Tweet
.:: ภาพเสริม GT300 ด่วนๆๆ ::.
จากกระทู้เก่า http://forums.overclockzone.com/foru...d.php?t=590808
ภาพการทดสอบ >> ภาพวีดีโอที่แอบถ่าย http://www.ustream.tv/recorded/2256749 << (ดูยังไงยังไงก็แอบถ่าย)
พระเจ้าต่อได้แค่จอเดียว!!!
เวลาการจัดงาน
3:56
We are here and ready to get started.
4:00
It hasn't started yet but we should be under way any moment.
4:08
We have video, though no idea what you'll see.
4:13
We'll have a better view once it starts.
4:15
Show starting late obviously...
4:20
Dan Vivoli of NVIDIA. How is audio?
4:26
If you want the scoop on today's announcement, see this: NVIDIA Fermi Next Generation GPU Architecture Overview
4:30
Hey all, for video, head to http://nvidia.com/gtc It is much better and I can hold my laptop straight.
4:31
Don't worry, I'll still be here to take photos and walk you through as it goes.
4:33
Webcast seems to work really well now: http://nvidia.com/gtc scroll down a bit and click "View webcast"
4:37
I guess I can give you guys audio only.
4:39
Wow, good job NVIDIA. can't take photos of the screens because they are blurry without the 3D glasses.
4:39
Jen-Hsun is talking about real-time ray tracing and showing it on stage.
4:41
Claiming a speed up of 48-50x over traditional ray tracing.
4:43
We are know hearing about 3D Visin, 3D cameras and other technologies NVIDIA has been promoting.
4:47
Announcing the Fuji 3D camera. Claims to be the firts consumer 3D camera.
4:51
So far, I agree Gabe.
4:51
[Comment From GabeMcG]
This 3D camera is also a rehash from Quakecon, and also very, very lame
4:52
Now for some new physics demonstrations.
4:54
Pretty imrpessive fluid dynamics demo being shown. No mention of particle counts, etc.
4:55
Those demos were actually using PhysX to be exact.
4:58
Now the demo is throwing bodies at a wodden wall. To be honest I am not that impressed.
4:58
[Comment From dreamer]
will ATI ever get into the physics with their GPUs or would they add that to thier core chip or onboad processor,hmm?
4:59
They seem to be going the route of waiting for other software companies to make the physics engines that are GPU acelerated. Who knows how long they are going to be waiting...
4:59
We are now moving pats "visual computing" and into "parallel computing"
4:59
Prepare for CUDA talk!
5:02
Expand
5:02
NVIDIA CEO in real-life 3D!
5:04
Now keep in mind that my photos of the projector screens are going to be inherintely blurry thanks it running in 3D...
5:04
Expand
5:04
Here is an updated version of the NVIDIA NVRT ray tracing engine at work.
5:05
Expand
5:07
This shot shows a similar scene, but after a real-time rotation. The "point" effect is caused by the engine catching up. It looks very similar to our recent Caustic real time ray tracing demo: http://www.pcper.com/comments.php?nid=7801
5:08
Expand
5:09
This image is of the green-liquid demo they were showing - again the blurriness is NVIDIA's damn fault, not ours. But the effect was cool, just not anything truly ground breaking we haven't seen before.
5:09
[Comment From thank you]
the lack of graphics speed and fidelity talk here or in the new pcper article, combined with huang's comments about the future of nvidia == cuda+tegra, Im not confident in FERMI blowing our minds
5:10
It would seem NVIDIA is either in trouble or possibly just far enough behind that they are holding card close to their chest.
5:11
We are back talking about heterogenous computing - a very worthy goal.
5:17
Time for more physics simulations
5:18
These were mostly shots from movies like ILM and Paramount.
5:18
[Comment From Zeiva]
If we dont see any speed or fidelity news here today would you say it would be worth to go and get a 5850 instead of waiting for a comparable nVidia card to be released? (Comming from a 8800GT 512MB BTW)
5:18
Good question Zeiva: i'd say probably yes. Did you read our HD 5850 and HD 5870 reviews?
5:21
[Comment From hblima]
For a person with an i7 920 system just waiting for a video card, would you say it would be better to get something like a 9600gt/9800gt and wait for Fermi, or get a 5850/5870? (from someone without a video card)
5:21
I think you would be hard pressed to find a downside to getting a 5800 card now.
5:23
[Comment From janis]
come one aske the question will gt300 get 75fps in crysis 1900x12000 4xaaf
5:24
Janis we have absolutely no way of guessing performance yet.
5:24
[Comment From TMarques]
CUDA programmer here. What they're announcing with Fermi is *very* exciting for us(L2 cache? just what I needed!) but I think it will hurt them in the graphics market quite like the GT200 did. The384bit bus on 4.8GHz GDDR5 is enough for pretty much everything.
5:26
Thanks for the comment TMarques; do you mean it is enough for everything CUDA but not gaming?
