'Powerful' hardware vs 'efficient' hardware

So, I brought this up over on the Destructoid forums a little while ago, and got some interesting responses, so figured I'd bring it up here as well and see what the Escapist community thinks.

See, when it comes to gaming hardware, there seems to only be one metric which gamers use to decide whether a piece of hardware is great or shit: Power! Whether it's Bits, or GFLOPS or clockspeeds, we gamers seem to let hardware live or die purely based on how powerful it is.

Now, my contention is that this is a rather bizarre mindset to be in. If you look at any other field of technology analysis, nothing is ever rated purely on how powerful it is, but on a whole category of requirements that are all measured up. What a piece of technology is worth is how well all those factors come together.

The example I used was cars- right now, gamers seem to be stuck in the Jeremy Clarkson mindset of:

image
image

..wherein the perfect car would look probably something like this:

image

However, Jeremy Clarkson knowingly portrays himself on Top Gear as a bit of an idiot, and would readily admit that you can never judge a car purely on how much horsepower it has. Yes, a car might have 800bhp, but if it handles like a turd, rides like a trampoline made of grit, and drinks up fuel like an alcoholic at Oktoberfest, then it's not really all that good of a car. Great automobile engineering is about bringing together a whole host of different things like reliability, efficiency, handling, weight, and so on, and making a car work well within the design limits you set it.

To illustrate this, the following is one of the most acclaimed cars of the past twenty years. There are few cars that can keep up with it on the race track, and it has won plaudits and awards all around the world.

It also only has 134hp.

By any modern metric, the actual power of the engine in the Lotus Elise is quite modest. Even the 'supercharged' Exige model only offers 192hp, which isn't bad, but certainly nothing compared to a Lancer Evo or a Nissan Skyline, let alone something like a Pagani Zonda or a Mclaren Mercedes SLR. But if the engine is so weak, why is the Elise regarded as so good? It doesn't have any POWAH! after all.

Because while the engine is fairly lower powered, the car itself weighs next to nothing, and has been designed entirely around offering unparalleled handling. Top Gear themselves called the Elise the best handling car in the world, and its acceleration is also phenomenally fast (0-60 in 5.8 seconds). If anyone wants proof of what a phenomenal piece of engineering the Elise is, especially the Exige model:

Ok, so enough wittering on about cars j-e-f-f-e-r-s, what does this have to do with games?

Well, it seems to me that more and more of us are falling into the Jeremy Clarkson mindset of "More POWAH!" except without the self referential irony. We have an increasing tendency to praise hardware that's seen as powerful, even if it's actually quite badly put together, while sneering at hardware we deem less powerful, even if its put together and runs with incredible efficiency. I'll use a couple of examples, first of which is:

image

The Gamecube.

When it came out, the Gamecube's on-paper specs were pretty modest. They were better than the PS2's, sure, but they were also absolutely dwarfed by the Xbox's. The Xbox not only had a higher clocked CPU (733MHZ), it had an internal hard-drive... in fact, you can read an in-depth specs analysis sheet here. Barring a few exceptions, the on paper specs of the Xbox blow the doors off the Gamecube in pretty much every category. More RAM, more memory bandwidth, the whole shebang.

So in theory, that means the Xbox should have had noticeably better looking games than the Gamecube, right? Actually, no. That wasn't the case. Despite its lower specs on paper, the Gamecube was also one of the most efficiently engineered consoles the industry has probably ever seen. Not only were games such as Star Fox Adventure notable for having advanced graphical effects despite being launched early in the console's life, but Rogue Squadron 3 has the highest polygon count of any sixth-generation console game, while also running at 60fps. Wind Waker was one of the most notable games for its advanced graphics. Despite being a cel-shaded game, and therefore ostensibly being simpler, Wind Waker had a whole host of ridiculously advanced programming going on under the hood: early tesselation, running two lighting engines simultaneously, and other neat graphical tricks. Despite having specs that should have resulted in lesser looking games, the Gamecube was able to match the Xbox with games like Metroid Prime and RE4, and even outperform it with RS3.

Even better, the Gamecube was made with a prime lode of Nintendium. Not only was it a well engineered piece of kit, it was also the smallest of the sixth-gen consoles, and nigh on indestructable. In fairness, the old Xbox was also a durable old machine, but in this video here, the Gamecube is the only console to survive having weights dropped on it, being hit with a sledgehammer, and dropped from a 1 storey height. Not only did Nintendo engineer the Gamecube to work with an incredible degree of efficiency, they also made sure it could take one hell of a beating.

Conversely, let's look at a system that was praised for its specs at the time, the Xbox 360. Despite being somewhat short on RAM when it came out, tech-heads everywhere praised it for the fact that it had a hefty tri-core CPU, an advanced GPU, and its own built-in tessellation unit. Purely based on metrics of POWAH! the 360 was a superb piece of kit for the time. What happened next? Well...

image

As it turns out, while the 360 may have been a 'powerful' piece of hardware, the first wave of models were also terribly designed and badly put together. To this day, there still hasn't been a conclusive reason discovered for why so many consoles fried themselves. Some believe it was the poor quality soldering used to connect the components, others believe it was Microsoft cutting corners on the GPU which led to excessive heat production. However you look at it, though, the 360 was a very shabbily put together piece of hardware. High-end PC components were chucked onto a motherboard without consideration of how much heat they'd produce. Extra large fans were then bodged together to deal with that heat, without consideration of how much noise they'd produce. Even if your 360 is working to this day, there's no denying that it's a loud machine which sits in the corner making Concorde impressions every time it fires up.

Even worse, that tesselation unit that's sitting in the 360? For the most part, it's redundant. Some games like Halo: Reach managed to make use of the tessellation unit to render prettier looking oceans, but for the vast majority of games, the tessellation unit has been sat there collecting dust on the motherboard. The consensus apparently being that while in theory it's nice to have such a unit in the console, in practise Microsoft made it too difficult to use properly to make it worthwhile.

