PlayStation Vita has swiftly established itself as the premier platform for the most technologically advanced mobile games, boasting a simply superb hardware spec, a beautiful screen and an excellent array of input controls. The only thing that disappoints is Sony's historical annoyance in insisting upon proprietary memory cards in an environment where almost everyone else seems to get by just fine using SD cards and their micro equivalents. This article asks if they are needed, and whether the big price premium translates into a better experience for the gamer.
Sony says that the cards are required for security concerns (an odd excuse for a system that lets you back up all content to PC or Mac, and not a good enough reason to shift the cost onto the consumer) and also says that a set performance level is required. Bearing in mind how many games stream in assets while you play, this second reason makes a lot more sense. At Digital Foundry, what we wanted to know was how fast these proprietary cards are, how they compare to what's already on the market, and to discover what the performance level is that Sony deems appropriate to justify a proprietary set-up.
It's fair to say we didn't go into this in a particularly optimistic frame of mind: when it takes 9.5 minutes to back up WipEout 2048 from Vita to PC (a game with a download size of 1.6GB, suggesting a paltry 2.87MB/s transfer speed), we aren't particularly optimistic about performance here, but we wanted hard numbers across a range of scenarios.
A good test is sequential read and write speed. We tried to measure this by using a 425MB 720p video file in the h.264 format, implementing the Content Manager to stream the file from computer to Vita and back again, repeating the test three times. We asked four users - two using the 8GB card and two 16GB owners - to help us out here to give a range of results.
First up, a look at read performance - probably the most important element for games.
"When it takes 9.5 minutes to back up the 1.6GB WipEout 2048 to your PC, it's safe to say that we didn't go into these tests with much optimism."
|8GB Card 1||8GB Card 2||16GB Card 1||16GB Card 2|
|Read Test 1||6.44MB/s||6.16MB/s||7.20MB/s||7.46MB/s|
|Read Test 2||6.34MB/s||6.54MB/s||7.33MB/s||7.46MB/s|
|Read Test 3||6.25MB/s||6.44MB/s||7.33MB/s||7.87MB/s|
Read speeds are hardly colossal for flash media, bearing in mind that even a cheapo USB flash drive should be able to provide a 10MB/s throughput, and a carefully chosen bargain stick such as the ByteStor Data Ferry can achieve over 15MB/s with ease. Class 10 microSDs can easily outstrip the read speeds here too and are a quarter of the price of the Vita cards. In terms of other comparisons, the PS3's Blu-ray drive can achieve a maximum throughput of about 8MB/s, so the Vita gets close with the 16GB card but falls short on the 8GB variants we tested.
Now onto the write speeds, where things take a turn for the bizarre, and where performance falls through the floor. What we should emphasise here is that all runs were consecutive with no breaks and the data clearly indicates that caching isn't going on - as results can get worse on later runs, rather than better. What we'll also point out is that this is one single file being transferred. Transferring scores of photos across to the Vita sees performance fall still further - dramatically so in fact, which may explain why WipEout and other titles take so long to back up onto your PC.
|8GB Card 1||8GB Card 2||16GB Card 1||16GB Card 2|
|Write Test 1||5.45MB/s||3.63MB/s||4.47MB/s||7.08MB/s|
|Write Test 2||4.34MB/s||5.0MB/s||5.18MB/s||7.72MB/s|
|Write Test 3||2.81MB/s||4.21MB/s||7.33MB/s||7.08MB/s|
The results here are eye-opening, indicating significant disparities not just between 8GB and 16GB cards but also between individual users. It's safe to say that write speeds - a traditional weakness of flash RAM - were all over the place, particularly poor on the 8GB stick with only our top-performing 16GB card showing any kind of consistency on both read and write tests. Quite why we see so much inconsistency between one test and the next is completely unknown to us - it may well be an issue with the Content Manager for all we know, though read speeds perhaps suggest otherwise. These are fairly consistent between users, but again show a clear advantage to the 16GB cards completing the read task seven to ten seconds faster on our test file.
"We see a clear performance increase between our 8GB and 16GB test cards and data from Japan suggests that a 32GB upgrade boosts speeds still further."
We also asked three 32GB users in Japan to contribute data. While the tests were not quite as exhaustive, we saw better results again. Read times varied between 8.3MB/s for one lucky user, with the slowest result coming in at 7.32MB/s. Write speeds saw a max of 8.3MB/s for the user with the 32GB "golden card", averages of 7.08MB/s for another and a slowest recorded speed of 6.07MB/s.