5:27
Jen-Hsun is now interviewing a rep from a tumor detection software/hardware company about how GPU computing can increase performance and reliability.
5:30
Hey hey, they are talking about Fermi now!
5:31
Get the details here, btw: http://www.pcper.com/article.php?aid=789
5:32
JHSUN says that they started from scratch with this GPU.
5:32
He is getting alot of applause from the developers here for 8x peak DP performance increases, ECC support
5:34
He just showed Fermi card and they have a working system as well.
5:35
Telsa C1060 getting 3.52 FPS in a simulation using double precision
5:35
Fermi gets about 18 FPS
5:35
About 4.5-5x faster
5:38
I have some photos of the first Fermi card to share in just a moment. Looks basically like exisiting graphics cards....
5:39
Expand
5:40
Expand
5:40
Just one DVI output - keep in mind this is NOT a gaming card, but the Tesla model for super computing.
5:41
Expand
5:43
Expand
5:43
On the left, current Tesla running the demo at 3-4 FPS. The new Fermi is running it on the right at 18-20 FPS. Double precision computing work being done here.
5:47
Well we saw a GeForce slide. It basically says to wait you'll hear more later.
5:49
Expand
5:50
Adobe is on stage now, hopefully some big news about Flash and GPU acceleration.
5:53
Sounds like they announcing acceleration for netbooks - hopefully for ION and not Tegra again.
5:54
Yup, its running accelerated Flash on ION!
5:55
Expand
5:55
Full 1080p HD flash video running on the ION with an Atom CPU - no release date for the player mentioned.
5:56
"Now Jen-Hsun, Jen-Hsun, Imma let you finish, but AMD has the best GPU of all time!!"
5:56
Sorry, had to to do it.
5:58
Now on stage - iRay. Internet-based simulation. Think of OTOY or OnLive for CAD designers rather than gamers.
5:59
Nevermind - this is more like the caustic demo, but over the internet rather than local.
6:04
Very impressive demo, just too long.
6:04
[Comment From Rob-Galesburg]
Do they expect ION to compete with pineview in the future, or are they just kinda doing a victory lap?
6:04
A good question. We have heard rumors about "ION 2" but no news yet.
6:05
Now we are looking at augmented reality demonstration with tires, like car tires.
6:10
This is pretty cool - they are simulating wheels on Ferrari tires using just the tire.
6:14
Jen-Hsun is off the stage looks like we are winding down. Dan Vivoli is on stage agan to close us out.
6:15
I'll be shutting down. Thanks for attending our live chat!!!
ผลการทดสอบ Fermi !!!
ภาพการทดสอบ >> ภาพวีดีโอที่แอบถ่าย http://www.ustream.tv/recorded/2256749 << (ดูยังไงยังไงก็แอบถ่าย)
พระเจ้าต่อได้แค่จอเดียว!!!
เวลาการจัดงาน
3:56
We are here and ready to get started.
4:00
It hasn't started yet but we should be under way any moment.
4:08
We have video, though no idea what you'll see.
4:13
We'll have a better view once it starts.
4:15
Show starting late obviously...
4:20
Dan Vivoli of NVIDIA. How is audio?
4:26
If you want the scoop on today's announcement, see this: NVIDIA Fermi Next Generation GPU Architecture Overview
4:30
Hey all, for video, head to http://nvidia.com/gtc It is much better and I can hold my laptop straight.
4:31
Don't worry, I'll still be here to take photos and walk you through as it goes.
4:33
Webcast seems to work really well now: http://nvidia.com/gtc scroll down a bit and click "View webcast"
4:37
I guess I can give you guys audio only.
4:39
Wow, good job NVIDIA. can't take photos of the screens because they are blurry without the 3D glasses.
4:39
Jen-Hsun is talking about real-time ray tracing and showing it on stage.
4:41
Claiming a speed up of 48-50x over traditional ray tracing.
4:43
We are know hearing about 3D Visin, 3D cameras and other technologies NVIDIA has been promoting.
4:47
Announcing the Fuji 3D camera. Claims to be the firts consumer 3D camera.
4:51
So far, I agree Gabe.
4:51
[Comment From GabeMcG]
This 3D camera is also a rehash from Quakecon, and also very, very lame
4:52
Now for some new physics demonstrations.
4:54
Pretty imrpessive fluid dynamics demo being shown. No mention of particle counts, etc.
4:55
Those demos were actually using PhysX to be exact.
4:58
Now the demo is throwing bodies at a wodden wall. To be honest I am not that impressed.
4:58
[Comment From dreamer]
will ATI ever get into the physics with their GPUs or would they add that to thier core chip or onboad processor,hmm?
4:59
They seem to be going the route of waiting for other software companies to make the physics engines that are GPU acelerated. Who knows how long they are going to be waiting...
4:59
We are now moving pats "visual computing" and into "parallel computing"
4:59
Prepare for CUDA talk!
5:02
Expand
5:02
NVIDIA CEO in real-life 3D!
