Radeon HD 6870 could widely outperform gtx 480

[flippin_waffles]

Just to keep things in perspective in the context of the other thread. :rollseyes:

http://www.kitguru.net/components/gr...k-performance/

[Helloworld_98]

isn't kitguru that site which made up a load of BS news late last year?

I seriously doubt these results, especially seeing as the 6870 will be such a minor improvement over the 5870 and there is 4 months left to write drivers

I guess that the actual results for the 6870 in vantage will be maybe 500 higher than GTX 480 max, and the 480 will still win when both are overclocked

edit: just read it properly, they think NI will come out in Q4 2010 in 28nm, and think that if AMD get their 28nm ready by then that NI will be released this year, so in essence they're saying AMD will scrap SI and will magically tape out and be able to sell 28nm parts by the end of this year

[Olivon]

Kit ... What ???

[SKYMTL]

LOL. I love it when sites wave their hands in the air screaming "Me too!!!" after either Fud or Charlie publishes an article. It's hilarious.

[Vardant]

In other news, the sun could rise again tomorrow.

[Frontl1ne]

They're talking about NI though, which is supposed to be ATi's new arch right? The one on 28nm, which is due out next year.

Edit: Well, they seem to think that NI will be coming out this year....... IMO, it should be expected that NI will out perform the GTX 480 by a fairly large margin because by then, Nvidia should have their second gen fermis ready to roll out.

[flippin_waffles]

isn't kitguru that site which made up a load of BS news late last year?

I seriously doubt these results, especially seeing as the 6870 will be such a minor improvement over the 5870 and there is 4 months left to write drivers

I guess that the actual results for the 6870 in vantage will be maybe 500 higher than GTX 480 max, and the 480 will still win when both are overclocked

Kitguru are the guys from driverheaven, which was also quite credible. I'd put more faith in their sources than Fuad's.

[RPGWiZaRD]

This is more speculation than news. Ofc HD 6870 should be faster than a GTX 480 if there's at least some sense in AMD/ATI, the interesting part is exactly how much?

But a relatively small leap for NI is prolly to be expected, everything points to it. I'd be satisfied when taking all stuff into account, seeing around 20% perf advantage over GTX 480 (or a bit more in tesselation heavy scenarios which is what I expect them to focus on), however I wouldn't be suprised if average advantage is below that. We have power consumption and other factors also, if it could outperform GTX 480 by 20% at-very-likely-to-be-below GTX 480 power consumption levels, that would be a good achievement in this day and age for ATI/AMD for example even if on paper a 20% perf advantage would sound quite small for a new series and all, it would still be a huge win for AMD/ATI.

[Dimitriman]

Wow.. am I really reading Nvidia boys saying that the 6000 series will barely outperform a GTX480?

You guys do realize that building a graphics chip is not playing with lego right? That there are millions in RD costs and that even this "tweaked" version needs to be a project with expected returns that outweighs its cost?

I don't think AMD will release a chip marginally faster than its 1 year current architecture, even if this is a tweak, a single chip performance +/- 10% that of a 5970 should be expected.

Unless their directors do not want to sell cards anymore and are just in a strip club somewhere blowing all their cash on things even hotter than a GTX480.

[570091D]

Wow.. am I really reading Nvidia boys saying that the 6000 series will barely outperform a GTX480?

You guys do realize that building a graphics chip is not playing with lego right? That there are millions in RD costs and that even this "tweaked" version needs to be a project with expected returns that outweighs its cost?

I don't think AMD will release a chip marginally faster than its 1 year current architecture, even if this is a tweak, a single chip performance +/- 10% that of a 5970 should be expected.

Unless their directors do not want to sell cards anymore and are just in a strip club somewhere blowing all their cash on things even hotter than a GTX480.

and you, of course, realize this news came from kitguru... the same site that saw a custom gtx480 pcb and declared the coming of the 512sp gtx485.

and yes i do think that amd would ship a modified version of evergreen with better tessellation performance and only a modest improvement for all other aspects of the chip.

[Dimitriman]

and you, of course, realize this news came from kitguru... the same site that saw a custom gtx480 pcb and declared the coming of the 512sp gtx485.

I haven't even looked at the kit guru article. I'm just making an educated guess based on what I understand about the chip so far.
And based on the likely strategic decisions of any successful business in this scenario.

If AMD releases a 6000 series that dissapoints in performance, they will not have made a good strategic decision. Unless of course they call it 5890, supposed to simply match GTX480 performance, and leave 6000 series as a real product for next year.

