Gadgetory

we have brand-new testing that you're seeing T is now and it brings some competition to the space of console benchmarking using gnz rebuilt console testing software we're benchmarking the PSone classics at frame time consistency and gameplay quality today particularly against the backwards compatible ps3 play of the same games the PlayStation Classic is a mini console using a modern arm 1.5 gigahertz quad-core CPU and a pre-loaded selection of 20 games that are about two decades old the price of this combination including two controllers and no power adapter is $100 it's a product targeted at collectors and that nostalgic once owners of the ps1 to collectors go for it I guess but for people who just want to play original ps1 games again another option exists and that option is a $15.00 used ps3 and backwards compatibility that's what we're benchmarking today before that this video is brought to you by the EVGA CLC 280 liquid cooler people ask me how I keep cool during the summer with all this hair well I've tried a lot of different products and few do exactly what I need many of them cause tangles or worse EVGA CLC 280 helps keep my core temperatures low they're in hot benchmarking sessions the CLC 280 is price competitive and focuses on performance for value offering a 280 liquid cooler at an affordable price get yours at the link in the description below hair mounting kit sold separately we already have a teardown of the ps1 classic on the channel if you want to check it out and see what the hardware is inside of this thing it's very small but it's playing games that are about 24 years old so form factor isn't much of a problem because the components inside of it are more or less mobile components now as for alternatives this isn't the only way of course to play ps1 games the ps1 classic itself uses an emulator an open-source one and you can get that emulator if you wanted to so this isn't the only way to do it but another legitimate way to play old ps1 games would be basically any PlayStation console ps2 ps3 for example used for a lower price than you could get this so have you get the PS we used about 15 $20 something like that depending where you are and it too can play ps1 games so we just need to look at which one is the better solution if all you care about is playing those games and as close to the original experience as you can get because even though this looks like a smaller version of the ps1 it is a very distant from the original experience as is the ps3 emulating both are emulating so if you really wanted the original experience you have to buy the original console but no one wants to do that so we're gonna see how close you can get with these two devices and also here we're doing testing with a new frame time overlay approach for console games who knows maybe we'll use the software in the future but this is important to go over so if you're not familiar with our previous frame time testing for consoles we've advanced it significantly since the last time we did this there's another issue that's especially relevant to the PlayStation Classic games with frame times that spike up and down at regular intervals can appear to have a constant FPS when averaged there are 50 Hertz games included on the 60 Hertz classic which must therefore output frames at different intervals to run at the correct speed for example running on the PlayStation classic which outputs frames every sixteen point six seven milliseconds 60 fps an NTSC game with constant frame times of 50 milliseconds or 20 FPS would display as a AAA BBB CCC and those are frames that are new versus repeat frames while its powell counterpart p al with constant frame times of 60 milliseconds sixteen point six seven FPS would be something more along the lines of a AAA BBB CC CC DD d ee ee ffff and so forth because 50 is a clean multiple of sixteen point seven and sixty is not however depending on the size of the sample window the game could appear to run at a constant and smooth frame rate on both platforms as opposed to frame times which are spiking all over the place NTSC and PAL are technically standards for color encodings but they've become synonymous with a refresh rates which are generally 60 Hertz for mts North American and 50 Hertz for PAL it's not much of an issue anymore but historically has caused problems with porting games since the easiest way to convert NTSC developed games for release in PAL regions was to just slow them down so in the past developers have just slowed them down to 5/6 the speed because it's 50 Hertz versus 60 Hertz and of course again there are other differences there too but much of the criticism around the PlayStation Classic has revolved around Sony's decision to include the slowed down and inferior PAL version of some games so that's one of the things we can be looking at here today with the developer mode enabled for some CPU load metrics as well here's a quick tutorial on the interface at the bottom you'll see a frame time plot that has a playhead centrally at 0 seconds with plus 1 and minus 1 second indicators on either side the line graph is a frame time metric presented in increments of sixteen point seven milliseconds as we're working with a 60 FPS capture and a console that outputs 60fps ideally the frame times are perfectly flat and consistent that would be a good experience and for perspective 16.7 will be 60 FPS 33 is 30 the reported values it must cleanly be divisible by 16 point 7 milliseconds as that is the maximum baseline capture speed the top left of the interface shows the frame time of the frame currently being displayed if you ever see this jump significantly it is likely that you will also notice a stutter in gameplay that's all we're doing for now we'll expand more on this later but let's start testing with GTA then Ridge Racer and then Syphon Filter with GTA K and play testing on the ps1 classic we see that the frame time plot bounces rapidly between 50 milliseconds and 60 point seven millisecond deliveries for perspective 60 FPS would be sixteen point six 67 milliseconds constantly with no deviation from that output in this output at sixty six