5:04
Now keep in mind that my photos of the projector screens are going to be inherintely blurry thanks it running in 3D...
5:04
Expand
5:04
Here is an updated version of the NVIDIA NVRT ray tracing engine at work.
5:05
Expand
5:07
This shot shows a similar scene, but after a real-time rotation. The "point" effect is caused by the engine catching up. It looks very similar to our recent Caustic real time ray tracing demo: http://www.pcper.com/comments.php?nid=7801
5:08
Expand
5:09
This image is of the green-liquid demo they were showing - again the blurriness is NVIDIA's damn fault, not ours. But the effect was cool, just not anything truly ground breaking we haven't seen before.
5:09
[Comment From thank you]
the lack of graphics speed and fidelity talk here or in the new pcper article, combined with huang's comments about the future of nvidia == cuda+tegra, Im not confident in FERMI blowing our minds
5:10
It would seem NVIDIA is either in trouble or possibly just far enough behind that they are holding card close to their chest.
5:11
We are back talking about heterogenous computing - a very worthy goal.
5:17
Time for more physics simulations
5:18
These were mostly shots from movies like ILM and Paramount.
5:18
[Comment From Zeiva]
If we dont see any speed or fidelity news here today would you say it would be worth to go and get a 5850 instead of waiting for a comparable nVidia card to be released? (Comming from a 8800GT 512MB BTW)
5:18
Good question Zeiva: i'd say probably yes. Did you read our HD 5850 and HD 5870 reviews?
5:21
[Comment From hblima]
For a person with an i7 920 system just waiting for a video card, would you say it would be better to get something like a 9600gt/9800gt and wait for Fermi, or get a 5850/5870? (from someone without a video card)
5:21
I think you would be hard pressed to find a downside to getting a 5800 card now.
5:23
[Comment From janis]
come one aske the question will gt300 get 75fps in crysis 1900x12000 4xaaf
5:24
Janis we have absolutely no way of guessing performance yet.
5:24
[Comment From TMarques]
CUDA programmer here. What they're announcing with Fermi is *very* exciting for us(L2 cache? just what I needed!) but I think it will hurt them in the graphics market quite like the GT200 did. The384bit bus on 4.8GHz GDDR5 is enough for pretty much everything.
5:26
Thanks for the comment TMarques; do you mean it is enough for everything CUDA but not gaming?
5:27
Jen-Hsun is now interviewing a rep from a tumor detection software/hardware company about how GPU computing can increase performance and reliability.
5:30
Hey hey, they are talking about Fermi now!
5:31
Get the details here, btw: http://www.pcper.com/article.php?aid=789
5:32
JHSUN says that they started from scratch with this GPU.
5:32
He is getting alot of applause from the developers here for 8x peak DP performance increases, ECC support
5:34
He just showed Fermi card and they have a working system as well.
5:35
Telsa C1060 getting 3.52 FPS in a simulation using double precision
5:35
Fermi gets about 18 FPS
5:35
About 4.5-5x faster
5:38
I have some photos of the first Fermi card to share in just a moment. Looks basically like exisiting graphics cards....
5:39
Expand
5:40
Expand
5:40
Just one DVI output - keep in mind this is NOT a gaming card, but the Tesla model for super computing.
5:41
Expand
5:43
Expand
5:43
On the left, current Tesla running the demo at 3-4 FPS. The new Fermi is running it on the right at 18-20 FPS. Double precision computing work being done here.
5:47
Well we saw a GeForce slide. It basically says to wait you'll hear more later.
5:49
Expand
5:50
Adobe is on stage now, hopefully some big news about Flash and GPU acceleration.
5:53
Sounds like they announcing acceleration for netbooks - hopefully for ION and not Tegra again.
5:54
Yup, its running accelerated Flash on ION!
5:55
Expand
5:55
Full 1080p HD flash video running on the ION with an Atom CPU - no release date for the player mentioned.
5:56
"Now Jen-Hsun, Jen-Hsun, Imma let you finish, but AMD has the best GPU of all time!!"
5:56
Sorry, had to to do it.
5:58
Now on stage - iRay. Internet-based simulation. Think of OTOY or OnLive for CAD designers rather than gamers.
5:59
Nevermind - this is more like the caustic demo, but over the internet rather than local.
6:04
Very impressive demo, just too long.
6:04
[Comment From Rob-Galesburg]
Do they expect ION to compete with pineview in the future, or are they just kinda doing a victory lap?
6:04
A good question. We have heard rumors about "ION 2" but no news yet.
6:05
Now we are looking at augmented reality demonstration with tires, like car tires.
6:10
This is pretty cool - they are simulating wheels on Ferrari tires using just the tire.
6:14
Jen-Hsun is off the stage looks like we are winding down. Dan Vivoli is on stage agan to close us out.
6:15
I'll be shutting down. Thanks for attending our live chat!!!
ผลการทดสอบ Fermi !!!
Comment