The Fermi Thread - Part 3

[MrDiaz]

I was at my local supplier this past weekend and they told me the GTX 480 will cost around $700(canadian). Not sure how accurate that is but just thought I'd put it out there.

They don't know crap man don't take their word for granted.

[Levish]

They don't know crap man don't take their word for granted.

If I was a business and there is any sort of limited availability, I'd be foolish not to capitalize on this opportunity. They are run on money, not love and hugs.

This doesn't mean that I want prices to go up, it just means that its in your interest as a consumer to get a product and at the lowest price possible, but its in their interest as a reseller to charge as much as they can get for whatever stock they have.

[Picao84]

You need to make up your mind, does Charlie post news or not? It seems when it's news you like, you're fine calling it news. Selective reading 101.

It really depends on what he talks about. Everyone knows he spot on about these type of stuff (tape outs), but when he talks about chip architecture... Forget it... The software tesselator thing is a good example of him talking sht..

[Picao84]

Very scalable architecture indeed!

I remember when the whitepaper came out all the fantards were all over that word. Irony.

Where do you see the architecture is not scalable? If from here they can get all derivatives out of the door by years end, its a good execution, IMO. Keep in mind GF100 had problems they had (still have?) to iron out. Probably only after that they could proceed with next tape outs...

[MrDiaz]

And I'm fine with that, like you said... it's a business. I was merely focusing on the fact that they have no idea what pricing will be

[Sn0wm@n]

Where do you see the architecture is not scalable? If from here they can get all derivatives out of the door by years end, its a good execution, IMO. Keep in mind GF100 had problems they had (still have?) to iron out. Probably only after that they could proceed with next tape outs...

if fermi is so much scaleable why didnt they release it before that bloated big ass die that they call GF100 ???? marketing reason??? if so another fail reason by nvidia .... release what you can especially since you know fermi will be a problematic gpu ....

what i meant by fermi would be the low end part gf108 gf 106 and gf104


GF108 would have been the easiest one to do right .... yet it wont be out till late summer .....

[Picao84]

if fermi is so much scaleable why didnt they release it before that bloated big ass die that they call GF100 ???? marketing reason??? if so another fail reason by nvidia .... release what you can especially since you know fermi will be a problematic gpu ....

what i meant by fermi would be the low end part gf108 gf 106 and gf104


GF108 would have been the easiest one to do right .... yet it wont be out till late summer .....

Since when does a low end part gets released before a high end? (dont tell about 5870 vs 5970. 5970 is just two 5870 chips). The high end
is always the "proof of concept". Before it gets done and released, it would be foolish to go ahead with lower end parts.

Easiest to do? Why is that? Just because its smaller? LOL.. really LOL..

[Macadamia]

Where do you see the architecture is not scalable? If from here they can get all derivatives out of the door by years end, its a good execution, IMO. Keep in mind GF100 had problems they had (still have?) to iron out. Probably only after that they could proceed with next tape outs...

Good? Wow, you're clearly funnier than Charlie. Congrats!
Let me tell you what's proper good execution:

Enthusiast, Highend, Mid-high, midrange, midrange OEM, low-mid (and its OEM), lowend parts ALL in 6 months when the process is ing up. Oh, and 3 months before the competitor's highend part even shows up.

If the base architecture was SO scaleable why are they struggling to get ANY part out? Suuuuuuuuuuuure, big GPUs are "ing hard to build" according to nVidia, but smaller Fermis are NOT big GPUs.

They're much smaller, so why are they ed up too?

Every GPU architecture is scalable. But in reality it seems ATI has way less problems scaling their designs up OR down.

[trinibwoy]

I see we have yet another insider that not only knows how long Fermi derivatives took to develop but what state they're in too. You guys should really start your own sites and make some ad revenue for your trouble.

[Picao84]

Good? Wow, you're clearly funnier than Charlie. Congrats!
Let me tell you what's proper good execution:

Enthusiast, Highend, Mid-high, midrange, midrange OEM, low-mid (and its OEM), lowend parts ALL in 6 months when the process is ing up. Oh, and 3 months before the competitor's highend part even shows up.

If the base architecture was SO scaleable why are they struggling to get ANY part out? Suuuuuuuuuuuure, big GPUs are "ing hard to build" according to nVidia, but smaller Fermis are NOT big GPUs.

They're much smaller, so why are they ed up too?

Every GPU architecture is scalable. But in reality it seems ATI has way less problems scaling their designs up OR down.

