Printable View
THANK YOU! Who f'ing cares what it's called?Quote:
Originally Posted by Kunaak
Let me just add to this by saying that they aren't two separate cards, they are simply two separate names. NVIDIA decided to change it late in the development cycle, to coincide with the chipset naming scheme.
All this BS about NVIDIA holding out on the "real" G70 needs to stop. The "real" G70 and NV47 are one in the same. People are reading WAY too much into this.
That's your opinion sir, and you gotta admit that something weird is going on.....
Why you should care....would you care about a possible 90nm 32pipe nVidia part waiting to be released after R520?
signed....... :confused:
i won't even get 11K SLI at stock. :(
Quote:
Originally Posted by Turok
It also means prices will be lower.
Not necessarily....it just means there will be competition. That competition could bring along anything, but prices are largely controlled by other factors until a product's EOL. Of course, if the price is above MSRP, competition (with a large enough supply) will lower it....we don't have that situation though.
I think prices can only get better as a 20 pipe 7800 GT can be had for only $374 shipped from newegg. We've already seen that no one wants to pay $600+ for a video card so you can be fairly certain that the next gen cards will not exceed that mark as sales will suffer. Of course you will always have those few vendors that just don't get it and will seek to price gauge the consumer.
yes but does it really matter? as long as it performes way over any other card to the day i dont care if the core is called "heheheh9595959"
but nice find ;)
buhu.. haha i mean cmon... :stick:Quote:
Originally Posted by slider99
this answers explains what i totally think about this theory. this theory (for my opinion) is true but as the Cybercat told there is no secret G70. nv47 and g70 are same only reason nvidia changed name nv47 to g70 is to get more attention.Quote:
Originally Posted by Cybercat
what are you talking about? We don't know of such a thing, and any speculation that the "real" G70 is some sort of 90nm 32-pipe 512-bit 1024MB dual-GPU 32x PCIe super card, or whatever you feel like adding to the specs, is just useless made-up hype.Quote:
Originally Posted by Tim
We have the G70 already. Whether it was originally called the NV47 or not really doesn't matter.
lol you must have a decent amount of free time to even consider giving half a crap about this ;)
-CaT
wow, easy there man, no need to get all cranky on me...I was simply saying that nVidia probably has something up their sleeve...made up hype...definitely not...do I need to remind you about the pulled shamino article..? I hope not.Quote:
Originally Posted by Cybercat
Well he said that he was all day burning CD's...so yeah that gives him some time....you probably enjoyed reading every bit of it....I sure did...was a nice read...and interestingly enough....Quote:
Originally Posted by CaTalyst.X
Just be glad he took the time
If this is true, I think there is going to one hell of a showing in the next couple of months...
It's not the fact they did it that I care about, it's the fact that they tried so hard to cover it up that makes me wonder...Quote:
Originally Posted by CaTalyst.X
Quote:
Originally Posted by Cybercat
Again, just like I stated in the first post, TSMC's fabs are in taiwan, the stamp on the chip says korea, but IBM produces chips in korea, and IBM didn't make these chips... Also, why be internally coded as NV47, if they're stamped G70? Why take all the time and effort of telling everyone it's going to be the G70 when it's really the NV47?...You don't do that just because you felt like it, you do that when you have something to actually gain from it! The real question here is WHAT IS THAT THEY HAVE TO GAIN?
Yes, you can say "well, they parted out the NV47", inwhich case I'm gonna tell you to have some common sense, and remember that they're coded near the end of the process, NOT AT THE BEGINNING! Using a transistor from the NV47 wouldn't give it the name NV47, now would it? Think back to windwithme's fx-55 that was coded to read as an FX-57...That wouldn't occur just because they used a basic design, that's occuring because they are built from the ground up as a NV47.
Finally, yes, I had alot of time when I wrote this article. My main job is handling studio recording and mass production in my hometown(meanwhile, I end up looking up stuff like this during my side-job), inwhich case there's times(like the day in question), where I'm literally burning thousands of cds... That gives you more than a bit of time to look a few things up when something looks suspicious to you, wouldn't you agree?
