Semiaccurate says: "[Kepler] will win this round on just about every metric."

If the new Nvidia card out performs the 7970, is it really "beating" it since it is coming out so much later?

I don't know, is the 7970 really "beating" the GTX 580 since it came out so much later? GCN and Kepler are the same generation, no matter when they come out, so that is the competition.
 
I really hope Nvidia integrates PAP into their flagship GPUs. This has been the only thing keeping me from considering their cards as my computer doubles as my HTPC.
 
my point is that it doesn't matter that AMD releases first if Nvidia's offering totally crushes them...it only really matters in the long run if they are really late (AMD will get more sales this way even if they are the inferior card because people are impatient)...releasing their card first only serves to make their cards appear to be really amazing

I think it does more to saturate the high-end market than make their cards seem fantastic.

"Well, I just bought this card 3 months ago... Can't justify another half grand for 25%".
 
I don't know, is the 7970 really "beating" the GTX 580 since it came out so much later? GCN and Kepler are the same generation, no matter when they come out, so that is the competition.

That thinking can backfire as well like 1800xt.

The real lesson is most techs think in generation. The longer after one of the party waits to release their generations competing product, the less likely people are going to make the switch or hold out for it. Kelper can win the single GPU performance crown. But if they are within 10% and available 3-4 months earlier and then able to drop prices as production has already picked up, the harder it will be for Kelper to sell well.

Which you kind of have to feel bad for Nvidia this round. AMD made a dramatic change to the design of the CPU's and combined it with a massive jump in transistor size and still got it out and measurably faster then Nvidia's previous best. Kelper isn't going to be that revolutionary besides the big jump. Usually when something like this goes on its the other way around (the r600, FX5k, and a few others). This GCN card might not be the fastest card ever, but all things considered, even if GCN cards don't outsell Kelper cards, it will be fantastic start to a completely new architecture.

I would like to note that I don't know enough about Kelper, I don't know how much it deviates from Fermi.
 
An advantage AMD has is they can just release the HD7990 when the kepler arrives and still keep the performance lead. Or a revised HD7970 with much higher clock speeds.
 
if it does come out in june like kyle predicted
then many people will just wait out another 6 month for 8xxx series from amd
kepler dose need to come out within 3 month or less ;/
 
As much as I love to speculate and argue these points incessantly,

1) Semiaccurate is aptly named, and
2) NVIDIA has released no new information on Kepler.

Do I suspect, given industry history, that NVIDIA's flagship part will be faster? Yes. But I would be surprised to see them win efficiency, power consumption AND performance. That would be a significant engineering improvement over the last 2 generations of NVIDIA cards, compared with AMD.

So until NVIDIA drops more information on Kepler, it's all purely guesswork. At the end of the day, I'll buy whatever is faster or suits my needs the most.
 
if it does come out in june like kyle predicted
then many people will just wait out another 6 month for 8xxx series from amd
kepler dose need to come out within 3 month or less ;/

No need to buy the 8000 series or GTX 700 series, they'll be refreshes of the current arch and only 15-20% faster.
Whatever I buy this gen, I'm sticking to it until both companies release a new arch again.

New arch: Buy
Refresh: Skip
 
my point is that it doesn't matter that AMD releases first if Nvidia's offering totally crushes them...it only really matters in the long run if they are really late (AMD will get more sales this way even if they are the inferior card because people are impatient)...releasing their card first only serves to make their cards appear to be really amazing

Exactly... heck, when a company fails to deliver a good driver for an excellent product, it doesn't matter to me if it's a couple months later that the superior one comes out... I bought a 7970 and it's going back now to wait on Kepler. The two months "suffering" (lol) with my 570 SLI will go by like that and I'll get a nice, working set of SLI Kepler "560-570 equivalent in line-up price-wise" cards day 1 soon enough that'll blow away what the 7970 could do for only a little more with any luck. Most rumors currently are toward end of march, sometime in April, for Kepler... it's worth the wait for a WORKING product.
 
if it does come out in june like kyle predicted
then many people will just wait out another 6 month for 8xxx series from amd
kepler dose need to come out within 3 month or less ;/

I'm going to assume that June will be when it should be in high volume, and April will essentially be a soft launch, so you'll still be lucky to get one. As you mentioned however, people will be waiting for something new. It's just the name of the game, there's always going to be something better around the corner. Unless it something like less than say two months away I personally wouldn't sweat it. Either way you still are going to have buyers remorse within a month of your purchase.
 
No need to buy the 8000 series or GTX 700 series, they'll be refreshes of the current arch and only 15-20% faster.
Whatever I buy this gen, I'm sticking to it until both companies release a new arch again.

