The Playstation 4 Hardware Information Thread
- Thread starter Troublegum
- Start date
Operating the system with your voice
http://manuals.playstation.net/document/en/ps4/basic/voice.html
Seems to go beyond the SDK, too.
Which bring me to my next comment.... I remember mentioning this to a certain someone and that Yoshida was probably saving this information for a later date.
http://manuals.playstation.net/document/en/ps4/basic/voice.html
Seems to go beyond the SDK, too.
Which bring me to my next comment.... I remember mentioning this to a certain someone and that Yoshida was probably saving this information for a later date.
Yeah seems easily fixed, although a little concerning that it is wreaking HDMI cables until it's fixed. I imagine that many HDMI cables are going to get fubar'ed through people not knowing to push the peice of metal back down first.
http://www.twitch.tv/thejustinflynn
This guy is reporting his disc drive is making loud noises and he swears it's going to die soon.
This guy is reporting his disc drive is making loud noises and he swears it's going to die soon.
http://www.chipworks.com/en/technical-competitive-analysis/resources/blog/inside-the-sony-ps4
http://www.ifixit.com/Teardown/PlayStation 4 Teardown/19493
PS4 revealed to have 20 CUs, 2 disabled for yields. More tech info incoming.
Yes they decapped it already.
http://www.ifixit.com/Teardown/PlayStation 4 Teardown/19493
PS4 revealed to have 20 CUs, 2 disabled for yields. More tech info incoming.
Yes they decapped it already.
So chipworks measured the PS4 SoC at 348 sq mm, which at 28nm is quite a bit more than 3 bil transistors. According to chipworks the GPU accounts for 33% of the SoC.
Hotchips conference claimed Xbox One is 363 sq mm and 5 bil transistors. Those might be estimates or rounded up/down, we'll have to wait for the delidding.
Assuming those mm/transistor counts are right, PS4's SoC is almost as large as XB1, without 32MB ESRAM taking up 1.6 bil transistors.
Hotchips conference claimed Xbox One is 363 sq mm and 5 bil transistors. Those might be estimates or rounded up/down, we'll have to wait for the delidding.
Assuming those mm/transistor counts are right, PS4's SoC is almost as large as XB1, without 32MB ESRAM taking up 1.6 bil transistors.
Going a bit more in depth with the decapping results. Based on the 20 CUs w/ 2 disabled, PS4 is between a 7850 and 7870 w/ GPGPU customizations added on.
7850 = 7870 w/ 4 CUs disabled
PS4 = probably based on a 7870 w/ 2 CUs disabled
7870 = 0 CUs disabled
The decapping also puts the "14+4, 4 CUs can only be used for compute" to rest, as each CU is clearly identical.
If PS4 has a similar transistor count as XB1, it goes a long way to explaining why PS4 had more room to add 6 extra CUs, 16 more ROPs, more GPGPU support, etc. as it doesn't have 1.6 bil of ESRAM eating into the budget. That is a rather unfortunate design choice for the Xbox in hindsight.
This also means that the PS4 SoC has about the same yield rate as XB1, probably a little better because it's not quite as big. Given the difference in supply, it lends credence to the theory that PS4's production started earlier while MS was farther behind for whatever reason (my guess was they planned to launch in 2014).
7850 = 7870 w/ 4 CUs disabled
PS4 = probably based on a 7870 w/ 2 CUs disabled
7870 = 0 CUs disabled
The decapping also puts the "14+4, 4 CUs can only be used for compute" to rest, as each CU is clearly identical.
If PS4 has a similar transistor count as XB1, it goes a long way to explaining why PS4 had more room to add 6 extra CUs, 16 more ROPs, more GPGPU support, etc. as it doesn't have 1.6 bil of ESRAM eating into the budget. That is a rather unfortunate design choice for the Xbox in hindsight.
This also means that the PS4 SoC has about the same yield rate as XB1, probably a little better because it's not quite as big. Given the difference in supply, it lends credence to the theory that PS4's production started earlier while MS was farther behind for whatever reason (my guess was they planned to launch in 2014).
Last edited:
1) It was known since January that each CU was identical.
2) It's still balanced for rendering at 14CU, with an additional 4CU's that don't add much when used for rendering...those 4CU's are there for GPGPU tasks. So no, it doesn't change the discussion on that topic in any way whatsoever unless you find clinging to strawmen arguments enlightening.
