• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

New Amd Card Specs

My point is that you don't need to be a practicing expert on an issue to hold an opinion about it.
And my point was, that the experts view the term "patent troll" as a fad term from a year or more ago, and not a valid one. So my real, overall point is, we, the uninformed, make poor judgements and therefore flawed opinions.

I'll give an example of this in a different setting: I have a friend who was not incredibly hardware savvy. He wanted to build his own PC and came to me for advice, I asked him what his goals were and he said he wanted the fastest processor with the fastest video card. Fair enough. Then he said, "I want Intel, I want the best, I want a Pentium." I just kind of stared at him, as this was at the tail end of 2010. I then explained to him that Pentiums were low end chips, and that Intel has the Core ix line, and to look at that ect. He had an opinion and general idea of the hardware world, and a fuzzy idea about what works and that lead to his opinion. That Pentiums were the best processors you could own in 2010. So yes, you do not need to be an expert in a field to form an opinion on it, but its most likely you will not make the best judgement. Hence why I would take a patent lawyers opinion on the current state of patents, than computer joe's opinion on it.
 
Last edited:
Hence why I would take a patent lawyers opinion on the current state of patents, than computer joe's opinion on it.

Wouldn't that depend on the action you wanted to take with said opinion?
 
And my point was, that the experts view the term "patent troll" as a fad term from a year or more ago, and not a valid one.

Given the landscape of patent litigation, I'm not sure how someone could come to that conclusion? What term would you apply to companies like Intellectual Ventures or NTP? Their entire business model is to make money filing suit with their portfolio of questionable patents (most of them obvious, trivial, or with many examples of prior art) against companies they claim are infringing. That's all they do. They don't make anything and they don't own any real assets. Patent trolling describes what they do perfectly.

This isn't a new phenomenon or some fad term either. This type of thing has been going on for decades. It started in other industries and migrated to technology when the number of patents filed ballooned in the mid-1990's. It's now gotten severely worse with the exponential onset of software patents.

What Rambus was doing arguably wasn't patent trolling in the beginning. They at least had developed similar technologies and had acquired some valid patents on it that could have conceivably been protected in court. What they ended up doing though was trolling with their patents once they saw the DRAM market winds blowing the other way from their business.
 
Patent trolling describes what they do perfectly.
Then those specific, few examples, are trolls.

This isn't a new phenomenon or some fad term either.
Google: "Term: Noun: A word or phrase used to describe a thing or to express a concept, esp. in a particular kind of language or branch of study." The TERM "Patent Troll" is a fad. Patent trolling is not, however, a fad. It's basically the purpose of patents anyway. You misunderstand my use of the word term and went on a tangent here. I am not arguing with you on your point there, I agree.

What Rambus was doing arguably wasn't patent trolling in the beginning. They at least had developed similar technologies and had acquired some valid patents on it that could have conceivably been protected in court. What they ended up doing though was trolling with their patents once they saw the DRAM market winds blowing the other way from their business.
Then that is smart business for them until they could find a way to produce something in a field they specialize in. Protect their assets until they could make use of them in the best manner possible. Would you do any different if you were in their shoes???
My point is, the TERM "Patent Troll" carries too much of a negative connotation and its really beginning to annoy me, it is used too often and in many cases for the wrong reason. Yes, there are lots of patent disputes going on. What did people expect after the patent system started being used in particular fields? Once a field reaches a mature stage, it is only logical that patent disputes would begin being common place. It's almost unavoidable to begin suing others over patent infringement. They are not being trolls, they are protecting their business.
When a business is nothing but patents without product, then yeah, I guess they are a troll.
 
thread_direction.gif
 
I like those TDP ratings getting 470 performance for 90 Watts will be really awesome.
 
No, they will support it, but PCIe versions are backwards compatible. They will work fine in PCIe 2 boards.
 
It won't make it out before BF3, therefore 6950 x 2 inc.

I'm not worried about it. I just keep trolling NE's Open box for a 6950 2GB and I swore peeps are fast as hell when they come in :(
 
I need to upgrade my 4830 512 for Skyrim. Was thinking about just grabbing the Asus 6950 DCUII 2gb and unlocking the 6970 within... however if I wait for the 78xx series to come out, I can grab the 69xx series at drastically reduced prices OR grab a 78xx...

OH THE INDECISION!
 
always wait!

even if the cards disappoint, the drop in price on the previous gen means you win

information = $$$
and we have no info at this point
 
I just got off the phone with my mate, he has a sapphire 6950 (the original PCB with dual-bios switch) that he's looking to get rid of for ~$215...

INDECISION OVERLOAD.
 
This one I know for a fact is less than 6 months old (I was there when he bought it... from my store) and he today confirmed the dual-bios switch, which means it can most likely unlock additional shaders, and it's a sapphire, which is one of two brands I'll ever trust, and its the 2gb model.