New arch: Buy
Refresh: Skip

hehe i typically like to do the opposite
because refresh drivers are always SO much better than new arch drivers - so that part of the "new GPU" headache is gone
 
No need to buy the 8000 series or GTX 700 series, they'll be refreshes of the current arch and only 15-20% faster.
Whatever I buy this gen, I'm sticking to it until both companies release a new arch again.

New arch: Buy
Refresh: Skip

what
do you still have that 570 gtx in your sig?
that's refresh of 400 series
 
Last edited:
Exactly... heck, when a company fails to deliver a good driver for an excellent product, it doesn't matter to me if it's a couple months later that the superior one comes out... I bought a 7970 and it's going back now to wait on Kepler. The two months "suffering" (lol) with my 570 SLI will go by like that and I'll get a nice, working set of SLI Kepler "560-570 equivalent in line-up price-wise" cards day 1 soon enough that'll blow away what the 7970 could do for only a little more with any luck. Most rumors currently are toward end of march, sometime in April, for Kepler... it's worth the wait for a WORKING product.

I've had 0 driver issues since reformatting, but then again I've only been playing BF3 and SS2, can't speak to anything else.

Should I have to reformat in a perfect world, just to get drivers working? No, but I've been in this game a long time and I'm used to display drivers getting moldy, from both the green and red teams. I spent less time reformatting then I did tweaking my overclock, so for the best performance possible, it was totally worth it. YMMV.
 
all these OMG AMD drivers suck i'm sending mine back , post pics or it didn't happen :). 7970's are sold out so your negative comments didn't have much effect go cry in a corner with your favourite Nvidia GPU telling it that one day it will all be better !
 
I think Charlie is trolling. This isn't him. He absolutely hates nvidia, i'm guessing he's being sarcastic based on some sort of inside info.
 
all these OMG AMD drivers suck i'm sending mine back , post pics or it didn't happen :). 7970's are sold out so your negative comments didn't have much effect go cry in a corner with your favourite Nvidia GPU telling it that one day it will all be better !

7970's being sold out dont mean much like you think.
 
The longer after one of the party waits to release their generations competing product, the less likely people are going to make the switch or hold out for it.

How long is "acceptable" between releases from the two companies though? It seems if they don't launch on the exact same day everyone goes bezerk. The 7970 isn't out two weeks yet and is in short supply. A bit too early to panic.
 
I find it odd myself of why people don't do a switch out the right way and bitch about a driver.. most think windows 7 is god and will not reset the cmos for a new DMI data pool update for the new hardware they just placed in the PCI Express slot..
 
read the comments on that story, guys.

I think Charlie is trolling. This isn't him. He absolutely hates nvidia, i'm guessing he's being sarcastic based on some sort of inside info.

Likely just posting FUD to see where it ends up and who's taken the bait.
 
I think Charlie is trolling. This isn't him. He absolutely hates nvidia, i'm guessing he's being sarcastic based on some sort of inside info.

agree and i think it has more to do with people overhyping GK104 which we all know is the replacement for the GTX 560TI and not the GTX 580. so the odds of it out performing the 7970 is a complete joke.. thus you get this nvidia troll article from Charlie which it deserves.

the GK100 which IS the replacement gpu for the GTX 580 won't be out any time soon and Kyle's predicted June release is more then likely correct.
 
I think Charlie is trolling. This isn't him. He absolutely hates nvidia, i'm guessing he's being sarcastic based on some sort of inside info.

As much as I think Charlie is a nut sarcasm isn't his style, he usually calls it the way he sees it, though I've usually that he was high when he saw it. Maybe he is high this time but my guess is that nVidia's mid-range part will be on par with AMD's high-end part.
 
shhhh!

watching people actually take this troll bait is hilarious, let alone quote the article in a thread title HAHAHAHA
 
shhhh!

watching people actually take this troll bait is hilarious, let alone quote the article in a thread title HAHAHAHA

Again, just not his style and the article has a number of specifics, like RAM size, ports and so forth. Indeed it would be very odd for Charlie to help out nVidia because that would only reinforce people waiting for Kepler. He's never held back before where it concerns nVidia, so why now?
 
Because it doesn't make any sense. At 256-bit bus and 2GB memory and the fact that this card is supposedly a midrange card it's somehow supposed to compete, and thrash, the top end AMD cards?

He doesn't hold back regardless of whom he's talking about, but the above makes no sense whatsoever. He's posting FUD to see where it will end up and who runs with it, and apparently the answer is a whole lot of you.
 
