PowerVR To Make Mobile Graphics, GPU Compute a Three-Way Race Again

← Back to Stories (view on slashdot.org)

PowerVR To Make Mobile Graphics, GPU Compute a Three-Way Race Again

Posted by timothy on Sunday June 17, 2012 @04:42AM from the good-kind-of-instability dept.

MojoKid writes "For over 10 years, the desktop and mobile graphics space has been dominated by two players: Nvidia and AMD/ATI. After 3dfx collapsed, there was a brief period of time when it looked as though Imagination Technologies might establish itself as a third option. Ultimately, that didn't happen — the company's tile-based rendering solution, Kyro, failed to gain mass-market support and faded after two generations. Now, there's a flurry of evidence to suggest that Imagination Technologies plans to re-enter PC market, but from the opposite direction. Rather than building expensive discrete solutions, IT is focused on deploying GPUs that can challenge Nvidia and AMD solutions in tablets, mobile phones, and possibly netbooks. Over the past two weeks, Imagination Technologies has announced new, higher-end versions of its Power VR Series 6 GPU, claiming that the new Power VR G6230 and G6430 go '"all out," adding incremental extra area for maximum performance whilst minimising power consumption.' There's a new ray-tracing SDK out and a post discussing how PowerVR is utilizing GPU Compute and OpenCL to offload and accelerate CPU-centric tasks." Update: 06/17 17:53 GMT by T : Related: An anonymous reader adds a link to a new project from the FSF to reverse engineer the PowerVR SGX.

74 comments

Min score:

Reason:

Sort:

Umm, no by TheRaven64 · 2012-06-17 04:49 · Score: 4, Informative

The mobile graphics space has been dominated by one player: PowerVR. ARM and nVidia are more recent entrants. AMD doesn't yet have anything in this space, although that will change very soon.

--
I am TheRaven on Soylent News
1. Re:Umm, no by dmesg0 · 2012-06-17 04:58 · Score: 4, Informative
  
  AMD used to have mobile core, but they sold it to Qualcomm. It's now called Qualcomm Adreno.
2. Re:Umm, no by Anonymous Coward · 2012-06-17 05:16 · Score: 0
  
  AMD used to have mobile core, but they sold it to Qualcomm. It's now called Qualcomm Adreno.
  Qualcomm Adreno shipped the most GPUs in 2011.
  http://www.reuters.com/article/2012/02/25/idUS12516+25-Feb-2012+BW20120225
3. Re:Umm, no by dmesg0 · 2012-06-17 05:23 · Score: 3, Informative
  
  Their statistics are per SoC vendor, not by GPU core vendor. E.g. Apple and TI SoC come with PowerVR, Samsung SoC contain Mali (on most popular devices) or PowerVR.
4. Re:Umm, no by hairyfeet · 2012-06-17 09:40 · Score: 1
  
  Which shows what a piss poor previous CEO they had by selling it. But other than Nvidia PowerVR has pretty much ruled the roost in mobile, but now you have Nvidia, Intel looks to be tailoring the new Atoms for that market, and I wouldn't be surprised if AMD is cooking up a version of Bobcat or quietly marrying a couple of ARM cores to one of their Radeons instead of just tacking on an ARM to their X86 BD designs.
  Honestly? I don't care WHO wins because as long as their is growth and competition we the consumer benefit. For ages the GPU in mobile devices were frankly piss poor, just as the IGPs in x86 laptops and desktops were piss poor. Nvidia and AMD made having a decent GPU a selling point and suddenly everyone is talking GPUs which makes me very happy, hell AMD seems to be betting that a "good enough" CPU with an awesome GPU is the way to go and I can see no fault with their logic. There is simply many jobs that consumers want to do today that benefit from having a decent GPU so having the entire market focused on making good GPUs is frankly wonderful news to me, I for one welcome PowerVR into the fray.
  
