[TT]NVIDIA Tegra 4 processor details leaked

[Jaydip]

http://www.tweaktown.com/news/27343...8nm_six_times_the_power_of_tegra_3/index.html

 

[thilanliyan]

I wonder if Tegra will ever be used in Windows Phone or blackberry?

Those are the only 2 OSes I'm even remotely interested in.

 

[ShintaiDK]

Issue is that Windows Phone 8 was just released. And before that it only supported singlecores.

But Windows Phone and Blackberry. You sure like niche markets.

 

[thilanliyan]

But Windows Phone and Blackberry. You sure like niche markets.

Lol yes, my current phone is a Meego OSed Nokia N9. Honestly it is the most intuitive phone I have ever used...the Swipe UI on it is just untouchable, and I'm not an app junkie so it doesn't bother me that much that there are not that many of the popular apps available for it (there is a good dev community so important stuff does come out for it).

Since Nokia killed Meego, the closest gesture based UI is BB10 (I have a Playbook and it is really nice and easy to use IMO) and possibly Jolla's Sailfish OS which is why I'm keeping my eyes on them. The only reason I would even consider WinPho is that Nokia makes some amazing phones...otherwise I wouldn't care much about it.

 

[JohnnyChuttz]

In for new Nexus 7 with Tegra 4.

 

[bigrash]

Nice. I'm very happy with the performance of my Tegra 3 tablet, and can't wait to see what Tegra 4 brings.

 

[Jaydip]

Lol yes, my current phone is a Meego OSed Nokia N9. Honestly it is the most intuitive phone I have ever used...the Swipe UI on it is just untouchable, and I'm not an app junkie so it doesn't bother me that much that there are not that many of the popular apps available for it (there is a good dev community so important stuff does come out for it).

Since Nokia killed Meego, the closest gesture based UI is BB10 (I have a Playbook and it is really nice and easy to use IMO) and possibly Jolla's Sailfish OS which is why I'm keeping my eyes on them. The only reason I would even consider WinPho is that Nokia makes some amazing phones...otherwise I wouldn't care much about it.

Oh boy it was a gem.We say AMD is mismanaged, sometime we should look at Nokia.

 

[tviceman]

I like and prefer companies comparing their new products to their own previous products rather than to competitor pproducts. It makes the performance claims more credible, IMO.

 

[thilanliyan]

Oh boy it was a gem.We say AMD is mismanaged, sometime we should look at Nokia.

I know...Meego and the Swipe UI is amazing (and garnered universal praise from the tech websites when it was released which was a rare thing for Nokia at the time), and if they had brought it out say in 2009 rather than 2011, they might have had some better fortunes. Nokia has fallen so far from where they were...I guess that happens when you get complacent.

 

[Jaydip]

I honestly beilve Elop is on MS's payroll rather than Nokia's

 

[thilanliyan]

I honestly beilve Elop is on MS's payroll rather than Nokia's

Well, he did come from MS...so maybe he still is.

I think he realized Nokia no longer had the pull to go at the OS and hardware challenge by itself like it had done for so long with Symbian and so had to align itself with one of the major players in the OS industry...and I'm sure MS gave some lucrative terms for Nokia to be an exclusive partner. I'm actually glad Nokia is not an Android OEM...there's just not a lot to differentiate between phones in the Android world IMO to make much of an impact. Hopefully Nokia does make a comeback...their stock has risen quite a bit lately...

 

[-Slacker-]

Heheyyyy 6 times more powerful due to 6 time s more cores! 'Cause that's exactly how it works.

 

[Acanthus]

A9 or A15?

Looks like A9.

 

[Saylick]

A9 or A15?
Looks like A9.

Looks like A15.

 

[ShintaiDK]

Heheyyyy 6 times more powerful due to 6 time s more cores! 'Cause that's exactly how it works.

GPU "cores" do scale like that.

 

[TakeNoPrisoners]

Heheyyyy 6 times more powerful due to 6 time s more cores! 'Cause that's exactly how it works.

I think you have the GPU confused with the CPU.