I simply cannot see a 6000 series GPU with very marginal improvement over the current, especially in areas such as Tesselation.

[RPGWiZaRD]

OK, I'll throw you something to think about.

- NI 40nm is only meant to be a stopgap until SI on 28nm is coming. IE, a short-lived chip.
- NI have had very little development time and has a very tight schedule => no time for drastic improvement/changes. I personally think this might even have been a last-minute decision cuz of complexities that 28nm manufacturing or the SI architecture itself had encountered so they needed more time.
- GTX 480 already has a significant perf advantage over HD 5870.
- NI's purpose is just to beat Fermi in performance and even better, at lower power consumption making Fermi look even worse. There's no point in making this card significantly faster, it's better to strike harder with SI 28nm. Making fast card now can be countereffective later on.

[Starscream]

In other news, the sun could rise again tomorrow.

Source?

[PrometheusCon]

Sounds to me that this is what the 4890 is to the 4870.

[MrMojoZ]

Sounds to me that this is what the 4890 is to the 4870.

And if so, jumping to a 6xxx model number would be a giant mistake.

[Sam_oslo]

LOL. I love it when sites wave their hands in the air screaming "Me too!!!" after either Fud or Charlie publishes an article. It's hilarious.

+1. LOL they have been really fast to counter Fud. LMAO.

This GPU-business is getting more like monkey-business. It is getting really difficult to discuss GPU in these days, too much PR and propaganda, including in this forum.

[informal]

OK, I'll throw you something to think about.

- NI 40nm is only meant to be a stopgap until SI on 28nm is coming. IE, a short-lived chip.
- NI have had very little development time and has a very tight schedule => no time for drastic improvement/changes. I personally think this might even have been a last-minute decision cuz of complexities that 28nm manufacturing or the SI architecture itself had encountered so they needed more time.
- GTX 480 already has a significant perf advantage over HD 5870.
- NI's purpose is just to beat Fermi in performance and even better, at lower power consumption making Fermi look even worse. There's no point in making this card significantly faster, it's better to strike harder with SI 28nm. Making fast card now can be countereffective later on.

"- GTX 480 already has a significant perf advantage over HD 5870."
Define significant. If it's 15% on average(as it is) you need better term : minor. And given the die sizes of both chips it's NV that's in trouble with perf./mm2 efficiency,not AMD.

[Helloworld_98]

all this Northern Islands Southern Islands crap is confusing, you never know which is which

[Kuroimaho]

Kidguru, now that's a source to reckon with...

- NI's purpose is just to beat Fermi in performance and even better, at lower power consumption making Fermi look even worse. There's no point in making this card significantly faster, it's better to strike harder with SI 28nm. Making fast card now can be countereffective later on.

I doubt they would bother or would aim for that, they could have raised the clocks of the 5870 if they really wanted to half a year ago. More likely they test out the new arch and since the difference isn't that significant they could take over the 480, but as a new rev and 485 is coming this would be aiming at a moving target.

[SKYMTL]

Kitguru are the guys from driverheaven, which was also quite credible. I'd put more faith in their sources than Fuad's.

Some blokes from DH broke off to form their own website. How does this give them cred?

To me, this is just a bait and switch in order to get traffic to a new website.

[flippin_waffles]

Some blokes from DH broke off to form their own website. How does this give them cred?

To me, this is just a bait and switch in order to get traffic to a new website.

Thanks. Maybe you'd like to compare the amount of credible articles on fudzilla to the amount of credible articles on kitguru and DH? Didn't think so.
Wasn't that Zardon bloke owner or part owner of DH?

[BababooeyHTJ]

Kitguru are the guys from driverheaven, which was also quite credible. I'd put more faith in their sources than Fuad's.

Oh come on Intel's dual IGP is right around the corner. Fud is very credible.

[Metroid]

So much hype for nothing. This website is doing what FUD has been doing for years.

[SKYMTL]

Thanks. Maybe you'd like to compare the amount of credible articles on fudzilla to the amount of credible articles on kitguru and DH? Didn't think so.

You're still comparing KG to DH. Why? One is a publication that posts straight up articles while the other is a rumor mongering site with actual articles thrown in for a touch of legitimacy.

I'd actually say the score between Fudzilla and KG since KG opened its doors is about the same. Fudo publishes just as many PR pieces, reviews and actual News articles.

The main thing you have to remember is that articles such as this are traffic generators; hence why certain sites do it.

[eric66]

well of course it must do that whoever thinks that amd will waste there precious 1 year just to add few minor upgrades to evergreen core must ........ hmm well i don't know what to put here