  --
  ACs don't waste your time replying, your posts are never seen by me.
5. Re:Umm, no by greg1104 · 2012-06-17 13:38 · Score: 2
  
  Intel is starting to provide more serious competition to both NVIDIA and AMD/ATI too, on the laptop end of mobile computing. The latest rev of their graphics chipset, the HD 4000, is more than enough GPU power for many people.
6. Re:Umm, no by Anonymous Coward · 2012-06-17 14:10 · Score: 0
  
  And Qualcomm has been poaching AMD engineers like mad ever since.
7. Re:Umm, no by viperidaenz · 2012-06-17 15:18 · Score: 1
  
  They also stuck a PowerVR in their latest Atom CPU's
8. Re:Umm, no by TheGratefulNet · 2012-06-18 03:38 · Score: 1
  
  marshalltown? and is there much linux (accel video) support yet? is booting into text mode working, at least?
  I did have to do some kernel command line video parms to get my amd fanless board to even init video well enough to see text on the console. without it, you have to install with another video card or install from some other system and move the disk over.
  anyone know what version of kernel is promised for the marshalltown series?
  
  --
  
  --
  "It is now safe to switch off your computer."
9. Re:Umm, no by hazydave · 2012-06-18 04:31 · Score: 1
  
  Yup... PowerVR has long been the dominant GPU in mobile devices. You'd have to say that just because they're in every iPhone and iPad, even if they weren't also in many other SOCs, like those from TI, Samsung, Marvell, NEC, and Renesas. And Intel.
  Yup, they've been in netbooks... the PowerVR SGX535 usually goes by the alias "Intel GMA500" or "Intel GMA600" or whatever, but they've been the low-end default among Intel systems for awhile. That's probably changing, with Intel doing their own actual GPUs here and there.
  As mentioned, AMD/ATi sold off their mobile GPU to Qualcomm, which calls it Adreno these days. And there have been some signs Qualcomm may be looking elsewhere, as they have yet to make that technology convincing. nVidia, too, hasn't been their usual desktop selves in the mobile space; the fastest graphics in mobile devices for the last year have been dual-core PowerVR in the Apple products, unless you count the quad-core PowerVR driven Sony Playstation Vita. ARM's MALI is also fairly new... ARM looking to capitalize on their crazy-successful processor core business with a whole soup-to-nuts solution.
  
  --
  -Dave Haynie
10. Re:Umm, no by hazydave · 2012-06-18 04:47 · Score: 1
  
  Medfield uses the same PowerVR SGX540 found in the TI OMAP 4xxx series, and going back to the original Samsung Galaxy phones. The forthcoming Clover Trail (Atom Z2580) goes to the dual core PowerVR SGX544MP2, for next year's tablets (maybe phones too?), which is basically a Direct3D 9.3 upgrade to the SGX543MP2 in the iPhone/iPad.
  
  --
  -Dave Haynie
11. Re:Umm, no by TheGratefulNet · 2012-06-18 04:58 · Score: 1
  
  searching find this: http://www.phoronix.com/scan.php?page=news_item&px=OTYyNw
  This means that for the Linux 3.1 kernel we could see the Cedar Trail graphics code merged, but more than likely it will still not be accelerated and only just be useful for kernel mode-setting. Before the Cedar Trail hardware is even publicly released we'll see the Linux 3.2 kernel and possibly even the Linux 3.3 release, so at least there's more time to hopefully better the open-source Cedar Trail capabilities.
  so, looks like its still not ready for prime time yet.
  
  --
  
  --
  "It is now safe to switch off your computer."
12. Re:Umm, no by phoenix_rizzen · 2012-06-18 20:59 · Score: 1
  
  The GPU core in the mobile (as in phone/tablet) Atom (aka Medfield) is a PowerVR SGX540. And the GPU core in several mobile (as in netbook) Atoms are PowerVR as well. There's only a couple of versions of the Atom CPU that include Intel GPUs.
13. Re:Umm, no by phoenix_rizzen · 2012-06-18 21:01 · Score: 1
  
  It'll be interesting when Qualcomm's S4 Pro SoC is released, as it combines the CPU power of the Krait with the GPU power of the Adreno 300-series.
Must have by Anonymous Coward · 2012-06-17 04:49 · Score: 1

Imagination Technologies has announced new, higher-end versions of its Power VR Series 6 GPU, claiming that the new Power VR G6230 and G6430 go '"all out," adding incremental extra area for maximum performance whilst minimising power consumption
I've been waiting for a GPU that goes "all out" for as long a I can remember. The nine tenths out that you get from a typical NVIDIA card just isn't enough any more.
1. Re:Must have by rsilvergun · 2012-06-17 06:46 · Score: 3, Funny
  
  I'm holding out for 11/10ths.
  
  --
  Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
2. Re:Must have by hazydave · 2012-06-18 04:48 · Score: 1
  
  Hell with the fractions. I want one that goes all the way to 11.
  
