PlayStation 4 (codename Orbis) technical hardware investigation (news and rumours)

Status
Not open for further replies.
To be fair... German electricity prices are round about 6 or 7 times that (I am paying about 24cents/kwh).

But still... Sony announced that the "secondary chip" was supposed to allow just that... and now it really doesn't do it... yet. The standby mode is still buggy for me (haven't tried recently, so it could be fixed by now). Adding downloads via the website of PSN doesn't lead to the console actually downloading it, only when switching it on does it start... which is a bit... backwards it seems (PS3 can do it, when you enable PS+ auto update).

Didn't have a problem like that with mine... Downloads seem to be working pretty well when the consolde is off, by adding them on the iphone app for example. Maybe there's something you need to set up on the PS4 to allow that?
 
To be fair... German electricity prices are round about 6 or 7 times that (I am paying about 24cents/kwh).

But still... Sony announced that the "secondary chip" was supposed to allow just that... and now it really doesn't do it... yet. The standby mode is still buggy for me (haven't tried recently, so it could be fixed by now). Adding downloads via the website of PSN doesn't lead to the console actually downloading it, only when switching it on does it start... which is a bit... backwards it seems (PS3 can do it, when you enable PS+ auto update).

Didn't have a problem like that with mine... Downloads seem to be working pretty well when the consolde is off, by adding them on the iphone app for example. Maybe there's something you need to set up on the PS4 to allow that?

Same here, works great for me. And yes, you have to enable downloads in standby for that to work. I think its enabled by default for NA systems, disabled by default for certain other regions. But I may not be remembering that correctly. Anyway, see here and here for instructions.
 
I have set it up... but I haven't really checked for it since... around launch I guess... Maybe it was fixed long ago and didn't notice^^
 
I have set it up... but I haven't really checked for it since... around launch I guess... Maybe it was fixed long ago and didn't notice^^
Yeah they never say anything when they patch bugs in the firmware, it's always the generic statement "System software stability during use of some features has been improved."
 
Where on the APU die are the 8x ACE?

You can see the 18CU are ringed by blocklike µarchitecture, many of which are 32 ROPs. AMD Diagrams representing ACE show them to be relatively small compared to CU and located near the primary gpu command processor. Speaking of which where is the command processor on the floor plan? For the PC gpus Both AMD & Nvidia command processor is easy to locate on the transistor layer, its sandwiched between half of the CU/streamprocessors.

But on all apu die shots regardless if they are llano or kavari or the xbox one or ps4 I cannot find them.
 
Last edited by a moderator:
AMD Diagrams representing ACE show them to be relatively small compared to CU and located near the primary gpu command processor.
Don't rely on the logical block diagrams to tell you anything about the physical components and layout of the chip.
 
Don't rely on the logical block diagrams to tell you anything about the physical components and layout of the chip.

That would be the initial thought for anyone 1st looking at certain diagrams, and that is what I initially thought, and if someone initially asked me I'd consider them ignorant for thinking you could. But on closer examination of some of the diagrams (not all of them) you still can take some facts away. In a some of the diagrams the creator tries to have just a small degree of attempt not to be toooo far off in blocks sizes and shape for the known locatable components in the die shots. Knowing that they put this effort into the compoennts widely known by this threads regular posters including u and I, we can hopefully use this small bit of info to help us locate the unknown location components. In the various shape&size semi accurate diagrams of ACE they are always smaller than compute units.

Thats all I'm taking away.
 
Last edited by a moderator:
Haven't had that issue myself. Everything downloaded as expected so far.
 
The standby automatic downloading has been working for me since launch

Yeah, works fine for me. It downloads firmware updates, I initiated few downloads manually via android app and web interface, and it continues to download games when I initiate them via PS4 store and then put console into standby.

Zero issues so far, although it is a shame that it cannot maintain game state during standby mode. Hopefully they will enable that during 2014.
 
I'm hoping the reason they can't go with low power network standby mode and suspend / resume is simply a software problem and not a hardware design problem, which would be unfixable without a hardware revision. I'm hoping otherwise, but it may be that first line of PS4's won't be low-power operational? ("fixing" these wouldn't create backwards compatibility problems, btw.)

I have no problems with downloading during standby, btw.
 
I'm hoping the reason they can't go with low power network standby mode and suspend / resume is simply a software problem and not a hardware design problem, which would be unfixable without a hardware revision. I'm hoping otherwise, but it may be that first line of PS4's won't be low-power operational? ("fixing" these wouldn't create backwards compatibility problems, btw.)

I have no problems with downloading during standby, btw.

I dunno. If like we can assume the low power ARM is downloading only compressed data, I presume it would need a much powerfull CPU than the weak ARM to uncompress it in order to install the stuff properly?

In that hypothetical case to decompress the data then 1 or 2 cores + the GDDR5 ram would always be needed and powered. That would explain the 70W figure when PS4 is standby-downloading.
 
I dunno. If like we can assume the low power ARM is downloading only compressed data, I presume it would need a much powerfull CPU than the weak ARM to uncompress it in order to install the stuff properly?

In that hypothetical case to decompress the data then 1 or 2 cores + the GDDR5 ram would always be needed and powered. That would explain the 70W figure when PS4 is standby-downloading.

I see no reason why the ARM core cannot decompress the data, it is likely to decompress it slower unless it has some sort of fixed function accelerators. Given that background downloading and installing most likely occurs over a large amount of time the ARM core being massively slower shouldn't really be a problem..
 
I see no reason why the ARM core cannot decompress the data, it is likely to decompress it slower unless it has some sort of fixed function accelerators. Given that background downloading and installing most likely occurs over a large amount of time the ARM core being massively slower shouldn't really be a problem..

Curious. Is decompression faster than a typical download? Surely it will change according to the country, but surely the processor can decompress faster than an average 1 or 2MB/s connection? In that case it doesn't matter how fast the decompression is, as we're still limited by our download speeds?
 
It'd have to be a hell of a slow CPU to not be able to decompress a couple MB/s realtime. Basically this is almost impossible, late 90s CPUs were fast enough for that. Some might recall that Metroid Prime on gamecube loaded compressed assets off that mini-DVD disc and did realtime decompression while the game was running (using leftover CPU cycles.)
 
Maybe it's a security problem, not performance?

All access to the 16bit DDR3 can be read with a 32 channel logic analyzer, and most EE students have access to a lab with one. So I suppose it can't be easy to keep all keys secure in this situation, for both the PSN protocol and the disk encryption. The entire south bridge memory would have to be considered insecure. :???:
 
Status
Not open for further replies.
Back
Top