The ramblings of a Linux geek

Sep 3, 2009 - 15 minute read - Comments - Linux

New LINUX Gaming Rig Review

Finally! It’s about time I get this to you guys right?? :). Well here it is at last, the full review of the Budget Gamer that I told you about in a previous post.

Here is the list of hardware components again for you:


Case CM Storm Scout
PSU Cooler Master UCP 700W 80 PLUS Silver
Motherboard MSI 770-C45 AM3 AMD 770 ATX
Processor AMD Phenom II X4 945 3.0ghz
Memory OCZ Reaper HPC 4GB (2x2GB) DDR3 1600 (PC3 12800)
Hard Drive Western Digital 500GB Sata 3.0Gb/s 7200 RPM
Video Card Sapphire Vapor-X 4890 1GB, factory over-clocked at 870MHz/1050MHz.


That’s it! My gamer system that came in at around $740 with all parts purchased through Newegg. I’ve had this system for just about 3 weeks now and it has definitely been a pleasure to work with.

I’m certainly not the biggest gamer (or best at reviews for that matter), and this is the first PC that I have owned that could be considered to be even remotely close to a (mid-)high end machine. Being that, I of course love this system since it’s so much better than anything I’ve ever had previously :D. It’s also the first complete system I built from scratch. It makes me just a tad biased I think ;).

But, raw numbers don’t lie – so while I personally love my new computer because of it’s inherent awesomeness :), I will be posting benchmarks with numbers so that you can peruse the intrawebz and compare what I’m getting with other relevant reviews to your section of interest.

Only three components really warrant reviews in my opinion, and that’s the case, the CPU, and the video card. I don’t have any kind of tools or know-how to “benchmark” the memory, I am ill-equipped to properly benchmark or analyze the PSU, and benchmarking the motherboard, well.. I have no idea how you would do that.

For those components I’ll say these things:

1) The computer is running, evidently the PSU works well enough. (Though I am Hypothesizing it is preventing me from reaching full O/C potential on my Radeon 4890).

2) I had to change settings, timings, and other stuff in the motherboard BIOS to get 1600 from my DDR3 memory. It defaulted to 1066.

3) On Ubuntu Linux, my free -m is showing me 3900MB memory. This is less than 4GB, but when I googled it seemed normal? I still want to look into this.

Alright then, now that all of that is out of the way, lets get this party started shall we?

Cooler Master Storm Scout

Features & Specifications


Storm ScoutImage on, see their review here

CM Storm Scout Features
Model COOLER MASTER Storm Scout SGC-2000-KKN1
Type Mid-Tower ATX
Motherboard ATX, Micro ATX
Color Inside: Black, Outside: Black
Fans/Cooling Included:
120mm Rear Exhaust
140mm Top Exhaust
140mm Front Intake
Replace top 140mm with 120mm
120mm x 2 mounts on Side Window
Drive Bays: 3.5″ Internal HDD: 5
2.5″ Internal HDD/SSD: 1 converter mount provided (uses 1 3.5″ slot)
5.25″ External: 5
3.5″ External: Can be converted from 1 5.25″ bay
Expansion Standard: 7
Special USB Security System: 1

Cooler Master Storm Scout

Charts & Thoughts


Below you will see the temps of the GPU/CPU/HDD’s while running in this case. These temps were not taken in any special way other than visually monitoring the lm-sensors and ati’s aticonfig tool. Idle temps were taken after having the system running for 30 minutes with all applications closed and screensaver off. Load temps were taken after running through 3 Doom 3 benchmarks, 3 Prey Benchmarks, and 2 Left 4 Dead benchmarks. In between each Benchmark I noted the temps of all components and listed the highest one reached here.

My case has the 2 120mm aftermarket fan’s – one is an Antec 3-speed set on maximum and the other is a thermaltake single-speed. Both have blue LED’s which fit with the GFX card, but not with the default red LED’s on the case :D, I have a bit of a hybrid at the moment. I noted that both fan’s lowered the temps of the GPU/CPU approximately 3 degrees each (around 6-8 degrees total), but as would be expected I didn’t see much change in the HDD temps (they already have the front 140mm fan blowing directly on them).

The other thing that will affect temps is I keep my GPU fan running at 70%. The fan on the GPU is near silent until about 65%, 70% is noticeable but not annoying. Any higher and you start to get the vacuum cleaner sensation.

AMD Phenom II X4 945 3.0ghz

Introduction & Benchmarks


Image from, see their review here

Welcome to the AMD PII 945 section of this review. There isn’t much to say here other than when I purchased this setup, it was simply cheaper when including the price of the CPU + Motherboard to go with an AMD chip instead of an intel.