 

[Acanthus]

Looks like A15.

Hmm i don't see any clues in the slide regarding that.

Is that the "eagle" core they refer to?

 

[BenSkywalker]

A9 or A15?

A15, we've known that for well over a year now(this leak has more to do with some final specs, the overview of what the chips was has been known for quite some time).

 

[Red Hawk]

So, 76 graphics shader cores? Is this architecture closer to Kepler, Fermi, or pre-Fermi? Seems to fall somewhere between the 48 shader Geforce GT 520 and the 96 shader Geforce GT 430. And that's not necessarily bad for a phone graphics chip, I guess. I think would be at a distinct disadvantage against Ivy Bridge and Trinity in larger form factor situations. Doubly so against Haswell and Kaveri. But what about Jaguar or Clover Trail? How do "four+plus+one" 28 nm ARM cores fair against four 28 nm ultra low power x86 cores? That's a showdown I would be looking forward to.

 

[Tuna-Fish]

Heheyyyy 6 times more powerful due to 6 time s more cores! 'Cause that's exactly how it works.

In graphics, yes, that is exactly how it works. The reason more cores don't buy you linear increases in power is because the problems don't have enough parallelism. In graphics, you fire up a shader program for each and every pixel you draw on the screen, and, with few exceptions, they can all work in parallel. So, at a reasonable resolution, you don't run into problems until you have a million SPs.

 

[Red Hawk]

In graphics, yes, that is exactly how it works. The reason more cores don't buy you linear increases in power is because the problems don't have enough parallelism. In graphics, you fire up a shader program for each and every pixel you draw on the screen, and, with few exceptions, they can all work in parallel. So, at a reasonable resolution, you don't run into problems until you have a million SPs.

There's other factors, like memory bandwidth, texture units, ROPs, rasterizers, etc. These don't necessarily scale up along with "graphics cores". For example, the Radeon HD 7970 has literally twice the amount of shaders as the 7850 does (2048 vs 1024). Yet the 7970 does not have twice the graphics power of the 7850 at the same clock speed in practice.

 

[BenSkywalker]

There's other factors, like memory bandwidth, texture units, ROPs, rasterizers, etc. These don't necessarily scale up along with "graphics cores".

Actually, outside of memory bandwidth they do, they do precisely that. A graphics core is a set combination of ROPs and shader units(ROP is a raster op pipeline that handles texture/raster ops). The interesting part about Wayne is that it is Kepler core versus G70 based derivatives in the older Tegra parts, it is possible nV is being rather conservative with their 6x estimate(although, real world performance could well be in line with that even if theoreticals put it at closer to 10x).

 

[Saylick]

Actually, outside of memory bandwidth they do, they do precisely that. A graphics core is a set combination of ROPs and shader units(ROP is a raster op pipeline that handles texture/raster ops). The interesting part about Wayne is that it is Kepler core versus G70 based derivatives in the older Tegra parts, it is possible nV is being rather conservative with their 6x estimate(although, real world performance could well be in line with that even if theoreticals put it at closer to 10x).

If so, that would be very interesting, but that doesn't seem to justify the 6x increase over Tegra 3, which has 1/6th the amount of shaders. If they were Kepler cores, I'd assume we'd get >6x the performance. But yes, perhaps nVidia is being super conservative. Source?

 

[StrangerGuy]

I believe it when I see the actual results. Until then its NV SoC hypeeeeee as usual.

 

[-Slacker-]

GPU "cores" do scale like that.

I think you have the GPU confused with the CPU.

In graphics, yes, that is exactly how it works. The reason more cores don't buy you linear increases in power is because the problems don't have enough parallelism. In graphics, you fire up a shader program for each and every pixel you draw on the screen, and, with few exceptions, they can all work in parallel. So, at a reasonable resolution, you don't run into problems until you have a million SPs.

*Rubs forehead in homicidal frustration*

And are all shaders equivalent to one another? Are the shaders on a kepler based gpu the same as the ones on fermi? Is a gtx 680 3 times more powerful than a gtx 580?