  --
  -Dave Haynie
I hope they can do better on drivers by PhrostyMcByte · 2012-06-17 04:50 · Score: 2

The PowerVR GPUs integrated into Intel's Atoms are great -- in theory. The drivers are so terrible they can't even run Windows 7's aero at acceptable FPS, let alone a game. They also don't bother to support 64-bit, or any x86 Linux other than 32-bit MeeGo.
I don't know if it's PowerVR or Intel, but someone needs to get their drivers in order before they'll have a chance of encroaching on any of the existing players.
1. Re:I hope they can do better on drivers by guises · 2012-06-17 05:04 · Score: 1
  
  PowerVR could compete if they opened their drivers, or at least released specs so that other people could write them. They've been extremely reticent about this.
2. Re:I hope they can do better on drivers by fuzzyfuzzyfungus · 2012-06-17 05:25 · Score: 4, Interesting
  
  I'm inclined to wonder if their 'tile-based rendering' scheme for cleverly throwing away work that doesn't actually have to be done is more driver-based than some of the competing GPU vendor's schemes, making them rather touchy about opening that.
  
  It could also just be that they have some sort of inertial paranoia thing going on as a company; but it certainly seems like it might have had to be something good if Intel, Chipzilla himself, couldn't wring decent drivers out of them for their GMA500-based parts.
  
  That isn't exactly a spat on the debian mailing lists over firmware-linux-nonfree, that's a potentially huge design win that ended up sucking fairly hard wherever it showed its miserable face...
3. Re:I hope they can do better on drivers by Anonymous Coward · 2012-06-17 05:25 · Score: 0
  
  PowerVR graphics suck on Linux, FreeBSD, Windows...they just aren't good hardware, or they're just severely hampered by poor quality software. Might be fine in a phone, I don't care.
4. Re:I hope they can do better on drivers by 91degrees · 2012-06-17 05:39 · Score: 1
  
  I'd have expected these to be really good at dealing with aero, because transparency is one of the areas tiled architectures work well. You're doing a lot of reading and writing to/from the framebuffer. When your framebuffer is a 64x64 tile then the entire thing can be kept in a fast segment of memory.
  
  If it is a driver issue, then it's rather sad to be letting poor software ruin a good overall design.
5. Re:I hope they can do better on drivers by nogginthenog · 2012-06-17 06:12 · Score: 2
  
  But it kicked ass on the Dreamcast back in 1999!
6. Re:I hope they can do better on drivers by wampus · 2012-06-17 06:40 · Score: 1
  
  Just like AMD?
7. Re:I hope they can do better on drivers by guises · 2012-06-17 07:07 · Score: 1
  
  Just like AMD?
  That would be fine, though they could certainly go further. What are you getting at?
8. Re:I hope they can do better on drivers by wampus · 2012-06-17 07:09 · Score: 1
  
  AMD sucks butt on Linux despite them releasing specs, whereas nVidia is as close to "just works" as is practicable using their binaries. You know, the exact opposite of what you are claiming.
9. Re:I hope they can do better on drivers by Anonymous Coward · 2012-06-17 07:32 · Score: 0
  
  AMD sucks butt on Linux despite them releasing specs, whereas nVidia is as close to "just works" as is practicable using their binaries. You know, the exact opposite of what you are claiming.
  Yes, and with nVidia stopping to support Linux (cf. Optimus), it "just stops working" too.
10. Re:I hope they can do better on drivers by Creepy · 2012-06-17 07:42 · Score: 1
  
  PowerVR has always taken the cheap/power friendly road, building for smaller devices and low end graphics needs. They also typically have been a generation or two behind on die shrinks, letting nVidia and AMD absorb the cost of creating the new processes and methods. Compared to AMD and nVidia, they are terrible, but compared to Intel GMA, they aren't all that bad (especially when Intel was doing transformation and lighting on the CPU). They sucked on Linux/FreeBSD because the OSS OpenGL drivers were never very good (at least when I used them) and on Windows because nVidia and AMD are so much better. On phones, though, where low power consumption > graphics throughput, they aren't bad. I certainly wouldn't play a 3D game on mine (it's 2 years old in August), but it plays Angry Birds just fine.
11. Re:I hope they can do better on drivers by Vanders · 2012-06-17 07:52 · Score: 1
  
  If it is a driver issue, then it's rather sad to be letting poor software ruin a good overall design.
  That wouldn't be anything new. S3 managed to ruin their hardware with poor software their entire lives.
  