If you don't like the article, there's no need to flame. No need to say "you're wrong", without any sort of proof. No need to post "well, I don't care what they call it", when you didn't even read thru it, it's not just the fact that they 7800gtx turns out to be merely the NV47, but why nV went thru all this simply to make it appear NOT TO BE the nV47.
You're asking a fundamental question that no one but NVIDIA themselves can answer. Your question essentially comes down to, why use the "G70" name in the first place? Why would NVIDIA change their naming scheme? Well for whatever reason they saw it as being more marketable and that it made sense at the time. Exact reasons are still unknown, but again, I don't think it really matters. NVIDIA would like it to be called the G70, so we'll call it that.Quote:
Originally Posted by DilTech
I don't think anybody in their right mind is going to try to argue that this is wrong. I think the evidence is pretty damn consistent with your theory. For all practical knowledge, at one point the G70 was called the NV47. Multiple sources confirm this, and until equally convincing and concrete evidence proves that this is just a glitch or doesn't hold water, like I said, no one in their right mind will try to argue against it.Quote:
Originally Posted by DilTech
The only thing I'm arguing against is the speculation that branches off of this fact. Speculation such as that the real G70 is a 90nm 32-pipe card, and that NVIDIA held out on it. Or that NVIDIA jipped us of the real G70, which was a 32-pipe card. Frankly, all that speculation sounds completely bogus to me. I really don't see how it could be any more straight forward. The core was originally to be called the NV47 but NVIDIA later decided to change it to the "G70". No big deal.
As for TSMC not producing the chip... Well I don't think there's enough information to form ANY sort of conclusions from that. It sounds very probable that perhaps TSMC didn't produce the chip. What does that imply? No one can honestly say. For all we know there could be multiple revisions of the G70, some made in TSMC, some not. Are ALL G70 cores stamped with the Korea locale? I certainly don't know. Nor do I see the issue as any sort of significance until more info is revealed.
Problem starts with the NV47 being cancelled, as well as NV50, RSX being Playstation only, and RSX being 90nm G70. So, uh, who's fabbing the Sony chips?
You know cadaveca, I never even thought about that one... The RSX IS a 90nm G70, which thus gives further reason to believe that NVidia may have held back, again though, I'm not saying that is true... It is a VERY good point though.
As for who's fabbing them, I'll have to look into that one, it's a very good question indeed.
As for cybercat, google a search on 7800gtx reviews, every picture seen says korea on the core... That's pretty good odds that all of them are being made there.... Also, I'm not saying any of my theories are correct, other than that the G70 is the NV47, this one is a definite. The only *real* question is why try and cover it up, that part doesn't make sense, as most consumers have no clue what core is what anyway, they just look at the box, so I don't see how saying it was a completely new architechure would really help them sell the cards.
omg where to start... Prescotts dont have more piplines, just a longer one. 31(IIRC) "stages" if you will, compared to the old Northwoods 20, and A64s 14(?). Look at it like an assembly line, where the pipeline equals a stage in the line, and clock speed as how fast the product moves stage to stage thru the assembly line. Lets say the product is a Car. So if it takes A64 only 14 steps to compete the car, it can (and has to) run at a slower speed than a pressie that takes 31 steps, but at a much faster speed. thats why p4s are clocked higher and can OC to much higher Mhz levels, but the A64 can do the same amount of work at a lesser clock speed. Such as an A64 4000+ running very similarly to a P4 at 4 Ghz.Quote:
Originally Posted by Turok
As for video cards, the number of pipes DOES seem to make the biggest difference. If clock speed mattered more than archetecture (like you seem to think it does), then my old 9600XT at 500 core/600 memory speeds, should have been able to out-do a 9800XT that was clocked at 412/700-800(?dont remeber). The difference is pipelines. my 9600XT had 4, and the 9800XT had 8. It worked faster thru paralelism, being able to do more per clock cycle due to the incresed number of pixel pipelines and vertex shaders. Its the same princeple as dual core cpus being able to do more in less time thru paralelism. If you want to see the difference, search futuremarks ORB or just read reviews. an x800 Pro is the same chip as an x800XT, except for 4 locked pipelines (12 vs. 16) You can OC a pro to speeds faster than a stock XT, and still have much less performance. Look back at past GPUs. theyve all stayed at 200-500 core clock speed since at least the geforce 4 series. The performance gains are coming from added pipelines and better archetecture. Im not sure how many pipes the older cards had, but the GeforceFX series, and the Radeon 9200 thru 9700 had 4 pipes, then the 9800s had 8. now the X800 pro and 6800 plain have 12. the 6800GT and Ultra and X800 XT and X850s have 16. and the very latest generation, the 7800 series, including the GT and GTX Versions have 20 and 24 pipes respectivley. Looking back, it becomes less of a mystery as far as GPU performance goes, doesnt it? :hrhr:
You don't? :stick:Quote:
Originally Posted by DilTech
I think something new would sell better than something old and rehashed.
