are too many licensees producing chips competing for market share in whatever form they can get it. They might manage to hit that sort of high price point for server-oriented designs but that is a completely different story.

Right now, the only major threat to Intel is Intel itself: pricing themselves out of the market.