point seven milliseconds we're getting a new in-game frame for every four frames of video playback meaning that we get three repeated frames for every new frame in a traditional GPU this would mean that the front buffer and back buffer haven't flipped yet and the frame is still being polled and represented because that no new frame is yet ready to be presented we define this as a stutter every now and then GTA on the ps1 classic and counters massive spikes 283 milliseconds it would be easy to put these in FPS because that number it looks atrocious but it would kind of be disingenuous because to distill this into fps means that it would average out between 67 milliseconds 50 millisecond frame times and then the the major 83 millisecond frame time as well and because frame at throughput isn't constantly 83 milliseconds 100% of the time it's not really proper to look at it as an FPS number this is a rare occurrence but one that's important what's really happening is one frame is taking 83 milliseconds to create as opposed to a desirable 33 or 50 millisecond output if you had a consistent 82 millisecond output for 107 frames it would be acceptable to call that 12 fps but this is one frame repeated a few times before a new frame is presented so it's not really appropriate to average this out into an FPS number the more accurate description here might be stutter or hitch if you prefer where the front buffer is spitting out a new frame after an 83 millisecond pause these 83 millisecond spikes seem to spaced exactly one second apart from one another when they happen if you look at our frame time plot we don't know 100% for sure why they happen but we have a few ideas let's start by looking at CPU utilization to try and figure it out in the bottom right you'll see a number that's bouncing around between something like 12 and 20 that's a percentage and it's the CPU percent utilization it does not appear that we're running into this any kind of strain on the CPU even during these spikes we don't have a firm conclusion of why this happens again but our summation is that this could be a CPU timing loop or timing synchronization issue between the internal clock on the modern ps1 remake and the original ps1 where the modern equipment may be more accurate and a self-correcting another theory that we'll get into a moment is NTSC versus PAL and anyway we get one new frame three repeat frames two new frames and three repeat frames and then cycle the result is still a one to two ratio it just seems like there's a synchronization issue that resolves itself at the end of a one-second interval as illustrated in the pattern showed earlier where we dip to two steps down to sixteen point six seven milliseconds for good measure we also rewrote our software to divide against a 20 millisecond interval instead of sixteen point six and better matching a pal output and that resulted in similar behavior being shown just at slightly different intervals given that the behavior repeated in both instances and having confirmed through Richard Ledbetter's articles that the device outputs at 60 Hertz we believe that this is a synchronization issue between a 50 Hertz game and a locked 60 Hertz device we were able to validate this issue of stutter output visually by stepping through the video frame by frame this is an objectively poor experience much of it is to do with a 50 Hertz game on a 60 Hertz console the ps3 emulating the ps1 is on the screen now you'll instantly notice that the line is now perfectly flat which is how a console should always look that's the whole point of a console the point of a console is that it's a single set of hardware that allows for a consistent experience and hopefully consistent frame rate because the game developers that know what they have to work with on the ps3 the experience is superior than the ps1 classic it's still 50 milliseconds so that's not great but we can actually average this one out to FPS as it's a flat line the entire benchmark the average FPS for GTA on the ps3 is about 20 which just isn't a great experience on either device the ps3 doesn't run into the same issue of frame time of spikes because it's a native NTSC version of GTA remember the ps1 classic is preloaded so you can't choose the native NTSC version of the game you can force it into NTSC mode via a dev console which we did and that helps a bit but you'll still get more frame time spikes than on the ps3 the ps1 classic also seems to undergo corrective spikes downward following the spikes upward as of trying to maintain a target frame time and again we have footage of NTSC forced on and you see some similar behavior so doesn't really solve the issue because it's not the native version of the GTA NTSC game next is Ridge Racer which runs NTSC on both devices we tested both demo mode and an actual race with demo mode showing the ps3 at 33 milliseconds flat for 100% of the race the end result is easily calculable into FPS as it's fully consistent at 33 milliseconds in terms of frame rate we're looking at 30 FPS average for Ridge Racer throughout the demo there was 1/2 interval spiked up to 66.7 milliseconds on the ps3 in the middle of our two minute captured there was another single spike 250 milliseconds at around 1 minute 45 an hour capture but neither of these is particularly bad because it's not a consistent issue on the ps1 classic we observed spiky behavior at 33 point 3 millisecond baseline deviating from the baseline to hit 50 milliseconds then 16 point 7 milliseconds correctively after that at 32 point three three milliseconds the frame output behavior is summarized as hit miss hit miss hit missed four frames where a hit is a new frame and a miss is a replayed frame that's normal behavior that's what you would expect for 30 frames per second given that there are 60 frames in the capture file when we encounter one of these spikes distanced exactly one second apart from one another the pattern changes to hit miss hit miss miss hit hit miss hit the net is zero and it seems like the device is struggling to maintain a target frame time and correcting upon those spikes like the ps3 there's still the occasional