It's Capt'n Corrupt.

I'm talking more about pixels that polygons, but yes there are many other factors that limit practical throughput.

I haven't been able to find any links on the Tegra T20 performance. Care to share?

Thanks for the link!

Whoops, sorries. Capt'n Crunch sounds like a cereal, but it does have that nice ring to it

Do you mean theoretical maximum pixels per second?
Because that is usually blown out of proportions and is not a good indication of the device's graphic's potential. I think the max polygons/sec (aka triangles/sec) are more indicative, but even that is flawed.
Unless you mean Pixel fill rate, then I understand what you mean, but that means of measurement is also flawed.

For an analogy, its like looking at a list of cars. You have the max HP (polygons/sec), but that doesn't mean the car is faster. It usually is, but there are several other governing factors. And likewise, the pixel fillrate is like the max torque. The real indication is their real performance, eg "0-100" = "sustained polygons/sec" or even "fps in a 3D game".

It was difficult but I did manage to find some thoughts on the Tegra2 specs, and it's (theoretically) max output is below the SGX540, but in practical/real world testing it should perform slightly better than the Hummingbird chip. I will post these up when I find them again.

On another note, anyone remember the Advent Vega?

Its a pretty hi-end tablet that runs Vanilla Froyo on the Tegra2.... there's some new info/video on it.
http://www.engadget.com/2010/10/17/a...t-might-get-it

Anyone notice that the representative said during 3:20 - 3:35, that the screen is above 720p but below 1080p. This implies 1366 x 768 (which I pray it has so I can buy it in a heartbeat).

BUT hold on a sec, wasn't the speccs leaked saying it comes with 1024 x 600?
And it makes sense for 600p because that is the maximum resolution Android supports (according to Google). With Gingerbread, Google is planning on starting their tablet revolution and one of their plans is is to increase the max resolution to 720p.

If you think this is a non-issue, I tell you it most certainly is. If the Vega has a 720p screen but that is limited by software to 600p I will be a happy chap = the community can upgrade it to Gingerbread and remove that limit. If it has a 600p screen, I will think twice about this product.
Why is this such an important difference?
With 720p available it means you can watch HD television, HD videos (think bluray/online).
It also means that you don't have to convert your HD media from 720p down to 600p (which would be a big pain in the).
And last but not least, the difference in pixel density is quite apparent (119PPI low VS 157PPI medium), and this means less "clunky" and much smoother graphics.

Yes, yes. But I'm curious about the practical pixel throughput in relation to accelerated UIs which may only map 2-10 quads/triangles and mainly manipulate on-screen graphics with shader/texture combinations. In this case the Polys/m capability doesn't matter near as much.

I've just spent ~1 hour trying to find the page from my history and I honestly couldn't find it. I also google'd for it for about another hour and still couldn't find it within the search results. If I find the source later I will remember to post it here. The blog was a bit speculative but it used some logic to explain why the Tegra 2 should perform slightly better than the Hummingbird when it comes to graphics, but if given the actual requirements for the chips the "theoretical" performance of the SGX540 was noticeably higher.

You can take this as a grain of salt, personally I think its accurate, but none of these matters when the OEM provides shoddy drivers/sloppy software (like in the case of Galaxy S and the homebrew lag fix).

Also, I think it's going to be hard to find the practical pixel throughput for the devices/chips on the internet because that's specific information. You may have to wait to buy them and benchmark them yourself ... i know it sucks, but thats the case until there are enough Tegra2 devices and benchmarking tools out there.

Okay I sort of found it. It was in an article talking about the Galaxy S core I think, and someone replied to it, so it wasn't in the actual article and they had a link to these showing neocore benchmarks:

Tegra 2 = 27.4 fps
http://www.youtube.com/watch?v=uJav9ns6b4o

SGX540 = 55.7 fps
http://thatsbadass.com/samsung-galax...ore-benchmark/

The comment went on to say how having 1GB of RAM in Tegra 2 devices should remove the bottleneck and greatly improve 3D performance. As well as saying the Galaxy S had much less (62%) polygons/pixels to produce at WVGA (800x480) compared to the 1024x600 that's running on the Tegra 2. Made sense to me.

I'd like some benchmarks when something like these get released:
http://androinica.com/2010/07/27/rum...d-of-the-year/
http://www.pahing.com/2010/09/viewso...vidia-tegra-2/

Just watched this

http://www.youtube.com/watch?v=_Z_H_-0TIYA

It's a shame that Toshiba decided to take a dump on such a great hardware by giving it a hideous physical styling (large bezel, cheap plastic feel etc) and crossed it with shoddy software (no flash out of the box, flash issues, bad user interface, no multitouch in certain areas, Apps don't become fullscreen, Apps don't work properly etc).

I've only got hopes for the Advent Vega and I'm crossing my fingers that it has 1GB RAM and 720p (as spoken from the representative) display. The community can develop custom ROMs for it that's tablet-friendly and possibly do the market-hack.

And I've just noticed another contender in the tablet ring.

The ViewSonic G-Tablet:
http://www.viewsonic.com/gtablet/spec.htm

Will they fail like Toshiba did, or will ViewSonic find the G-spot?
We'll have to wait and see, but enjoy this video for now:

http://www.youtube.com/watch?v=aL1b109GB2c

Just to follow up on this post. The Galaxy Tab scored an impressive 52.9 fps on the NeoCore benchmark (SGX 540 @ 1024x600):
http://www.youtube.com/watch?v=uL8XXTDvNCQ