I believe these two consoles make a very direct contrast to each other, and to me, they highlight the disconnect that is becoming more apparent in this industry. Praising something purely for the sake of POWAH! is pointless if efficiency and reliability are thrown out of the window. Conversely, hardware that is put together with the focus on utmost efficiency can be surprisingly effective at outputting great looking graphics, while also offering the possibility for greater reliability, smaller power consumption, and better affordability.

I know there are quite a few PC gamers on here who like to bemoan the current lack of POWAH! in today's gaming machines, but I also believe there are quite a few like myself, who would like to see the emphasis in the industry change from GFLOPS and clock speeds to overall efficiency and cohesive engineering.

Your thoughts? Anyone believe that the industry should focus on powerful hardware specs over other facets of hardware engineering?

It's not the size, it's how you use it.

To go out on a limb, I reckon I speak for most of us here when I say that; I'd go further to suggest that technology is actually probably one of the least limiting aspects in game development at the moment.

Of course the availability of that technology is something else in itself, and so is the choice of technologies used to develop. And programming skill. And, oh bollocks there goes my point.

Pink Gregory:
It's not the size, it's how you use it.

To go out on a limb, I reckon I speak for most of us here when I say that; I'd go further to suggest that technology is actually probably one of the least limiting aspects in game development at the moment.

I agree, and there are a lot of mistruths that are currently spouted about gaming hardware. According to some PC enthusiasts I've read on these very forums, the PS3 and 360 are incapable of running games above 30fps. Which is actually bollocks. Ninja Gaiden II, Bayonetta and Metal Gear Rising are all games that have great graphics, while also runing at 60fps. COD has been running at 60fps for years and years. Hell, my old Xbox has games that run at 60fps.

The problem right now is developers not working within the boundaries they're given. They spend millions and millions and millions pushing graphics technology as hard as they can, then have to spend more money winding it back when they find out they've gone beyond the boundaries of what the hardware sets. The reason a studio like Platinum is able to make such silky games is because they know the boundaries of the hardware, and they know how best to utilise the hardware up to those boundaries.

There were games of enormous size and scope back in the sixth-gen days. One of my favourite games, Otogi, has a final boss fight where you fight against a flying sorcerer on a series of floating gardens that hover in space over a giant purple supernova. If that's the sort of imagination developers managed to show off on a system with 64mb of RAM, then something like the 360 should be a playground of the imagination.

Instead, developers take up too much space with high-res stubble textures and bump-mapped manly muscles, then complain that the techology is stifling there imaginations. It's bollocks.

This, in a nutshell, is exactly what I find so irritating about games nowadays. The graphics and POWAH push stems directly from the collective gaming industry hard-on for "realism," and they're only shooting themselves in the ass. I was recently able to finish Windwaker for the first time, courtesy of Dolphin, and IMO it's tied with Journey for most beautiful game in the industry. Then I popped in Twilight Princess, and shut it off within the hour for how awful it looked.

Granted, there are games like Halo 4 that have managed to get close to that level of beauty, but that's only because they've had eight sodding years to pull it off. It really bugs me that devs don't seem to design for what they're given anymore, or that they take a relative eon to do it.

Vuliev:
This, in a nutshell, is exactly what I find so irritating about games nowadays. The graphics and POWAH push stems directly from the collective gaming industry hard-on for "realism," and they're only shooting themselves in the ass. I was recently able to finish Windwaker for the first time, courtesy of Dolphin, and IMO it's tied with Journey for most beautiful game in the industry. Then I popped in Twilight Princess, and shut it off within the hour for how awful it looked.

Nintendo have always been really good at getting amazing visuals from hardware that theoretically should be sub-par. Like I said in the OP, there's a ridiculous amount of advanced programming they did for Wind Waker. You can read an advanced analysis here, but a few of the more notable features included real-time cloth physics, having characters/objects and buildings lighted by two completely different lighting engines (buildings have a softer, more realistic lighting engine compared to characters using the iconic cel-shade lighting engine), and the fact that Link would dynamically move his feet to suit whatever terrain he was in.

Granted, there are games like Halo 4 that have managed to get close to that level of beauty, but that's only because they've had eight sodding years to pull it off. It really bugs me that devs don't seem to design for what they're given anymore, or that they take a relative eon to do it.

I thought Halo 4 was overrated. There were some nice graphical effects, but the game was far too dependent on the orange-and-teal colour scheme that's currently ruining creativity in visual media. That, and I thought a lot of its visual elements had already been done better in Metroid Prime (Promethean Knights/Space Pirates, Forerunner Vision/Visor Modes, etc).

j-e-f-f-e-r-s:
I thought Halo 4 was overrated. There were some nice graphical effects, but the game was far too dependent on the orange-and-teal colour scheme that's currently ruining creativity in visual media. That, and I thought a lot of its visual elements had already been done better in Metroid Prime (Promethean Knights/Space Pirates, Forerunner Vision/Visor Modes, etc).

Having also played Prime recently, I have to say that I like the design of the Prometheans better than I do the Space Pirates. The Knights are actually intimidating, both in presence and in combat ability, and their combat cues are much more subtle than the Pirates. When I fight Knights, especially when they have support from Watchers/Crawlers, I have to play carefully and neutralize targets efficiently. With Pirates, I feel like I'm just running down a checklist:

1) Normal? Missile x2, dead.
2) Camo? Thermal Visor, Wave Charge x2, dead.
3) Beam-specific? Switch beam, Charge Beam until dead, hop around as necessary. Power Missile against Power, maybe Ice Spreader against Ice.

Granted, some of that is simply because of Samus's intrinsic power through her suit, but the Chozo Ghosts still manage to intimidate and impress despite being moderately easy to defeat, and Space Pirates in the 2D Metroid games also have those same qualities.

Visually, I think the angular features of the Knights, alongside their enormous stature (even relative to the Chief), are what give them their intimidating presence. The Pirates just...don't have that. To me, Pirates are much more like the Covenant Elites--fast, agile, reasonable threat level, all relayed through their equal-ish size and curvier/organic shapes--but still just not as commanding as a Knight. And Knights have speed, agility, and mobility that greatly belie their size. Pirates, not so much: normal ones are fast with faster/lighter attack, big ones are slow with slow/powerful attacks. Almost as straightforward as you can get, really.

