- Dec 19, 2004
Release in Jan, sounds like a ploy to get tempted 5xxx Xmas buyers to wait on the possibility its a better card......
Doubtfull, it's simply the earliest Nvidia can tell at what clockspeeds the gpu's can run. Not many people will be waiting for Fermi performance numbers, because those who know that the numbers are coming, also know Fermi most likely won't be avaiable for quite some time, and that even if it's better, it will also be more expensive/cost the same as a similar ATI-card. Nah, money will be spent during xmas either way, Fermi or not...Release in Jan, sounds like a ploy to get tempted 5xxx Xmas buyers to wait on the possibility its a better card......
It's certainly within the realm of plausible values, but it can be as little as two months depending on a few factors (mentioned below).Well, it's actually 4 months and try to assign dates to that and you'll see it is not realistic.
Do you agree?
Would you, please, give me one example of a new (as a new generation not a refresh) GPU launch that was less than 4 months from a tape out to actual product launch?It's certainly within the realm of plausible values, but it can be as little as two months depending on a few factors (mentioned below).
Did you/do you put much stock into the rumors that NV had amassed 9000 wafers banked at the point in the flow where A2 stepping diverged from A1?
If you believe NV did that then you'd be compelled to consider that an Oct release was possible, and indeed planned (why else amass so many wafers?) with the working assumption on NV's part being that A2 quick-fixes were all BEOL (power dist).
However, yeah if you did not believe the 9000 wfr bank rumor and you made some worst-case assumptions about how early in the A1 stepping the A2 stepping was going to diverge then you could easily walk it out and make some 6 month worst-case scenario predictions.
Given the driver situation with G80, DX10 and Vista I kinda wonder just how "mature" fermi's DX11 Win7 were at the time the go/no-go decision was made on A1 for production.
It kind of makes me think of what Yoda would sound like if he ate some of those mushrooms in that forest of his.lol no shit. I just re-read that post three times, and gained nothing but a larger headache for trying to decipher that madness.
I'd rather wait until we get closer to launch time. As of right now, we have absolutely no info worth it's salt. A sticky right now would be just full of nothing. And to merge all the speculations threads would be downright confusing to many.keysplayer, or any other mod in these parts, have you considered sorting out a sticky thread for all stuff relating to Fermi release dates?
Rather than constant FUD vs everyone else threads.
That would be me, I guess.Who had the Nemesis translator from the other thread?
Ya know. There are 2 sides to every story. I like your version and point of view! It reminds me of a young fool I knew very well. He was a nice guy but his head got stuck so far up his own ass its a mircle that he ever seen the light again.
Even tho 5800 are hard to find doesn't mean Bad for poor ATI/AMD . Sure it would be nice to have 60% + yields. The point of fact is this. ATI is getting chips at At least @ 40% yield and they are flying off the shelves giving many the appearance of availability issues. There Availbable ya just have to get in line . Demand is so high even at 80% yields availability would be low. These things are Best VPU ever built. and it shows in its shear overpowering performance compared to it competitor . NV present generation cannot compete so they had to move to a new arch . and they waited to long and will pay dearly for things like this .
I noticed you like to talk cuda or C for cuda and how cuda has native C++ lol .
You really haven't a clue of whats going on in other areans .
Lets take for Example Intel Ct with is 100% compatable with AMD products. You havent a clue what Ct is do ya .
You don't care because your only interested in 1 companies tech.
Well Ct is the C for cuda killer its not even a contest. You add RapidMinds into the mix and its all over with. I told ya this game isn't about hardware its about software and with Ct/Rapidminds this game is over befor it gets started. Compilers Its the everthing of Universial transitor language usuage . Who ever can compile any language at runtime wins. I give you Intel Ct/Rapid Minds . Soon to be shown . As soon as NV shows. As I said CT is open to all x86 processor. Boris is a pure genius.
Dear Wreckage, I acknowledge the fact that there are 2 sides to every story, and I honestly like your version and point of view! Honestly, I like it - because I find it very humorous, you remind me of a young fool I know. Nice guy and all, but he's one of those people who have their head stuck up their ass its a miracle they ever see the light again.
Even though the 58xx cards are hard to find (a fact that I easily concede to) , that doesn't mean it's the end of the world for poor ATi/AMD. While it would have been ideal to have 60%+ yield at this point, the fact is that at least ATi is getting around 40% yield and their 40nm products are flying off the shelves, which merely aggravate the appearance of availability issues. But the products are available, you just have to be patient, get in line. In my opinion, demand for these 58xx parts are so high that even at 80% yields, we would still be experiencing these “availability issues”. These things are simply the best GPU ever built (I also like to call them VPU for Video Processing Unit, for obvious reasons), and it shows in a lot of reviews and benchmarks where they pretty much slaughter most of the competition in raw performance, and almost absolutely annihilates everything in a price/performance point of view. The present generation of nVidia simply cannot compete, so they have to move to a new architecture (the fabled Fermi, of course), but they've been so delayed that they will no doubt pay for it dearly – it's never been a good idea, business-wise, to be late in the game. Also, they will pay dearly for shenanigans like this, where they try to pull a fast one on people:
Dear Wreckage (can I call you Wreck? Wrecky? Wrecky-boo? No? Ok, Wreckage then), I also noticed that you like to talk about CUDA or “C for CUDA” and how CUDA has native C++. Pardon me for a moment while I hold my sides laughing. L. O. L!
You really have no clue what happens in areas such as GPGPU computing. Let me try to impart some knowledge into you to try to educate you.
Let's take, for example, Intel's Ct, which I know for a fact is 100% compatible with AMD products. Yes, yes, I know you are confused, because you don't know what Ct is.
I assume you don't care about Ct because you are only interested in 1 company's products, that company, of course, being nVidia. But hush, listen and learn, I shall educate you on the matter regardiing Ct.
Think of Ct as the “C for CUDA”-killer, as it will most likely slay nVidia's precious GPGPU computing platform upon release. Ct is leaps-and-bounds better than CUDA/C for CUDA that it is no contest. Not surprising since it has the Intel juggernaut behind it. And as if that weren't enough, add another esteemed company in this area, RapidMinds, into the mix. Boom! It's all over for nVidia before it even started. You see, the breadth of my understanding on this subject leads me to believe for a fact that compilers are the key to everything in the GPGPU area. Think of it as the “Universal Translator” for languages. Whoever has the best compiler wins. And Intel/RapidMinds (actually, for the grossly uneducated, RapidMinds was acquired by Intel) has borne for us Ct, coming soon, as soon as nVidia presents Fermi in all its glory, Ct will be shown to upstage it. As I said, Ct is available for all x86 processors, so think about it. Pure genius, all the way.