Who the hell said GF108 is bad too? Like I said, it would be foolish to advance to low end parts before the high end is finished. No one does that! And for you smart ass:

GF100 is done in clusters as you know. If GF108 is one part of those clusters, in this scalable architecture, and if the clusters have/had problems in GF100, do you think it would make sence to advance for GF108??

But Ill stop feed the trolls, anyway. /Ignore

[flippin_waffles]

It really depends on what he talks about. Everyone knows he spot on about these type of stuff (tape outs), but when he talks about chip architecture... Forget it... The software tesselator thing is a good example of him talking sht..

What it really depends on, is whether or not you like what he posts, plain and simple.

[Picao84]

What it really depends on, is whether or not you like what he posts, plain and simple.

Really, just stop trolling. It was you who started the argument, bye. /Ignore

[lowfat]

Has there been any picture floating around of 470/480's w/ DisplayPort? This is a requirement for my display.

[Macadamia]

No one does that!

R600 - May 2007
RV630 - June/Jul 2007
RV610 - Aug 2007

But the competition did it just fine. RV630 and 610 would probably be even earlier if UVD wasn't problematic (it wasn't present on R600).


regarding silicon logic compartmentalization:

And there's no difference to it largely speaking. (GT200 to GT21X, Nehalem to Arrandale- yes Nehalem does it organized too) GT21X does scale down proportionately to desired performances, but it took them over a year.

You have all this fluff talk based on the same whitepaper but the dates and history (no not some revisionist history) shows otherwise.

[SirKronan]

Has there been any picture floating around of 470/480's w/ DisplayPort? This is a requirement for my display.

Yeah, I was wondering about that myself. If they can do display port on the 9400 chipsets, why not on their higher end GPUs???

[Behemot]

This is not pointed to you personally, but to the whole buggus in question.

"Windows 32bit" will only see and utilize no more than 4 gig (just a little under), but all the other programs installed will use eveything available. So when playing a game, it is not windows 32 managing the memory, but the game itself and the hardware in question (which is sort of a program on its own).

So basically you guys are half right and confuse.

Everything will be in use, but windows 32 will only use what it can, 2 different thing.

This is the first time I heard about games managing HW and memory. What is the OS good for than? I though the OS manages all HW resources including memory. If the OS cannot see it, how it could give the memory to the program?!

Second thing, the program itself is 32bit too. It can no way see 4 GB RAM and another GB's VRAM. We all know SW under 32bit OS could use up to 3 GB's and now you say it can use even 8 GB's? Or 16 GB's??

[Zed_X]

gtx 480 clocks are great, like i said! tdp is good, its far away from 298!

[Sn0wm@n]

gtx 480 clocks are great, like i said! tdp is good, its far away from 298!

proof to back up those claims ?????

[Olivon]

gtx 480 clocks are great, like i said! Tdp is good, its far away from 298!

700+ ?

[trinibwoy]

700+ ?

Lol, you're asking about clocks and not TDP right? It's hard to tell nowadays!

[Musho]

Lol, you're asking about clocks and not TDP right? It's hard to tell nowadays!

700w+ TDP would be a bit insane, wouldn't it?

[hennyo]

This is the first time I heard about games managing HW and memory. What is the OS good for than? I though the OS manages all HW resources including memory. If the OS cannot see it, how it could give the memory to the program?!

Second thing, the program itself is 32bit too. It can no way see 4 GB RAM and another GB's VRAM. We all know SW under 32bit OS could use up to 3 GB's and now you say it can use even 8 GB's? Or 16 GB's??

The 4GB Ram is possible on a 64bit version of windows running a 32bit program with special tags. Also the 8GB is possible on both 32bit and 64bit server enterprise edition, tho the 32bit 8GB address comes with significant speed penalties. The 16GB 32bit program access is possible but ONLY on 64bit datacenter editions of windows, which cost what someone well paid makes in a year, maybe even more.

[Olivon]

Lol, you're asking about clocks and not TDP right? It's hard to tell nowadays!

Come on trinib'

Of course that clocks, 700+ was originaly aimed by nVidia if I remember well.

[Marios]

Zed_X, in case you know something share it with us in a form like ... 480SP: 650/1500/4000 250 TDP

[annihilat0r]

Zed_X, in case you know something share it with us in a form like ... 480SP: 650/1500/4000 250 TDP

NDAs exist for a reason