As for Promethean Vision/Visors, eh, I don't really get the beef. The Visors are needed for more than just seeing enemies around corners, and Promethean Vision isn't too far removed from other vision enhancers in other modern FPSs. Now, if it's that Promethean Vision is so horribly overshadowed by Autosentry/Jetpack/Dodgepack and is therefore a "poorer implementation of the Visors," I could see that, but it's still not much of an argument--just means that 343i didn't have the Vision as a central mechanic, and therefore didn't spend the time on it that Retro did with the game-critical Visors.

I'll grant you the orange/teal thing, but only barely. Halo had been using that long before others jumped on, and it's one of the few game franchises that still manages to do it well in spite of that market saturation. The Forerunner installations you visit in the other Halo games already had that visual style in place to distinguish Forerunner from Covenant, and Halo 4 really only sharpened that distinction.

It's more about optimization than power, I think. The biggest difference between this console generation and the Gamecube/PS2/Xbox generation, in my opinion, is that a lot of the best games, as far as looks and performance go, were console exclusives. Okami, Final Fantasy XII, Zelda: The Wind Waker, Metroid Prime, Halo 2, Tales of Symphonia, Tales of the Abyss, Kingdom Hearts, Fable, Jet Set Radio Future, Metal Gear Solid 3, Shadow of the Colossus, Ninja Gaiden, Otogi, these were all games that really showed off what their respective consoles could do and were (at least at the time) exclusives. Although in the case of Shadow of the Colossus it also showed off what the PS2 couldn't do, but that's beside the point.

I believe the people who made those games generally had a much greater idea of what they were working with, because they weren't trying to program for three or four different systems with completely separate development code simultaneously.

And it happens a fair bit on PCs today, too. The consoles use DirectX 9 and OpenGL for their APIs, so a lot of developers probably don't know jack shit about DirectX 11 and implement it horribly on the PC, leading to situations like Dragon Age 2 or Batman: Arkham City where even some of the more powerful single DX 11 cards struggle to keep a steady 60 FPS with everything cranked up.

shrekfan246:
It's more about optimization than power, I think. The biggest difference between this console generation and the Gamecube/PS2/Xbox generation, in my opinion, is that a lot of the best games, as far as looks and performance go, were console exclusives. Okami, Final Fantasy XII, Zelda: The Wind Waker, Metroid Prime, Halo 2, Tales of Symphonia, Tales of the Abyss, Kingdom Hearts, Fable, Jet Set Radio Future, Metal Gear Solid 3, Shadow of the Colossus, Ninja Gaiden, Otogi, these were all games that really showed off what their respective consoles could do and were (at least at the time) exclusives. Although in the case of Shadow of the Colossus it also showed off what the PS2 couldn't do, but that's beside the point.

I believe the people who made those games generally had a much greater idea of what they were working with, because they weren't trying to program for three or four different systems with completely separate development code simultaneously.

To an extent, but there are exceptions. Splinter Cell: Chaos Theory on Xbox still looks exceptional even today, and it was a multi-platform title. I'm still amazed at just how good that game looks, and it's over ten years old and doesn't even have a trippy artstyle like Okami or Wind Waker. I've seen very few modern military games which look as good as that opening lighthouse level. Mmm... that lighthouse level...

Similarly, Burnout 3 and Burnout Revenge also looked absolutely unbelievable on PS2 and Xbox, despite being multiplats. Though you could chalk that up simply to Criterion being absolute monsters at getting good visuals... Actually, speaking of Criterion, Black was another phenomenal piece of game tech, and again, I think that was a multiplatform title as well.

Do PC games count? Because Doom 3 came out on both Xbox and PC, and looked phenomenal on both (though of course the PC version looked better).

What else was there... actually, I remember Metal Arms: Glitch In The System being a very impressive piece of visual programming, and I'm pretty sure that game came out on all three systems. It had a very cartoony look, but had incredibly detailed mechanical parts on the various robot characters, and had an insane amount of destructible scenery and particle phsyics for the time.

I'll agree that it's to do with how well developers are able to get inside the hardware, but that doesn't always mean exclusivity. Ubisoft had two SC games prior to Chaos Theory, which gave them plenty of time to get to grips with the Xbox hardware, even though they were making multiplatform games. I guess in common parlance, it just had to do with how not-shit the developers are at their jobs.

That said, I am always in awe whenever someone like Naughty Dog or Ninteno EAD shows off what they can do with their chosen piece of hardware. I just wish Naughty Dog would stop moving closer towards the brown-and-gritty formula of third-person shooters.

And it happens a fair bit on PCs today, too. The consoles use DirectX 9 and OpenGL for their APIs, so a lot of developers probably don't know jack shit about DirectX 11 and implement it horribly on the PC, leading to situations like Dragon Age 2 or Batman: Arkham City where even some of the more powerful single DX 11 cards struggle to keep a steady 60 FPS with everything cranked up.

Isn't part of the problem as well that developing for PC, you're working with a nigh-infinite set of possible variables, and trying to cover them all is a total bitch? I can only imagine trying to work with the dozens and dozens of different CPU, GPU and RAM configurations amongst the userbase is a total headache for any developer. You can't exactly fine tune a game to make use of certain hardware specifications if you don't know if half your potential audience will even have that particular piece of hardware.

EDIT

Also, the fact you got an Otogi reference in there warms the cockles of my heart.

Vuliev:

Having also played Prime recently, I have to say that I like the design of the Prometheans better than I do the Space Pirates. The Knights are actually intimidating, both in presence and in combat ability, and their combat cues are much more subtle than the Pirates.

I really, really didn't like the Knights. I found them, in my playthroughs, to be little more than bullet sponges. They didn't seem particularly smart, and when facing them in Spartan Ops, me and my co-op buddy just learned to wait while they teleport, wait for them to re-appear, then empty a clip into them. Wait for them to teleport, wait for them to re-appear, empty clip, rinse, repeat, rinse, repeat.

I was udnerwhelmed by the Prometheans in general, but the Knights were the worst for me. The Elites in previous games felt like genuinely intelligent threats, I never got that same feeling with the Knights.

