KitGuru KitGuru.net – Tech News | Hardware News | Hardware Reviews | IOS | Mobile | Gaming | Graphics Cards
Nvidia Corp. is set to unveil its new-generation high-performance graphics solutions based on the code-named GM204 graphics processing units in about two weeks from now, but so far there have been no leaks indicating performance of the GeForce GTX 970 and the GeForce GTX 980 graphics cards. Luckily, a staff member of Overclockers.co.uk decided to shed some light on what to expect from the novelties.
Initially Nvidia’s partners will bring GeForce GTX 970/980 graphics cards with 4GB of GDDR5 onboard, but eventually they will also release versions with 8GB of memory and custom coolers, according to Andrew “Gibbo” Gibson, a purchasing manager at Overclockers UK, who shared his educated guesses at OcUK forums. Presumably, there will be factory-overclockers versions of GeForce GTX 970/980 graphics cards as well.
The amount of rumours spread about performance of graphics cards based on the GM204 graphics chip (which were first known as the GeForce GTX 880/870/860 and later it turned out that Nvidia will call them GeForce GTX 980/970/960) was rather incredible this spring, but none of the rumours was close to reality. As expected, the GM204 is set to replace the GK104 graphics processing unit (which powers GeForce GTX 780/770/760/680/670 solutions), which is why it was naïve to expect it to outperform graphics boards based on the fully-loaded GK110 (GeForce GTX Titan Black, GeForce GTX 780 Ti).
“Speculation is [the GeForce GTX 980] replace[s] 780, not 780Ti,” said Mr. Gibson. “Performance wise, my lips are sealed on how it compares to 780Ti, but of course 980 is quicker than 780 for sure!”
The GeForce GTX 780 graphics card is based on a cut-down GK110 GPU that features 2304 stream processors (SPs), 192 texture units (TUs), 48 raster output units (ROPs) and 256-bit wide GDDR5 memory bus. The GeForce GTX 780 Ti powered by the fully-fledged GK110 graphics processor is a much more powerful graphics solution with 2880 SPs, 240 TUs, 48 ROPs and 384-bit GDDR5 memory bus.
Based on KitGuru’s analysis of what to expect from graphics solutions based on the GM204, the GeForce GTX 980 graphics card could be 30 per cent or even faster compared to the currently available high-end GeForce GTX solutions. However, this expectation might be too optimistic.
“Do not expect a huge leap in performance over current single GPU stuff,” said Mr. Gibson.
Nvidia did not comment on the news-story.
Comments of Andrew Gibson reflect his personal expectations and are not official statements from Overclockers UK.
Discuss on our Facebook page, HERE.
KitGuru Says: While graphics cards based on the GM204 may not represent a performance breakthrough, they will still be faster compared to today’s solutions not based on ultra-premium GK110 GPU with all the units activated.
Your numbers on here are wrong for the GTX780. It is based on GK110 and has 2304 shaders,
If its faster than the 780 “for sure” that leaves it pretty much guaranteed on par with the 780 Ti at worst, hopefully for $200 less.
GTX 780 is GK 110 and got 2304 stream processors with 48ROP and 192 TMU i think u mistaken 680/770 with a 780 ?
that means that there will be a 980Ti Later on 🙂
Figured as much 6 months ago, Need the proper die shrink.
They are just trying to pull another one over everyone’s heads, 680 fiasco all over again. Mid range rebadged as high end! What is more alarming is it’s even more dodgy as changing the name from 880 to 980 some uninformed folks out there may think it is the proper “big gtx” card and throw money at it.
Wait it out folks for AMD and Nvidia’s proper next gen with the die shrink, Unless you absolutely need an upgrade to get playable frame rates seems like a waste of money upgrading from anything newer than a 680.
Glad I skipped 680 went from 480’s to 780’s, Same thing here need to skip and wait for the proper full card.
http://wccftech.com/geforce-gtx-980-alleged-benchmark-tdp-170w/
it is said to be about 10% faster than the gtx 780 ti
Gibson never said the 980 won’t beat the 780 Ti. He said his lips are sealed on that, which actually gives reason to suspect that it MAY beat the 780 Ti. Otherwise he would just come out and say it won’t beat it.
I hope they don’t use those horrible looking plastic shrounds and stick with the Titan style cooler albeit with an updated aesthetic.
You’re late. Here’s a more recent derp-leak: http://wccftech.com/geforce-gtx-980-alleged-benchmark-tdp-170w/ 😛 (wonder how much time this one will last too)
What a shit of article:
Title: “Nvidia’s GeForce GTX 980 will not beat GeForce GTX 780 Ti – insider
”
Evidence about it:
“Speculation is [the GeForce GTX 980] replace[s] 780, not 780Ti,” said Mr. Gibson. “Performance wise, my lips are sealed on how it compares to 780Ti, but of course 980 is quicker than 780 for sure!”
Ouh, Was Mr. Gibson saying in some place of this sentence the claim of the article? NO. So you are misinforming your readers.
Plus, your ignorance, Anthony Silov, about this world is epic:
“GK104 graphics processing unit (which powers GeForce GTX 780/770/760/680/670 solutions)”
So, now, the GK104 is the GPU inside of the GTX 780… lol.
More:
“The GeForce GTX 780 graphics card is based on a cut-down GK110 GPU that
features 2304 stream processors (SPs), 192 texture units (TUs),
48 raster output units (ROPs) and 256-bit wide GDDR5 memory bus”
Ok, a very misinformation about the GTX 780, the card has a 384 bit wide bus, like the GTX 780 Ti. It’s imposible to disable a memory controler without disable some ROPs with the kepler arquitecture, too. So, if you have 48 ROPs, you have ALL the bus and memory controlers (and the L2).