Plus I'm in Australia and per Yahtzee's great beaches tax, the cheapest sapphire 6950 2gb is the Dirt3 edition, which is here.

and I have THREE dirt3 unlock codes on my desk right now ( I know a lot of AMD reps), so I am in no way tempted by the custom heat-spreader label.

All things considered, it's a pretty tempting deal.

To un-derail the thread

or to rerail...? it

Will the 78xx be just updated 69xx architecture or will it be new? I know it's a bit difficult to draw the line between new arch. and updated old arch., but if I recall, the 68xx and 69xx chips were just die-shrunk 58xx chips with clock adjustments. Is this something we can expect from the 78xx? or even the 79xx?
 
The way AMD is talking it's going to be all Graphics Core Next, which you can read up on at anandtech and PCPerspective. I'm guessing high end and midrange (7850-7990) And maybe even the 7700 and 7600 cards, though I wouldn't be surprised to see some tried and true VLIW-4 on the low end. The 6900 cards were more than just a die shrink, edit: In fact there was no die shrink, they're on the same 40nm process, due to TSMC cancelling/failing to meet expectations on the 32nm node. While the lineage was still apparent in the 6900 architecture they switched to VLIW-4 instead of VLIW-5, as well as doubling up some of the non-shader hardware compared to the 5800 cards. The 6800's got a re-engineered memory controller that was something like half the size of AMD's previous 256 bit controllers with around 80% of the throughput, big efficiency gains there. *edit: All the 6xxx cards except for the 6750 and 6770 (rebranded 5700's) also got fixed AF units that they somehow messed up on the 5000 series. The problem only seems to show up in trackmania and a few niche titles though so it's not that big of a deal.
 
Last edited:
If the replacement for the 6870 (7870?)is more then 70% faster and around $200 I am selling my 6870 and getting the new card. If the speed bump is less then that, I will be picking up a new 6870 at whatever reduced price for some CF love.
 
As far as those specifications are concerned, GCN is going to be confined to the HD7900 series only, and the others are going to be VLIW4 as per previous cards. Of course, no confirmation that any of this is true. I'm anticipating the HD7800s to be similar but not identical to the HD6900s in the same way that the HD5750 and HD5770 weren't just die-shrunk HD4870s and HD4850s.
 
aah.. the switch from my 4870 to 7870 will be glorious..
I wonder how these cards (if all true) with Rambus memory will be priced. Rambus likes to be expensive.
 
My point is that you don't need to be a practicing expert on an issue to hold an opinion about it. I'm not a baseball player, but I still have opinions about the game. Similarly, many people aren't patent lawyers, but they're still entitled to their opinions on patent law and its misuse.

Patent trolling is a real concern, and what Rambus did or didn't do with their patents is fair game for discussion. Especially since they brought the issue into a public courtroom.

You continue your apples to oranges comparisons...

Baseball is a publicly accessible sport, where we have nearly full access to any and all information that coincides therein.

Patent law, and the likes are completely different, where you might see 20% of the information or transactions that go back and forth between parties. You don't see/hear about all the stuff thats goes on pre-trial behind closed doors. You only see what happens once the company escalates it to the courtroom.

By this logic people are drawing conclusions on miss-information or partial information, rendering their opinions null and void. I think thats what DOOMHAMMA was trying to get across.
 
I'm personally getting a boner over the 7950 -- 1900ish cores, 2GB of memory and hopefully with an intro price of about 300 dollars -- yes please. I don't have the insane budget most would have on graphics, since it looks like we have to wait another 5 months or so that gives me time to save, still $350 dollars will probably be my upper limit.

I'm running my GTX280 and I game at 1920X1080 - so the 7950 would I'm sure let me instantly run BF3 maxed out in all its glory.

The thing Im most interested in though is its massively parallel processing for a few things I'm playing with in Elecomsoft's wireless security auditor and it's ability to take advantage of GPU processing.

Since performance on that is very similar to how Radeon's perform on bitcoin... it's going to be fun times. Just imagining two 7950's *drool*
 
I hope these things can deliver double performance over the 6970 at a minimum. A 125-130% increase would be fucking hot. I need more power for eyefinity, crossfire plain sucks in triple head so I need more single card juice.
 
Double performance over the 6970?!

Even expecting 50% more performance is probably wildly optimistic IMO. 30% is more is fairly likely.
 
Will the 78xx be just updated 69xx architecture or will it be new?
No one really knows. There have been rumors that it'll be a split arch. launch branding a la 6xxx series. There have also been rumors that it'll be a bottom to top new arch. launch.

My WAG is the former rather than the latter since GCN is supposed to be hotter and consume more die space then 6xxx style VLIW4 GPU's for the same amount of performance. Not as bad as Fermi but still worse. It wouldn't be very efficient or cost effective to replace every product range, especially the low end, with a bigger hotter GPU.

