The GK110 Arrives

GK110 hasn't arrived, the Tesla K10 is in fact GK104.

The GK110 based Tesla K20 is due at the end of this year. So it hasn't arrived :confused:


I wonder if a Geforce version of GK110 will be released?
 
I wonder if a Geforce version of GK110 will be released?

If AMD can not bring anything at least 30% faster than the GK104 and a card that can beat the 690, we may never see the GK110 Geforce. Maxwell (the next generation after Kelper) was slated for a 2013 release so they may just skip ahead.

muIXp.jpg
 
If AMD can not bring anything at least 30% faster than the GK104 and a card that can beat the 690, we may never see the GK110 Geforce. Maxwell (the next generation after Kelper) was slated for a 2013 release so they may just skip ahead.

muIXp.jpg

Old road map. Who knows what will happen at this point.
 
He was just posting some information. Why you gotta bring up fanboyism when no one else was?

I don't even see why he brought AMD up in the post above yours lol...he always does digs/trolls at AMD, I don't get it. Guess he really hates them. :confused:

I do hope that AMD steps it up. At least AMD is consistently on time with GPU's, hopefully this December will be better and a return to the glory days of the likes of x1900 or 5870 with Sea Islands. As of this round nvidia definitely won, for sure!
 

Where's the product? Until there is product on the shelf, it has not "arrived.":rolleyes:

All this thread is is one overly desperate fanboy posting an obscure marketing press release that he wasted hours of his life searching for that really doesn't tell anyone anything of significance. I also find it highly ironic that the OP lists an article about "stealth marketing" from AMD when his fanboyish antics on this forum provides free stealth marketing for Nvidia.
 
Tesla K10 is basically GTX690. I wonder why do a Tesla GK104 since its double precision performance is mediocre.
 
If AMD can not bring anything at least 30% faster than the GK104 and a card that can beat the 690, we may never see the GK110 Geforce. Maxwell (the next generation after Kelper) was slated for a 2013 release so they may just skip ahead.

muIXp.jpg

You are really quoting the same graphic that has Kepler in 2011?
 
Maybe there is no GK110 gpu for gaming. It's possible nvidia decided to get all the GPGPU stuff out to make a more efficient cooler less power consuming gaming chip and a beastly workstation card in Tesla.

If amd doesn't really push forward something amazing nvidia will probably sit content winning this gen with the GTX 600 series. AMD's head start is probably going to keep diminishing until they fix their drivers and price their products more competively against nvidia's offerings.

I like when both manufacturers are kicking ass as it makes things better for us the gamers. Better products, as both companies keep pushing each other. Usually better prices too.

I reaped the rewards last gen and had 2 monster unlocked 6950s running at near 1000 Core clocks for $300 each. Their drivers back then weren't as half assed as what people are experiencing with the new GCN architecture. It's pretty sad what's been going on this gen for AMD.
 
There will be a GeForce GK110. Nvidia announced at GTC the Tesla K20 for the end of this year and the GeForce version 'not before' Q1 2013.
 
http://www.brightsideofnews.com/new...ansistors2c-300w-tdp2c-384-bit-interface.aspx


This product is based upon new GK110 silicon, a 7.1 billion transistor manufactured using 28nm process node. According to Jen-Hsun Huang, this is the largest number of transistors on a single piece of silicon – Xilinx has the second place with a FPGA chip with a bit over six billion transistors. Third one is a GPU from AMD, the 4.31 billion transistor monster, and then we get back to GK104 with 3.1 billion transistors. Top 5 are rounded with a 10-cor Intel Westmere processor, a 2.6 billion transistor chip manufactured in 32nm technology.

GK110 i.e. Tesla K20 brings over 1.5TFLOPS of double-precision FP performance, i.e. 3x more than Tesla C2075 with its 515 GFLOPS. The GK110 GPU brings an unannounced number of cores, 384-bit memory interface and an undefined amount of onboard memory. The memory combinations possible are 6GB, 12GB or 24GB. Jen-Hsun mentioned that the amount of memory depends on the DRAM industry. We expect to see the 12GB version, though.

Over 2X as many transistors as the GK104 :eek:
 
Wow, prime got destroyed pretty fast in this thread.

and the GK110 most likely does exist ....










... on paper! :D
 
GK110 hasn't arrived, the Tesla K10 is in fact GK104.

The GK110 based Tesla K20 is due at the end of this year. So it hasn't arrived :confused:


I wonder if a Geforce version of GK110 will be released?

it will but as a GTX 700 series near the end of this year. probably around the same time AMD plans on releasing their 8k series aka GCN v2.0.
 
it will but as a GTX 700 series near the end of this year. probably around the same time AMD plans on releasing their 8k series aka GCN v2.0.