  --
  Syllable : It's an Operating System
12. Re:I hope they can do better on drivers by DarwinSurvivor · 2012-06-17 08:38 · Score: 1
  
  When a company drops support for their hardware, the drivers don't magically VANISH, you can still run them. You may need to use a downgraded kernel, but it will still work!
13. Re:I hope they can do better on drivers by guises · 2012-06-17 08:45 · Score: 1
  
  In my experience the RadeonHD driver has been pretty solid. It's open, requires no kernel downgrades, and will continue to be supported into the foreseeable future. Lack of 3D acceleration is the only issue that I'm aware of, but they're working on that. This is a big step up over Nvidia's approach (or PowerVR's).
14. Re:I hope they can do better on drivers by viperidaenz · 2012-06-17 15:21 · Score: 2
  
  Wait, your 3D graphics card has solid drivers but no 3D support?
15. Re:I hope they can do better on drivers by guises · 2012-06-17 16:37 · Score: 1
  
  My graphics card has solid drivers but no 3D support. There haven't been dedicated 3D graphics cards since the Voodoo 2 days. At least not mainstream ones.
16. Re:I hope they can do better on drivers by Anonymous Coward · 2012-06-18 02:36 · Score: 0
  
  I'm inclined to wonder if their 'tile-based rendering' scheme for cleverly throwing away work that doesn't actually have to be done is more driver-based than some of the competing GPU vendor's schemes, making them rather touchy about opening that.
  The "throwing away" has always been hardware based even with the original PCX1/PCX2 systems. The driver on those did have to decide which tiles each object would go into, but even that went into hardware in the next generation (e.g. Dreamcast's CLX2).
17. Re:I hope they can do better on drivers by hazydave · 2012-06-18 04:51 · Score: 1
  
  Most of the Intel GMA series ARE PowerVR cores. Just slighly higher performance versions, not identical to those found on smartphones.
  
  --
  -Dave Haynie
18. Re:I hope they can do better on drivers by Anonymous Coward · 2012-06-18 10:37 · Score: 0
  
  PowerVR has always taken the cheap/power friendly road, building for smaller devices and low end graphics needs. They also typically have been a generation or two behind on die shrinks, letting nVidia and AMD absorb the cost of creating the new processes and methods.
  You don't understand the semiconductor industry. nVidia, AMD, and Imagination are all fabless semi companies. None of them 'absorb the cost' of creating new processes, except in the sense that some of the money they pay to the fab-owning companies which build their parts is used by said fab owners to pay for process R&D.
  (Before you object, yes, AMD used to have fabs, but they haven't for a few years now. Besides, the graphics part of AMD is the former ATI, and ATI was a fabless semi company too. Even when AMD owned both ATI and fabs at the same time, ATI video chips were still made in outside fabs, probably because that's what all the designs in progress when AMD acquired ATI were targeted at. It's not easy to turn multi-year-long chip development programs around on a dime.)
Not news by dmesg0 · 2012-06-17 04:50 · Score: 3, Insightful

It's already more than 2 players in mobile space: ARM with its Mali core, Qualcomm with Adreno (former ATI/AMD), NVidia with Tegra and IT with PowerVR.
In addition, Intel already uses PowerVR cores in some Atom CPUs (targeted for tablets).
Since they are still talking about mobile, how is that news?
1. Re:Not news by UnknowingFool · 2012-06-17 05:01 · Score: 4, Informative
  
  Yes. It's not like PowerVR isn't already used in smart phones and tablets today. Most people just have not heard of them. If you are using an Apple mobile device, you are using PowerVR. If you are using a Sony PS Vita, you are using PowerVR. If you are using a Samsung Galaxy, you are using PowerVR.
  
