A new job and a dead GPU: An excuse for a new gaming PC

Something quite notable in my life has happened that I forgot to mention in my last post. After seven years in third-level education and just as much time spent in my previous job as a shop assistant in a petrol station, I’ve finally got a job that is relevant to what I’m studying and am most proficient at. I’m now working in enterprise technical support for Dell, which is quite a change, but both makes use of my technical skills learned both at DIT and the almost twenty years that I’ve spent playing around with computers in my own time and the customer service skills that I learned in my last job. Notably, the new job comes with a considerable increase in my pay; while the two-and-a-half times increase per annum comes mostly because of the fact that I work five days a week now, I am still making more now than I would have working full time previously.

Coincidentally, very recently, I experienced some bizarre glitches on my primary desktop computer, where the X Window System server on Linux appeared to freeze every so often, necessitating a reboot. Resolving the cause of the problem took some time, from using SSH to look at the Xorg logs when the crash occurred to discovering that the issue later manifested itself occasionally as graphical glitches rather than a complete freeze of the X Window System, then later experiencing severe artifacting in games on both Linux and Windows. In the end, the diagnosis led to one conclusion – my five-year-old ATI Radeon HD 4890 graphics card was dead on its feet.

Fortunately, I had retained the NVIDIA GeForce 8800 GTS that the computer had originally been built with, so I was able to keep my primary desktop going for everyday tasks by swapping the old GPU in for the newer, dead one. However, considering the seven years that I’ve got out of this computer so far, I had already been considering building a new gaming desktop during the summer to upgrade from a dated dual-core AMD Athlon 64 X2 to something considerably more modern. The death of my GPU, while not ultimately a critical situation – after all, I did have a replacement, a further three computers that I could reasonably fall back on and five other computers besides – did give me the impetus to speed up the process, though.

After looking into the price of cases, I decided that I would reuse an old full-tower case that currently holds my secondary x86 desktop (with a single-core AMD Athlon 64 and a GeForce 6600 GT), adapting it for the task by cutting holes to accommodate some 120mm case fans and spray-painting it black to cover up the discoloured beige on the front panel. Ultimately, this step will likely cost me almost as much as buying a new full-tower case from Cooler Master, but will at least allow me to keep my current desktop in reserve without having to worry where to find the space to put it. A lot of the cost comes from purchasing the fans, adapters to put 2.5” and 3.5” drives in 5.25” bays and selecting a card reader to replace the floppy drive that will be incompatible with my new motherboard. Nevertheless, the case is huge, has plenty of space for placing new components and should be much better for cooling than my current midi-tower case, even considering the jerry-rigged nature of it.

I had considered quite some time ago that I would go for a reasonably fast, overclock-friendly Core i5 processor and have found that the Core i5-4690K represents the best value for money in that respect – the extra features of the Core i7 are unnecessary for what I’ll be doing with the computer. To get the most out of the processor, I considered the Intel Z97 platform to be a necessity and was originally considering the Asus Z97-P before I realised that it had no support for multi-GPU processing. To be fair, I haven’t actually used either SLI or CrossFireX at any point, but do like the ability to use them later if I wish, so eventually, I settled on the much more expensive but more appropriate Asus Z97-A, which has capacity for both SLI and CrossFireX, the one PS/2 port I need to accommodate my Unicomp Classic keyboard without having to use up a USB slot and which seems to have sufficient room for overclocking of the i5-4690K.

To facilitate overclocking, I have also chosen to purchase 16GB of Kingston 1866MHz DDR3 RAM and an aftermarket Cooler Master Hyper 212 Evo CPU cooler to replace the stock Intel cooler. I’m not looking for speed records here, but would like to have the capacity to moderately overclock the CPU to pull out the extra operations-per-second that might give me an edge in older, less GPU-intensive games. I’ve also gone for some Arctic Silver 5 cooling paste, since cooling has been a concern for me with previous builds and I’d like to make the most of the aftermarket cooler.

Obviously, being a gaming desktop, the GPU will be a big deal. I had originally looked at the AMD Radeon R9 280X as an option, but the retailer that I have purchased the majority of my parts from had run out of stock. As a consequence, I’ve gone a step further and bought a factory-overclocked Asus Radeon R9 290, hoping that the extra graphical oomph will be useful when it comes to playing games like Arma 3, where I experienced just about adequate performance with my HD 4890 at a diminished resolution. The Arma series has been key in making me upgrade my PCs before, so I’m not surprised that Arma 3 is just as hungry for GPU power as its predecessors.