As for Promethean Vision/Visors, eh, I don't really get the beef. The Visors are needed for more than just seeing enemies around corners, and Promethean Vision isn't too far removed from other vision enhancers in other modern FPSs. Now, if it's that Promethean Vision is so horribly overshadowed by Autosentry/Jetpack/Dodgepack and is therefore a "poorer implementation of the Visors," I could see that, but it's still not much of an argument--just means that 343i didn't have the Vision as a central mechanic, and therefore didn't spend the time on it that Retro did with the game-critical Visors.

To me, my point of contention is that Halo 4 seemed to take a lot of inspiration from the Prime games (I know they had a couple of designers from Prime working with them), without really doing anything with it. The visors in Prime worked so well because, at the time, they seemed like such a cool new idea, and they added not only to the combat, but to the exploration feel as well. The Splinter Cell games also did that. Halo 4 felt like it had Promethean Vision because it had worked well as a mechanic in other games, not because it added something necessary to the game itself. Rainbow Six, Metroid Prime and Splinter Cell all managed to utilise alternate vision modes while still feeling unique and fresh. Same for Arkham Asylum and Detective Vision. Halo 4 just felt like it was hopping on a bandwagon regarding that mechanic. In my opinion, of course.

I'll grant you the orange/teal thing, but only barely. Halo had been using that long before others jumped on, and it's one of the few game franchises that still manages to do it well in spite of that market saturation. The Forerunner installations you visit in the other Halo games already had that visual style in place to distinguish Forerunner from Covenant, and Halo 4 really only sharpened that distinction.

Actually, from what I recall, there wasn't originally much orange at all in the Forerunner architecture in the earlier games. While Covenant always had that purple-colour scheme, from what I remember, Forerunner architecture was nearly always blue, with the occasional bit of purple thrown in to tie them in thematically with the Covenant. To me, the biggest colours in Halo's pallet were always green, purple and blue, with the odd bit of brown and grey thrown in for the humans. Green for the outdoor levels, purple for the covenant ones, blue for the forerunner ones. Oh, and white for Assault On The Control Room. That's why I found 4 to be such a visual let down. The previous games managed to utilise somewhat unique and vibrant colour schemes, whereas 4 ended up relying on the oldest colour-trick in the book. The last level, if I recall, was literally just nothing but orange, with blue elements brashly pasted on top.

I kind of expect Halo games to keep pushing unique colour pallets, not falling back on colour schemes that are the very definition of generic. It was bad enough when Battlefield 3 jumped in on that scheme.

j-e-f-f-e-r-s:

To an extent, but there are exceptions.

Snip

Well, yeah, I'd be remiss to ignore the exceptions, that just wasn't really the point I was trying to make. I didn't include PC games, because while there are Xbox/PC games that look amazing (such as Doom 3, Star Wars: Republic Commando, Chronicles of Riddick: Escape From Butcher Bay) on both systems, the PC version is still generally a better representation than the Xbox version.

Sure, in the end it mostly comes down to how good the developers are at their job, but I feel like it's easier for them when they're not working with multiple systems that have absolutely no resemblance to their competitors.

Isn't part of the problem as well that developing for PC, you're working with a nigh-infinite set of possible variables, and trying to cover them all is a total bitch? I can only imagine trying to work with the dozens and dozens of different CPU, GPU and RAM configurations amongst the userbase is a total headache for any developer. You can't exactly fine tune a game to make use of certain hardware specifications if you don't know if half your potential audience will even have that particular piece of hardware.

Yes and no.

Boiled down to the simplest end, a PC will typically have a variation of parts from three different companies that actually make a difference in the end-game: Intel, AMD, and Nvidia. I believe there are other manufacturers but those are the most common ones, and generally what will provide the average processor and graphics card. RAM doesn't have a huge impact on things like hardware compatibility from what I know, outside of "needing to have enough", and most PCs now come with a minimum of 4 GB, with a very small cost to upgrade past that relatively speaking.

You can see the split between AMD and Nvidia when it comes to optimization though. Because Nvidia has the market cornered on more realistic physics in games right now, a lot of companies sign deals with them and optimize much more efficiently for Nvidia graphics cards than AMD cards. Certain games are better-optimized for AMD cards, especially ones published by Square Enix lately. But most of the drivers are optimized by the hardware manufacturers, so the developers don't need to really account for the fact that one person might still have an Nvidia GTX 260 while another person has a GTX 680. Also, a lot of newer graphics cards are heavily based off of the designs of older ones, anyway, so they share a lot of similarities which I imagine would factor into how easy it is to develop for them.

I'm not sure I speak for anyone other than myself when I say this, but I think the reason people go for powerful hardware is it's easier to communicate (looks better on paper) and not that many people have the knowledge to determine whether something is being efficient or not. But that said, I'd choose a platform based on the games it's going to support sooner than how efficient it is.

shrekfan246:

j-e-f-f-e-r-s:

To an extent, but there are exceptions.

Snip

Well, yeah, I'd be remiss to ignore the exceptions, that just wasn't really the point I was trying to make. I didn't include PC games, because while there are Xbox/PC games that look amazing (such as Doom 3, Star Wars: Republic Commando, Chronicles of Riddick: Escape From Butcher Bay) on both systems, the PC version is still generally a better representation than the Xbox version.

Sure, in the end it mostly comes down to how good the developers are at their job, but I feel like it's easier for them when they're not working with multiple systems that have absolutely no resemblance to their competitors.

I can't argue with that. I wonder how much worse this generation has been for that sort of thing. I remember at the start of this gen, developers were complaining that the PS3 architecture was so bizarre, they really had trouble coding games for it. Even now we're still seeing problems like the Skyrim debacle on PS3. I know Sony initially wanted to make the PS3 hard to develop for 'on purpose', but I wonder if at some point they caved and got some middleware out to developers to make porting across from the 360 easier?

Yes and no.

