VERY common mistake by people here: nV doesn't have a 12 month cycle

[Uttar]

Hello everyone,

I've seen nearly everyone on this forum make this same mistake: thinking nVidia releases a fully new product every 12 months. That is not correct.

nVidia schedule is releasing a total new product every *18* months.

nVidia wanted to release the GF3 in end 2000 as everyone know i think, but due to delays, they released it in beginning 2001.

So... Original Product: 1st Half 2001. 1st Refresh: 2nd Half 2001. 2nd Refresh: 1st Half 2002. New product: 2nd Half 2002.

That's the CORRECT nVidia product schedule.

And i really doubt they want to do the same delay TWICE. But who knows. That's not the point of this post.


Uttar

[StealthHawk]

yes, it is very disappointing that nvidia has fallen to a 18 month cycle between new architectures.

that being said, now that ATI is finally on the ball it seems, let's hope this encourages nvidia to stop feeding us garbage "refreshes" and get back to a 12 month cycle.

edit: upon further study, Uttar is completely correct about the 18 month cycle. i retract my first "paragraph."

[Uttar]

Originally posted by StealthHawk
yes, it is very disappointing that nvidia has fallen to a 18 month cycle between new architectures.

that being said, now that ATI is finally on the ball it seems, let's hope this encourages nvidia to stop feeding us garbage "refreshes" and get back to a 12 month cycle.

Err, that isn't quite my point.

nVidia ALWAYS was on a 18 month cycle. However, for the GF3, it took 24 months between new products instead of 18 months.

18 months is pretty much fine, since it's not really possible to make MS update DX more often than that...


Uttar

[sbp]

Originally posted by Uttar


Err, that isn't quite my point.

nVidia ALWAYS was on a 18 month cycle. However, for the GF3, it took 24 months between new products instead of 18 months.

18 months is pretty much fine, since it's not really possible to make MS update DX more often than that...


Uttar It is-what about the Riva128 and TNT series?The Riva 128 first came out in Sept. 1997 anf it was 2 years later the GeForce came out in Sept. 1999.

[Bigus Dickus]

I've gone through the announcement and release dates of all NVIDIA products since the TNT, and they are on approximately an 11 to 12 months cycle between new cores (NV15/NV20, NV20/NV25, etc.) with a refresh stuck somewhere in between, varying from 4 to 8 months (NV16, NV20-Ti, NV28 etc.). The only real exception was from the GeForce to the GTS, which only took 8 or so months instead of 11.

So, I'm not sure what you are claiming. 18 months is too long between new cores (considering NV15 and NV25 to be different enough to be included as new cores) and it's too short between new DX architectures (NV10, NV20, NV30).

:confused:

[Uttar]

Originally posted by Bigus Dickus

So, I'm not sure what you are claiming. 18 months is too long between new cores (considering NV15 and NV25 to be different enough to be included as new cores) and it's too short between new DX architectures (NV10, NV20, NV30).

:confused:

It's not too short between new DX architectures. NV actual goal is 18 months between NV10 and NV20 as well as NV20 to NV30.

You don't think i'm right because there was 24 months between NV10 and NV20 - but that's because nVidia got a delay there.

nVidia goal was to have the NV20 out in 2nd Half 2000 and NOT 1st Half 2001.

So now, nVidia is trying to get back on schedule by having the NV30 by 2nd Half 2002 ( 18 months from 1st Half 2001 )

And this time they'll try way harder because else everyone will think their schedule is a 24 month one...


Uttar

[sancheuz]

It doesnt really matter, becasue there hasnt been games.... yet..... that are using up the geforce4 or geforce 3 power, that is till ut2k3 and doom 3 and splinter cell comes and deus ex 2 come out.

[Matthyahuw]

You're all wrong!

arguing about useless crap...gimme a break, if you like a card, buy it...if not, don't...

[Uttar]

Originally posted by sancheuz
It doesnt really matter, becasue there hasnt been games.... yet..... that are using up the geforce4 or geforce 3 power, that is till ut2k3 and doom 3 and splinter cell comes and deus ex 2 come out.

*cough* Morrowind *cough*

You shouldn't even hope for any AA/Aniso in Morrowind with full detail. Morrowind is VERY GPU intensive and i'd love to see a card being able to do 4x AA with it - Morrowind is, IMO, a game where you REALLY see the advantage of AA.