I’ve also gone for a solid-state drive for the first time in order to speed up both my most resource-intensive games and the speed of Windows. I’ve purchased a Crucial MX100 128GB 2.5” SSD, which should be adequate for the most intensive games, while secondary storage will be accommodated by a 1TB Western Digital drive for NTFS and a 320GB Hitachi drive to accommodate everything to do with Linux. I also bought a separate 1TB Western Digital hard drive to replace the broken drive in my external hard drive enclosure, which experienced a head crash when I stupidly let it drop to the floor. Oops. Furthermore, I’ve also gone for a Blu-Ray writer for my optical drive – I’m not sure whether I’ll ever use the Blu-Ray writing capabilities, but for €15 more than the Blu-Ray reader, I decided to take the plunge. After all, I’m spending enough already.

Last but not least is the PSU. “Don’t skimp on the power supply”, I have told several of my friends through the years and this was no exception. Taking in mind the online tier lists for PSUs, I considered myself quite fortunate to find a Seasonic M12II 750W power supply available for under €100, with fully-modular design and enough capacity to easily keep going with the parts that I selected. The benefits for cable management from a modular power supply can’t be overstated, which will be useful even with the generous space in my case.

Overall, this bundle will cost me a whopping €1,500 – almost double what I spent on my current gaming desktop originally. Of course, any readers in the United States will scoff at this price, benefited by the likes of Newegg, but in Ireland, my choices are somewhat more limited, with Irish-based retailers being very expensive and continental European retailers not being as reliable when it comes to RMA procedures if something does go wrong. Nevertheless, I hope the new computer will be worth the money and provide the sort of performance gain that I haven’t had since I replaced my (again, seven-year-old) Pentium III system with the aforementioned single-core Athlon 64 system.

I’ll be looking forward to getting to grips once again with another PC build. Here’s hoping that the process will be a smooth one!

Why I hope that SteamOS will be successful

I’m a Linux user. Linux has, for several years, been my primary operating system on nearly every computer that I own – I have run openSUSE on my desktops since before it was called openSUSE, I run various versions of Ubuntu on my laptops, Raspbian on my two Raspberry Pis and I even have Debian derivatives running on my Wii and PlayStation 2.

I am also a PC gamer, something which really shouldn’t come as a surprise given my history of video game reviews. I have been playing PC games since the mid-1990s, starting with various MS-DOS games such as SimEarth and Indiana Jones and the Fate of Atlantis, proceeding to Windows 95 with Command & Conquer: Tiberian Dawn, Sid Meier’s Civilization II and SimCity 2000 and continuing to the present day with my most recent acquisitions including The Elder Scrolls V: Skyrim, the entire Tropico series and most recently, Arma 3.

Unfortunately for me, these two facts do not reconcile very well. While gaming of some variety has been possible on Linux since before I started, many of the games available have been open-source projects, ranging from casual puzzle or card games up to the likes of NetHack and Battle for Wesnoth. Most commercial video games on Linux have been from indie developers whose audiences are committed enough to their titles to deal with any hiccups they might experience when dealing with Linux, while a few older FPS titles come courtesy either of id Software’s policy of releasing their engines under open-source licences a few years after their release or by extensive reverse-engineering of the game engines to allow the games to run under Linux.

A lot of the games in question are very impressive in terms of gameplay and are to be lauded for that, but being a Linux user has often meant some sort of compromise in gaming terms. In order to experience the same games as the mainstream audience, one either has to run Windows as a secondary operating system, with the commensurate use of disc space on a separate partition, faff around with Windows emulation, which falls short on the most recent games and on several older titles or simply buy a console at significant expense. I have traditionally opted for the former but consider it to be somewhat irritating in the face of disadvantages of Windows that led me towards using Linux primarily in the first place.

Until recently, the highest-profile company whose games were available under Linux were id Software, largely because of John Carmack’s insistence on the open-source availability of their engines. That has changed of the last year or so, when Valve Software announced the release of SteamOS, a custom distribution of Linux designed for playing games. Valve Software have been one of the poster children for PC gaming for several years. After coming straight off the starting blocks with Half-Life in 1998, they have barely put a foot wrong since then. The highlight of this streak of strong titles has been the groundbreaking Half-Life 2 in 2004, a game which proved that Valve’s original title wasn’t just a flash in the pan.