  --
  Well, there's spam egg sausage and spam, that's not got much spam in it.
2. Re:Not news by dmesg0 · 2012-06-17 05:12 · Score: 1
  
  Not entirely true, my Galaxy S2 has ARM's Mali 400 GPU. Only Galaxy S and Galaxy Nexus come with PowerVR.
3. Re:Not news by JAlexoi · 2012-06-17 11:51 · Score: 1
  
  S3 is Mali-400MP as well.
4. Re:Not news by phoenix_rizzen · 2012-06-18 20:53 · Score: 1
  
  Well, if you are using *some* Samsung Galaxy devices, you are using PowerVR. There are many variations of the Galaxy, Galaxy S, Galaxy SII, and Galaxy SIII, some with Adreno, some with Tegra, some with PowerVR.
Does Linus know about this? by Anonymous Coward · 2012-06-17 04:50 · Score: 0

After reading the Slashdot article about Linus' opinion of NVIDIA these days, I wonder if he knows about
this outfit and whether if IT will want to offer support for Linux in their product offerings...
1. Re:Does Linus know about this? by dmesg0 · 2012-06-17 04:55 · Score: 3, Informative
  
  PowerVR drivers are either closed source for ARM or nearly nonexistent for x86 (Intel Poulsbo) on linux.
2. Re:Does Linus know about this? by 0123456 · 2012-06-17 05:03 · Score: 2
  
  Nvidia must have paid PowerVR to start releasing mobile chips so Linus won't be able to say they're the worst hardware manufacturer for Linux any more.
3. Re:Does Linus know about this? by JAlexoi · 2012-06-17 11:52 · Score: 1
  
  Eh.. What do you think Orange SanDiego runs? http://www.gsmarena.com/orange_san_diego-4588.php
4. Re:Does Linus know about this? by Anonymous Coward · 2012-06-17 12:14 · Score: 0
  
  Also, good luck getting OpenGL support out of a powervr chip. Their specs claim OpenGL, but their Linux drivers only support OpenGL ES. Unfortunately, most Linux software that does hardware acceleration is still only OpenGL. Ubuntu on my pandaboard is still mostly software rendering.
  Apparently they have OpenGL drivers but they only give them if you license them as an extra fee, and nobody does.
5. Re:Does Linus know about this? by Anonymous Coward · 2012-06-18 01:33 · Score: 0
  
  In actuality, if you're using an Android device with an OMAP3/4 in it, you're using one of their parts...using Linux drivers to do it with. As for "worst", I'd hesitate to say that one.
Matrox by Anonymous Coward · 2012-06-17 05:00 · Score: 0

Owns the imaging market, medical, industry, framegrabbers,
Clue yourself in from the fact there has existed more than 2 players in the GPU market for a LONG time.
GPU != gaming.
1. Re:Matrox by fuzzyfuzzyfungus · 2012-06-17 05:32 · Score: 1
  
  Video input is certainly a whole different game; but I'd be curious to know how well Matrox is holding up in video-output land.
  
  Now that digital video interconnects are pretty standard outside of the cheapest seats, Matrox's historical reputation for shipping the quality DACs isn't terribly exciting, and 'mainstream' GPUs that cost essentially nothing, to the nearest thousand, seem to be sprouting additional heads faster than the Hydra.
  
  What is saving them from that?
2. Re:Matrox by Anonymous Coward · 2012-06-17 11:22 · Score: 0
  
  Very well on video output, ESPECIALLY VERY HIGH RESOLUTION and MULTI SCREEN which is required for medical, presentation and industry. These are not your 1080p monitors, think Retina displays everywhere, especially for medical.
  While NV are in the render farm business, they are nowhere to be seen nor heard in Matrox's domain as is AMD.
  Matrox is NOT a consumer GPU company they are real world GPU's for production and science, not animation farms or gaming.
3. Re:Matrox by Anonymous Coward · 2012-06-18 11:01 · Score: 0
  
  Matrox is NOT a consumer GPU company they are real world GPU's for production and science, not animation farms or gaming.
  Matrox wishes they were a consumer GPU company, though. They once were, and they got relegated to this small and shrinking niche because they couldn't compete with ATi or NVidia in the markets where the real money is.
  Aside from what Matrox used to hang its hat on (DACs), these "REAL GPU" things you mention are some of the easiest problems to solve in GPU design. Super high pixel counts are not rocket science, it's literally just widening the horizontal and vertical pixel counters in the state machines which control display refresh. Many-headedness is just a matter of adding more refresh engines and copies of the output interface circuits.
  Matrox is in a very risky position because their uniqueness is getting either less relevant or adopted by the mainstream. fuzzyfuzzyfungus already mentioned two such things (better DACs, multihead). The same thing is already happening to high res: if you didn't notice, Apple just shipped a 15.4" 2880x1800 notebook display driven by either Intel HD4000 integrated video or NVidia 650M. What happens to Matrox when you can get all the high res and multihead goodness you want with a real GPU behind it, not the weak GPUs Matrox offers because they lost that war and no longer have the capability to keep up?
PowerVR already in Intel Atom chips by CityZen · 2012-06-17 05:07 · Score: 1

