PDA

View Full Version : Falanx - ATI and Nvidia's new competitor


Pages : [1] 2

AthlonXP1800
08-23-05, 05:50 AM
http://www.hardocp.com/article.html?art=ODAy

Falanx's new graphic core chip Mali200 to be ready in 2006 is a tile based render, scalable, support new features beyond Shader Model 3.0, have physics engine as well as multi-cores and has Windows Vista drivers ready.

Desktop Mali200 cards with multi-cores could have as many as 16 cores will see as a serious threat to ATI future products, it make R500's eDRAM an expensive waste with huge amount of bandwidths. I think Nvidia knew about Falanx new graphics core for a year, Nvidia next G80 chip are incorporated physics hardware engine and probably brand new render design based on GigaPixel technology or maybe hybrid tile based render. :)

nutball
08-23-05, 06:51 AM
Forgive my cynicism but haven't we been here before? Bitboys spring to mind.

From what I've read Falanx are an IP company (along the lines of the PowerVR brigade) and so will be reliant on other people licensing their tech and producing chips. Given that there's already a company in that position who hasn't made any impact on the PC market... well wake me up when something exciting happens.

As for the tile-based stuff, that's in the same pigeon hole as ray-tracing and the Amiga as far as I'm concerned. Oh and Betamax is in there too, lol.

SH64
08-23-05, 06:59 AM
I'll believe it when i see thier products in the market & the benchmarks all over the net , otherwise its another XGI for me.

jolle
08-23-05, 07:50 AM
yeah history doesnt show any great examples of promising competitors breaking in on the market..
Parhelia was prolly my biggest disappointment, then we got S3 and XGI, and not to mention Bitboys (again) who had, seemingly, good hardware (12mb EDRAM for example) long ago, but never managed to get any of their chips to market.

If they bring out something good, thats just great, but Im not holding my breath anymore..

Lfctony
08-23-05, 08:02 AM
XGI comes to mind, don't buy a thing until I see an actual preview.

shabby
08-23-05, 08:48 AM
Should this 3 week old news really be i the rumor section? :)

Graphicmaniac
08-23-05, 10:20 AM
their tecnology remember me the last Vodoo videocards built with 1, 2, 4 and so on chips linked to each other.

rohit
08-23-05, 04:22 PM
yes, we need another competitor. i wish to see falanx in desktops.. and hope they are competitve!

Toss3
08-23-05, 04:54 PM
yeah history doesnt show any great examples of promising competitors breaking in on the market..
Parhelia was prolly my biggest disappointment, then we got S3 and XGI, and not to mention Bitboys (again) who had, seemingly, good hardware (12mb EDRAM for example) long ago, but never managed to get any of their chips to market.

If they bring out something good, thats just great, but Im not holding my breath anymore..
Yeah the Parhelia really sounded nice on paper, so I'm not getting my hopes up for this one(learned this through dissapointment)!

Zelda_fan
08-23-05, 05:31 PM
The Parahelia was pretty good when stacked against the Geforce 4. Only problem was that 4 days later ATI blew the lid off the 9700 which ate everything for breakfast.

fivefeet8
08-23-05, 05:37 PM
Tile based rendering sound nice. It should alleviate a lot of the huge memory bandwidth requirements of current and future video cards. Now if only they could actually bring it out and compete.

anzak
08-23-05, 06:34 PM
Tile based rendering sound nice. It should alleviate a lot of the huge memory bandwidth requirements of current and future video cards. Now if only they could actually bring it out and compete.

They claim that it will be able to do 4xAA without any performance hit. I find that kinda hard to believe.

H.263/H.264/WMA Acceleration and being Longhorn (err Vista) ready would make this a good graphics card for a multimedia computer.

I would love to see a real compeditor for ATI and nVidia. I haven't given up on XGI yet.

Fotis
08-23-05, 06:46 PM
Man, I wish there was a third competitor!More competition equals better prices,more options, and drives technology further.
I wonder why Intel doesn't give it a shot apart from their integrated graphics.With their manufacturing capabilities(fabs) they would be a great competitor. :eek:

AthlonXP1800
08-23-05, 07:26 PM
They claim that it will be able to do 4xAA without any performance hit. I find that kinda hard to believe.

It not that hard to believe, Nvidia had GigaPixel GP1 chipset that are tile based render, it ran Quake 3 with free 4xFSAA without performance hit but it doesnt not have TnL build in the core. :)

fivefeet8
08-24-05, 03:52 AM
They claim that it will be able to do 4xAA without any performance hit. I find that kinda hard to believe.


I find it a bit hard to believe too. But a tile based rendering engine could fare a lot better with FSAA.

