Sony's Next Generation Portable unveiling - PSP2 in disguise

[eastmen]

Yeah, I didn't expect big speeds, because the HDD in the PS3 simply isn't that fast either (doubt that even gets 30+) and the Vita memstick just seems significantly slower than that.


Can't agree with this though:



All of those other systems are hacked.

I don't recall any of them using the advance security features of SD however.

Not only that but SD cards are dirt cheap. I have a $15 micro sd card thats getting 20MB/s

 

[Arwin]

Yeah I know. I actually have a Sony DSC-WX10 that takes a regular SD card. I bought a 16GB card for that one too.

 

[Shifty Geezer]

To be fair, game cards and memory cards are two different things. Not saying one actually performs differently than the other on Vita, but it's not impossible.

The Wipeout test was from the download version on memory card, not on game card.

 

[Exophase]

The Wipeout test was from the download version on memory card, not on game card.

That's the point. They're testing memory card speeds, not game card speeds.. the question raised was about game card speeds. Just because you can run games from memory card doesn't mean the game cards have the same performance.

 

[Shifty Geezer]

That's the point. They're testing memory card speeds, not game card speeds.. the question raised was about game card speeds.

The question asked by DF was how fast the expensive memory cards were. Was someone asking how fast that game cards were?

 

[Exophase]

The question asked by DF was how fast the expensive memory cards were. Was someone asking how fast that game cards were?

You're right, the misunderstanding was mine.. I read the older posts but lost my direction once the comparisons to DVD/Bluray drives were made. Indeed the original question was why these proprietary memory cards were used, not referring to game cards. I'm still surprised Sony moved away from even their established format for this.

Still would be interesting if we could somehow get an idea of how the gamecards perform. Or maybe people have already noticed the load times being the same between them and memory card downloads?

BTW, who is DF..?

 

[Arwin]

The gamecards are known to be a bit faster, performance similar to or faster than the 16GB cards, while being 4GB cards generally themselves.

 

[Ryan Smith]

The gamecards are known to be a bit faster, performance similar to or faster than the 16GB cards, while being 4GB cards generally themselves.

Has anyone cracked open a game card yet? I'm curious as to just what they're using inside.

 

[Shifty Geezer]

BTW, who is DF..?

Digital Foundry, the technical arm posting on Eurogamer and headed by Grandmaster who doesn't post much on this board any more.

 

[dagamer]

Grandmaster has released an investigation. Doesn't look good, and once again Sony are making things complicated and confused. As I mention in the comments, their PC software can cripple transfer rates in their HD camcorders, so it's not conclusive, but the performance in a game load is curious.

I'm not seeing 30+ megabytes a second really happening though.

That test makes me wonder if there isn't some on-the-fly encryption going on by either the card or the OS because those speeds are pretty abysmal. Perhaps a use of one of the ARM cores in the system? Has anyone taken one of these memory sticks apart?

 

[Fafalada]

As I mention in the comments, their PC software can cripple transfer rates in their HD camcorders, so it's not conclusive

It's not PC software - in case of Vita PC side is doing pretty much nothing anyway, and you can easily verify you'll get same speed between Vita and your PS3.
It can easily be a limitation of how system handles UBS transfer though - PSPGo's USB transfer clocks the same ~7MB/sec, and the real-world numbers for its Flash memory speed are about 3x higher then that. Hell my old Samsung phone had USB transfer speed of 1MB/sec where the SD card I used in it would run around 10MB/s by itself - and both of these cases are direct transfers to/from USB device in Windows, no vendor software involved outside the device.

Anyway DF should know better then to make all sorts of conclusions with no real facts to go on.

but the performance in a game load is curious.

Which part? Times mean nothing out of context - I've seen PC games loading minute+ on SSDs that are an order of magnitude faster then Flash cards.
And the differences between different capacity cards aren't unheard of, although I'm not convinced they really exist based on method of measurement either.

 

[Lazy8s]

A question to IMG: does PowerVR's usage of a tile buffer necessitate that internal color operations are performed at full precision? Or, to achieve the original dithering when emulating PSP games, can Vita use a 16-bit tile buffer?

 

[Fafalada]

PSP emulation still runs games dithered, so obviously they "can" do 16bit some way (doesn't mean you need native color-buffer support to do it, but other options might be prohibitively expensive).
And frame-buffer color-depth aside, the math that some games do would require 16bit color AND z for the emulation to work as well.

