The problem with Bloodborne's loading times between deaths is not a hardware one but obviously something to do with how they load the levels. Surely the game doesn't need to flush out and reload that much data in and out of the RAM? And surely they can fix it in an update? Where other games have no loading times at all, it seems like something is broken in Bloodborne.
Have you upgraded your PS4 HDD?
- Thread starter orangpelupa
- Start date
D
Deleted member 11852
Guest
I'm not suggesting Bloodborne loading times are bad because of a hardware limitation just that tests done thus far (including the previous Digital Foundry testing) demonstrate that it's not possible for SSDs in PS4 to get anywhere near their theoretical maximum performance. Mark Cerny has spoken several times about all games being compressed and the interface (USB IIRC) and decompression hardware are likely the bottleneck.
I wonder if the game is perhaps aggressively steaming in the environment rather than loading in a level and the assets required. Or perhaps the data is badly organised on disc so large reads have to take place to pick out the few bits the game needs. I.e. they've compressed entire asset paks and to pull a few assets from the middle or end of the pak you need to read from the start and uncompress most of the pak. It could be a lot of things - hopefully they'll explain what they did to improve it when they patch it.
It's probably less of an issue in Japan, they probably undertake the tea ceremony after each death.
those with SSD should also enjoy faster suspend/resume
D
Deleted member 11852
Guest
How so? I thought it was preserving RAM rather than writing out to disk (suspend) and reading it back (resume). Otherwise the power usage in rest mode shouldn't have changed.
my PS4 makes heavy HDD load sound when suspend and resume so i assume they write something to disk. Maybe like the small pagefile that was on windows 8? Its only appear if you turn on HybridSleep/Fastboot
Silent_Buddha
Legend
In theory they could. But I'm guessing one of the reasons that it uses USB is that the custom APU has no SATA interface in order to save a bit on die space. If that's the case there's no reason they couldn't put it back in, but then again there's also no reason why they should put it back in (from Sony's POV as the USB interface doesn't necessarily limit the speed of any conventional HDD they put in).
There's likely a lot of things that exist in standard AMD APUs that just don't exist in the PS4 (and XBO as well for that matter) APUs. You don't need a lot of PCIE lanes for example or support for memory interaces to anything other than the memory that you use as another example.
Regards,
SB
D
Deleted member 11852
Guest
The PS4 has a SATA controller, otherwise SATA HDDs and SSDs wouldn't work at all. I think the reason they chose to connect SATA devices over USB was TrustZone.
loekf
Regular
Makes no sense to me. I would say an USB 3.0 host controller and a SATA interface both sit on the same high-speed bus. I can't imagine they hooked up fast peripherals on a AXI / AMBA (whatever it's called these days) bus exclusively connected to the ARM CPU. That would have major speed and latency bottlenecks.
It's an interesting design choice by Sony.
On topic: upgraded to a 1.5 TB HGST 2.5" when I got my PS4 (end of 2013). Never looked back.
The usb-to-sata bridge on the PS4 should be 3 times faster than a mechanical HDD, and an SSD is doing barely 1.5x on the PS4 so the bottleneck is probably elsewhere.
- Could be a software throttling of the speed for game testing consistency.
- The hardware decompression and decryption could be the bottleneck as everything is compressed.
- They might force the controller on SATA150 for increased signal integrity since the HDD isn't faster anyway.
- The south bridge could have been designed just powerful enough to handle an HDD, for cost saving and power saving.
D
Deleted member 11852
Guest
I agree and both buses would be handled by the South Bridge, which even on PS4 I believe is a separate chip, so the APU is irrelevant.
But the reason is, as I speculate above, for the TrustZone implementation. Having worked on several TrustZone systems I can see why it would be preferable to route all I/O handled by the South Bridge across USB so that the bridge only that bus has to be engineered to be subordinate to TrustZone. You then let to controller arbitrate traffic and the TZ solution becomes much simpler. You can certainly have TrustZone implementations which handle multiple buses simultaneously but I don't see why would want to over engineer this on a games console.
Wow. So I've started backing up to my USB3 external drive.... 5 hours for 300GB worth of data?? Why so slow?
D
Deleted member 11852
Guest
That's a gigabyte an hour or around 16Mb/sec. What sustained write speed do you get from a PC/Mac?
Don't know. But USB3 is around 400Mb/s, isn't it? And no it's not a gig per hour
D
Deleted member 11852
Guest
Don't know. But USB3 is around 400Mb/s, isn't it? And no it's not a gig per hour
Sorry, a gigabyte a minute. There's no way you'll get 400Mb/sec write speed with a HDD except in a RAID array. This is why I'm curious what sustained write speed your backup device/disk can achieve on a PC/Mac.
Oh never mind. I just realised that even 400Mb/s would be 50MB/s. I'm getting around 20MB/s. Still not great but makes more sense now. Panic over. I'll just have to wait. And then wait all over again when I restore the backup. Maybe I should have deleted some stuff I haven't used for ages.
Similar threads