For the very high end its perfectly fine to do that since the people who go after those very high end GPU's usually aren't too concerned with heat or power usage. But for budget or mid range stuff a shrunk and improved VLIW4 makes lots of sense.
 
Double performance over the 6970?!

Even expecting 50% more performance is probably wildly optimistic IMO. 30% is more is fairly likely.

A jump from 4870 to 5870 pretty much double the performance.
 
Hush you.

I want these FOR my gaming machines and bitcoin miners. Win-Win!!!

I think what he means is that bitcoins themselves will be worth so little people will no longer bother with it. The value of a bitcoin is dropping faster than lindsay lohan's panties after 2 margarittas.

By the time the 7000 series release a bitcoin will not be worth the power you use to mine it.
 
A jump from 4870 to 5870 pretty much double the performance.
Nope. I'm sure you can cherry pick a few games where that is true generally the difference was more like 30-40% faster for the same settings. But why would you compare the 5870 to the 4870 when you had the 4890 to consider as the fastest 4xxx card? Its not like that card was unusual, hard to get or all that expensive either vs. the 4870.

But why even discuss that at all? 4xxx vs 5xxx performance tells you nothing about how 6xxx will perform vs the 7xxx series GPU's. After all the difference from the 5xxx GPU's to the 6xxx was totally different too, around 20-30% IIRC. Past performance has no bearing or predictive abilities for future performance even from the same manufacturer.
 
Hah. Love the Rambus hating...

For those not in the know: Rambus XDR memory has and is still to my knowledge being used as the main system memory in the Sony Playstation 3.

So it's not as if AMD would be jumping in headfirst by going with XDR2. Second gen XDR is a totally plausible alternative to DDR, or at least Rambus wants everyone to think that. I just hope AMD is good at hiding latency.
 
I think what he means is that bitcoins themselves will be worth so little people will no longer bother with it. The value of a bitcoin is dropping faster than lindsay lohan's panties after 2 margarittas.

By the time the 7000 series release a bitcoin will not be worth the power you use to mine it.

You know, I for one would welcome some Lohan panty drop benchmarks on the front page.
 
2048 SPUs eh? Sounds like a bitcoin beast.

That's what I was thinking!

Oh look, more BC's per slot, oh and better graphics too!

WINNING!!!



(and if you have a hard time swallowing the whole bitcoin thing, replace "BC's" with "PPD")
 
Hah. Love the Rambus hating...

For those not in the know: Rambus XDR memory has and is still to my knowledge being used as the main system memory in the Sony Playstation 3.

So it's not as if AMD would be jumping in headfirst by going with XDR2. Second gen XDR is a totally plausible alternative to DDR, or at least Rambus wants everyone to think that. I just hope AMD is good at hiding latency.

XDR2 is technically sound but very expensive and doesn't have a big enough performance advantage over DDR5 which has yet to be maxed out. It also isn't used in the PS3 or hardly anywhere for that matter so supply is a real issue. XDR2 rumor is pretty widely recognized as BS at this point. For top end they'll likely go to 384 bit wide memory bus + faster DDR5 + bigger on die cache.
 
Nope. I'm sure you can cherry pick a few games where that is true generally the difference was more like 30-40% faster for the same settings. But why would you compare the 5870 to the 4870 when you had the 4890 to consider as the fastest 4xxx card? Its not like that card was unusual, hard to get or all that expensive either vs. the 4870.

But why even discuss that at all? 4xxx vs 5xxx performance tells you nothing about how 6xxx will perform vs the 7xxx series GPU's. After all the difference from the 5xxx GPU's to the 6xxx was totally different too, around 20-30% IIRC. Past performance has no bearing or predictive abilities for future performance even from the same manufacturer.

The 4890 was running on the same 40nm process as the 5870 and came along later in the life cycle (it was more or less a "4870 Ultra", all they did was shrink the die and and increase the clock speed), the 4870 to 5870 was a valid comparison as they were both launch products and the 5870 represented a shift in both process and architecture. The 5870 to 6970 was again on the same 40nm process, the performance gains were all from architecture enhancements and tweaks.

This time were doing a double die shrink jump going tom 40nm past 32nm down to 28nm. That's an absolute shit load more transistor space and it's not unreasonable to expect an 2 fold increase in performance this time around, especially if were looking at a major (working from the number of stream processors) update in the architecture.
 
A jump from 4870 to 5870 pretty much double the performance.

Firstly, no it was more like 80%. Secondly, that was the 5th best performing card of its time, being compared against the top performing card of its time. If you're going to do that you might as well say the HD7970 will be "twice" (80%) as fast as the HD6870. I think that sounds about right to me. In other words, about 40% faster than the current best (GTX580) seems about as much as can be expected.
 
Back
Top