An older video card which used TBR, the KyroII gets 2xAA almost completely free. And only loses 4 fps at 4xAA.
http://www.hardwareanalysis.com/content/reviews/article/1291.16/

takyon369
08-24-05, 03:25 PM
You did notice that this product is for Mobile computing. From the looks of it, its not Laptop either its actually for mobile CELL PHONES.

Even if these guys were for "real" this has no immidiate bearing on desktop markets.

Although Nvidia and ATI are obviously already branched into the mobile cell phone world, so this might be a good thing as an IP for Nvidia to purchase perhaps.

I'm unimpressed without any proof though.

Zelda_fan
08-24-05, 03:41 PM
It's a scaleable architecture where the chips are small enough to stack one right next to the other. The chips are extremely small btw. For exmaple, you could fit maybe 50-60 falanx chips in the size of the G70 chip. Each chip has 1 pipe so 50 chips on a die would = a 50 pipe chip which would blow away anything currently on the market. For a cell phone, you would just use 1 chip, and for a laptop maybe 20 chips.

Vapor Trail
08-24-05, 04:18 PM
I would really like to see a 3rd big vendor of video cards. Competition is always good. The big problem for me is drivers. It takes a while to make good drivers and it will even be harder for an entry company. Who is going to buy a new company's card with shaky drivers. Not me...

anzak
08-24-05, 05:57 PM
I find it a bit hard to believe too. But a tile based rendering engine could fare a lot better with FSAA.

An older video card which used TBR, the KyroII gets 2xAA almost completely free. And only loses 4 fps at 4xAA.
http://www.hardwareanalysis.com/content/reviews/article/1291.16/

Impressive. I'm a n00b in the world of computer graphics cards so I don't remember the Kyro II.

fivefeet8
08-25-05, 01:18 AM
Impressive. I'm a n00b in the world of computer graphics cards so I don't remember the Kyro II.

Ah. Tile Based Rendering was pushed by PowerVR. Incidently, the Kyro2 also uses technology from PowerVR. PowerVR also created the Tile Based Rendering technology behind the Sega Dreamcast.

Redeemed
08-25-05, 02:01 AM
Well, I was actually pleased with all my S3 chips, especially the ones from Diamond. Now that Diamond is back, I'm hoping that they'll be able to move up there with ASUS and other companies of that sort. ;)

I'm not to impressed so far about this new chip. When I actually see a review of this chip in a desktop and it is kicking around the current GPU like a ragdoll, then I'll be impressed. Untill then, I'll resort to merely "spectating".

nutball
08-25-05, 06:25 AM
It's a scaleable architecture where the chips are small enough to stack one right next to the other. The chips are extremely small btw. For exmaple, you could fit maybe 50-60 falanx chips in the size of the G70 chip. Each chip has 1 pipe so 50 chips on a die would = a 50 pipe chip which would blow away anything currently on the market. For a cell phone, you would just use 1 chip, and for a laptop maybe 20 chips.

That's not necessarily true. There's a recent David Kirk interview floating about where he talks about the mobile and embedded market. He made the point that the design decisions you make when creating a core for mobile/embedded are fundamentally different to those you make when designing a high-performance core for desktop use. It's not just a question of pipeline count, clock-speed, etc., it's the whole ethos of the thing.

I'm not sure that there's any reason to believe that this core would scale to fifty pipes any better than the current NVIDIA/ATI designs would scale to fifty pipes. Fundamentally NVIDIA & ATI aren't stupid, and they have a lot more experience in designing high-performance scaleable GPUs than Falanx.

Darkfalz
08-25-05, 03:38 PM
A couple of DC titles used the PowerVR tile based FSAA (which, presumably, had little performance impact) but it didn't look very good, in VGA mode anyway.

msxyz
08-26-05, 07:13 AM
If it's a IP company like VideoLogic then this project will remain a potentially good competitor... on paper.

Behind the Kyro GPUs there was ST (not exactly a small company, owning a good portion of the discrete semiconductor products and cheap custom integrated circuits market) and Guillemont. Despite good acceptance by the market, both investors withdrew only after one generation.

So, unless one big name (ie Intel, AMD) decide to license this technology I don't think we will see anything new in the graphics arena. The only other chance is that a big mobo/card manufacturer eventually decides to make a joint venture with some founrdy to launch its own line of graphics accelerators. But which is in the position and has the resources.

Superfly
08-26-05, 01:21 PM
Man, I wish there was a third competitor!More competition equals better prices,more options, and drives technology further.
I wonder why Intel doesn't give it a shot apart from their integrated graphics.With their manufacturing capabilities(fabs) they would be a great competitor. :eek:

no no - dont give Intel idea's

they have both the fab's and the money/marketing machine to destroy ATI and NVIDIA, I've long been of the opinion that they are intentionally staying out of the market for this very reason.

god help us all if Intel were to bring out a serious GPU.