If you look around the web at 945 reviews and benchmarks they show the cpu being almost equivalent to the Core 2 Quad 2.66ghz Q9450 area. When purchasing a Q9450 (or similar) + LGA 775 motherboard the cost would always come out higher with intel. The other reason I went with AMD is the AM3 chip is new and likely to be supported for some time now. When AMD released the AM3 chip they made sure it was backwards compatible with the AM2+ motherboards as well, which makes me hope that whatever the next evolution of processors are may be backwards compatible again. This way, if I ever upgrade the CPU I may not have to upgrade the motherboard as well. In the intel world, the LGA 775 is already dead, replace by the Core i7’s LGA1366 whose motherboards are even *more* expensive, and have no backwards compatibility.

Since there isn’t much I have to test out a processor except for whatever Phoronix Test Suite provided me, I am including this graph on the same page. Only one page for the processor :).

Sapphire ATI Radeon 4890



Images on, see their review with more benchmarks here

Welcome to the main event! This is probably the section of the review that the majority of readers care the most about: The video card! All benchmarks were run using Ubuntu Karmic Koala Alpha 4 with the 2.6.31 linux kernel. ATI drivers used were 9.8 on all 1680×1050 tests, but before I got to the 1440×900 tests ATI decided to pre-release a copy of 9.10 drivers for Koala users and Alpha 5 came out, so I used those here. In the charts on the following page I have organized them in a few different ways.

The charts that say “Native” in the title are games that ran natively in the Linux OS, whereas the chart entitled “Compatibility Layer” shows benchmarks of games that did not have a native Linux client and had to be run through one of 3 different WINE variants: Wine, CXGames or Cedega. I used PTS (Phoronix Test Suite) or Guru3D benchmarks in all scenarios except World of Warcraft, I started the timetest in StormWind and flew from StormWind to Booty Bay.

If any particular game ran on more than one wine variant, I have tried to include benchmarks of all variants to see which one provides the best performance. Wine was the only candidate that would successfully play Left 4 Dead or Fallout 3. I could not get CXGames or Cedega to work, which is a huge ++ for the free product, but in regards to Fallout 3 it was still a huge pain. Both CXGames and Cedega played WoW but I noticed no discernible difference in game play or bench tests.

I was able to get most benchmarks to run at 950mhz GPU clock for an O/C test on my graphics card, but some of them (like Doom 3) would lockup my system. The curious thing is I don’t think it’s my graphics card, by the way it’s acting. It only locks up if I have all 4 cores (in the main processor) set to “Performance” in order to by-pass AMD’s Cool N’ Quiet (yes, this had a tremendous impact on FPS in games, around 20+ for Doom 3) and running a more graphical involved benchmark. I think it might be my PSU that is giving me issues, as bringing the card higher requires more power. The heat on the GFX card when the fan is on 100% is never in critical levels so that’s the only thing I can think. I did include results at 950mhz for all benchmarks, but some of those results should be questioned as I had to do some weird stuff to get them to complete.

On all games I selected the maximum (or ultra) settings where possible using whatever built in slider bar the game had. I did not pick and choose various settings or change anything from what the game decides when you select it’s highest option. I forced 16X AF in the provided amdcccle control applet that comes with the catalyst drivers.

Most native benchmarks were run with the help of the Phoronix Test Suite, which is a truly awesome benchmark suite in my humble opinion. While it does have it’s issues, for the most part all free benchmarks setup and installed without a problem or user interference, while the paid games told you where to put certain files from the discs. Running the benchmarks with the included GUI was a very simple process but the one down side is you can’t select game quality on several of the benchmarks, so on these ones I just had to leave phoronix defaults.

Since I have not found anyway to measure FPS with a third party tool like FRAPS in Linux, all tests with wine were done with their built-in benchmarking tools (I tried using FRAPS in wine, wasn’t working right). Unfortunately, not Phoronix nor built-in benchmarks seem to give you the full Min/Max/Avg spread of FPS and only give you the Average. Average is nice but really the most noticeable framerate will be the Minimum. If you average 60 but your minimum is 5 you may get frustrated in a game (and it could get you killed), so I had intended to give the three numbers, but can only provide what I am given at this point.

Also unfortunately, Fallout 3 does not include any kind of benchmarking, or even any way to overlay the FPS that I can tell. Most people use FRAPS for this. Since I couldn’t properly benchmark it I just put it at 60 in the chart since the game, when it runs, is very playable and I experience very little jerkiness. It’s just the getting it to run that is a downright pain.

In Left 4 Dead I used the timedemo provided by Guru3D and had all settings on High, but disabled Multicore Rendering because that caused *severe* tearing when enabled. This could be due to the fact I was running in wine and not cedega. Unfortunately I was having trouble getting cedega to work due to the fact I didn’t have a CD to install it from – but general consensus online is that cedega is supposed to run L4D much better than Wine, so while the chart will show wine FPS, it should be higher if using cedega.

So now that I have bored you to death with all of the details ;), lets show you the charts!

Sapphire ATI Radeon 4890

Benchmark Results





Final Thoughts


I have been very happy with my new system in the near 3 weeks I have owned it, as in these 3 weeks I have really given it a beating running benchmark after benchmark for this review – even taking it in the trailer and going camping at one point :).