Boiled down to the simplest end, a PC will typically have a variation of parts from three different companies that actually make a difference in the end-game: Intel, AMD, and Nvidia. I believe there are other manufacturers but those are the most common ones, and generally what will provide the average processor and graphics card. RAM doesn't have a huge impact on things like hardware compatibility from what I know, outside of "needing to have enough", and most PCs now come with a minimum of 4 GB, with a very small cost to upgrade past that relatively speaking.

You can see the split between AMD and Nvidia when it comes to optimization though. Because Nvidia has the market cornered on more realistic physics in games right now, a lot of companies sign deals with them and optimize much more efficiently for Nvidia graphics cards than AMD cards. Certain games are better-optimized for AMD cards, especially ones published by Square Enix lately. But most of the drivers are optimized by the hardware manufacturers, so the developers don't need to really account for the fact that one person might still have an Nvidia GTX 260 while another person has a GTX 680. Also, a lot of newer graphics cards are heavily based off of the designs of older ones, anyway, so they share a lot of similarities which I imagine would factor into how easy it is to develop for them.

I know that the PC gaming hardware market is currently cornered by Intel, AMD and Nvidia, so no surprises there.

With all that said, doesn't the need to cater to minimum specs somewhat hamper what PC developers can do comparitively? For example, anyone developing for the 360 will know that the system has something like 480mb of RAM (once you remove the OS footprint), and that'll be standard across all units, so they can develop the game to use every last bit of RAM as efficiently as possible. Same for the PS3, though they'd have to deal with that 256/256mb split. With PCs, their customers could have 2GB of RAM, or they could have 16GB of RAM, or anything in between. Even worse, depending on what version of Windows they're running, the amount of RAM already taken up by OS space could vary to, well... anything, surely? How easy is it to code to the metal when you don't even know how much metal your average consumer is going to have? Do you just create a cutoff point and say 'anything below this point will not be catered towards'?

MeChaNiZ3D:
I'm not sure I speak for anyone other than myself when I say this, but I think the reason people go for powerful hardware is it's easier to communicate (looks better on paper) and not that many people have the knowledge to determine whether something is being efficient or not. But that said, I'd choose a platform based on the games it's going to support sooner than how efficient it is.

Well, naturally that's always the best idea. A great piece of hardware with no games is a pretty piss-poor gaming console, no two ways around that.

My issue is that if we're going to have this side of gaming that gets frothy about tech-born and hardware, it would be nice if we could balance things out and actually analyse all the functions of a bit of hardware, not just however many millions of polygons it manages to output.

For example: the Wii U. Loads of people have written the console off because they believe it's nothing more than current gen tech, and its 7 years too late to the party.

That's actually complete rubbish, for a number of reasons. First and foremost being that the Wii U is more powerful than current gen consoles by quite a margin. If you're interested, there's a very interesting thread over on NeoGAF where they got some up-close photos taken of the Wii U GPU. The photos show that Nintendo didn't get an off-the-shelf card, but ordered something completely unique and customised. There's about 30% of the GPU die which people still haven't been able to work out. Common consensus, however, is that it's a highly unique, highly engineered piece of tech, much moreso than the bog standard off-the-shelf parts in the 360 and PS3. The dude who took the up-close pics said he was incredibly impressed by what Nintendo had come up with.

Secondly, and perhaps even more importantly, is that the Wii U manages to utilise this more powerful hardware whilst at the same time drawing less than half the electricity of current gen consoles, and producing far less heat. That in itself is something worth applauding from a tech point. The entire console draws around 32W when playing games, compared to the 76W of the 360. Even more importantly, unlike the 360 it manages to stay perfectly cool while doing so. That's pretty damn neat, because one of the biggest causes in hardware failure is overheating caused by drawing too much power and things running too hot. If the Wii U is able to keep running cool and draw smaller amounts of leccy, that means it's likely to be far more reliable in the long run. And it still manages to do all this while using hardware that's more powerful than current gen systems.

That is the sort of engineering we should be seeing more of, in my opinion, and it's the sort of engineering we should be celebrating. I'd certainly put that sort of hardware over the sort of hardware that's liable to cook itself, and needs its own personal reactor just to output intense graphics. But as it stands, that's currently the sort of engineering liable to earn contempt from a large number of gamers, as we've seen, and that makes me sad.

j-e-f-f-e-r-s:
Do you just create a cutoff point and say 'anything below this point will not be catered towards'?

Well, yeah, that's pretty much exactly what they do. That's what minimum system requirements are for. Anything below them, and sure it's possible that you'd be able to run the game, but you wouldn't have the best possible experience. They shoot for an average level of hardware, because despite their differences both AMD and Nvidia cards have extremely similar performance levels, and then tell you you're SOL if you've still not upgraded.

Granted, I think that part of the reason PC game requirements steadily climb even while they're being made cross-console is partly because a lot of developers simply don't know how to optimize properly for the PC, but you can pretty plainly see the difference between, say, the console version of Batman: Arkham City that runs at 30 FPS and the PC version running at a smooth 60 FPS. Or Sleeping Dogs -- While I don't know what framerate it runs at on consoles I imagine it's probably 30 like most other open-world games, and they don't have the high-resolution textures the PC gets; And the game is fantastically optimized for even mid or high-end graphics cards from 2009 (at least AMD, I don't know about Nvidia).

Have you guys seen John Carmack's tech demo of Rage running at 60 fps on an iphone?

It should be easy to find on youtube.

The reason it's relevant is because in this day and age, even mediocre hardware has phenomenal power when you think about how many floating point calculations they can do per second... but the weakest link these days is absolutely the programming. In the 90s the hardware clearly could not keep up, now the reverse is true.

j-e-f-f-e-r-s:
I know there are quite a few PC gamers on here who like to bemoan the current lack of POWAH! in today's gaming machines, but I also believe there are quite a few like myself, who would like to see the emphasis in the industry change from GFLOPS and clock speeds to overall efficiency and cohesive engineering.

So... you'd like to see what Intel already did with it's Ivy Bridge CPUs? A modest increase in processing power over Sandy Bridge while reducing energy consumption by almost 20%. They could have rolled out the SB family of CPUs at all 4Ghz+ but instead shipped them as 3Ghz+ and shaved decent amount of the power draw instead... and if you don't give a toss about power efficiency, get a K variant and OC the balls off it. The advantages of a 10nm die-shrink.