Although I don't know which PC products include CedarView/CedarTrail (Atom D2500/D2700/N2600/N2800) chips, they are there (with SGX545).
1. Re:PowerVR already in Intel Atom chips by Anonymous Coward · 2012-06-17 05:28 · Score: 0
  
  And it sucks so bad I just use the serial console or ssh.
PowerVR! by hsa · 2012-06-17 05:28 · Score: 1

I had Asus EEEPC 901 and I currencly have Intel DC 2700 DC Atom-motherboard.
I just love their Linux support! On 2700DC it doesn't exist. Hell, it took them 6 months to get XP drivers out.
On GMA 950 (EEEPC 901) they decided quietly to downgrade OpenGL back to 1.x -versions, because they couldn't be arsed with maintaining the driver base.
Let me guess - that SDK is Windows only?
1. Re:PowerVR! by JAlexoi · 2012-06-17 11:54 · Score: 1
  
  And I quote "OS support for Windows, with Mac OS X and Linux to follow;"
Intel: 59% of market by vinn · 2012-06-17 05:29 · Score: 2

Not entirely sure what is meant by "dominated" - Intel has 59% of the market (source: http://hothardware.com/News/AMD-Grabbed-GPU-Market-Share-from-Nvidia-Intel-in-Q4/ ). I think what was meant was something like, "AMD and nVidia have dominated the GPU market for serious gamer geeks". The rest of us running our Latitude work laptops could care less what kind of GPU is in it because they've all been sufficiently powerful for years.

--
----- obSig
1. Re:Intel: 59% of market by Henriok · 2012-06-17 05:37 · Score: 2
  
  You can surely dominate in other respects than market share, no? Look att Apple who completely dominated the cell phone industry from the day they released the iPhone, with exactly 0% market share. Domination can be measured in the means others take to counter the attack. Everyone scuttles against Apple or is about to perish. nVidia and AMD surely dominates the desktop GPU market before Intel since it'sIntel that's playing catch up in means of power and features.
  
  --
  
  - Henrik
  
  - when the Shadows descend -
2. Re:Intel: 59% of market by UnknowingFool · 2012-06-17 05:38 · Score: 2
  
  Yes the point is a little muddled. NVidia and ATI are the two major players in the discrete GPU space. For mobile devices, very few discrete players exist as the GPUs are often part of the chip in a SOC design. For instance Qualcomm's Snapdragon, TI's OMAP, Apple's A4/A5, nVidia's Tegra. If anything PowerVR is more like ARM in that they license out their designs for others to build.
  
  --
  Well, there's spam egg sausage and spam, that's not got much spam in it.
3. Re:Intel: 59% of market by espiesp · 2012-06-17 06:53 · Score: 1
  
  I don't know about you, but my Latitude work laptop has an Nvidia GPU so I'm not sure what you're trying to say.
4. Re:Intel: 59% of market by TheLink · 2012-06-17 06:57 · Score: 1
  
  The desktops I build typically have motherboards with intel graphics bundled, even if they actually going to have an Nvidia/ATI card in them. It doesn't really cost much extra, and often those motherboards are actually the cheaper ones (more volume I guess).
  
  And if/when the nvidia/ATI card fails, you may still have a usable computer for nongraphics intensive stuff (like filling out RMAs, looking for replacements, reading Slashdot).
  
  So I wonder if that sort of thing would be 5% of that 59% or more?
  --
  
  Too many replies beneath your current threshold
5. Re:Intel: 59% of market by Creepy · 2012-06-17 08:04 · Score: 1
  
  Yep - only about 10.5% of gamers use Intel - source: http://store.steampowered.com/hwsurvey suggesting very few of this almost 60% care about graphics at all - all they want to do is bring up their Office spreadsheets or play Solitaire or Facebook games at best. Integrated graphics are extremely cheap to add and they often are added at almost cost to the motherboards and chips they are embedded on (because Intel cares more about selling Motherboards and CPUs). Also many people I know own a desktop for better graphics/games and a laptop or netbook for travel, and very few of these laptops have good graphics (mine and a friend of mine's wife, because she played WoW and is on the road a lot).
  I, however am in the other court - I can't even start my day without a GPU on my work laptop, and it needs to have at least OpenGL 1.5 on it. Anything less and I can't bring up CAD visualization software. Some of the newer software I'm working on actually support OpenGL 3 features (which is a huge step in the CAD world - OpenGL 4 would be really cool with hardware tessellation, but they aren't there yet).
S3 Graphics? by ELCouz · 2012-06-17 07:10 · Score: 1

