Gaming PCs: What to Buy and How to Build

#1 Edited by Geno (6479 posts) -

Table of Contents


A. Opening Comments

i. Disclaimers and Terms of Use

ii. Introduction

iii. Methodology 

B. Recommended Build

iv. Option 1

v. Option 2

vi. Option 3

vii. Option 4

viii. Option 5 

C. Resources

ix. Graphics Card Analysis

x. Processor Rankings

xi. Recommended Sites

xii. Compatibility Guide

xiii. How to Build Your PC

xiv. Recommended System Builders

xv. Useful Programs

xvi. Stress Testing

xvii. Peripherals Guide Useful Programs

xviii. Glossary 


A. Opening Comments

i. Disclaimers and Terms of Use

Disclaimer: This is merely a guide to what is currently available in computer hardware, it is up to you to buy, build, and use responsibly. Neither I nor anyone else in this thread will be held liable or responsible for any damages, monetary or otherwise, incurred as a result of the content in this thread to any party whatsoever.

No portion of this post is endorsed by any company, the opinions given forth in this post are purely my own, formulated from benchmarks and other resources with the intent to avoid bias.

All references and resources used are property of their respective owners.

Terms of Use: This information may be freely used and distributed, but it would be nice to get some credit.

ii. Introduction

This post is meant as a guide and resource for people planning on buying or building a gaming PC. It showcases the most value and performance that can be bought for a gaming PC at 5 price points, $600, $800, $1000, $1200 and $1400. It also serves as an educational resource for those not familiar with computer components. I’ve seen many people asking for help building a PC or buying parts at a particular price point, so hopefully this guide will prove useful. The following is my methodology, the recommended builds, and a resource center containing such information as a glossary, recommended sites, and how to build your machine. Please feel welcome to discuss hardware, ask questions or help others in this thread as well. If this is your first time here, please read the disclaimers, terms of use and methodology first.

(June 30, 2012 Changelog)

After a long hiatus, I've finally come back to update this guide. Sorry for the wait guys, I knew I said I would update earlier but I've just been severely busy.

Build section has been changed to a "core build" plus graphics options. Increase in price of HDDs and lack of mid end graphics options has forced this change; $600-$800 builds have much lower value than before, as $200+ components have replaced $100-$150 components as value leaders. Although more expensive in an absolute sense, you get much more performance per dollar and increased longevity of components (for instance an overclocked $200 Ivy Bridge quad core will be quite sufficient for gaming for something like 5 years; on the other hand a $100 dual core will become outdated within a year, if not already). Applicable games still remains.

Graphics Card Analysis graph also updated with all the new cards; note that cards more than 1 generation in the past no longer have a value measurement, because their prices are no longer established. Power efficiency and performance data for these cards are estimated as best as possible by scaling legacy data with new data. Also updated tier lists.

Added "Useful Programs" and "Compatibilities Guide" sections.


iii. Methodology

Builds are chosen with the intent of running the games listed under "Applicable Games" at their maximum quality settings with around an average of 60fps or above. The settings considered are 1920x1080/1200 (fairly standard resolution for most modern monitors), 4xAA (significant image quality improvement for relatively little performance cost), 16x AF (significant image quality improvement for next to no performance cost) and highest in-game settings. The hardware recommended is not necessarily required for gaming at lower settings, or with less performance demands, but are chosen due to the widespread availability of benchmark data tested at such settings, and because they are are the settings used by many PC gamers. Anything beyond these specifications, such as 8xAA, 2560x1600 resolutions or supersampling are considered "beyond" highest settings in the context of this guide.

Builds will not include the cost of a monitor, mouse, keyboard, speakers or OS as it is presumed the buyer will get these based on their personal tastes and needs. In general, the cost of a typical monitor is $200 and Windows 7 Home Premium 64-bit is $100. Mouse, keyboard and speakers will possibly be an additional $100 or more on top of that. Shipping and tax costs will also not be included since they will vary significantly for person to person, but 10-20% is a good ballpark. The final cost of builds may therefore be significantly more expensive than as listed. Factor in the previously mentioned costs as applicable. All pricing will be from, the largest North American e-tailer of computer components. Newegg has competitive prices, and a good reputation amongst customers. There are other places where you may find similar or the same items such as Amazon or TigerDirect.

