Predict: The Next Generation Console Tech

Status
Not open for further replies.
A 500Gb mechanical drive would be better, more storage to rip the whole game to the HDD if needed which would be much better for streaming.

SSD's are overrated.

I agree that copying the whole game to the HDD is better for streaming assets, but the problem is : 1- horrible waiting times for console gamers, 2- not enough HDD space to copy each time 25 Gb+ game.

SSD's are overrated.

I dont agree with that, do you use a ssd ? personally after using a SSD I cant go back to HDD, never.

SSDs are a must, maybe even for consoles if we want to do real impressive multitasking (a lot of parallel downloads, playing a game, video-audio chat, and browsing the internet and sending an HD video stream to other devices, and recording TV shows...all this at the same time, try to do that without a SSD and you will kill your operating system)
 
At which manufacturing process should a Cape Verde needs to be produced to run at 2 ghz speed like it runs at 1 ghz on 28nm process ? Or can it be clocked to 1.6 ghz which gives about 2 TF of performance !
 
SSDs are a must, maybe even for consoles if we want to do real impressive multitasking (a lot of parallel downloads, playing a game, video-audio chat, and browsing the internet and sending an HD video stream to other devices, and recording TV shows...all this at the same time, try to do that without a SSD and you will kill your operating system)

OT but everything you mention could easily be done with a normal hdd except for maybe recording (never done this so don't know how much of a strain that puts on your drive).
 
SSDs are a must, maybe even for consoles if we want to do real impressive multitasking (a lot of parallel downloads, playing a game, video-audio chat, and browsing the internet and sending an HD video stream to other devices, and recording TV shows...all this at the same time, try to do that without a SSD and you will kill your operating system)
I think those are pretty unrealistic requirements. You aren't going to be browsing the internet and gaming - you'll switch one out to do the other. Video and audio chat can be cached in RAM and doesn't need to access the disk drive. Sending a film remotely is just a little transmission at the BW of the video, so a 40 megabit video sucks 5 MBps from your BW, which is nothing and video is inevitably less than that. The only real issue is disk thrashing, moving the head around three files at a time (game/browser, film streaming and download), which can be managed with suitable system design (decent HDD cache either in RAM or on HDD). Certainly massive multitasking isn't reason to add an SSD next-gen. The only real reason is random-access performance for data streaming in games. There are so many smart devices out there, the Wii approach of a system farming out content to dumb terminals makes very little sense.
 
It has 1080 pads so I'd hope it's 1024 bits. Makes me think Sony will do the same if they can handle the heat on the RAM. Then they can use cheap RAM.
They clearly made a mistake in the article, I looked at the the real WideIO specs, and it's 44x6 microbump per 128bit channel, so it all adds up to 512bits needing 1056 pads. The Vita has a slightly non-standard version which is 45x6 microbump per channel. Looking at the die it's clear that's a 4 channel chip too.

It's very fast and low power either way :LOL:
 
AMD and Apple are in the HBM camp, while Microsoft and IBM are in the HMC camp.
According to this presentation (AMD TFE 2011), they see HBM as the direct successor to GDDR5 for GPUs. Diff-GDDR5 looks like it will never happen, nor GDDR6. They see GDDR5M as the interim solution to replace DDR3 in low cost situation before HBM becomes mainstream, and not DDR4. (why not ddr4???)
http://sites.amd.com/us/Documents/TFE2011_006HYN.pdf
So the AMD image of a GPU with two memory chips on an interposer, which we saw earlier, were most probably 2 HBM 1024bit, giving 256GB/s.
 
Last edited by a moderator:
HBM seems to be a more straightforward port of DRAM to a stacked interposer format.

The largest apparent difference that I can see is that there isn't a layer of logic at the bottom of the DRAM stack.
HMC provides controller logic and high-speed bidirectional links. A lot of other possibilities for topology, DRAM technology, and simplification of CPU and GPU memory controller logic can come with HMC.
It's possible the on-stack logic can affect some form of repair or defect compensation that HBM may not.

HBM seems to be okay if focusing on a limited subset of capabilities using existing DRAM tech, while HMC seems to offer more applicability, expandability, and future development.
 
Nope. Simple, highly parallelisable code working on fancy data structures. Xenon and Cell will be as good as any SIMD architecture in this regard. If the x86 processors in the next-gen consoles are less potent in the SIMD department (highly likely), there'll be no improvements in the physics department (except physics may shift to GPU). Furthermore, an x86 processor can be anything from i7 to Celeron, so that doesn't tell us anything about the performance. x86 only tells us the instruction set of the CPU, and some assumptions about architecture which can be ignored when we have specific details about the cores (AMD Jaguar or whatever).
Actually, what is the nature of the GPU? It is mostly SIMD computation logic. No cache, little branch prediction, very little instruction re-ordering. Thus it's exceptionally bad for anything except 3D Math. For instance Ray Tracing - you've got to go through all the vertices in a scene to see what the bullet intersects with, for each bullet. Hard to do without a cache! The cache greatly accelerates it - but storing it on the chip. Thus the Power7 is ideal, with up 80 MB of eDRAM cache.
 
They clearly made a mistake in the article, I looked at the the real WideIO specs, and it's 44x6 microbump per 128bit channel, so it all adds up to 512bits needing 1056 pads. The Vita has a slightly non-standard version which is 45x6 microbump per channel. Looking at the die it's clear that's a 4 channel chip too.

It's very fast and low power either way :LOL:

Thanks for correction and clarification. I expect it in at least one of the two next gen consoles.
 
You've got to look at this from the CPU's perspective. The extremely complex program that governs physics is decompiled to assembler code. The Intel chip - goes through the code analyzing for dependencies, storing often used variables first, and executing the code without dependencies first. Storing the assembler code in on-chip buffers. Thus - it can go through extremely complex code quickly, and with little effort by the developer. The GPU - no cache, no ability to analyze for dependencies, no out-of-order instruction execution, thus making it an order of magnitude slower for any code, except extremely simple SIMD (3D math).
 
I dont agree with that, do you use a ssd ?

I've used more SSD's then most people....

Had 4x OCZ Vertex drives back when SSD's hit the market... over £500 on drives.... my other half was not happy.... :D

Now I run an SSD for my OS and a couple of mechanical drives in RAID 0 for my games, the mechanical drives average ~235mb/s :thumbup:
 
You've got to look at this from the CPU's perspective. The extremely complex program that governs physics is decompiled to assembler code. The Intel chip - goes through the code analyzing for dependencies, storing often used variables first, and executing the code without dependencies first. Storing the assembler code in on-chip buffers. Thus - it can go through extremely complex code quickly, and with little effort by the developer. The GPU - no cache, no ability to analyze for dependencies, no out-of-order instruction execution, thus making it an order of magnitude slower for any code, except extremely simple SIMD (3D math).

3D_world, you're out of your element. I hadn't had a good laugh for the day, so thanks for your laughable post.
 
Status
Not open for further replies.
Back
Top