NV35 correction

[Johnny Rotten]

The way I look at it, NVIDIA needs two (2) new chips to compete against ATI, to capture both the middle and high-end.

For the middle end, you really cant over-market some new, spit shined AGP 8x Geforce4 to compete against the 9500 Pro, especially given DX9 and possible Doom3 texture/pass issues that may occur at the same time.

I dont agree with this. A 4600 calibre card would be a fine competitor to the 9500 pro over the time that it would need to be competitive (ie the first 2 or 3 quarters of 2003)

 

[Fuz]

I dont agree with this. A 4600 calibre card would be a fine competitor to the 9500 pro over the time that it would need to be competitive (ie the first 2 or 3 quarters of 2003)

I agree, in terms of performance they may compete, even though the 4600 will be faster in some situations and vice versa.
But what about price, will they compete at the same price point? If I remember correctly, the 9500pro will be released at $199, and thats RRP. What is a cheap 4600 worth? IMHO, the 4200 would be a better competitor to the the 9500pro in pricing and performance.

Edit: Having read sharkfood's comments about DoomIII got me thinking. If the game is released this year, or very early next year, then the 9500pro might have an advantage over the GeForce Ti range.

In other words, ATI's product line up is looking damn good for this christmas.

 

[jb]

Johnny Rotten


I think you should also weigh in the feature set as well. I know for some people the may not even use the "dx9 features" of the card before the upgrade again (there are the people that upgrade every year or sooner). However for those that plan to keep the card a bit longer, having DX9 just may be better in the long run. Besides it will allow them to watch cool tech dx9 demos as well as complete all of the test in the next 3dmark which we know is all so important And yes I know you said first few Qs.

Anyways I feel you have to factor in the features as well as the price/perfroamce of the card....

 

[Johnny Rotten]

I dont agree with this. A 4600 calibre card would be a fine competitor to the 9500 pro over the time that it would need to be competitive (ie the first 2 or 3 quarters of 2003)

I agree, in terms of performance they may compete, even though the 4600 will be faster in some situations and vice versa.
But what about price, will they compete at the same price point? If I remember correctly, the 9500pro will be released at $199, and thats RRP. What is a cheap 4600 worth? IMHO, the 4200 would be a better competitor to the the 9500pro in pricing and performance.

Well, the 4600 is nvidias fat cat in terms of margins right now. Obviously once the NV30 arrives (assuming it ever arrives ) the 4600's margins can be cut down to Ti4200 levels as the NV30 will slide into the high end/high margin product position.

 

[Sharkfood]

I dont agree with this. A 4600 calibre card would be a fine competitor to the 9500 pro over the time that it would need to be competitive (ie the first 2 or 3 quarters of 2003)

You and I know this... but the average flunky walking into BestBuy sure doesnt.

NVIDIA proved this isn't the tendency of the market with the Geforce3, despite initial unimpressive benchmarks compared to their Ultra line.

Same thing really, except it was NVIDIA vs NVIDIA. Ultra vs Geforce3.

At the time, DX7 vs DX8... and the marketing angle of DX8 games "on the horizon" was enough to sway consumers the direction towards the Geforce3, forgiving performance and pricetag.

If a new DX9 3DMark is released, or Doom3 benchmarks with some form of performance yield on the 9500 Pro.. this would pretty much bury the Ti4x00's that no marketing ploy could really save.

 

[KimB]

At the time, DX7 vs DX8... and the marketing angle of DX8 games "on the horizon" was enough to sway consumers the direction towards the Geforce3, forgiving performance and pricetag.

I don't buy it. The GeForce3 was just plain superior, due to the ability to do FSAA at reasonable framerates, and also anisotropic filtering.

If a new DX9 3DMark is released, or Doom3 benchmarks with some form of performance yield on the 9500 Pro.. this would pretty much bury the Ti4x00's that no marketing ploy could really save.

By the time that happens, nVidia will no longer be pushing their GeForce4 product line, except perhaps at the low-end (Even if you were to believe that the NV30 wouldn't be out until March this would still hold...).

 

[Randell]

I don't buy it. The GeForce3 was just plain superior, due to the ability to do FSAA at reasonable framerates, and also anisotropic filtering

It was way after launch that people begun to realise the Gf3 could do quality AF bearing in mind the need for a tweaker to enable it. Also nVidia pushed Quincunx AA, which was instantly dismissed as blurry and didnt push 2xRGMS. 2xRGMS and 4xAF was enough for most people to realise good performance & IQ with the Gf3, but I reckon that was a long time after inital reviews dismissed it as an expensive toy for being slower in some benches compared to the Gf2 Ultra.

 

[jb]

[quote="ChalnothI don't buy it. The GeForce3 was just plain superior, due to the ability to do FSAA at reasonable framerates, and also anisotropic filtering.[/quote]

Well I will give you 2 out of the 3. Back when it first was launched no one really new about AF as that did not seem to catch on until well after the GF3 was out. I can not remember if AF was exposed in the drivers way back then (yes your could force it with a tweaker, but the average Joe does not use a tweaker). People saw it was fast and had new features and thats why they bought it.

 

[Randell]

Well I checked Anand and there was no mention of AF in his Feb preview, March review or July round-up.

In fact i remember learning about AF on the Gf3 from John Reynolds endorsing the IQ with 2xRGMS and AF on a Ti200. I'm sure hardened Gf3 users had worked out AF before then though, but publicly I only recall seeing it talked about in the run up to the 8500 launch though I'm sure.

edit: clarification

 

[demalion]

There was actually a rumor of the sort concerning what nVidia would be introducing in the near future when we were discussing the NV18/28 etc. Personally, that makes sense to me, but what part number would it have if such a thing is in the works? The only number being thrown around is the NV31...could be to NV30 what RV200 was to R200, I suppose.

No, I'm not saying I have any info to support this guess, just that it is what makes sense to me for nVidia's profits in my knowledge vacuum of their chipset line up.

Well, it seems the NV28M is this cut down Ti 4200 chip, as mentioned here, among other places.

If this chip is cheap to produce, and it seems likely it is, nVidia has an answer to the 9000...it just seems they are unwilling to offer it in the desktop space (yet) to replace the NV18 (it would likely make the nv18 effectively a useless product).

If this is true, at that clock speed, and being "cut down" in some other way (likely matching the rumor in that nv18/28 thread), I'd think it would be slower than the 9000, and perhaps that's why it is only being offered as a "high end" mobile part (where unplugged performance or performance relative to the slower clocked 9000s could conceivably be different).

How do you think this reflects on the schedule of a nv35 release? I think it indicates nv31 being released later than I'd thought, and I'd think nv31 and nv35 would both be intended to be released at the same time (I'm assuming they are "teething" only on the nv30, and these lessons would be applied to the later parts...likely they are doing so right now if the nv30 issues are pretty much worked out by now).

 

[Joe DeFuria]

I think it indicates nv31 being released later than I'd thought, and I'd think nv31 and nv35 would both be intended to be released at the same time

That's pretty much mu assesment as well. Assuming the NV28M is pretty much just an NV28 with slower clocks...this would indicate to me that nVidia's "next gen" mobile part is still at least 6 months from being launched.

The only other explanation, would be that nVidia's next gen mobile part doesn't offer the same performance levels as the NV28M, so that the NV28 M will remain a viable "desknote" solution despite a newly updated mobile core. Note that's not TOO entirely out of the question, and actually could be the case.