I'm not sure why you reference the gamecube over the PS2 given PS2's far more modest power specs however putting out very pleasing games visually RE4 may not have been as good looking but it did look damn close.

Well I always like to say people are wrong when they just say PCs are more powerful therefore they're better. The reason PCs have more going for them is you can do more with them, I make a lot of digital art so having so I want a powerful PC and going the extra few steps for a gaming PC just works out just makes sense. Someone who just wants to game may be happy with a console but with what consoles have been doing lately these days I just don't have the interest beyond what I have.

j-e-f-f-e-r-s:
I know there are quite a few PC gamers on here who like to bemoan the current lack of POWAH! in today's gaming machines, but I also believe there are quite a few like myself, who would like to see the emphasis in the industry change from GFLOPS and clock speeds to overall efficiency and cohesive engineering.

We bemoan the lack of power as they aren't efficiently designed. If they were efficiently designed and could pump out 2 times as much for the same amount of power as a PC could, we'd be happy. Sadly, they're not, so we'll bemoan their lack of power 'cause whilst it may be Ok for now, its what we'll be stuck with for the next 8 years or so going by this generation, and really no matter the power or efficiency that's going to leave it way behind in another 2 years, let alone 8...

j-e-f-f-e-r-s:
Conversely, hardware that is put together with the focus on utmost efficiency can be surprisingly effective at outputting great looking graphics, while also offering the possibility for greater reliability, smaller power consumption, and better affordability.

I know there are quite a few PC gamers on here who like to bemoan the current lack of POWAH! in today's gaming machines, but I also believe there are quite a few like myself, who would like to see the emphasis in the industry change from GFLOPS and clock speeds to overall efficiency and cohesive engineering.

Your thoughts? Anyone believe that the industry should focus on powerful hardware specs over other facets of hardware engineering?

My thoughts, you use the word efficient a lot and never define it, there is no intuitive foolproof way to define it in such way that covers PC and the tree main console makers. Till i know what you mean by it any answer would be me guessing your mind. The only thing clear is that by efficient you don't mean reliable, efficient in power consumption per hour played or high in performance/price ratio, since you mentioned them as separate entities.

I guess you mean something like "capable of delivering good game experiences with dated hardware", whatever good game experience might mean.

Also, I believe that a segment of the PC game industry should focus on powerful hardware development, yeah.

I think the "POWAH!" mostly comes from PC gamers like myself. Because for PC's, power is exactly what you need the most. Why? Because games are almost always first and foremost optimized for a console, and only then put in a configuration for PC's. So, wanting better graphics (sometimes far better) while still being able to play as smoothly as on a console, requires a lot more power.

Consoles on the other hand don't need as much power. Because they only have one specific hardware set up and games are mostly optimized for them first. But because PC gamers are often used to 'MORE POWAH!' they project that school of thought onto consoles as well when seeing their specs.

That being said, the current specs known for the next gen consoles is still pretty damn low. I can understand wanting to cut manufacturing costs, but this is a bit over the top imo. Yes the next gen will be quite a bit more powerful than the current one. But PC's will remain so much more powerful that efficiency won't matter much. Unlike previous generations, the PC will remain on top in terms of graphics when the PS4 and Xbox720 come out. Something that consoles used to be better at when a new generation came out.

its because efficiency is hard to measure sure efficient is good there is no denying that but its easy to compare than 8 cores running at 1.3 ghtz or what ever in the PS4 than 16 cores running at 3.5 ghtz in a PC hell its hard just comparing 8 GbB of GDDR5 RAM against 8 GbB of DDR3 RAM added to the 12 GbB crosfire of GDDR5 you get in a PC (side note those GPU's use less power than last generation). anyway power is easy to measure and compare so we can have our EPEEN length battles over hardware that is not out on the market yet or otherwise unavailable to benchmark besides any moron can make use of raw powwa and that seems to be what we have for game devs at the moment morons and clowns
and looking at that post even i am getting confused what chance does a console gamer have at understanding if you add arcane concepts like efficiency in

EDIT some hards should have been easy

I'm just wondering why consumers aren't taking a bigger interest in the lifespan of their technological devices, including the possibility of upgrading parts without much technical expertise. You can teach virtually anyone how to replace a graphics card or expand RAM in their PC. Consoles don't really have that option, so they have to be designed perfectly from the get-go. Upgradeable consoles should be the future. Maybe consoles will even become gaming modules in home computing systems that control lighting, tv, PC use, gaming, water and electrical management etc. If all components would be durable, and interchangeable it could mean a much more consumer friendly way of choosing hardware.

Assuming responses are intended to be from the stand-point of hardware only in which case the only goal of hardware is to be as powerful as possible. Optimization (efficiency) is done almost entirely on the software end of things. Id rather game developers have to work harder to use the hardware then the hardware not being capable of doing what its needed for.
Now things like a poorly designed motherboard that overheats are a quality issue not a power issue.. if youve built something that will destroy itself then you havent built anything that should be considered market ready, youve built a failed prototype.

xDarc:
Have you guys seen John Carmack's tech demo of Rage running at 60 fps on an iphone?

It should be easy to find on youtube.

The reason it's relevant is because in this day and age, even mediocre hardware has phenomenal power when you think about how many floating point calculations they can do per second... but the weakest link these days is absolutely the programming. In the 90s the hardware clearly could not keep up, now the reverse is true.

That Rage clip is bit of a cheat because an iPhone isn't that far behind a PS3 or 360, in fact the iPhone probably has more RAM. I always look for a balance of power and efficiency when I upgrade...in my last major upgrade I got two GPUs that individually outperform my old card yet together consume less energy and produce less heat. It is true that PC programming is rarely as efficient as for consoles , but there is oodles of power to spare.

j-e-f-f-e-r-s:

Pink Gregory:
It's not the size, it's how you use it.

To go out on a limb, I reckon I speak for most of us here when I say that; I'd go further to suggest that technology is actually probably one of the least limiting aspects in game development at the moment.