Component choices will be based on their stock performance, value and futureproofness. I will do my best to provide a fair conclusion for performance based on benchmarks and reviews. 

B. Recommended Build

Prices and recommendations are as of August 17, 2012

iv. Radeon HD 6850/6870 ($140 - $170)

Applicable Games (will be optimal for anything equal to or less demanding than the following at 1920x1080, 4xAA and max settings with 40-60+fps average performance target):

Assassin's Creed series

Call of Duty series

Civilization IV

Older Codemasters racing games (Dirt 1, Dirt 2, GRID, Fuel)

Dawn of War II series

Dead Space series

Diablo III

Far Cry 2

Guild Wars

HAWX series

L.A. Noire

Mount and Blade series

Sniper Elite V2

Street Fighter series

Supreme Commander series

Telltale games (Sam and Max, Monkey Island etc.)

Most Unreal Engine 3 games (Batman: Arkham series, Mass Effect series, Mirror's Edge, Gears of War, Duke Nukem: Forever, Alice: Madness Returns etc)

Valve games (Team Fortress 2, Left 4 Dead series etc.)

World of Warcraft 


This build will play most of today’s games at highest settings at the indicated performance levels. 

v. Radeon HD 7850 ($250)

Applicable Games (will be optimal for anything equal to or less demanding than the following at 1920x1080, 4xAA and max settings with 40-60+fps average performance target):

Alan Wake

Dragon Age: Origins

Fear 3

Mafia II (without PhysX)

Red Faction: Armageddon

The Elder Scrolls V: Skyrim

Spec Ops: The Line

The Witcher 


This build will play most of today's major games with little to no difficulty and will also hold its own in some more demanding titles. 

vi. GTX 660 Ti ($300)

Applicable Games (will be optimal for anything equal to or less demanding than the following at 1920x1080, 4xAA and max settings with 40-60+fps average performance target):

Battlefield: Bad Company 2

Civilization V

Crysis 2 (DX9)

Dirt 3, Dirt Showdown

Dragon Age II

F1 2010

Fable 3

Just Cause 2

Lost Planet 2

Max Payne 3

Risen 2: Dark Waters

Saint's Row the Third

S.T.A.L.K.E.R. Call of Pripyat 


This build will play the vast majority of today's titles at excellent performance and quality settings. The most demanding games still won't have perfect performance, but will be playable even at very high detail settings. 

vii. Radeon HD 7850 Crossfire ($500)

Applicable Games (will be optimal for anything equal to or less demanding than the following at 1920x1080, 4xAA and max settings with 40-60+fps average performance target):

Aliens vs. Predator

Anno series (Dawn of Discovery, 1701 etc.)

Borderlands (with AA enabled through driver)

Ghost Recon: Future Soldier

Grand Theft Auto IV

Mafia II (with PhysX, if Nvidia listed)

Splinter Cell: Conviction

Starcraft II (with AA enabled through driver)

World in Conflict 


This build will play almost all games with perfect performance at their highest settings. Other than a handful of demanding titles, no game proves a challenge. 

viii. GTX 660 Ti SLI ($600)

Applicable Games (will be optimal for anything equal to or less demanding than the following at 1920x1080, 4xAA and max settings with 40-60fps average performance target):

Arma II series


Crysis and Crysis: Warhead

Crysis 2 (DX11)

Metro 2033

S.T.A.L.K.E.R. Shadow of Chernobyl and Clear Sky

Total War series (past Medieval II)

Witcher II 


This build represents the reasonable maximum that anyone should try to attain in terms of game performance. Even the most demanding games will perform excellently at maximum quality settings with this build. In addition, it will be comfortable with more advanced and demanding features such as tessellation, supersampling and ambient occlusion. 


C. Resources

ix. Graphics Card Analysis


Last updated August 17, 2012


*Measurements for value, performance and power efficiency are relative and directly proportional, and use the left axis. Prices shown are based on the most common price for that graphics card, and use the right axis. This graph will be updated quarterly, or as deemed necessary.


