I think 3 SMX (and 128bit gddr5) should probably be enough to go against the 7770.
Though it depends what salvage parts nvidia is going to do. Some rumors saying a gtx655 (which could have 4 SMX then) might also be possible. In any case there's definitely enough room for two cards (both price and performance-wise) between gtx650 and gtx660 if nvidia so desires (amd still has room for one card there too but haven't heard much about a possible HD7790/HD7830 lately).
I've not yet figured out however why nvidia calls everything which is faster than a igp GTX these days. At this point they could really just drop the now meaningless prefixes and go with just the number.
If you look closer at some of the reviews, it really is quite shocking how some of them are getting away with what they are doing.
Hardware Canucks benchmarks some game called Wargame:EU Escalation -
If you ever wondered why they can come up with the 660 being faster than the 7870, now you know. One totally unknown game giving the 660 a 37% advantage over the 7870. Even over 10 games that's a huge swing in favour for just one game.
I suppose the moral is, if you desperately need to find the results you are getting paid to find, you will always be able to get hold of some utterly obscure game to do the job for you. And these sites wonder why they get accused of shilling?
Hardware Canucks benchmarks some game called Wargame:EU Escalation -
If you ever wondered why they can come up with the 660 being faster than the 7870, now you know. One totally unknown game giving the 660 a 37% advantage over the 7870. Even over 10 games that's a huge swing in favour for just one game.
I suppose the moral is, if you desperately need to find the results you are getting paid to find, you will always be able to get hold of some utterly obscure game to do the job for you. And these sites wonder why they get accused of shilling?
If you look closer at some of the reviews, it really is quite shocking how some of them are getting away with what they are doing.
Hardware Canucks benchmarks some game called Wargame:EU Escalation -
If you ever wondered why they can come up with the 660 being faster than the 7870, now you know. One totally unknown game giving the 660 a 37% advantage over the 7870. Even over 10 games that's a huge swing in favour for just one game.
I suppose the moral is, if you desperately need to find the results you are getting paid to find, you will always be able to get hold of some utterly obscure game to do the job for you. And these sites wonder why they get accused of shilling?
Is it really all that obscure? I'd never heard of it, but apparently it's made by the developers who made R.U.S.E., and has received pretty good reviews.
Except both Anandtech and Hardware Canucks also put the card in-between the 7850 and 7870. Close to the 7870, but HardOCP puts them in the same range as well. I just don't see much disagreement here.
If you look closer at some of the reviews, it really is quite shocking how some of them are getting away with what they are doing.
Hardware Canucks benchmarks some game called Wargame:EU Escalation -
If you ever wondered why they can come up with the 660 being faster than the 7870, now you know. One totally unknown game giving the 660 a 37% advantage over the 7870. Even over 10 games that's a huge swing in favour for just one game.
I suppose the moral is, if you desperately need to find the results you are getting paid to find, you will always be able to get hold of some utterly obscure game to do the job for you. And these sites wonder why they get accused of shilling?
I dont know why he use this game, its not a really known games but why not..
( Lol sorry at first i was do a mistake with Oil Rush )
Last edited by a moderator:
And it just happens to be 37% faster than the competition. You need to ask yourself how he even came into knowledge of this title in the first place. Look at the rest of his games - they are all AAA titles. And then...this.
http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/56674-nvidia-gtx-660-2gb-review.html
Batman AA
Battlefield 3
Crysis 2
Deus Ex
Dirt 3 (notice it's not showdown)
Metro 2033
Shogun 2
Skyrim
Wargame Escalation
Witcher 2.
Why is this game being included? Is it just coincidence that it's massively faster on Nvidia cards and gives them a huge overall total increase? I don't believe in these kinds of coincidences.
Last edited by a moderator:
The 660 is 1% faster at 1080p at Hardware Canucks, and 4% faster at 1080p on Anandtech.
It's not a lot no - but when the rest of the tech press has the 7870 some 5% to 12% faster on average you have to wonder where this 15% or so is disappearing to on these sites.
It is actually quite interesting to analyze. The radeons don't exactly do great without msaa, but they get totally outclassed with msaa. It looks like with msaa the benchmark is completely limited by memory bandwidth (660 and 660ti are very close together, so are 7850 and 7870 and those have an unusually large difference to 7950). For some reason it just seems like the geforces are way way more bandwidth efficient in this title though I wouldn't know why - that could range from radeon driver bugs (e.g. issues with buffer compression) to game coding issues or it could be a entirely legitimate difference.Hardware Canucks benchmarks some game called Wargame:EU Escalation -
My guess would be that it's a really obscure title that Nvidia could "quick fix" somehow to get better performance out of it. AMD probably didn't have the resources.
While I believe that this kind of thing should be noted, I don't believe it should count 4%+ towards the end totals in a benchmark suite that is otherwise entirely made up of AAA titles.