3) We also knew about the 2CU's disabled but good to get confirmation I suppose.
The 'more GPGPU support' is the 4CU's. And MS has their GPGPU bet on low latency as opposed to more general purpose CU's (they tested and discovered latency was much more problematic for typical GPGPU tasks than a perceived lack of parallelism). We will see how that shakes out over the cycle but nothing right now is suggesting either design is favored.
It is presumably better than X1's yields because the eSRAM is mighty tricky to manufacture. Just a guess, but most of us have presumed for a long time now that this was the case. Also, MS never planned anything but a 2013 launch. That's ignorant FUD which is conclusively falsified by the Yukon leak.
Also, before anyone gets excited about a 1.8GHz CPU you'll see ppl taking at face value on GAF, it's from a website that is guessing and did so back in October. That info is likely made up (or more likely a typo).
Wasn't this debunked? Its 256mb ddr3
Ok, it does say that in the tear down but its 2gigabit so that's 256mb
Last edited:
PS4 APU Apparently Has 128 Hidden Stream Processors – 10% More Power
AMD has delivered its APU to both the Xbox One and PS4 in the form of Bonaire and Pitcairn GPUs and Jaguar Cores. The PS4 according to last known reports had 1152 SPs and 8 Jaguar Cores, but guess what, information has surfaced that shows 128 SPs completely unaccounted for.
128 SP Unaccounted for in PS4 GPU ‘Pitcairn’ Raising the Total Upto 1280 SPs
Well before everyone blows the gaskit, keep this in mind: Even though the logic of the source is undeniable it could very will be that the SPs are dead and left over in the fab process. Meaning that they are as good as useless. However i will admit this that the probability of them merely inactive is far greater than permanently deactivated . Ok here it is. The leak source this time is MyDrivers.com. They apparently took a tear down of the PS4 APU and noticed something peculiar. There are 20 groups of 64 SPs. 20×64 = 1280. That is 128 more than the official record of 1152.
Plain View of the Transistor Level Kernel
Notice how there are 20 AMD Radeon Graphic Cores. We know each core is a group of 64 Stream Processors/Shader Cores/ALUs(Whatever it is you want to call them) and 20×64 =/= 1152. Infact we previously thought there would be only 18 cores, but as you can see there are 20. So we basically have 2 Full Cores out of 20 unaccounted for. That is 10% of Pitcairn’s power right there.
An Update by Sony could Unlock the remaining 128 SPs or are they already working?
There are basically three possibilities. 1) The 128 SPs are ‘dead’ (Yield) and permanently deactivated. There are ways to do so. 2) Those SPs are active and are hidden to provide extra juice (unlikely). 3) Those SPs are inactive and can be reactivated by Sony. Now it is option 3 that takes our fancy. If the going gets tough later down the road, Sony can just release a ‘miraculous’ patch that will boost the PS4′s GPU ‘Pitcairn’s power by 10%. Which is quite a decent amount we might add.
AMD has delivered its APU to both the Xbox One and PS4 in the form of Bonaire and Pitcairn GPUs and Jaguar Cores. The PS4 according to last known reports had 1152 SPs and 8 Jaguar Cores, but guess what, information has surfaced that shows 128 SPs completely unaccounted for.
128 SP Unaccounted for in PS4 GPU ‘Pitcairn’ Raising the Total Upto 1280 SPs
Well before everyone blows the gaskit, keep this in mind: Even though the logic of the source is undeniable it could very will be that the SPs are dead and left over in the fab process. Meaning that they are as good as useless. However i will admit this that the probability of them merely inactive is far greater than permanently deactivated . Ok here it is. The leak source this time is MyDrivers.com. They apparently took a tear down of the PS4 APU and noticed something peculiar. There are 20 groups of 64 SPs. 20×64 = 1280. That is 128 more than the official record of 1152.
Plain View of the Transistor Level KernelNotice how there are 20 AMD Radeon Graphic Cores. We know each core is a group of 64 Stream Processors/Shader Cores/ALUs(Whatever it is you want to call them) and 20×64 =/= 1152. Infact we previously thought there would be only 18 cores, but as you can see there are 20. So we basically have 2 Full Cores out of 20 unaccounted for. That is 10% of Pitcairn’s power right there.
An Update by Sony could Unlock the remaining 128 SPs or are they already working?