Overall results are conclusive despite some of the strange data: the bigger the card you buy, the better the performance you will get, and there are obvious speed advantages in buying the 16GB card, with less variance in write speed if you get the 32GB offering (still unavailable in Europe).
What isn't conclusive is whether Content Manager has issues resulting in the poor write speeds, and whether encrypting/decrypting content might have an impact. This may explain poor sync times for game content, but why would Content Manager bother encrypting a simple video file? It doesn't make sense. Regardless, we wanted to see if the disparity extended to gameplay, so we did further testing with the same cards. Suffice to say that the difference we saw in the read/write tests extends to game loading times too.
Real Life Testing: WipEout 2048
WipEout 2048 takes a long time to load levels, even in its new patched form, which we found to lop off around 10 seconds of loading time on a 16GB card (based on the Race 1/Race 2 tests below). We decided to compare the patched game across the four memory cards we had available for testing. Alas, our Japanese 32GB testers didn't own WipEout so couldn't take part.
The results once again show a level of disparity between all testers, but once again there are comfortable advantages for the 16GB card owners.
Indeed, on the Race 1 result we see the best 16GB card outperform the poorest 8GB card by over 35 per cent. Quite why we are seeing this is unclear: all of the cards contained varying amounts of content and in an ideal world we'd have been running WipEout and nothing else, but the notion of more free space buying faster speeds doesn't correlate with our data when the "faster" 8GB stick actually had more content on it than the slower one...
|8GB Card 1||8GB Card 2||16GB Card 1||16GB Card 2|
|Race 1||50 secs||47 secs||45 secs||37 secs|
|Race 2||43 secs||40 secs||37 secs||34 secs|
|Race 3||30 secs||28 secs||28 secs||28 secs|
|Race 4||39 secs||37 secs||35 secs||36 secs|
|Average||40.5 secs||38 secs||36.25 secs||33.75 secs|
Fascinating stuff. Previously we'd compared the extended load times of WipEout 2048 in its unpatched form operating from the memory card compared to its retail "cartridge" companion. Across the run of results we saw that sometimes the download was a couple of seconds ahead of its retail counterpart, but sometimes the reverse was true meaning no overall load time win for either form.
"WipEout's recent patch lopped off up to ten seconds of loading time, but we still saw noticeable performance differences between our 8GB and 16GB Vita test cards."
All of which leads us to wonder whether the proprietary cards are actually justified. After all, Xbox 360 gets by quite nicely by allowing you to use virtually any USB flash storage medium even for complete game installs, simply refusing to operate if your storage isn't fast enough - so why not do the same on Vita?
In comparison to microSDs out there, we see the Vita cards operating at the equivalent of Class 2 to Class 6, depending on what kind of mood you catch it in (the class system is defined by write speeds). Read times are not really covered by the SD Class system, but in our experience they're either the same or quite a bit better - flash memory in general tends to be more challenged when it comes to writing data as opposed to reading it back.
The Vita cards seem to max out at somewhere in-between 6MB/s to 8MB/s, generally speaking - quite a large delta. At the time of writing, a 16GB Vita card costs around £40 while a Class 10 MicroSD with a reputable brand, offering the same storage, can be purchased from Amazon for as little as £10.
So does the Vita cards' premium price-point buy you a good performance level? Seemingly not, but we have to make do with the hand we've been dealt, and it seems to be the case that the higher capacity the card, the better the performance level. Try as we might, we can't see more empty space buying better read/write times and again it would go against the ethos of the proprietary system guaranteeing a certain performance level. Were it not for the eye-watering £40 price-point, the 16GB card would be an instant recommendation - if only for the fact that there are so many good games, that 8GB card is going to fill up pretty quickly...
Will you support the Digital Foundry team?
Digital Foundry specialises in technical analysis of gaming hardware and software, using state-of-the-art capture systems and bespoke software to show you how well games and hardware run, visualising precisely what they're capable of. In order to show you what 4K gaming actually looks like we needed to build our own platform to supply high quality 4K video for offline viewing. So we did.
Our videos are multi-gigabyte files and we've chosen a high quality provider to ensure fast downloads. However, that bandwidth isn't free and so we charge a small monthly subscription fee of $5. We think it's a small price to pay for unlimited access to top-tier quality encodes of our content. Thank you.Support Digital Foundry