NVIDIA G92 : Pre-review bits and pieces
- Thread starter Rys
- Start date
I did see it. I took it to read "tradition", which is a fine way of saying "because we've always done it that way". This probably seems like I'm making a big deal out of it more because you guys keep arguing with me about it than anything else.
Btw, are you suggesting that 8800 GTX is spinning up to 85%+ but we aren't hearing it because of the heavier design?I did see it. I took it to read "tradition", which is a fine way of saying "because we've always done it that way". This probably seems like I'm making a big deal out of it more because you guys keep arguing with me about it than anything else.
I took it similarly, and I kind of agree with Geo.
If you were to design a safe startup sequence, why not start the card in "2D" mode, handshake the successful startup of the cooling solution, then crank up the frequencies? If the design decision is really "my god, it might break without the fan on", then, err, shouldn't you actually check that the fan is running first, prior to slamming the metaphorical foot down on the metaphorical megahertz accelerator?
That's all doable, but that would all increase costs and that was not the aim of the GT methinks And it's not tradition, it's just the common way of doing it. Dunno about the G80, maybe it's not needed because of the heavier cooler or they invested more in that area there or whatever. We do the same thing in cars though, when the rpm sensor fails for example.
dnavas, how will you check the fan if the chip still isn't running? The startup sequence of the PC does not turn on the gfx chip first, that's just the gfx BIOS sensing power-on at first and spinning the fan per default I suppose.
And it's not tradition, it's just the common way of doing it. Dunno about the G80, maybe it's not needed because of the heavier cooler or they invested more in that area there or whatever. We do the same thing in cars though, when the rpm sensor fails for example.dnavas, how will you check the fan if the chip still isn't running? The startup sequence of the PC does not turn on the gfx chip first, that's just the gfx BIOS sensing power-on at first and spinning the fan per default I suppose.
Sure. And the shuttle engines run at 109% normal, too
I thought Geo's point was that "normal" fan speed should cover "normal" cooling. So, if you turn the fan "on", absent an "afterburner" signal from the gfx chip, the fan runs in half-speed mode. Of course, that's absent a lot of knowledge on my part as to how fan speeds are controlled in a PC. I assume there's a PWM controller somewhere, and that you could default said controller to 50% for power-on.
I don't think you worry about checking the fan without the chip up, you just bring the chip up in 2D/Mhz-constrained mode, then check everything is good. [That's to control for the case where you restarted a "hot" video card, which I would think would be the only worrisome case on startup.]
Well, it *sounds* reasonable, anyway
Some of the articles I've read notice that the chip gets a wee bit hot, so the startup of a hot video card may be a noticeable issue which prevents this sort of noise abatement. Of course, that's a guess layered on top of a series of unfounded suppositions, so you should take it for granted that I grew up in NY where they mined salt by the truckload....
It's not like EATM Pauly. It doesnt adjust LoD. Its simply more accurate form of alpha to coverage. Has alot less blending artifacts.
First thing I checked when the first driver enabling it hit the street; I'll say bravo NVIDIA as it really makes TSAA redundant at least in all my corner cases I had tested in the past.
You know my opinion on the initial TMAA they had; this one truly deserves the term transparency antialiasing
Yes. If you check Multisampling you get automatically the new transparency mode if you have the newest drivers installed.
Just finished reading Josh's latest on penstarsys, summary ..
- G92 is 384bit
- Nvidia shaved off 80-100 million transistors (ala G71), some undisclosed/unknown feature taking up the space
- Not sure if G92 has DP support
- AMD partners happy with the way they handled R600 EOL (2900 Pro)
- RV670 has fixed AA resolve that plagued R600
- Nvidia will refresh their line up in November
8800GTX (new) - 700MHz Core, 128SPs @ 1800MHz, 384-bit
8800GTS (new) - 650MHz Core, 112SPs @ 1600MHz, 320-bit - GX2 in Q1 2008 - 600MHz Core(s), 2 x 128 SPs, 1.5GB
- Possibility of dual-slot RV670 at 1GHz to combat some of these refreshes (GTS), also possible that RV670 is 512-bit
- 2 x RV670 card in Q1 2008
That little tidbit keeps popping up, but R600's lead designer (and Dave as well) stated AA resolve through the shader processors was a design decision, not a bug.
The only way to "fix" resolve in this case would be to add hardware AA resolve back in.
I don't know if I can believe the rumor mill over the guy who helped make the thing.
Rumors were of a "speedpath" issue for shader AA resolve.
I firmly believe he's wrong on, hmm, well at least the vast majority of those points. I've already said what I believe on G92 SKUs in the past, and I'll reiterate it in an article very soon, so I'll just leave it at that.
I wouldn't believe everything that's written in that article. For instance:
He thinks RV670 has more than 800M transistors while we know now that it has even less than the R600 and G80 at 666M. And there are some more flaws in there.
2900guy: GPU isn't a punnet and TMUs aren't apples. You can't just put them in. You would have to increase number of whole quads and it's dispatch logic, which would be quite expensive transistor-wise. And RV670 is mainstream. If nVidia needs 320mm2 GPU to compete ATis 194mm2 GPU, why to change anything on ATis GPU?
The new TMAA should be available to all Geforce cards that support the feature.
Chris
Thnx for the "confirmation" Chris..
