nVidia GTX 680 Specs and Images

I just posted this today... I dunno how I managed to miss this thread, my bad.

 

No, it's a guess, but there is no DX12 in W8, so likely W9 now, and since W8 came three years after W7 i would guess 2015 at the earliest.

I know you can't play 3D Mark 11, my point was that there is no point moving to the next api when we have only started to properly use the existing one.

 

Something doesn't add up for sure.

How can there be posts like this:-
kepler-will-be-unbeatable

Yet this is only on par with ATI's offering.

I was really hoping for the 680 to be worth upgrading from my OCed 580.

 

you clearly said foreseeable future. Samaritan isn't a real game, its a tech demo. So i don't see you having enough horsepower for the current foreseeable future. That was my point with FFXIV being a DX11 game that taxes current hardware, regardless of this if a brand new FPS came out using Samaritan as its base none of us would be able to play it smoothly on our hardware full quality.

 

I clearly said I have enough horse power for the forseable future. that in now way implies that I could max out everything that can come out in the future. dont take this the wrong way or anything but I am going to ignore your posts from now on.:nerd:. and lastly I dont think any games that look like Samaritan will be out for quite a while

 

I understand what you are saying man. dx11 has not even been fully tapped yet, probable due to all the dx9 ports

 

That's not what i said, better read again

And since when were NV unified gpus based on AMD's VLIW4?

G80 core


G200
http://images.hardwarecanucks.com/image/skymtl/GPU/EVGAGTX280/GTX280-85.jpg

GF110 core (2 disabled its from 448core ed.)



AMD used this similar SIMD core cluster approach, like nvidia does and its been doing it since G80 and that was my whole point about it.



vs older VLIW4
http://techgage.com/images/news/amd_hd6970shader_122810.jpg



And AMD's GCN SIMD (aka nvidia approach) vs VLIW4 diagram



and again that's what i meant when i said AMD "copying" Nvidia. They had to use such approach if they wanted to rival Nvidia professional compute market.

http://www.anandtech.com/show/5261/amd-radeon-hd-7970-review/3

 

but people are projecting new consoles by 2013 :3eyes: and Hl3 is already in the works. You must not follow game trends that much. Your "foreseeable" future is a grim one.

 

6Pin are 75watts each, and that is just a spec not an actual limitation. a card can draw more wattage through the 6pin power if necessary

 

the other 75watt comes from the m/b 150+75=225

 

Wii U is expected to be out within the next year and it's possible the next XBox will as well. Sony isn't expected to "update" until at least 2014. If Sony sticks to what they recently said, it will be 2016.

You tell me to read again, then go on to compare G80 (VLIW4) and GF110 (VLIW4) to AMD's GCN (SIMD)...which is NOT a VLIW architecture. GCN, by design, is closer to a CPU than it is a VLIW based GPU. If you want an accurate comparison, put GCN against Intel's Larrabee prototype as that was also a SIMD architecture...as is Kepler.

 

Makes sense, sony has been trailing in the game market for years now. Wouldn't surprise me if they just focused on the Vita for the next two years and then released ps4 during DX12's reign.

 

Sony said they intended the PS3 to last 10 years when it was launched 6 years ago, which would put it's next update in 2016. Waiting that long, I expect to see Sony start losing share in the console market.

 

10 years was also so people would pre-maturely buy the ps3 with little-to-no games for it. I see nintendo delivering, i see M$ delivering and I see sony trying to pick up the slack again.
But really, sony was always a gadget based company so the vita will be their big seller for quite a while now.

 

Are you serious? G80 and GF110 are both using VLIW compute units, even though its a AMD thing with what 3000 - 6000 series? ok.. xD


have a look here, its SIMDs, AMD's GCN uses similar SIMDs too.. Yeah similar to larrabee and yet similar to nv with unified SIMD approach.

So were ATI 4000-AMD 6000 series but those had VLIW in it..

http://www.behardware.com/articles/772-6/nvidia-fermi-the-gpu-computing-revolution.html


FERMI part
Fermi, each of the 16 multiprocessors has a double scheduler that runs at the low frequency and 4 execution blocks at the high frequency:

- two 16-way SIMD units (the 32 “cores”): 32 FMA FP32s, 32 ADD INT32s, 16 MUL INT32s, 16 FMA FP64s
- a quadruple SFU unit: 4 FP32 special functions or 16 interpolations
- a 16-way 32-bit Load/Store unit

you see its SIMDs but you say it VLIW, because of those 4 execution blocks? lol ok, im not gonna debate further.

 

More rumours:
http://www.legitreviews.com/news/12616/

 

They should pull an intel and put two under the same heatspeader

 

You're just lost.... It's not my fault you don't understand it. The SIMD units are necessary for GPGPU functionality, but the architecture itself is a MIMD architecture based on VLIW4. Graphics processing by SIMD would be slow (think trying to run full DX10 on a GF2 MX200) as only 1 graphics instruction would be processed at a time whereas with MIMD, multiple instructions are processed simultaneously.

That would be interesting to say the least.

 

Likewise with your "GF110 is based on VLIW too".


btw

 

^^wheres the link for that info?^^