Because it doesn't make any sense. At 256-bit bus and 2GB memory and the fact that this card is supposedly a midrange card it's somehow supposed to compete, and thrash, the top end AMD cards?

He doesn't hold back regardless of whom he's talking about, but the above makes no sense whatsoever. He's posting FUD to see where it will end up and who runs with it, and apparently the answer is a whole lot of you.

It might be garbage, but it's garbage that's favorable to and helps nVidia, that's why I don't think it's complete bunk. Charlie does EVERYTHING to hit nVidia usually and certainly nothing to help it.
 
supposedly a midrange card it's somehow supposed to compete, and thrash, the top end AMD cards?

It's possible. Until we see the entirety of the architecture we can't say for sure that nVidia's next mid-range card won't be able to compete with AMD's latest top-end card.
 
Because it doesn't make any sense. At 256-bit bus and 2GB memory and the fact that this card is supposedly a midrange card it's somehow supposed to compete, and thrash, the top end AMD cards?

He doesn't hold back regardless of whom he's talking about, but the above makes no sense whatsoever. He's posting FUD to see where it will end up and who runs with it, and apparently the answer is a whole lot of you.

You know, I just re-read his article, and at no point does he actually say GK104 (as opposed to the GK100 or whatever the flagship-to-be is) will beat a 7970, what he says is that nVidia will beat AMD this round, and that he has seen 2 GK104s. You guys seem to be confusing the 2 points.

1) I don't think he's totally BSing, because he wouldnt do nVidia the favor of showing them in a good light

2) I DO think he's trolling, trolling everyone who interprets the piece to say GK104 will beat the pants off tahiti. Read it more closely and see that while you could construe it that way, at no point does he actually say that.

3) What he's saying is that the nVidia lineup will beat the AMD lineup, and that the GK104s he's seen look in good shape.
 
Last edited:
Because it doesn't make any sense. At 256-bit bus and 2GB memory and the fact that this card is supposedly a midrange card it's somehow supposed to compete, and thrash, the top end AMD cards?

The 6970 manages to compete with the GTX 580, and it only has a 256-bit memory bus. Bus width isn't everything - maybe there are architectural changes in Kepler than reduce the impact of memory bandwidth.

That being said, I agree with Satrafton - the article says that Kepler will beat GCN, not that GK104 will beat the 7970. I think everyone expects GK100 to beat the 7970.
 
It's possible. Until we see the entirety of the architecture we can't say for sure that nVidia's next mid-range card won't be able to compete with AMD's latest top-end card.

nVidia releasing its mid-range stuff before the high end means one of two things. Kepler blows chunks or kicks serious ass.
 
I think everyone expects GK100 to beat the 7970.

Not in terms of pure performance but it could be that nVidia's performance to price ratio might be compelling, kind of like AMD and nVidia switching places this time.
 
Last edited:
amd has been using 256bit memory since 9700pro and all of the sudden it sucks?

wow.
 
So Kepler is going to be slower than the 7000 series?

the answer to all your Kepler questions is as follows: "you'll know when it comes out"

everything else is biased noise developed to get hits for websites
years of video card released have taught me that

Again, just not his style and the article has a number of specifics, like RAM size, ports and so forth. Indeed it would be very odd for Charlie to help out nVidia because that would only reinforce people waiting for Kepler. He's never held back before where it concerns nVidia, so why now?

dude - the article DRIPS sarcasm
any credible news would be in a proper, somewhat-journalistic article
don't you think?
 
dude - the article DRIPS sarcasm
any credible news would be in a proper, somewhat-journalistic article
don't you think?

Yeah I think, and I know Charlie a little, I've exchanged emails with him in the past. Your first mistake is this whole credible and proper news thing, I mean the name of the site is Semiaccurate!

Again, either Kepler sucks or kicks ass. I don't know and neither do you but this is TOTALLY outside of Charlie's nature, that much can't be argued.
 
Yeah I think, and I know Charlie a little, I've exchanged emails with him in the past. Your first mistake is this whole credible and proper news thing, I mean the name of the site is Semiaccurate!

Again, either Kepler sucks or kicks ass. I don't know and neither do you but this is TOTALLY outside of Charlie's nature, that much can't be argued.

shoot dude, email him again and ask! haha
 
nVidia releasing its mid-range stuff before the high end means one of two things. Kepler blows chunks or kicks serious ass.

They are releasing the midrange first because the gk100 had major problems. That is why there is a gk110, the respin of the gk100. If they didn't have to do this the high end cards would have come out first.
 
Back
Top