Originally posted by Matthyahuw
You're all wrong!

arguing about useless crap...gimme a break, if you like a card, buy it...if not, don't...

What are you talking about? nVidia product cycle is 18 Months, and that's all. If you don't trust me, i can always try to find a link to proof it.

And i still don't understand how this fits in the context - the NV30 isn't released yet.


Uttar

[Nv40]

Guys what you need to understand is that NVidia Release cycle was always based on ...

1)the Competitions avaible at the time
2)a new Microsoft DIrect 3d
3)OEM manufacturers

So you cannot release a brand new totally technology until
microsoft release another direct3d .. see ?

Thats why Geforce3 and Radeon2 (direcxt8) have almost same technology , and Nv30 and Radeon9700 (direcx9) too.. see ?

The thing is that While ATi release only a new product every time
Microsoft up the specs in Direct3d ,Nvidia do refresh in between
which believe it or not ,its a good thing for gamers ,because
every new product lower the prices of older ones !!! ;)

StealthHawk :yes, it is very disappointing that nvidia has fallen to a 18 month cycle between new architectures.


it could be very disappointing for you ,if you buy every time Nvidia release a minor refresh ,or any single modification .
if you are so dissapointed with Nvidia you can jump the side
of the ATi betatesters and wait with patience for months for new drivers to let you play latest games and enable harware that often ships with broken features . ;)

if you buy the best product for your money by the end of every year or closer +/- ,your experience with Nvidia will be no less that
Quantum alien new technology improvements in speed and quality ..:)

1)Tnt2 ultra
2)Geforce2 gts
3)Geforce3 ti500
4)Nv30 (christmas of 2002) +/-

or

1)Tnt2
2)Geforce1 DDR
3)Geforce2 ultra
4)Geforce4
5)Nv30


Remember that there is 12 months +/- between each one of those .:)
So i dont know why anyone would complain about more than 2x times
increase in performance with more eye candy ,because since Tnt2 this
is what Nvidia have been doing every year .. ;)

[sbp]

Originally posted by Matthyahuw
You're all wrong!

arguing about useless crap...gimme a break, if you like a card, buy it...if not, don't... Your right. We should be arguing abut which beer is best. http://sbp777.homestead.com/files/beerchug.gif

[Sgt. Slaughter]

If you're in the Northwest, nothing beats Black Butte Porter. My all time fav beer

[-=DVS=-]

DUFF BEER

Duff Man owns you all :D :D

[DadGT]

Duff sucks. Drink Fudd!

[Uttar]

Well, since this is the Rumor Mill...

I think a new beer is coming soon! It'll, like, own every other beer. Biggest revolution ever!


Uttar

[gravioli]

Originally posted by Sgt. Slaughter
If you're in the Northwest, nothing beats Black Butte Porter. My all time fav beer

It's not bad if you have it with food, but I still prefer Pyrimid Apricot Ale. Heck, have a six pack, and you have all the servings of fruit you need for the day :D.

Anyway, I agree with Matthyahew; length of development cycles is useless IMO.

[StealthHawk]

Originally posted by Nv40
it could be very disappointing for you ,if you buy every time Nvidia release a minor refresh ,or any single modification .


that's not the point. for example the gf3ti500 was a complete joke as far as anything is concerned. the RAM had the same spec as that used with the original gf3, only it was clocked higher. theoretically, all gf3 should be able to clock the memory as high as the gf3ti500 with no problems at all. i would have a little more respect if nvidia had introduced the ti500 at say $350, but asking $400 for an overclocked gf3?

you're telling me that nvidia couldn't find slightly better ram and then in turn clock the core higher too? i think i can safely say that almost every gf3 could clock as high as or close to a gf3ti500 without any extra cooling. i really don't think the same can be said for any other nvidia product. ie, no TNT could be clocked to TNT2Ultra levels normally, no gf2 could be clocked to gf2ultra etc.

[Cotita]

Actually the geforce3 ti500 and ti200 are the best card nvidia has ever released. From a financial point of view.

nVidia had the geforce4 waiting in the wings in case the radeon8500 was a geforce3 killer. We know how that story came out.

nvidia made a good choice, why would they release a new chip when they can just speed up the old one? Nvidia had the chance to sell its geforce3 ti500 chips at a higher price and they knew yields were good enough. Why risk shifting to a new core with lower yields and higher cost?