Knowing that you unknow the basic about the GTX 780, you said after the previous misinformation:
“fully-fledged GK110 graphics processor is a much more powerful graphics
solution with 2880 SPs, 240 TUs, 48 ROPs and 384-bit GDDR5 memory bus.”
It’s a more powerful card (around 20% at best), but NO in the sense that you are saying. You are saying that the GTX 780 is a a) GK104 and b) a very much castrated version of GK110. Both are lies.
And with this lies you are making a sci-fi article. Genious.
Compared to my 670 4gb, it should be a massive performance boost.
two words: mind blown!
I knew this all this time pretty much…they are delivering Maxwell which is focused on being energy efficient…not Volta or Pasqual.
You can expect each card to be about 5-15% faster than their respective 700 series counterpart. But the exciting thing about this is how energy efficient they are going to be.
A Maxwell based Titan would probably only consume 70-50% of the power that a Kepler based Titan would consume…
That is the whole deal about Maxwell… the ENTIRE 900 series will be outright “made” for ITX builds and energy efficient gaming PCs.
Calling it now…the 980 will only be 5-15% faster than the 780. Thus not a real 780 Ti contestant(it will likely be VERY slightly below the 780 Ti).
What it WILL do is consume only a fraction of the power. Keep in mind that this release is Maxwell. A card focused on efficiency…not Volta/Pasqual…which will be the next series focused on actual performance improvements.
Maxwell will only be faster based on the fact that there is far less energy consumed and thus it can clock higher.
I expect the 980 to consume anywhere between 30-50% less power than the 780.
Gibbo actually said GTX 980 will def beat GTX 780 like it would be no problem. Since there is only 10% between GTX 780 and 780 Ti one cant help but wonder that the 980 will beat 780 Ti too
I hope the high-end part is soon to follow then.
And overclocking
Yes it should overclock really well because of the high energy reserves and low heat. I wouldn’t be surprised if the 980 is powered by a single 8pin.
The only thing that perplexes me is that even though initially they will release models with 4GB of Vram…models with 8GB are planned….like what the hell.
AMD was just showing that by using realtime texture compression you can easily game in UHD with 2GB Vram without performance loss (you can even lower the bits of your interface to below 300 and it won’t affect performance) so 8GB seems quite over the top (and it will unnecessarily eat up energy that could be used elsewhere)
Where do you get your facts from, article author.. out your ***? The 780 = GK110, 384-bit mem bus.
it will be stupid if gtx 980 is less than 10% faster than gtx 780
The 780 has a 384 bit bus just like the TI, it’s just clocked to 6008MHz instead of 7000MHz
There have been some unofficial benchmarks and comparisons: http://forums.anandtech.com/showthread.php?t=2398674
I was thinking the same about these cards, GTX 780 and its variants are all powered by GK110 but the 780 and Titan are powered by the revision A1 and 780ti and Titan Black Revision B1 of Kepler Gk110 GPU.
More sensationalist bullshit headlines that have NO basis in what was ACTUALLY said
You have it all wrong ! GTX 760/770 is GK104 !! GTX 780/780ti is GK110 ! Twat noob! 😛
I wouldn’t jump on the band wagon of hype for these cards. There’s been rumors that it won’t beat the 780ti. Nvidia released the 780ti after the original 780. Its highly likely that nvidia will do the same thing and release a fully fledged gm204 chip in the form of the 980ti. Heck, the 980 supposedly has just a 256bit memory bus.
As this 9 series hype ! GM204 is just a GK104 rewamp to milk the cow even more! Wait for GM210 for the fully loaded board (4000 cuda cores) !
When the CUDA Cores are 40% better than the 780’s, and higher clock speeds the 980 will run circles around the 780 Ti…
slightly stronger, but uses half power. amazing.
Thanks 4the heads up Shane!
Necro to prevent misinformation:
Its now January and you’ve been shown wrong. No idea why people try to write an article with a conclusion based solely on general specs without a thought to architecture. Thanks for cluttering the internet up with more erroneous “information”.
For anybody who came here looking for real information:
— The 980 was a leap over the 780, not the 780 Ti. Yet it still outperforms the 780 Ti while running more efficiently and being priced considerably lower for its respective release.
— Summary? 980>780 Ti based on performance/efficiency/price/memory size/anything else
— Find something else to write about Anton Shilov, tech stuff doesn’t seem to be your thing.
The 800 series was mobile primarily, for some odd reason. It may have had to due with the thermal and energy goals of Maxwell. The 980 ended up beating the 780 Ti despite being heavily overclocked and thus no room for growth and overclocking a GTX 980, not difficult due to drastically reduced thermal/energy constraints lets it thoroughly thrash the GTX 780 Ti. Maxwell is doing amazing things despite a lack of a die shrink. A GTX 680/770/780 will get anything to be playable, at current, at extremely high settings 1080p. You would only want a 900 series GPU if you want to utilize downsampling or play at higher resolutions with crazy settings, maybe multi-monitor or 3D if that is someone’s thing. 680 was known to be mid-range as was the 780. The sad part was their “high end” GPUs such as the Titans which were a miserable mess. The 980 is actually a mid-range GPU as well. A die shrink to 16nm seems to be the core goal for both Nvidia and AMD at this point.