- Each category below is a binning of approximately 36%, that is, the total performance difference from the weakest to the strongest card in each category is about 36%, as is that from the average card in each category. This value was derived using the fifth root of the performance multiple between the strongest card and what is commonly considered the mid end card.

- The following is based purely on performance, with the strongest at the top, weakest at the bottom. Look above for exact relative performance. 


GTX 690 

Very High End

GTX 590

HD 6990

GTX 680

HD 7970 

High End

GTX 670

HD 5970

HD 7950 

GTX 660 Ti 

GTX 580

HD 7870 

Upper Mid End

HD 6970

GTX 570

GTX 480

HD 7850

HD 6950 2GB

GTX 295

HD 4870X2

Mid End

GTX 560 Ti

HD 5870

GTX 470  

HD 6870

HD 5850

GTX 285

HD 6850

x. Processor Rankings

There are dozens of different processors within each tier, so the CPU rankings are not comprehensive. They simply list the most popular and/or most worthwhile processors at that tier and are meant to give you an idea. To construct the chart I used my own experience looking at CPU reviews, as well as Passmark data as a point of reference. Processors are listed alphabetically within each category (or close to it; no performance ordering is implied except for the categories). Like the GPU section, each category is separated by about 20% relative performance in games. 


Intel Core i7 3700/3900 Series

Intel Core i5 3400/3500 Series

Intel Core i7 2600 Series

Intel Core i5 2500 Series

Intel Core i7 965-995X series

AMD FX-8120 

Very High End

Intel Core i5 2300/2400 Series

Intel Core i7 920-960 Series

Intel Core i7 860/870

Intel Core i5 750/760

AMD FX-6120

AMD Phenom II X6 1055T-1090T 

High End

Intel Core i3 2100 Series

Intel Core 2 Q9000 Extreme Series

AMD Phenom II X4 960-980 Series

AMD Phenom II X6 1035T-1045T 

Upper Mid End

AMD Phenom II X4 945/955

Intel Core 2 Q8000 series

Intel Core 2 Q9000 series 

Mid End

AMD Athlon II X4 640/645

Intel Q6000 series 

It should be noted that differences in CPU performance after a certain point (around the High End-Very High End area) are largely academic - that is, the extra performance makes little to no difference in actual gameplay since the framerate will be above 60, the refresh rate of most monitors. The GPU is the bottleneck for the vast majority of games, so the benefits of a top-end CPU are usually realized in only a handful of RTS games during certain heavy load scenarios. That is why, given the choice, it is almost always smarter to spend more money on the GPU than the CPU after a certain point. Tom's Hardware recommends spending no more than around $200 on a gaming CPU as stated in their monthly CPU articles. 

xi. Recommended Sites



These sites provide valuable price vs. performance articles, game performance analyses, and industry insights.





Tom's Hardware 


These sites house the most knowledgeable community, and are a good source of leak info and hardware discussion.

Beyond3D Forums

EVGA Forums (Nvidia-centered)


techPowerUP! Forums

Xtremesystems Forums 


These sites offer up to date news for your hardware and other tech needs.




These sites offer reviews across a wide range of games with a large range of cards.







The Rest of the Best

These are the sites not mentioned in the above categories, but are still high quality and deserve a look.





Hot Hardware

PC Perspective

Pure Overclock


X-Bit Labs 

xii. Compatibility Guide

Here are important things to look for regarding compatibility

1. Motherboard-CPU. Different motherboards use different CPU sockets. Sandy Bridge and Ivy Bridge processors use LGA 1155, last gen Core i5 processors use LGA 1156, and Nehalem Core i7 processors use LGA1366. Check that your CPU and motherboard are both of the same socket.

2. Case-Motherboard. A midtower will not fit some of the larger motherboards. Make sure that both your motherboard and case are of the same size standard. This will be “ATX” the vast majority of the time, but it’s wise to double check.

3. RAM-CPU-Motherboard. LGA 1156, LGA1155 and AM3 use dual channel RAM. LGA 1366 uses triple channel RAM and LGA 2011 uses quad channel RAM. Make sure that your RAM, CPU and mobo are all compliant with one another in terms of RAM channels. Other specs to look out for are DDR# and number of pins. These should all be provided in the motherboard spec, just find the RAM to match.