Here's a picture of another TSMC 110nm part, nv44Quote:
Originally Posted by DilTech
http://www.overclockercafe.com/Revie...6200A/1008.htm
which country is on the packaging ? ;)
Regards
Andy
DIltech...who made the hardwarew to test WGF 1.0? Uh, that would be the 6 series...
what happened to WGF 1.0? It got delayed due to M$ and , well, you know...WGF1.0 was rolled into DX9.0c.
Who's hardware meets WGF 2.0 specs? Not nVidia's.
Who's chips stand to not be up to standard when it comes to WGF 2.0?
for that, i'll point you here:
http://www.bit-tech.net/bits/2005/07...terview/4.html
And finally, to lead you a little further in your investigation, DilTech...
http://www.penstarsys.com/editor/com...idia/g70_spec/
When you put it all together...
:clap:
Again...x800XTPE...bet ya it had it soft ground in the pipeline..
R520 and RV530...soft ground in pipes? Again? different design teams, same issue?
uh, HELLO?
mr_knowitall15 - your 9600 only has a 128bit memory bus, otherwise it would be quite well matched to 97/98 series cards (which are all 8 pipe, not 4 unless a quad is disabled to make it a 9500np).
Neither does the R520... The NV50 was suppose to be NVidia's WGF 2.0 card, which ironically enough the G80 now is suppose to be. That card is already taped out ready to go for when ATi launch the R580.Quote:
Originally Posted by cadaveca
Either way, I'm pretty sure I'm onto something with this, it's just a matter of what it is.... The fact that the RSX is a G70 on the 90nm process just points in the right direction even further.
Actually, if the R520 has a unified shader, pipeline, it's good to go for WGF 2.0...and i know that the Xenos chip 100% is unified shader.
r500= xenos
rv515= x1300
r520= x1800
RV530= x1600
to me, they should all be unified shader pipelines. the r520 may not have this functionality enabled...but the r580 should, i would think. they are all the same gen...and ATI is very simple in thier naming scheme...
R= Radeon
RV= Radeon Value
5= fifth generation.
r5xx= "xx"= chip desgin #.
R1800 is WGF 2.0 compliant, by the bittech article, if you read the whole thing. other sources point in this derection as well. Only lately has it been said that the chip features standard pipelines..probably because they have fixed assignments for now until Vista is released, and games can actually take advanatage of the unified shader pipeline. becasue Vista is slated for next year, this makes sense for the r580 to be unfied-enabled, as should be the next-gen chip from nvidia, to be released in the spring, i think.
The NV47 was changed to G70 naming scheme when they decided on the final chip that RSX would be. becaseu RSX is NV47 on 90nm, but intended for the consoles, it stays as RSX nv47, while the desktop pc market gets the G70 NV47...the naming difference is becasue of the design differences, although they do feature the same funtionality....although if you ask me, the GSX will be faster due to less load demand from the memory controller, and the slower membus...which should allow them to clock the chip much higher than it's desktop counterpart.