this result in Physx Fluid, Driver?
Printable View
this result in Physx Fluid, Driver?
I thought there was a new integration node with 8800GTX..,. as well as a totally different architecture. Let alone how crappy SLI was those days.... bad example, mate ;)
lulz? Where do you see a 35% difference in performance? I see 42% difference...
What is there to see in that picture?
prava 83 -> 118 = 35 :D
Guys, are we really discussing results obtained in a CPU bottleneck?
42% faster = 1.42. Let's get back to Titan ;)
Math is a wonderful science isn't it ? :D
Anyone else feel the supposed boost clocks seem a little high?
This is off topic but what memory speed are the radeon hd 7000 cards?
yes. if 680 boosts around 100mhz to 1100mhz-1200mhz, then a chip 2x bigger boosting to 1000mhz+ does seem aggressive.
the only clock speed I trust so far is the opencl one below. 875mhz is around 100mhz more than the k20x, and this makes sense. trade some efficiency for more performance in this enthusiast card, and boost will take it to 900-something mhz, very close to 1000mhz, maybe even a little over. I will be slightly surprised if titan really boosts over 1000mhz, but not very surprised if it is just 1005mhz or whatever. I just think ~950mhz makes more sense. this is still looking to be a 250W card easily
Stock was 5500mhz , GHZ it is 6000mhz.
Anyway, yes boost seems too much high for 240W, and the slide is strange ( 2 colors font, font dont completely match, strange way of write the specs for the memory ( no space between the / and have you allready see a gpu brand write "b" for bit .. )
Cant say if it is a true one or not ( depend who have done the slide who is similar of the 680 slide )
Considering 28nm is a mature process at this point both AMD and Nvidia should both have an idea how best to utilize the transistor designs for 28nm to optimize their gpus for power and performance.
Just because 1st get 680 consumed xxx power for xxx clock speed doesn't mean new designs on the same process will perform identical even on the same node.
I'm sure there is allot learned as far as what kind of transistor to use where to optimize power efficiency as general gpu design optimization for a particular node.
For the most part I'll reserve my thoughts until I see real numbers.
Thing is, thr 580 didn't want to clock that well, 950 was very good, 1000 was stellar and took a lot of juice and cooling, aaaand custom PCB to make it comfortable. Stock vGPU they were nothing overly special, 900 - 925 MHz. This is another power hungry (probably hot with the same size cooler to dissipate 230+ w) card limited with (so we're told) no voltage adjustment and 8 phase 'nvidia spec' VRM, I honestly think a moderate at best OC is expected. That said they should be powerful even stock, so with all that firepower for gamers I doubt overclocking is even neccessary unless using huge resolution/cracked hardcore GFX games.
real benchmarks:
http://www.hardware-infos.com/news/4...ark-score.html
http://www.techpowerup.com/gpudb/b14...GTX_Titan.html
GPU Name: GK110
Process Size: 28 nm
Transistors: 7,080 million
Die Size: 502 mm?
Render Config
Shading Units: 2688
TMUs: 224
ROPs: 48
SMX: 14
Pixel Rate: 51.2 GPixel/s
Texture Rate: 205.0 GTexel/s
Floating-point: 4,919.04 GFLOPS
Graphics Features
DirectX: 11.1
OpenGL: 4.3
OpenCL: 1.2
Shader Model: 5.0
Released: February 18, 2013
Bus Interface: PCIe 3.0 x16 ?
ASUS Part #: GTXTITAN-6GD5
Clock Speeds
GPU Clock: 915 MHz
Boost Clock: 1019.5 MHz
Memory Clock: 1502 MHz (6008 MHz effective)
Memory
Memory Size: 6144 MB
Memory Type: GDDR5
Memory Bus: 384 bit
Bandwidth: 288.4 GB/s
Reference Board
Slot Width: Dual-slot
TDP: 235 W
Outputs: 2x DVI
1x HDMI
1x DisplayPort
Good finds.... :up:.
your welcome!
guys come on it is the same Result from Obr-Hardware That appeared two days before:)
http://i.imgur.com/dtAs2QE.png
http://www.obr-hardware.com/2013/02/...l-numbers.html
http://www.xtremesystems.org/forums/...=1#post5169787