4. Motherboard-GPU. Graphics cards require PCIe 2.0 or 3.0 slots. In addition, you will need at least 2 for multi-GPU, and they will need to be at least 8X speed each to avoid bottlenecking your cards. Some motherboards have 2 or more slots but the slots beyond the first one are at 4X or below; these you should avoid.

5. Motherboard-GPU (2). Nvidia uses a specific chipset for SLI. You should check whether the motherboard is SLI-compatible if you intend on using it for SLI. Generally, any board with 2 PCIe 2.0 slots can run Crossfire, and most modern motherboard chipsets like Z68 can run both SLI and Crossfire.

6. Power Supply-System. Add up the total system wattage at peak load. The GPU and CPU are the main power users in the PC, so a good approximation is to add up the max wattage of the GPU and CPU, then add about 100W for the other components. On top of that, add another 100-200W for headroom. Your final tally should be the size of the PSU that you will be using for the system.

7. Power Supply-Graphics Card. Graphics cards require some multiple of 6 and 8 pin connectors in order to have enough power to run. Check that your power supply has all of the necessary 6 and/or 8 pin power connectors. 8 pin connectors are generally just 6 spin with an additional 2 pins attached loosely (so you can use it for either 6 or 8-pin connections). 

xiii. How to Build Your PC

(Video from HardwareCanucks )

xiv. Recommended System Builders

There are many services to cater to those that have their PC in mind, but for one reason or another choose not to build it themselves. For a fee, you can pick out your parts and have a professional build it for you. Sometimes this is a wise decision, as the package usually includes not only building the PC, but also system testing, professional wiring, overclocking, technical support and a warranty. I've included the Resellerating for each company, which are aggregate reviews for each company (similar to how metacritic or gamerankings are average reviews for a video game). That way you can compare things like reliability and customer service as well.

Best overall system builder:

CyberPowerPC: Following the theme of this post, CyberPowerPC offers the best value in terms of build quality and price. They also have one of the widest selections of components and builds that I’ve seen. In some cases (during sales and such) a custom built PC may even be the same price as one that you build yourself, making this a smart option. Resellerating

Best extreme system builder:

Digital Storm : If you’re looking for an extreme end system, Digital Storm is where you want to go. They offer incredible deals on Core i7-class systems, overclocking, liquid cooling, and the best quality control in the business. Their builds are high-priced but even higher powered. Resellerating

Other system builders:

Alienware (RR)

AVADirect (RR )

Falcon Northwest (RR)

IBuyPower (RR)

Puget (RR)

Velocity Micro (RR)

xv. Useful Programs

The following programs are for protection and general maintenance of your PC that I recommend.

Avast Antivirus. One of the best free antivirus out there. Constantly updated, attractive interface, comprehensive protection and quick scanning.

Defraggler. Good interface, easy to use.

MalwareBytes Anti-Malware. Constantly updated, easy to use interface, extremely effective detection, and relatively quick scans.

Microsoft Security Essentials – One of the best security softwares – and it’s free.

FRAPS – good for taking game screenshots, recording video or monitoring FPS.

RealTemp – for general monitoring of CPU and GPU temperatures

MSI Afterburner – for overclocking, fan control, and monitoring of graphics cards

Nvidia Inspector – for deep driver tweaking with Nvidia cards

Radeon Pro – for driver tweaking with ATI cards 

xvi. Stress Testing

Upon building or buying a new PC, it is always a good idea to stress test it. Beyond speed and power, reliability is also a very important aspect of how your PC will perform. Those that buy their machines from reputable builders will have less of an issue with this, as their systems are already pre-stress tested, but doing so yourself never hurts either.

Computer components are extremely complex pieces of hardware. Even the best company with the best quality control can’t avoid defective products. On average, about 2-3% of all shipped products may have some defect that prevents it from working properly. If you happen upon one of these products, it is good to activate an RMA (return merchandise authorization) from your retailer immediately, or claim your warranty if applicable.

Stress testing typically takes a day to complete, but you can save yourself a lot of headaches in the future by doing it. You don’t actually have to sit around for the whole day either.

There are 3 main components that can be stress tested directly, the CPU, the RAM and the graphics card(s). They should all be tested separately, as well as all at once. 