What’s important to note here, though, is that Valve have also been a strong force for promoting independent game design. Steam, released in late 2003, has been the most notable example of a content delivery system done right. Among Steam’s features are the automated installation of patches, several community features allowing coordination of gameplay with friends and publication of screenshots and videos and a cloud storage system allowing save files and achievement progress to be distributed quickly to several different systems.

Valve are also known for their strong commercial and distribution advantages. The Steam store frequently has sales on various game titles, occasionally offering extensive discounts on games – my purchase of The Elder Scrolls V: Skyrim for €3.74 in June this year has been a particular highlight for me. They also promote independent game studios and offer a strong alternative to traditional publishers. Recent additions to Steam include the Steam Workshop and Greenlight, the former a way of quickly distributing user-created content, thus promoting one of the biggest advantages of PC gaming and the latter being a way for independent game developers to promote titles they may want to be made available on Steam. A number of “Early Access” titles in alpha or beta form are available through the Steam store as some forms of PC game development proceed towards a more audience-oriented method of bug-testing.

Valve have already done a great deal of work in promoting Linux as a gaming platform, having ported all of their own titles to Linux and selling or distributing hundreds more through the Steam store. Valve may well be the vanguard for making Linux gaming a viable alternative to Windows and offer the strong selection of their own titles along with the notable advantages of Steam as a game distribution platform. The proof in the pudding, however, will be when more mainstream developers see fit to release Linux titles as a consequence of Valve’s own orientation towards the platform.

As for the statement posited in the title, I prefer Linux as an operating system environment. I prefer the way that, even with a hefty desktop environment like KDE, my computer will feel quicker and less prone to hiccups in utility software when running Linux versus Windows. I prefer the flexibility to change parts of Linux as one sees fit, running different desktop environments or window managers as desired. I prefer the free and open-source nature of Linux and while Steam won’t offer much in the way of software that is “free as in freedom”, most of my utility software will remain free for me to modify as necessary – or even as desired. I’ve also grown used to the idiosyncracies of a Unix-like environment, from the file system to the command line – I can use Windows perfectly fine, along with a host of other graphical user interfaces, but my growing experience with Unix-like systems gives me a sense of familiarity that I find more pleasing.

PC gaming has, by and large, required me to use Windows. I find Windows works perfectly fine when I run games on it – they run smoothly right up to the point where the graphics card or processor cries uncle. I don’t find that sort of smoothness with utility software. Mozilla Firefox hiccups and splutters, frequently lacking response. Windows Explorer isn’t much better and in any case lacks some neat features from Konqueror on KDE or even Thunar for XFCE, including tabbed browsing. These sorts of hiccups may be down to the fact that my installation of Windows 7 really needs reinstallation, but there are too many idiosyncratic solutions I’ve had to make to get modded games running for me to do that. Then there’s the fact that I have to pretty much install a Unix-like environment through Cygwin if I want to have a programming environment like I’m used to. None of this software has regular updates through a package manager like I’ve become accustomed to on Linux either.

As a consequence, having to switch between the two operating systems between playing games and running utility and programming software is awkward and in any case, running Windows feels like a chore. I’ve said in the past elsewhere and I’ll say it again: Find me a way to get my game library running on Linux without much more effort than it takes to run the games on Windows and it will be difficult to find a reason for me to use anything else.

To close this article, I’ve recently reinstalled Steam on Linux with the aim of experimenting on how well games actually work. Installing Steam wasn’t too difficult – I just had to find a separate package for openSUSE as the package on the Steam website is designed for Ubuntu, Debian and other Debian derivatives with an APT package manager. I tested the original Half-Life, which ran pretty much perfectly – not a surprise, as I already knew that Quake III worked properly. I then installed Half-Life 2: Deathmatch, the smallest Source engine game in my collection.

After a bit of searching on Google to find the workaround to a problem involving a certain proprietary bit of software known as S3TC – one of those patent-related exploits which is unpatentable under the superior European patent system – I was able to get the game running. While I shouldn’t have been surprised, given the age of the Source engine, I had almost top settings straight off the bat with reasonably smooth performance using open-source ATI drivers. This was a pleasant surprise as I had expected stuttering, even given that my Radeon HD 4890 is easily capable of running Half-Life 2. Valve have clearly put effort into making sure that their Linux ports work, which is good to see. If Valve can succeed with Linux and convince other mainstream game companies to follow in their wake, we could see a viable alternative to Windows in yet another way.

