Sorry your browser is not supported!

You are using an outdated browser that does not support modern web technologies, in order to use this site please update to a new browser.

Browsers supported include Chrome, FireFox, Safari, Opera, Internet Explorer 10+ or Microsoft Edge.

Geek Culture / Larrabee Cancelled

Author
Message
Leadwerks
17
Years of Service
User Offline
Joined: 29th Jan 2008
Location:
Posted: 14th Dec 2009 05:09 Edited at: 14th Dec 2009 05:09
Link
http://www.bit-tech.net/news/hardware/2009/12/07/intel-larrabee-cancelled/1

I guess raytracing is still the technology of the future, for the foreseeable future.
Back to top
Profile PM
bitJericho
22
Years of Service
User Offline
Joined: 9th Oct 2002
Location: United States
Posted: 14th Dec 2009 05:48
Link


sad, I can't wait for raytraced games!! We'll have to wait a while longer it seems.

Back to top
Profile PM Email Website
Lemonade
16
Years of Service
User Offline
Joined: 10th Dec 2008
Location:
Posted: 14th Dec 2009 07:46
Link
I read about this a week or two ago. Sad.
Back to top
Profile PM
Duke E
16
Years of Service
User Offline
Joined: 10th Mar 2009
Location:
Posted: 14th Dec 2009 09:05 Edited at: 14th Dec 2009 11:39
Link
Sprung a lot of rumors that Intel was to buy NVidia (stupid bloggers) =P.
Back to top
Profile PM
Syncaidius
20
Years of Service
User Offline
Joined: 22nd Mar 2005
Location: United Kingdom
Posted: 14th Dec 2009 11:31 Edited at: 14th Dec 2009 11:38
Link
I'm glad its cancelled to be honest. They would have only ripped us off with the price anyway, just like with their I7s.

By the time Intel got this out OpenCL will have had a stronger userbase anyway. And for those of you who don't know what that is, it's a programming language that lets you run programs accross GPUs and CPUs. So, whats the point in larabee if we have this? :p

EDIT:
Another fail:
Quote: "Justin Rattner, recently showed Larrabee could crack the 1TeraFLOP mark at SC09 using a standard HPC benchmark (SGEMM 4Kx4K calculation). Compare this to ATI's Radeon HD 5870, which hits only 544GigaFLOPS, and Larrabee is clearly a HPC monster. That said, the Radeon has been available for nearly two months already. Nvidia claims its forthcoming Fermi chip is capable of between 900GFLOPs to 1.2TFLOPs, although no actual benchmarks have yet been shown."


Even though you see everywhere that the HD 5870 is capable of around 1.5TFlops, Nvidia's latest card being somewhere around that mark too. So again, whats the point in Larabee if we have more powerful tech already out? :p


Forum Thread
Back to top
Profile PM Email
Leadwerks
17
Years of Service
User Offline
Joined: 29th Jan 2008
Location:
Posted: 14th Dec 2009 17:13 Edited at: 14th Dec 2009 17:14
Link
A GPU architecture is not the same as a CPU. Cuda has been a tremendous failure because all it's good for is making photoshop filters. I know the new Ferni chip is supposed to change all this, but of course NVidia will tell everyone that.

I think the GPU will always be best for graphics and the CPU will always be best for logic, and all attempts for one to do the other will fail.
Back to top
Profile PM
lazerus
17
Years of Service
User Offline
Joined: 30th Apr 2008
Location:
Posted: 14th Dec 2009 17:29
Link
Dont know what it is, but this made me laugh...




http://lazerus-reborn.deviantart.com/
Back to top
Profile PM Email
budokaiman
FPSC Tool Maker
15
Years of Service
User Offline
Joined: 24th Jun 2009
Playing: Hard to get
Posted: 14th Dec 2009 17:34
Link
I remember writing a paper on this not long ago. I'm glad it's cancelled, it limits the power that can be used by the computer and with the complexity of most programs now, it would be useless in a few months anyway. It's only use could be for netbooks, but those really don't need to be any smaller.

This signature is legen-wait for it... dary };]
Back to top
Profile PM Email Website
Syncaidius
20
Years of Service
User Offline
Joined: 22nd Mar 2005
Location: United Kingdom
Posted: 14th Dec 2009 19:59 Edited at: 14th Dec 2009 20:03
Link
Quote: "A GPU architecture is not the same as a CPU. Cuda has been a tremendous failure because all it's good for is making photoshop filters. I know the new Ferni chip is supposed to change all this, but of course NVidia will tell everyone that."


Not to sound fanboish or anything but Nvidia has never been good at anything really. Whenever they do make a decent graphics card, they put the price up high enough that only the pure Nvidia fanboi's will buy their cards (the people who will pay £500 on a graphics card that will be at least 1/2 the price 6 months later).

So far, PhysX, complete failure really, I hardly see any games support it, and when they do, they don't make use of what Ageia originally made into its main selling point, hardware fluids. Havok have caught up to the point where they have amost identical fluid systems (if not better performing) without the need for an graphics or PPU card to run it. Also, the fact that my old Aegia PPU will no longer work unless I spend money on buying an Nvidia card to run with it (or use the really old Aegia PhysX drivers), because I'm using a ATI HD 4870 instead. To hell with that.

CUDA, well that was their attempt at making their own unified coding language that would (but doesn't) run well across GPUs and CPUs. This pretty much failed the same way ATIs CTM (ATIs version of CUDA)language failed. Theres no point in any of these now since we have OpenCL which is open source, free for anyone to use and of course runs across more platforms than just Windows. AMD have already decided to support OpenCL on their DX11 cards, so I guess its just Nvidia that needs to tear their heads off and glue them back on the right way round.