Prime95 is the staple application in testing CPU stability. The Small FFTs test pushes the CPU to 100% load, essentially a “worst-case-scenario” test. It can also test a mix of memory and CPU using the Blend Test. Since the program can push both CPU and RAM usage to 99-100% load, any real world application such as gaming will run stable if the computer runs stable in Prime95 (gaming will stress the computer significantly less than Prime95). Prime95 should be run for at least a couple of hours, most people prefer to run it overnight or even 1-2 days (for the really diligent). If any errors occur, then the CPU is not completely stable.

If you get an error or a system crash while stress testing an overclocked CPU, trying slightly increasing the core voltage, or decreasing the clock (as always with overclocking, learn what you're doing beforehand and do so at your own risk). If you get an error or a system crash with a stock CPU, then it is most likely faulty and you should RMA it for a new one. CPUs at stock clocks with stock cooling should not be crashing under Prime95, since they are designed to be fully loaded at those specifications.

Monitor temperatures with program such as RealTemp at all times. A CPU at maximum load at stock should not be exceeding 70-80 degrees Celsius. With a good heatsink, it should only be 50-70 degrees Celsius. If it’s exceeding these temperatures, slightly push the heatsink against the motherboard and see if the temperatures decrease. If they do, then you haven’t affixed the heatsink tightly enough. If it doesn’t, you either have inadequate case cooling, an inadequate application of thermal paste or a faulty CPU. The TjMax for most CPUs (maximum temperature before the CPU shuts off or throttles itself to avoid damage) is 100 degrees Celsius, but you should not be approaching it. 


RAM is typically tested with Memtest86. It can be used in conjunction with Prime95 to isolate CPU or RAM problems. You will need to burn it on to a CD or USB and then boot your computer from CD or USB (typically by pressing F8 while the computer is booting up and choosing which device you want to boot from). There should be zero errors while running this test. If overclocking, you could try to increase the memory voltage slightly (again, overclock only when you fully understand the process, and do so at your own risk). If it’s at stock and experiencing errors then you should return the RAM. Even a single memory error in Memtest can prove problematic in usage. 


There are two main factors you should stress test regarding GPUs, performance and temperature.

Update first to the most recent drivers for your system, and using any heavily GPU-oriented benchmark such as Unigine Heaven or 3DMark, you should compare it to the results of a website that tested your card using the same benchmark at the same settings. The numbers should be reasonably similar to each other (within 5%).

Temperatures can be tested by a program called Furmark (also known as Kombustor). It is essentially the Prime95 of graphics cards, and will push your card or cards to the max. The temperatures reached in Furmark will exceed the temperatures that you will get in any game, since no game will put 100% load on your cards 100% of the time. Furmark should be ran until card temperatures stop increasing (typically just a couple of minutes), or until you deem it necessary to stop them from experiencing damage (read below). Cards in Crossfire or SLI may run 5-10 degrees higher than their single configuration variants; e.g. if a single card runs at 70 degrees C at load, then two of them in SLI may run at 80 degrees C in SLI at load. Also, one card may be slightly hotter than the other due to proximity to the CPU, motherboard heatsink and RAM.

Google your card SKU (e.g. “GTX 460”) and Furmark together and search for a review that benched their card in Furmark, and compare. Results shouldn’t differ more than 5 degrees C from one another. Also in general, most cards shouldn’t exceed 90 degrees Celsius in single GPU configuration. The maximum safe temperature for most cards is 105-120 degrees C (at which point they will either shut off or throttle themselves to avoid damage), but ideally you should avoid getting anywhere near those temperatures since you may damage the surrounding components from the heat output.

Judging from the above, if anything looks off you should consider RMA-ing your card or getting better case cooling depending on the size of the deviation.

Finally, it is possible to test all 3 component areas at once by running the Blend Test in Prime 95 and Furmark at the same time; if your computer can survive a few hours of this then it is most likely completely stable. 

xvii. Peripherals Guide

Unlike most other PC components, when it comes down to peripherals such as mice, keyboards and headsets there aren’t any definitive benchmarks, so it really comes down to taste. There is also a lesser correlation between price and performance; while quality generally increases along with a product’s cost, it’s also well within the realm of possibility for a $70 headset to sound and perform better than a $200 headset. Therefore, this section is not split into tiers nor will it even make recommendations, it will simply provide a stepping stone to help you determine which product is right for you.