 

[Shifty Geezer]

Which part?

Well there's a lot of performance variation. It's unclear whether Race 1, 2, 3 and 4 are on the same map or different maps. If the same map (my initial assumption) then the wildly varying load times shows software is playing a big role. If different maps, then we are at least seeing varied results across cards that are supposed to be uniform in experience.

I'll grant that there's no enough information to know what comparisons are being made here. A little clarification would be useful. The real test of IO performance on the cards needs a proper hardware test, and not something dependent on PC software or card readers or USB controllers. Although even if the cards are fast, with the Vita contorller chip can't support that, it's kinda moot and they still should ahve gone wtih SD cards for performance and cost.

 

[Fafalada]

the wildly varying load times shows software is playing a big role.

That's the case in most games - even in those that actually do utilize I/O well (which is lest common then one would think in modern software).

Although even if the cards are fast, with the Vita contorller chip can't support that, it's kinda moot and they still should ahve gone wtih SD cards for performance and cost.

Do we have information about the capacity of controller chip?
Anyway the cost argument is moot - the mobile devices with biggest marketshare mostly use proprietary flash (be it internal or otherwise), priced at least as high (often higher) than Vita cards. It's something that market has shown the willingness to (over)pay for, manufacturers are just taking advantage of that.
On the other hand, the way Sony locked down the Vita, SD would be out of the question regardless of the pricing scheme anyway.

 

[Arwin]

On the other hand, the way Sony locked down the Vita, SD would be out of the question regardless of the pricing scheme anyway.

That's what I thought too, but apparently many people (including Digital Foundry) seem to be dismissive of that aspect of it.

 

[Lazy8s]

Digital Foundry gets some details from Studio Liverpool on WipEout 2048.

Digital Foundry: Lighting can look quite beautiful in WipEout 2048 - can you give us some technical background there?

Chris Roberts: The lighting on PS Vita is virtually identical to the lighting on HD/Fury. Ships are still lit using image-based lighting with blended diffuse and specular probes that we precompute along the track, and weapon effects use the same vertex-based lighting system, although on PS Vita both of these effects are handled by the GPU rather than SPUs as they were on PS3.

It's easier to talk about the differences - the real-time shadows on PS Vita are almost identical with the exception that we decided to use anti-aliased colour buffers for shadow rendering rather than shadow depth buffers.

This allows us to have transparency in our shadows and the memory cost of anti-aliasing is mitigated by the colour buffer only requiring 8 bits per pixel, so a 4xMSAA colour buffer uses the same memory as a 32-bit depth buffer, however it did mean we had to sacrifice self-shadowing on the ships.

The post-effects are slightly improved in WipEout 2048 - tone mapping is more 'correct' because the PS Vita has efficient hardware supported buffer formats, so the exposure control and bloom effects in 2048 are generally better-quality than the HD/Fury counterpart. Also, in 2048 we have longer tracks that tend to require a lot more lightmap texture space - fortunately the SGX supports texture compression with higher compression rates, so even taking the longer tracks into account, there are also improvements in pre-computed lighting resolution.

http://www.eurogamer.net/articles/digitalfoundry-wipeout-2048-tech-interview

 

[Tabris]

Sega said the bottleneck of Vita is the CPU
(Japanese)
http://www.4gamer.net/games/134/G013437/20120402067/
They use all core of the CPU when console's one only use 1 core
They also change the CPU based skinning to GPU based

 

[Deleted member 13524]

So 1 CPU is locked for OS, one is available for gaming.. what of the other two?

Is Sony holding back their hardware again, like they did with the original PSP?

 

[french toast]

So 1 CPU is locked for OS, one is available for gaming.. what of the other two?

Is Sony holding back their hardware again, like they did with the original PSP?

That is dumb as hell...unless the power consumption is bad with all 4 on the go..and they are waiting for node decrease before activation?

Either way if they are not in use....and they have been advertising 'quad core' then consumers are getting ripped off!

On the plus side, it shows mobile graphics power is better than alot of peeps thought, graphics look awesome on Vita..considering Vita uses the best of 'current gen' and we are nearly at the end of that cycle...good times indeed!