I think its funny that people complain about new cards coming out every 6 months or so and people complaining that developemente cycle is too long.

You can't please everyone

[StealthHawk]

i understand fully that the Ti series was a good move for nvidia. i am also not disputing the merits of the Ti series. however, i still feel that the ti500 was a rip off for consumers, especially when the R8500 was priced much, much lower.

are you actually disputing whether or not nvidia could have produced a faster clocked gf3 than the ti500? something more along the lines of the gains seen from gf2gts->gf2ultra or even gf2pro would have been a decent "refresh." the gf2ultra was functionally exactly the same as the gf2gts after all if i'm not mistaken. in 6 months all they could come up with is something i could overclock to with no additional cooling?

[Cotita]

with the ti500 the nv20 was already maxxed out, overclocking the core requires a very good cooling. As for the memory it was the fastest available at the time and if you remember the first cards could hardly overclock memory a but only a few mhz.

But then again why would nvidia invest in developing a new nv20 core stepping with the nv25 already finished.

Also you forget that the radeon8500 and the ti500 cost very much the same when they fist came out. When ATI couldn't deliver full performance they were forced to lower the radeons price, nvidia had no need to do so, they had the fastest card.

One more thing. When the geforce3 ti200/ti500 came out, the original geforce3 was phased out.

in 6 months all they could come up with is something i could overclock to with no additional cooling?

They had the geforce 4 done already. You should thank ATI for not having the geforce4 released earlier.

[Bigus Dickus]

Originally posted by Cotita
Also you forget that the radeon8500 and the ti500 cost very much the same when they fist came out. Really? I seem to remember the two performing very close with about a $100 price difference from the very beginning (the day the 8500 was available).

[Cotita]

Retail prices were 299 and 349 for the radeon and ti500 respectively. a $50.00 diff. well worth the extra performance and driver stability IMO.

[StealthHawk]

Originally posted by Cotita
with the ti500 the nv20 was already maxxed out, overclocking the core requires a very good cooling. As for the memory it was the fastest available at the time and if you remember the first cards could hardly overclock memory a but only a few mhz.

wrong. i believe that the memory on both cards was speced the same. in essence, the gf3 had "underclocked" memory since the memory used on both the gf3 and gf3ti500 had the same rating of 3.8ns. i don't know if the core was really "maxxed" out as you say, it was 20% higher than the original gf3's. however, the memory was not clocked 20% higher, so it would make little sense to clock the core higher when little to no performance difference would be realized. the only other purpose would be marketing, like the gf2gts which had a woefully inadequate memory clock compared to core clock(but that was done to inflate the texel fillrate and make the gf2 seem like a big deal compared to the competition)

One more thing. When the geforce3 ti200/ti500 came out, the original geforce3 was phased out.

hence why i said the ti series was not bad as a whole. besides the fact that the gf3ti200 had great performance with a great price.

They had the geforce 4 done already. You should thank ATI for not having the geforce4 released earlier.

i've heard this before, but do you actually have any hard proof of it? or speculation done by a reputable website. before i go on repeating such things with certainty, i'd like to know where the information came from

Retail prices were 299 and 349 for the radeon and ti500 respectively. a $50.00 diff. well worth the extra performance and driver stability IMO.

hmm, it seems you are right about the ti500 having an MSRP of $350. that's weird. i always thought it was $400. in that case it looks a little better in my eyes :o

[sbp]

Its puzzling this view that coming out with a product that costs less and run faster than its predecessor is a bad thing.

"theoretically, all gf3 should be able to clock the memory as high as the gf3ti500 with no problems at all." Theoretically was one thing, the reality was another.

Don't forget about better yields playing a part folks.

{goes drinks a Guinness}
http://home.earthlink.net/~sbp777/pics/guinness4.gif

[StealthHawk]

$350 is still expensive. you're telling me ATI can afford to sell the R8500 at $300(and it was usually found much cheaper) when it is a new card, and nvidia can only give us a mature board at $50 less? i don't think so. there were some minor changes to the layout of the Ti series as well, which probably lowered production cost. additionally, was it really "cheaper" or had gf3 prices already come down to $350.

[quote]Theoretically was one thing, the reality was another.[/quote

they were both rated the same. whether or not in actuality the memory used with the ti500 was "better" doesn't matter. besides which, i have heard of many, many people overclocking their gf3's to ti500 levels