I agree, and there are a lot of mistruths that are currently spouted about gaming hardware. According to some PC enthusiasts I've read on these very forums, the PS3 and 360 are incapable of running games above 30fps. Which is actually bollocks. Ninja Gaiden II, Bayonetta and Metal Gear Rising are all games that have great graphics, while also runing at 60fps. COD has been running at 60fps for years and years. Hell, my old Xbox has games that run at 60fps.

The problem right now is developers not working within the boundaries they're given. They spend millions and millions and millions pushing graphics technology as hard as they can, then have to spend more money winding it back when they find out they've gone beyond the boundaries of what the hardware sets. The reason a studio like Platinum is able to make such silky games is because they know the boundaries of the hardware, and they know how best to utilise the hardware up to those boundaries.

There were games of enormous size and scope back in the sixth-gen days. One of my favourite games, Otogi, has a final boss fight where you fight against a flying sorcerer on a series of floating gardens that hover in space over a giant purple supernova. If that's the sort of imagination developers managed to show off on a system with 64mb of RAM, then something like the 360 should be a playground of the imagination.

Instead, developers take up too much space with high-res stubble textures and bump-mapped manly muscles, then complain that the techology is stifling there imaginations. It's bollocks.

I think you might be taking the PC user statement out of context. The current Generation consoles couldn't run any of the big named console shooters at 60 fps at the resolution the typical monitor has these days, which is at least 1320x796, without resorting to upscaling a lower resolution or cutting back significantly on effects and texture detail. In fact, that is the problem development for consoles currently has: After picking the most efficient algorithms and getting the programming out of the way, does one run it at a slightly higher resolution, or downscale the resolution and run it with more special effects?

I agree that efficient hardware is the clear winner. But pc gamers are not shouting MORE POWAH. Infact most PC gamers research thoroughly before putting together their rig.

What we do moan about is that, even if the current consoles were engineered in a super efficient manner(which you said yourself they aren't) They are still incapable of keeping up with most recent gaming PC's in terms of performance (both efficiency and power). We see most devs designing games for the consoles and porting them over. (Understandable, the 360 is apparently very easy to make games for.)

So we have games being designed for consoles that are using out of date hardware. The complaint is that there is a potential for better things and we aren't taking advantage of it. New consoles using more recent(not necessarily more powerful, though it will be more powerful) hardware would help negate this.

j-e-f-f-e-r-s:
Snip for a fantastic post.

First off, really, really good post. I very much agree that hardware should be efficient rather than powerful. Sure the power should be there, but the optimization should be there as well.

I'd just like to answer one question that you asked earlier.

j-e-f-f-e-r-s:
Isn't part of the problem as well that developing for PC, you're working with a nigh-infinite set of possible variables, and trying to cover them all is a total bitch? I can only imagine trying to work with the dozens and dozens of different CPU, GPU and RAM configurations amongst the userbase is a total headache for any developer. You can't exactly fine tune a game to make use of certain hardware specifications if you don't know if half your potential audience will even have that particular piece of hardware.

Like Shrekfan said, it's a yes and no. He already pointed out the hardware side of things, so I'll take the software side of why it's a yes and a no.

The reason that there really isn't any issue, is due to APIs. APIs are Application Program Interfaces, which basically serves as a translation layer for all of the calls to be made between the raw code and specialized hardware (by which I mean, specialized code on that hardware). One such example is DirectX, which is used in most Windows games out there today. That API is also used on the Xbox 360, which makes that one of the reasons that the 360's easier to develop for, alongside the computer architecture inside the console. Similar APIs translate over and make the porting process easier.

That isn't just a console to PC and reverse thing. This also applies to all of the GPUs that Nvidia, AMD and Intel push out. They don't push out updates for every single GPU out there, that's asinine to all hell. What they do is release updates for GPUs that are the closet in architecture, like the GTX 600 series, and optimize it (the update or patch) from there. Saves a lot of time and money in the long run. I'm pretty sure that you know why newer cards run faster after that, having faster clock speeds and more VRAM and all of those nice hardware things.

So there's an explanation of why it isn't that bad of a nightmare to optimize for PC. If you need more answers, just feel free to quote me or PM me. Just don't ask me hardware related things, it seems you're much better than me at explaining the hardware.

Snotnarok:
I'm not sure why you reference the gamecube over the PS2 given PS2's far more modest power specs however putting out very pleasing games visually RE4 may not have been as good looking but it did look damn close.

I referenced the Gamecube because, to my mind, it's one of the best examples of ingenius console engineering. Lemmer explain: The PS2 had some great looking games. That is an absolute fact. However, the very modest specs of the PS2 compared to its fellow sixth gen consoles meant that technically, games tended to look far worse on it. If you ever compare a PS2 game with an Xbox game, the graphics are far more... papery, I guess would be a good term. Things are more pixelated, textures are lower quality, there are more jaggies, less AA, things like that. Looking at the PS2's graphics, you could see it had less powerful technology.

Again, that's perfectly fine, and it still had wonderful looking games, but them's the specs.

Now, the Gamecube also had specs which, on paper, were lesser than the Xbox. Unlike the PS2, however, the Gamecube was still able to output visuals which matched the Xbox punch for punch. I already mentioned that RS3 had the highest poly count of any sixth gen game. In fact, Factor 5 did a whole host of ridiculously clever visual programming with the RS games, from high level draw distances to really sweet bump mapping. StarFox Adventure was able to carry off the same sort of impressive fur rendering damn well near the launch of the Gamecube that later games on the Xbox would also do.

Here's a comparison: The Conker remake on Xbox:

image

Now here's Star Fox Adventure for the Gamecube:

image

Both games were developed by Rare, which no doubt explains all that graphical goodness. Now, that fur rendering going on there was incredibly advanced for the time. And Rare were able to do that for a Gamecube game that came out only one year after launch. Despite the fact that the specs of the Gamecube were far inferior on paper, it was so well engineered that it could use the same graphical tricks that the Xbox could with all its superior hardware.