Forum Thread
Back to top
Profile PM Email
NeX the Fairly Fast Ferret
20
Years of Service
User Offline
Joined: 10th Apr 2005
Location: The Fifth Plane of Oblivion
Posted: 14th Dec 2009 20:36
Link
Quote: "So far, PhysX, complete failure really"


Not really. It's good on low-end CPU systems. It really takes quite a load off the CPU in physics intensive situations. In the few games that support it, I get 5-10 extra FPS when enabling it on my 9500GT. In nVidia's intensive fluid demos, the framerate jumps from under 1 frame per second to nearly 10.

Athlon64 2.7gHz->OC 3.9gHz, 31C, MSi 9500GT->OC 1gHz core/2gHz memory, 48C, 4Gb DDR2 667, 500Gb Seagate + 80Gb Maxtor + 40Gb Maxtor = 620Gb, XP Home
Air cooled, total cost £160
Back to top
Profile PM Website
Benjamin
22
Years of Service
User Offline
Joined: 24th Nov 2002
Location: France
Posted: 14th Dec 2009 20:40
Link
Quote: "In the few games that support it, I get 5-10 extra FPS when enabling it on my 9500GT."


From... what frame rate?
Back to top
Profile PM Email
NeX the Fairly Fast Ferret
20
Years of Service
User Offline
Joined: 10th Apr 2005
Location: The Fifth Plane of Oblivion
Posted: 14th Dec 2009 20:48 Edited at: 14th Dec 2009 20:50
Link
25-30FPS improves to 30-40FPS. My system is seriously CPU limited. There are also less pauses as physics threads and loading threads fight for control of a single core as I walk around. Test case is Unreal Tournament III with the Extreme Physics mod pack on full settings @ 1280x1024.

Athlon64 2.7gHz->OC 3.9gHz, 31C, MSi 9500GT->OC 1gHz core/2gHz memory, 48C, 4Gb DDR2 667, 500Gb Seagate + 80Gb Maxtor + 40Gb Maxtor = 620Gb, XP Home
Air cooled, total cost £160
Back to top
Profile PM Website
Leadwerks
17
Years of Service
User Offline
Joined: 29th Jan 2008
Location:
Posted: 14th Dec 2009 21:18
Link
Quote: "Not to sound fanboish or anything but Nvidia has never been good at anything really."

Well, yeah, I don't think any of their non-graphics diversions will ever turn a profit.

Everyone is going on and on about parallelism, but I think it is mostly a pipe dream. I'm glad we have dual cores, but I don't think applications will ever make much use of quad or more cores.

Consider a long-division problem. Can you break the problem apart and give one part to each person in a group, have them solve their part, then come back and put them together to get the answer faster? No. And no amount of paradigm shifts will ever make that possible.
Back to top
Profile PM
Syncaidius
20
Years of Service
User Offline
Joined: 22nd Mar 2005
Location: United Kingdom
Posted: 14th Dec 2009 21:25 Edited at: 14th Dec 2009 21:27
Link
Quote: "Not really. It's good on low-end CPU systems. It really takes quite a load off the CPU in physics intensive situations. In the few games that support it, I get 5-10 extra FPS when enabling it on my 9500GT. In nVidia's intensive fluid demos, the framerate jumps from under 1 frame per second to nearly 10."


You missed my point though, what I meant was havok will run on anyones computer, regardless of PPUs or Nvidia cards present. PhysX forces you to have an Nvidia card if you want hardware accelerated physX, and the PPU cards that Ageia made don't work for anyone without an Nvidia card. So again, why bother with PhysX when you can use Havok and cover both ATI and Nvidia card users without any bother?

Imagine if you was an ATI user, you try to use hardware PhysX fluids but you get an "in your face" message saying you have to have an Nvidia card that supports PhysX.. you wouldn't be to pleased either...

Pretty much the main reason DarkPhysics was a complete waste of money after Nvidia ruined it. I bought the PPU which cost £180 at the time, worked like a charm on Cell factor, GRAW and some vehicle combat game I can't remember the name of. Then comes long Nvidia and forces you to have an Nvidia card present for the PPU card to work.

Its easy for all of you who already have an Nvidia card to disagree, but look at it from an ATI users point of view. Why bother buying a whole new Nvidia card just to use hardware physics, when instead Havok is right there waiting to be used.

Need more reasons why it fails? :p


Forum Thread
Back to top
Profile PM Email
NeX the Fairly Fast Ferret
20
Years of Service
User Offline
Joined: 10th Apr 2005
Location: The Fifth Plane of Oblivion
Posted: 14th Dec 2009 21:27 Edited at: 14th Dec 2009 21:29
Link
Quote: "So again, why bother with PhysX when you can use Havok and cover both ATI and Nvidia card users without any bother?"


Because PhysX works just fine with an ATi card? It's not accelerated, granted, but it works. Better to have the option of hardware accleration on selected systems than not to have it at all.

Quote: "Consider a long-division problem. Can you break the problem apart and give one part to each person in a group, have them solve their part, then come back and put them together to get the answer faster? No. And no amount of paradigm shifts will ever make that possible."


Perfectly true, but if you're doing a set of mixed questions (which is what most games are) and inverse quadratics come up you're best off offloading it to someone in your group good at maths rather than struggling through it yourself, if, say, you're no good at them.

Athlon64 2.7gHz->OC 3.9gHz, 31C, MSi 9500GT->OC 1gHz core/2gHz memory, 48C, 4Gb DDR2 667, 500Gb Seagate + 80Gb Maxtor + 40Gb Maxtor = 620Gb, XP Home
Air cooled, total cost £160
Back to top
Profile PM Website
Back to top

Login to post a reply

Server time is: 2025-05-25 15:18:36
Your offset time is: 2025-05-25 15:18:36