It will be interesting to see if Nvidia has finally hit the wall on transitors. I understand that 22nm is relatively new, has many problems and is steadily improving but I wonder if Nvidia bit off a bit more then they could chew with the GK110/100 design?

It looks like AMD didn't even want to attempt this.
 
It will be interesting to see if Nvidia has finally hit the wall on transitors. I understand that 22nm is relatively new, has many problems and is steadily improving but I wonder if Nvidia bit off a bit more then they could chew with the GK110/100 design?

It looks like AMD didn't even want to attempt this.

who knows.. it could of been the reason GK110 ran into so many problems. but my guess is most of the problems they ran into had to do with heat and power usage given the size of the actual gpu in comparison to GK104.
 
who knows.. it could of been the reason GK110 ran into so many problems. but my guess is most of the problems they ran into had to do with heat and power usage given the size of the actual gpu in comparison to GK104.

My guess is they couldn't get much of anything to work on the chip. I wonder if Nvidia told TSMC to stop production after intial attempts too. I mean that is A LOT of transistors.

I can't say I would be suprised if they redesign the chip to be smaller in the end to meet their expected Q3/Q4 release.
 
It will be interesting to see if Nvidia has finally hit the wall on transitors. I understand that 22nm is relatively new, .

Small Nitpick. 28nm

Good post though. It may be right on track with what went down.
 
That's why there's so few 680s and 690s. nVidia's saving the perfect GK104s for their Tesla K10 cards.
 
there is a K10 and a K20.

the K10 is really just a GTX690 so dual GK104 chips.

the K20 is using GK110 though.
 
Finally a real 50-100% upgrade instead of rip off place holder we were given for $500.

I guess my GTX 670 will be worth what it should be $250 any day now.
 
So the GK104 was a midrange chip after all. So I'm supposed to spend $630 for a speed binned gk104 with a 256bit memory bus and 4gb vram? Sounds kind of pricey for what you get. Wouldn't I be better off spending the same ~$630 and getting a 7970 with a 384bit memory bus and 6GB of vram? That sounds to me like it will have longer legs to Q1 of 2013 when the real high end Nvidia chip is released. I'll play crazy rez and insane modded Skyrim until then. I suppose SLI 670 4G EVGAs for ~$900 or a 690 for $1200 might feel like a GK110 if you close your eyes.

Seriously what are your plans because I want a GK110 in August and thats is apparently not happening.
 
Since this was an apparent false flag when the shit is the real Big Kepler coming out?
 
Finally a real 50-100% upgrade instead of rip off place holder we were given for $500.

I guess my GTX 670 will be worth what it should be $250 any day now.

really? where? I don't see it, all I see in this thread is reference to a tesla card.
 
Nvidia chip aims to power fastest supercomputer
http://www.bbc.co.uk/news/18033970

For those too lazy to read the article..Tesla K20 will be $1,500 - $2,000 so expect the desktop version to be pricey too.
The US government plans to install nearly 19,000 of Nvidia's new GPUs into its Titan supercomputer later this year. Chip maker Nvidia has revealed details of a new graphics processing unit (GPU) which it says will create the world's most powerful computer. Thousands of the firm's Tesla K20 modules will be fitted to an existing supercomputer at Oak Ridge National Laboratory in Tennessee, US. Nvidia says the machine will run nearly eight times faster than at present, carrying out up to 25,000 trillion floating point operations per second. Nvidia says the addition of its chips should allow Oak Ridge's Titan system to leapfrog from the world's third fastest supercomputer to the top spot.
But the extra speed comes at a cost. The upgrade is expected to involve the addition of almost 19,000 Tesla K20s. Each is set to have a list price of between $1,500 and $2,000 (£930-£1,245), although the laboratory will get a discount for buying in bulk. However, the investment will be partly offset by the fact that the machine should burn up less energy.
 
Nvidia chip aims to power fastest supercomputer
http://www.bbc.co.uk/news/18033970

For those too lazy to read the article..Tesla K20 will be $1,500 - $2,000 so expect the desktop version to be pricey too.
worskstation cards usually cost way way more than desktop versions. by the time this comes out for desktop it could easily be a $600 card or maybe less. of course that depends on other factors too such as what AMD will have to offer at that time.
 
I still think that the GK110 will be vastly different between the Tesla and Geforce version.

Regardless, the article doesn't say anything about the GK110 other then the expected release date, which is still far into the future. My crystal ball still says Nvidia will cut down on core size as well.
 
I think going forward this is what we will see. Fermi was a disaster from a engineering standpoint (though Nvidia actually made good money out of it due to good contract negotiation with TSMC) and I dont think they want to repeat it. From now on all consumer grade parts will be based on the GF104 design - cut down for gaming only. Professional parts will be like this - full FP32 performance. Essentially just two totally different market segments that do not overlap at all. Makes sense to me.
 
Back
Top