Roundups are where websites take a good variety of products in the same category and compare them. A good way to use this information is to first look at where a product compares overall against its competition, and then search for individual reviews of that particular product to get a more in-depth feel for its strengths and weaknesses. Remember, no product is perfect and although one may be rated “best of the bunch” you may find that its features are not to your personal liking or needs, so therefore it’s worthwhile to get more information. 


Gaming mice can significantly improve your gaming prowess. Faster response times, greater tracking, button macros and ergonomic design can go a long way. When considering gaming peripherals, a good mouse should be a top priority. One good research method that works particularly well with mice is to search for video reviews on Youtube; if not for anything else, it’s to see the mouse in “3-D”, e.g. rotated about its axes and held in people’s hands so that you can get a better idea about how it would look and feel in real life.

Established brands (in alphabetical order):





New or alternative brands to consider (in alphabetical order):










Mousepads are more based on preference than any other gaming product; very few technical specifications go with them except for physical dimensions. Qualitatively, most mousepads are hard (plastic feeling), soft (cloth feeling) or both (either feels midway between soft or hard, or is soft on one side and hard on the other). Some mousepads claim greater tracking ability; while this may be true to some extent, keep in mind that mousepads are a mature technology and most good mousepads perform similarly. As mentioned before, the main difference you want to look out for is hard vs. soft. You may also want to consider the dimensions of your mousepad depending on how much free space you have on your desk, and how large you like to make your sweeping motions. As seen in the following roundup by Tom’s, even a magazine can function in a pinch.

Established Brands (in alphabetical order):



New or alternative brands to consider (in alphabetical order):




Tom's Guide



When buying a headset, you should consider whether you want it for gaming, music or both. A gaming headset can still be good for music and vice versa, but a specialty headset will always be the best at its respective area.

Gaming headsets are generally very percussive and allow you to hear gunfire and footsteps quite clearly and from far away. They usually also carry good surround sound effects so that you can determine direction accurately.

Audiophile headsets are good for watching movies and listening to music. They have a nice “full” midrange sound, which is the range where most voices and instruments tend to belong.

Other headsets are great all around, and work well for both gaming and music, but being the best in neither.

Generally though, buying any set of quality headsets from a reputable company can’t steer you too wrong, and unless you’re a professional sound designer, you will most likely be unable to notice any small deficiencies between one good headset and another. Thus I recommend simply reading a few reviews, finding one that is well-liked and that has features and aesthetics you’re comfortable with, and going with that one.

Established Brands (in alphabetical order):










Turtle Beach

New or alternative brands to consider (in alphabetical order):








In the area of keyboards, you should search for reliability and aesthetics. Keyboards are arguably the least valuable gaming asset; that is, a gaming mouse will likely perform significantly better than a non-gaming mouse, but a gaming keyboard will probably only perform marginally better than a conventional office keyboard. Therefore I believe it is wise to go on Amazon or a similar site and read the reader reviews to see the “breakage rate” (e.g. how long it takes before it becomes unusable, which is outside the realm of critic reviews), and just choose one that has aesthetics to your liking.

Established Brands (in alphabetical order):





New or alternative brands to consider (in alphabetical order):

Wolf King






xviii. Glossary



ATX – Form factor of motherboards. Typically the larger the form factor, the larger the amount of ports and slots there are to use. ATX is the standard size which will fit in pretty much all cases, and other common types include MicroATX (smaller size board typically designed for very specialized uses) or EATX, which is a larger board that can generally only fit in full towers. Examples of such boards are the EVGA X58 Micro and the EVGA SR-2.

CPU – Central Processing Unit. The “brain” of a computer, it essentially directs all of the actions of the other components. In gaming, it’s usually the GPU that is most important in determining performance but an abnormally weak CPU will bottleneck the performance of the graphics card.

GPU – Graphics Processing Unit. This powers your graphics card and specializes in processing intensive 3D graphics in real time. In contrast to CPUs, they are highly parallel, with lower clock speeds but many more individual processing units. The most important factor in graphics performance.