Last edited by a moderator:
There's something definitively strange with the AA on this title.. allready the performance are benefits to Nvidia with 1x MSAA.. looking this setting dont exist in the Nvidia CP or AMD CCC, i can imagine this 1xAA is set in-game ..
then you have the opposite of what we can imagine with knowing the 600series and MSAA, specially on a RTS games when forcibly you should have a lot of objects, borders and the view is a lot larger of any fps or 3third person shooters games..
I will say the problem is how this MSAA is coded in the game... and not a bandwith efficient things..
then you have the opposite of what we can imagine with knowing the 600series and MSAA, specially on a RTS games when forcibly you should have a lot of objects, borders and the view is a lot larger of any fps or 3third person shooters games..
I will say the problem is how this MSAA is coded in the game... and not a bandwith efficient things..
1xAA is just AA disabled.
This is just for say the game is using his MSAA, and looking the results, we can imagine it could be strangely coded in the engine. ( in other bench, he write 0xMSAA )
( for be clear, i have nothing against the fact he use this game, specially when we see how perform the AMD vs Nvidia in Sniper Elitev2, Sleepdogs, Dirt showdown, all have their " games ", and ratio % are really just indicative, anyone should look the bench results by game and not only thoses one. )
Last edited by a moderator:
Blazkowicz
Legend
the letter X got cheaper.
there's a simple meaning though, I see it as "you can buy this for gaming".
the cards are faster than any laptop (barring exceptional gaming models) and are quite well perceived by people upgrading from a 5x slower GPU.
The 0xMSAA in the 8xMSAA graphs is rather the typo, since 1xMSAA or 1xAF are the correct shortcuts for no multisampling or no anisotropic filtering respectively. There's no such thing as 0x samples.
As for Warfare Escalation I don't know either what the big deal seems to be. I wasn't aware of the title and I might have a look at it. Instead of including in reviews the exact same set of game benchmarks I consider it refreshing to see something new for a change irrelevant on which GPUs it fairs better. Hardwarecanucks final rating puts the 660 29% ahead on average of a 560Ti which doesn't sound like unrealistic to me.
If a current 560Ti owner puts the above into perspective and especially the price difference between the latter and a new 660, I'm not so sure the latter is that much of an upgrade after all. Higher end Kepler SKUs obviously pull quite a bit more ahead, yet also at quite higher prices.
Blazkowicz
Legend
ouch, some test results of the GTX 650 are out (weird benchmarks on tomshardware, it's a joint review with the gtx 660 with two different mixes of card depending on settings)
it doesn't catch the 550 ti, so I agree in the end that the "X" is too much for this one.
it runs at only 47°C and uses a lot less power than even the 7750, so I wonder about overclocking.
it doesn't catch the 550 ti, so I agree in the end that the "X" is too much for this one.
it runs at only 47°C and uses a lot less power than even the 7750, so I wonder about overclocking.
GTX could be OK if it is restricted to GDDR5. GTX 550 Ti was available in 128-Bit DDR3-1066 versions...
Not really; since just whether a card has GDDR5 does not by itself say anything about performance, and most of the public has no idea what GDDR5 even is.
The second digit already outlines a card's performance position within its family, there is no need for a prefix to display that same information.
The only way I could see it make some sense is if it stood for how many auxiliary PCI-E power connectors the card needed. E.g. GT = none, GTS = 1, GTX = 2+.
Blazkowicz
Legend
GTX 650 could be OK if it is restricted to GDDR5. GTX 550 Ti was available in 128-Bit DDR3-1066 versions...
I'm glad I never saw these turds. I thought ddr3 GTS 450 were bad enough. actually a GT430 64bit would piss me off already and I was amazed at 64bit geforce 4MX 440 whereas there was a 420 name already for cards with slow ram.
anyway the GTX 650 can be attractive. if you're building e.g. a cheap silent mini-ITX box (with a two slot case) it's a no-brainer.
or pair it with a celeron or pentium G and you've got a crazy low watt, and cheap computer. who cares about the bar lengthes, the computer is half the price of your gaming ring and the power bill is low.
Last edited by a moderator:
Even the gddr5 550Ti didn't deserve the GTX moniker. The devaluation of that prefix began before that, but that pretty much was the final nail in the coffin for any meaning of that moniker imho. While GTX was creeping down to more cards before, these were at least faster than some of the previous generation cards carrying it (e.g. GTX 460 wasn't using top chip but still faster than say GTX 275). But GTX 550Ti undercut the previous generation slowest gtx card (the 460) by like 25% (ok the 460 768MB by a bit less but still definitely slower). So what wasn't good enough to carry the "ultimate GPU for gamers" flag (nvidia's words not mine) in the previous generation suddenly was good enough in the new generation?GTX 650 could be OK if it is restricted to GDDR5. GTX 550 Ti was available in 128-Bit DDR3-1066 versions...
So the gtx 650 just continues that sillyness (the only thing new is that now all current-gen chips are eligible for it).
But other than that it is indeed quite an ok card.