On The Gaming PC Upgrade Cycle and The Future

It has been an often-perpetuated myth, with little backing in reality, that personal computer gaming is a ludicrously expensive business, with people having to spend thousands of dollars every six months to stay on top of the curve. This myth contains a lot of exaggeration; personal computer gaming is expensive, but upgrade cycles come every two to four years for most computer gamers, depending on their tolerance for lower resolutions, and the “sweet spot” for computer design right now is somewhere around the €750 mark, with a €500 budget yielding a still-reasonable machine.

Indeed, over the last three years or so, desktop computer design has reached a point where most computers on the market with any sort of discrete graphics card can play the majority of modern games acceptably. Games consoles are a major contributor to this situation, especially with Sony’s insistence that the PlayStation 3 will last for a decade (perhaps as a budget option beside the PlayStation 4, but they’re dreaming if they think that it’ll last as a vanguard in the console war).

Consoles are considerably less powerful than gaming computers. The graphics that they generate are fairly impressive, but they’re not generating 1080p resolutions on most games – the graphics are upscaled for higher-resolution displays. As consoles have become the predominant platform for graphics-intensive games, the graphical quality of multi-platform games seems to be limited by the lowest common denominator, the Xbox 360. Because it doesn’t seem to pay much to try to push a computer to its limits, the most graphically-intensive game in common knowledge, Crysis, dates back to 2007.

All of this has made me consider the gaming computer in this context. If consoles with hardware which would be considered as mid-range in the PC market when the consoles were released are going to limit graphical quality on personal computers, there’s little point in spending huge amounts of money on a ridiculously powerful monolith of a machine. An AMD (previously ATI) Radeon HD 5970, the most powerful graphics card available, really requires three monitors, preferably with 2560×1600 resolutions, to demonstrate its power properly. This isn’t performance to turn one’s nose up at, but not many people have €8,000 or so of money to spend on a computer and three monitors simply to get the best out of games which already look impressive at less-demanding resolutions. Even my own machine, with a Radeon HD 4890, is a bit over-kill for the native 1280×1024 resolution of my monitor.

That’s a pretty obvious conclusion, but there are still elements of the gaming PC which don’t make much sense in context. Whenever people ask for recommendation for the specifications that their own designs for gaming PCs should follow, there will usually be a big discussion on the power supply unit. There’s a good reason for this, as the PSU is one of the most likely components of a personal computer to fail, and I advise never going cheap on the component, but a lot of the suggestions on the internet recommend a 750 watt power supply.

I’ve started to wonder of late why a personal computer should need 750 watts of power to sustain its processes, enough to light a house full of non-CFL lightbulbs, and several houses with energy-saver bulbs. Even with the slight inherent inefficiency of power supplies, and assuming that the PSU will run with 80% efficiency at maximum load, that’s still 600 watts required by the internals of the computer.

Graphics cards are a major culprit in this scenario. As the power of a graphics card increases, the amount of power it requires will also increase, sometimes to slightly absurd levels. With the market for expensive, high-end GPUs closing up as specifications for games stay relatively level, perhaps it’s time for graphics card manufacturers and developers to start considering how to increase the power efficiency of their products.

AMD has taken a small step towards this goal with their Radeon HD 5xxx series of cards, with the 5770 producing about as much graphical potential as a previous-generation Radeon HD 4870 card, but with less demand for electrical power. Yet, a discrete graphics processing unit at idle speeds still sucks up a lot of power, sometimes in the region of 100 watts. My HD 4890 would go unused most of the time if I didn’t run an instance of Folding@home on it in the background while doing less computer-intensive tasks, sucking up power while only rendering Windows 7 Aero effects. Perhaps it’s time to try working on graphics cards with separate units for slower and faster graphical settings, with the ability to hot-swap automatically between them. This capacity has been demonstrated on laptops, where power consumption is a big deal, but it also needs to be demonstrated on desktops.

The recent developments of SLI and CrossFire by both of the main graphics card developers has caused another problem. Yes, I understand that NVIDIA and AMD need to get rid of that lot of mediocre graphics cards you have left over somehow. That doesn’t excuse them from trying to dump this technology on us. If they want to sell dual-graphics card systems, at least make it so that the end-user gets graphical potential more equivalent to the amount of electrical power that gets used up. Less dual-graphics card systems, less high-end graphics units which serve more to show off the potential of the company than to do any real tasks, and more decent, power-efficient mid-range components which make financial sense, please.