They used to make good graphic cards (Savage series especially, not as good as ATI/NVIDIA but close) but i wonder what happened to them?
1. Re:S3 Graphics? by Anonymous Coward · 2012-06-17 07:45 · Score: 0
  
  Gobbled by VIA. They used it as the integrated graphics on their nano boards. Although they sold the patents off to HTC lately.
2. Re:S3 Graphics? by Anonymous Coward · 2012-06-17 08:53 · Score: 0
  
  Confusing, cause HTC is a subsidiary of VIA.
  On a side note, I'd much rather have a phone labeled "VIA" than "HTC".
3. Re:S3 Graphics? by Anonymous Coward · 2012-06-18 06:53 · Score: 0
  
  Trident and Cirrus Logic also made 3D PC chipsets.
  If we're ignoring the 3D requirement, Tseng Labs, Oak (they may have possibly done a 3D card, who knows)... IBM and Hercules. :P
  I'm sure I'm missing someone here...
PowerVR PCX2 by Hsien-Ko · 2012-06-17 07:40 · Score: 1

are still their glory days to me.

4-bit alpha blend precision in 24-bit color? Hell yeah!
1. Re:PowerVR PCX2 by Guppy · 2012-06-17 14:32 · Score: 1
  
  PowerVR PCX2
  I had one, a Matrox m3d -- Looked great paired up with a Tseng Labs ET6000. Performance wasn't that great though -- by the time the drivers matured, it was already verging on obsolescence.
Mailing list already closed due to legal troubles by Anonymous Coward · 2012-06-17 09:13 · Score: 1

The mailinglist is closed already cause of a possible lawsuit. Let me
quote Bob Ham:
> The GNU lawyers have apparently stated that this PowerVR reverse
> engineering project should not be hosted by the GNU project (on
> savannah.nongnu.org) as there is a risk of the GNU project being the
> subject of a lawsuit. Yay for lawyers!
>
> Unfortunately, that means this mailing list is now closing. I
> apologise for the inconvenience.
citing this OpenMoko mailing list message:
http://lists.openmoko.org/pipermail/community/2012-June/067130.html
Legal patent bullshit impedes reverse-engineering: by crush · 2012-06-17 13:29 · Score: 1

And it looks like the FSF project can't be hosted on Savannah due to potential "legal problems"
Never had 11/10ths by Taco+Cowboy · 2012-06-17 16:43 · Score: 1

I've used many many graphic cards from many many vendors, on many many platforms, from monochrome to cga to vga to whatever-ga that we have now
I must have missed something. I never had the pleasure to use any GPU that gave me 11/10ths performance

--
Muchas Gracias, Señor Edward Snowden !
1. Re:Never had 11/10ths by Anonymous Coward · 2012-06-17 18:37 · Score: 0
  
  Maybe you didn't configure your drivers properly, try renaming the file to quack.exe!
Discussion time by Anonymous Coward · 2012-06-17 21:02 · Score: 0

An anonymous reader adds a link to a new project from the FSF to reverse engineer the PowerVR SGX.
One problem with the FSF is that it often ends up attacking its allies (e.g. they attack the one GPU which actually offers enough low-level documentation to be used on things like BeagleBoards). Discuss.
GMA 500 drivers written by 3rd party by Sits · 2012-06-17 23:53 · Score: 1

The drivers for the GMA500 parts were written by Tungsten Graphics rather than Intel or Imagination Technologies. In a roundabout way other people did write the drivers :-).
Re:Mailing list already closed due to legal troubl by Anonymous Coward · 2012-06-18 01:30 · Score: 0

It WOULD help if you'd have quoted the other message from Bob Ham:

I feel I should clarify this. I was told that the lawyers said it would
be "not a good idea" if the project was hosted by GNU and they said "the
risk was small but that it would be better to avoid it entirely". It's my
inference that the risk they're talking about is of a lawsuit (although I
can't imagine what else they would mean).
From http://lists.openmoko.org/pipermail/community/2012-June/067132.html