I will probably be getting a new PSU, one with higher amp rating on the 12v rails (mine has 4x12v @ 19A), probably an 800W+ just to see if that helps me take my Radeon a little higher. The Radeon Toxic edition is clocked at 960mhz/1050mhz, but also takes 1 6-pin and 1 8-pin, so if I can get mine to run fully stable on all benchmarks at 950mhz I would be happy.

I think it’s clear from the last set of graphs that those people that say Linux can’t game are completely full of crap. Linux absolutely has the capability for companies to release very good games that run under linux with very good framerates. The native Linux games like Doom 3 and UT2004 obviously ran best, but using the wine variants gives people a *chance* to play their favorite winblows game under Linux without using a virtual machine or emulation (since “Wine Is Not an Emulator”). However, wine/cedega/cxgames results can vary from running perfectly smoothly (WoW) to running decently with minimal effort (Left 4 Dead), to running but with a massive amount of head pounding and throwing the monitor through the window (Fallout 3), to not running at all properly (Crysis).

Some people will say using software like wine/cedega/cxgames is cheating and will only keep the gaming companies ignoring Linux – since we’ll run it some other way. I do agree on the fundamental point they are making however I also live in the real world. Fact is, we want to game now, not when the companies wake up and make universal games. Also, it is extremely difficult for gaming companies to reach the Linux community and make a profit. The cost to benefit of this, as far as I know, has been extremely low on all companies that have tried it.

For example, EVE Online tried pre-packing a cedega build that they would officially support and offered it as a Linux package for download on their site. This, I think, was a very awesome move and they deserve commendation for trying it (even if it wasn’t “native”). However, after some time of offering it they eventually pulled it with a letter to the community that they simply weren’t getting the return from the Linux clients that they were for the Mac and Windows groups. However, last I read, they do make sure that when they release patches and updates, it will not break Wine/Cedega variants, which again is fine by me.

Another game worth noting is that Unreal Tournament 3 (UT2004 had a linux binary on the dvd at release) stated that it would ship with the Linux binary. Then just before the launch they said it would come shortly after. Then they launched the game and the Linux community waited patiently. And waited. And waited. And … yup.. still waiting. The game has been out since November of 2007 and near 2 years later still no linux client, but they do keep saying “It’s still being worked on.” Prime example that gaming companies simply don’t have the motivation required to keep (or even really get) Linux gaming to the high-end levels.

If we could get Wine or Cedega to the point that they could run any windows based program, or at least the gaming companies to take care to not break wine/cedega compatibility on updates, then that would be a perfectly fine solution as well (to me). We do need to get Wine/Cedega to be a little more streamlined so that framerates are a little better (in some games, there are benchmarks in others where wine surpasses windows), and gaming companies to be careful when updating their clients that they don’t break their Linux fanbase. A few years ago WoW learned that lesson the hard way and Linux users went ape-shit on Blizzard, ever since I don’t think there’s been any real problems directly related to WoW.

As it is, for the mass market, Linux gaming probably isn’t there yet simply due to lack of titles. This is not a fault of Linux, just of Linux adoption across the country. If you are a computer novice that just likes to game, unless your games have a native linux binary or are certified to work on wine, cedega or cxgames – then unfortunately I don’t see you removing your winblows partition completely to make way for Linux. I am very saddened by this, but it’s the most likely outcome.

On the other hand, for those of us that like to do real work on our computers without the interference of the operating system trying to force itself onto you at every turn ;), it is nice to know that buying higher end parts for our system isn’t a complete waste. On top of our productivity with Linux, we do have at least some really good games that will play on our operating system. The big three being Left 4 Dead, UT2004 and Doom 3. I hear people have gotten F.E.A.R. to run as well, which is awesome. I won’t include Fallout 3 on that list until getting it to install and run is a little more streamlined, and it still completely baffles me that the appdb rates this game as gold simply because once you DO get it running, it runs great. Oh, except for the random lockups, hanging, and the 4 hours it took you to get it running. Game itself is kinda cool though.. maybe thats how it works – they rate the game and not how well wine actually plays it :D.

The Wine appdb rating system is a bit weak in my opinion, if you have to download and compile wine from sources and apply a patch simply to get one game to run – that deserves a bronze at best. Talk about modifying all your shortcuts just to run two different versions of wine to run different games… PAIN.

But, as much as I harp on how difficult it was to get Fallout 3 to run on my system; that was 1 game out of 8 benchmarks, and the others more or less “just worked”, which is a pretty good outcome actually in my opinion. Left 4 Dead did require some tweaking, and some registry keys to get it to play fluidly on all max settings, but nothing I did was difficult (or help the Multicore Rendering option, unfortunately).

So, there you have it – a full review of a system running the Radeon HD 4890 Vapor-X GFX card on Linux 2.6.31 with the latest Catalyst drivers: And it rocks! Thanks for taking your time to read through my long-winded review of my new system. Hopefully the different sections will provide at least some people information to help them out.

See you next time!