That is what you can do when you focus on clever engineering over pure power. The Gamecube could never stand up to the Xbox in a pure GFLOPS dick-swinging contest, but when it came to actual games, developers were able to create much of the same visual quality. That, to me, is what gets missed in all these pissing contests about which new console is using which GPU and which CPU. What something means on paper and what it can mean in practise are two very, very different things.

Tanakh:
My thoughts, you use the word efficient a lot and never define it, there is no intuitive foolproof way to define it in such way that covers PC and the tree main console makers. Till i know what you mean by it any answer would be me guessing your mind. The only thing clear is that by efficient you don't mean reliable, efficient in power consumption per hour played or high in performance/price ratio, since you mentioned them as separate entities.

When I'm talking about 'efficiency', I'm talking about all those things together. A piece of hardware that draws a small amount of electricity, has greater reliability, and is still able to output hi-res visuals is a far more efficient piece of kit than a something which can output more polygons at higher res, but draws a huge amount of electricity, generates a huge amount of heat, and runs a high risk of melting itself in order to do so.

The point I was trying to make is that gaming seems to be the only area where technology is seemingly judged on one criteria (POWAH!) rather than a bunch of different criteria. It seems to me that as a community, we would do far better to actually judge new bits of kit based on a range of factors (such as the previously mentioned reliability, power draw, etc), rather than simply writing it off based on whether its got the latest GPU/CPU combo or not.

You only have to look at how a lot of people are writing off the PS4 for being 'underpowered' regarding its CPU and GPU, despite the fact that we don't yet know how those two things are engineered into the console, how they interact with the RAM modules, what the latency is like between components, etc etc.

I guess you mean something like "capable of delivering good game experiences with dated hardware", whatever good game experience might mean.

That can be a part of it, but certainly isn't a necessity. A console could come out with cutting edge parts that all work together beautifully, in which case it would be up-to-date and efficient. Such a console is unlikely, however, given that it's the newest components which are the most expensive, and no sane company is going to spend a lot of money on high-end parts, only to spend even more optimising them to the Nth degree. Hence why Sony spent $400 million developing the CELL, but didn't spend any money actually making it easy for developers to work with.

Older hardware tends to work better for optimisation, as the fact that it can be bought for a lesser amount means companies can then spend money tweaking it and optimising it. Nintendo have done this time and again. Rather than spending huge amounts on bleeding edge hardware, they utilise older hardware which can then be subjected to a whole load of R&D to get it working just how they want. The Gamecube is one example of how this can work very well on a tech level. The DS is another. The Wii U, for all that people slag it off for having less power than the PS4/Nextbox, could well be another example. Again, I recommend you check out the Wii U GPU thread on NeoGAF. The unit is so highly customised, no-one's been able to fully make heads or tails of it.

This is the sort of thing you miss if you focus purely on "WAH! MASSIVE FLOATING OPERATIONS! TERAFLOPS! TERAFLOPS!"

thesilentman:

j-e-f-f-e-r-s:
Snip for a fantastic post.

First off, really, really good post. I very much agree that hardware should be efficient rather than powerful. Sure the power should be there, but the optimization should be there as well.

That's something which I think has been missing in recent consoles. Microsoft did a good job at making a console that's easier to develop for, but they did an absolutely terrible job at testing to make sure it worked reliably. Even if the RROD fiasco is over, every 360 I've ever played has sounded like an asthmatic turbo-fan whenever you turn it on.

Conversely, Sony did a better job at reliability with the PS3, but they did an absolutely terrible job at optimising it for development. From what I understand, they developed the CELL, then just chucked it straight into the PS3 without doing any sort of modification to make it easier to actually program games for. Likewise, they went for a split RAM design without thinking about how that could actually affect games that prioritise world data and such over graphics *cough* Skyrim *cough cough* New Vegas *cough*

I mean, I know consoles back in the old days had their own quirks and failings (the N64 and it's lack of decent texturing springs to mind), but it really seems like the ball got dropped to an extent this time round. Or at least, somehow this mindset got fostered whereby graphics became the be all and end all, to the expense of other facets of game design.

SNIP

So there's an explanation of why it isn't that bad of a nightmare to optimize for PC. If you need more answers, just feel free to quote me or PM me. Just don't ask me hardware related things, it seems you're much better than me at explaining the hardware.

That makes a lot more sense now. Cheers for that.

j-e-f-f-e-r-s:
-snip-

Tanakh is critted with a wall of text!!

Over 9000!!!!

.
.
.
.

I will read it at some point, but mate brevity is the soul of wit :P

A lot of insightful answers, and I actually agree with @xdarc for once, but I think a lot missed why we funnel all the tech talk into a few specific benchmarks; marketing. They want to put the highest capacity numbers and fastest clock speeds because the field of PC computing has been dumbed down to the point of these generic labels.

Like the comparison you made of Jeremy Clarkson, which his target isn't necessarily the most educated, but the more enthusiastic automobile fans; PC specs for graphics have been reducced to the speed of the card and the speed of the memory.

Tanakh:
I will read it at some point, but mate brevity is the soul of wit :P

Wit fails to get the point that he's getting across. It'd be quite boring to read it short with the writing not full of life, no? :-D

Well, first off, XBOX games did in fact look better than Gamecube games. Don't get me wrong, I absolutely loved the Gamecube (it was the last great Nintendo console imo), but the games weren't graphically as impressive as those on on the XBOX. Artistically, games like Metroid Prime and Wind Waker were unmatched.

That being said, I find nothing wrong with being excited about new hardware. Console players should be excited. People that have owned a good gaming PC over the course of the last few years know how much better games can and do look when paired with a powerful rig. I was blown away when I first played Crysis 3 on high settings at 1080p. It's absolutely gorgeous. Consoles have never been anything other than graphics rendering devices, so it's no surprise that the primary differences between console generations are defined by computing power--they're computers!!

j-e-f-f-e-r-s:
-snip-

KK, after reading, I don't see the dichotomy you formulate. You are comparing good eng design vs bad eng design, saying that it's better when the console actually works, on that we all agree; the logic connections with the rest of your arguments seem to me a bit iffy, esp since I believe the difference of design between a gamecube and a xbox360 had more to do with the very different market segments each one was designed for.

 

Reply to Thread

This thread is locked