Tegra 4 slide leaks

[podspi]

I think the main advantage for all of these cores is responsiveness. We often forget most of these mobile devices are phones. Nothing is more frustrating than the phone lagging into the phone interface because another app is using up all the resources. That being said, I feel like memory plays a larger part in this than processing speed, since that can easily be taken care of by smart scheduling.

Mostly, I am disappointed at NVidia's reluctance to really push the bar when it comes to graphics performance.

 

[Idontcare]

I think the main advantage for all of these cores is responsiveness. We often forget most of these mobile devices are phones. Nothing is more frustrating than the phone lagging into the phone interface because another app is using up all the resources. That being said, I feel like memory plays a larger part in this than processing speed, since that can easily be taken care of by smart scheduling.

Mostly, I am disappointed at NVidia's reluctance to really push the bar when it comes to graphics performance.

Why is Nvidia slow in this dept? Do you think they underestimated the competition and got blindsided with their much faster graphics capabilities?

 

[Av9114]

I think the main advantage for all of these cores is responsiveness. We often forget most of these mobile devices are phones. Nothing is more frustrating than the phone lagging into the phone interface because another app is using up all the resources. That being said, I feel like memory plays a larger part in this than processing speed, since that can easily be taken care of by smart scheduling.

Mostly, I am disappointed at NVidia's reluctance to really push the bar when it comes to graphics performance.

My question is could they have just increased the size of the GPU or is there some other bottleneck limiting the graphics performance? If they could have just added GPU cores and they didn't for cost or power concerns, I think the additional CPU cores is a mistake. Does anyone disagree with that?

I can't help but think this is actually the OEMs fault. They have consistently displayed a sort of disinterest in paying for GPU performance. Nvidia ultimately is going to build with their customers want and not necessarily what they think is best.

I think the GPU makers and OEMs also need to do a better job of marketing. General consumers have been taught that more cores and higher clock speeds are better on the CPU side even though that's really not a great way to judge them. There's no simple marketing term for comparing GPUs and thus it's hard to sell them to the average consumer.

 

[Cerb]

My question is could they have just increased the size of the GPU or is there some other bottleneck limiting the graphics performance?

Memory. It's dual-channel, great...but if each channel is the same width as the Tegra 3, that's a whopping 64 bits. One channel worth of bandwidth, in PC terms, but double the potential IOPS.

More balls costs more money. More traces increases PCB area, and possibly layers, costing more money.

Why make a bigger GPU, if it'll just end up crippled by RAM like an AMD Zacate is?