big spike like at around 33 seconds in our test which we can show where we hit 67 milliseconds briefly then start a holding pattern at 50 milliseconds for an entire second then again for another entire second this is not good it's a big change in framerate fluidity it's for a long period of time and it impacts gameplay negatively we can play these side by side for a nearly perfect comparison versus the ps3 just to help make those differences more tangible there might be some model changes because it randomizes the cars but this happens consistently and across multiple passes the ps3 provides a better experience once again while the ps1 classic runs choppy and frame time consistency finally looking at CPU utilization the PSone classics elf reports a utilization of about 30 to 34 percent average with occasional spikes to 34 percent utilization we're far away from using all of the CPU for the record the non demo mode gameplay also exhibits the same behavior on both devices we can't synchronize these as easily naturally but the ps3 mains consistent and its frame times while the ps1 classic a bounces around regularly Syphon Filter was a game similar in some ways to metal gear but critically had integration of things like fog graphics and fire graphics there was a time when these were special the ps1 original console and all of its 33 megahertz glory had a dedicated solution for fog processing and that was a big deal Syphon Filter was - fog rendering what battlefield is - realtime ray tracing it's sort of leveraged as a technical demonstration of what the device is capable of putting the fire demo side-by-side with the ps3 we see that the ps1 classic again encounters positive and negative spikes ranging between sixteen point seven and sixty seven milliseconds there are occasional spikes - 67 milliseconds that aren't zero some like around the 23 second mark in the demo and these don't self correct they appear to just be straight dropped frames this is also repeatable behavior between multiple runs the ps3 handles this demo at a fixed 50 millisecond frame the frame interval which is what we would expect neither is a great experience by standards of what we would really want on modern hardware but the ps3 is doing a bit better CPU utilization on the ps1 classic demonstrated in this clip runs closer to 47 to 50 percent load that's high when considering it's a 24 year old game running on modern hardware but the emulation is likely the most abusive part of the games execution and that's why we see that high load also it's still not the best CPU in the world gameplay exhibits the same behavior on both devices for Syphon Filter on the ps1 we again see regular spikes and corrective dips and on the ps3 we see a flat line at 50 milliseconds and that's for real gameplay not just the demo so it's validated as for the PlayStation 1 classic versus the ps3 the ps3 is cheaper and it provides technically a better experience in just about every benchmark we did worst-case is it's about the same so if you just want to play ps1 games and you want to do it on a Playstation console buying an older console used would probably make more sense just make sure it's backwards compatible because some of the newer ones aren't as far as other options of course there's the digital marketplace that makes things easier but otherwise the main reason to get something like this is if you really want collectible value I suppose would be why you'd get it it's small it looks like the original ps1 and if you just like to have everything Sony or everything PlayStation or something along those lines then this is a bit smaller and a bit more appealing than a gross ps3 that you buy used for $15 on eBay so there is a there is a peel there but if you just want to play the games it's it's not objectively the best it they do some really strange things like the decision to run 50 Hertz games for example on a ps1 classic device that is outputting at 60 fps so very odd decisions by Sony here as for the rest one more note here on FPS FPS is an averaged number of frames rendered per second over an arbitrary window of time which becomes a problem when it's used for live updated counters in the corner of the screen or FPS overtime charts when an FPS overlay says 60 what does that mean when FPS overtime chart spikes down to 10 FPS for a single reported frame what does that mean the FPS at a precise moment in time is most likely to be calculated by counting the number of frames delivered in the previous second but what if no frames have been delivered in the past second the larger the window of time sampled the flatter the graph and vice versa calculate an average FPS over any length of time by definition smooths out the frame time spikes the best and most consistent way to use frame rate data is as an average over the course of a longer test but even that has limitations we do this for CPU and GPU reviews and again you can start doing things like percentile math you can take one percent point one percent that helps significantly and makes the data pretty usable but even still there are flaws as we've shown with our far cry 5 testing with the 9600 K for example so FPS is a an excellent and extremely useful metric it's important a lot of people understand it and it's easy to work with but also for these console tests we're probably going to stick to frame times because we're typically comparing two devices the biggest problem with frame times is that you can't really compress it into a big bar chart with 30 devices like you can with rate so that's a major benefit frame rate but for two device comparisons we're probably going to push pretty hard for frame times in our console testing anyway that's it for this one so let us know what you think about this thing like if you think it's actually worth it for example post a comment below about that we're curious to see what you think otherwise subscribe for more as always go to patreon.com/scishow gamers Nexus stops out directly or you can go to store die cameras next Annette let's pick up a shirt like this one or any of our other products thank you for watching I'll see you all next time you you