There are basically three possibilities. 1) The 128 SPs are ‘dead’ (Yield) and permanently deactivated. There are ways to do so. 2) Those SPs are active and are hidden to provide extra juice (unlikely). 3) Those SPs are inactive and can be reactivated by Sony. Now it is option 3 that takes our fancy. If the going gets tough later down the road, Sony can just release a ‘miraculous’ patch that will boost the PS4′s GPU ‘Pitcairn’s power by 10%. Which is quite a decent amount we might add.
We will wait for confirmation about the CPU clock rate. Though doubt we will get any and it won't make any difference anyways. PS4 is already comfortably more powerful than X1 with or without it.
There are no system bottlenecks that would give heavy diminishing returns on the last 4 CUs doing rendering work. 32 ROPs and GDDR5 width ensures that. CUs can be dedicated to rendering or GPGPU in any ratio. The main customizations are an added cache bypass bus to more easily allow idle CUs to do GPGPU work, better L2 cache support, and a significant increase in ACEs and compute queues. The GPGPU customizations are well documented and are already being used in games like Resogun.
http://en.wikipedia.org/wiki/PlayStation_4_technical_specifications#Graphics_processing_unit
http://www.vgleaks.com/orbis-gpu-compute-queues-and-pipelines
http://www.vgleaks.com/more-exclusi...mplementation-and-memory-enhancements-details
http://www.vgleaks.com/playstation-4-includes-huma-technology
XB1's CUs are capable of GPGPU tasks, but with 6 less CUs, only 2 ACEs/8 compute queues, and no GPGPU customizations that I'm aware of. I can hardly find any mentions of GPGPU by MS, let alone how "latency" gives them an advantage in it. RAM latency overall is a non-issue. If you have official sources or proof, feel free to post it.
PS4 will always have a ~50% stronger GPU, 16 more ROPs, better GPGPU support, and unified memory. It is a measurably more powerful system in pretty much any way that matters to real world games performance.
"conclusively falsified by the Yukon leak." I wouldn't put much stock in an old slide with estimated dates on it against other evidence, but okay.
We don't know the final CPU clock rate but vgleaks said 1.6, and I'm guessing it's still 1.6. It's nice to think about PS4 being OCed, but with that case size I doubt it. If they went with a bigger case and fan they might have had more room to OC.
http://en.wikipedia.org/wiki/PlayStation_4_technical_specifications#Graphics_processing_unit
http://www.vgleaks.com/orbis-gpu-compute-queues-and-pipelines
http://www.vgleaks.com/more-exclusi...mplementation-and-memory-enhancements-details
http://www.vgleaks.com/playstation-4-includes-huma-technology
XB1's CUs are capable of GPGPU tasks, but with 6 less CUs, only 2 ACEs/8 compute queues, and no GPGPU customizations that I'm aware of. I can hardly find any mentions of GPGPU by MS, let alone how "latency" gives them an advantage in it. RAM latency overall is a non-issue. If you have official sources or proof, feel free to post it.
PS4 will always have a ~50% stronger GPU, 16 more ROPs, better GPGPU support, and unified memory. It is a measurably more powerful system in pretty much any way that matters to real world games performance.
"conclusively falsified by the Yukon leak." I wouldn't put much stock in an old slide with estimated dates on it against other evidence, but okay.
We don't know the final CPU clock rate but vgleaks said 1.6, and I'm guessing it's still 1.6. It's nice to think about PS4 being OCed, but with that case size I doubt it. If they went with a bigger case and fan they might have had more room to OC.
Last edited:
Compared to high end PC GPUs? Sure, PS4 gets clobbered. But it still has that price/performance thing going for it that makes consoles so popular.
Nah its not bad at all in my opinion, just like I said in another thread a lot of ppl are so in love right now with the upgrade because its been 7 years. That's a long generation.
Yes it was. But damn, man. I remember it like it was yesterday.
Where the f*ck did those 7 years go? I didn't feel anything.
OfficialSono
XBL/PSN/Nintendo - OfficialSono.
How do you tell if your controller is charging? on the Playstation 3 you were able to press the 'PS' button and it would show a charge in the top right corner but that's not the case with the PS4. The only time I've seen a battery sign on the PS4 is when you initially power up the console.
Hold down that button for a second. It's on the bottom of that screen.