HDD - Hard Disk Drive. This is your mass storage device that usually runs on the SATA interface. Windows users are probably familiar with it, it is the drive usually labeled “C:\” in My Computer.

OS - Operating System. For gaming PCs this will be Windows. The newest version of Windows is Windows 7, and it along with Windows Vista supports DirectX11.

PCIe - PCI express. This high bandwidth slot is where your graphics cards are inserted. The current version is 2.0, with a 2.1 revision. 3.0 may be arriving soon.

RAM - Random Access Memory, also referred to as just “memory”. These act essentially as worktables for the applications that you run, storing the data of currently active applications to be used. The most demanding games currently consume around 1.5GB of RAM, and the Windows OS with typical background programs consumes another 0.5-1GB. It is thus recommended to use at least 4GB RAM at this point in time. RAM is also present in graphics cards in the form of VRAM, and acts as local, speedy memory for the GPU. Currently, DDR3 is standard for system RAM and GDDR5 is the standard for VRAM. There are also currently dual channel and triple channel variants, with the triple channel RAM only usable on the LGA1366 platform.

Tower – Otherwise known as a “case”, these will come in mid-sized and full sized varieties. Mid-sized towers are good for most configurations, but for hot running high end systems or systems containing large components like the higher end Radeon 5000 cards, a full tower or a large sized mid-tower is better for air flow, space, and cooling. Larger cases also allow for easier further expansion such as larger motherboards, additional hard drives or disc bays, and liquid cooling. It is recommended to get a good full tower if you can, since it will be the component that stays with you the longest. 


Bottlenecking – When one component significantly limits another. Components depend on each other to operate, and if one component reaches its maximum performance level, the corresponding performance level of a component that depends on it will be “bottlenecked”. Example: Game X has a maximum framerate of 100fps using an extremely fast processor and graphics card, and a framerate of 20fps using an extremely slow processor and the same fast graphics card. The graphics card is capable of attaining 100fps, but can’t due to the slow processor, and is thus being bottlenecked. The extra graphical power it is capable of is essentially being wasted. Thus in system builds it is important for components to be both powerful and balanced; if one is completely overpowered compared to another, then a lot of potential is being wasted.

Derivative – Any product coming out after an original hardware launch that is based on the same architecture, but has different specs thus delivering different performance. Generally referred to lower end models of the product line.

Frequency - Also referred to as the "clock". This pertains to the general speed of a component, or how many operations it can execute a second. The basic unit is the hertz (hz) and you will see them described with your processor, graphics card and even your memory. Generally the higher the hertz the better; the actual range of values that the hertz will be in is dependent on the component type and the architecture. For example, CPUs can have frequences over 4Ghz (Giga = billion) with overclocking, and over 3Ghz at stock values. Graphics cards on the other hand currently have hz measurements only in the 600-900Mhz (Mega = million) range, however they have hundreds or even thousands of smaller processors that do work in parallel (in comparison to the 4-8 that exist on a modern CPU). The architecture of the component will also affect speed; for instance, quad core intel CPUs based on the Nehalem architecture are faster than corresponding AMD quad cores at the same clock speeds. The GTX 295 with a core clock at 576 Mhz is actually almost twice as fast as the GTX 275 at 633 Mhz, since the 295 is dual GPU and has much more mini-processors (known as shaders and raster operators, and stream processors in ATI's cards etc). While a high frequency is indicative of speed, it always comes down to the real world benchmarks when measuring performance.

Multi GPU – Either SLI or Crossfire configuration; when 2 or more graphics cards communicate to each other over the PCI express lanes, achieving a theoretical x-scaling in performance where x is the number of graphics cards. In reality you usually see less than the theoretical scaling, but the performance benefits from multi-GPU can still be very significant, typically resulting in a 70-80% performance boost depending on the game. 


Ambient Occlusion – Technique that provides realistic lighting effects on irregular surfaces. Generally very GPU-intensive.

Anisotropic Filtering – Filters textures so that they look more accurate over a distance at certain angles. A cobblestone road without AF will look like a muddy road in the distance; with AF turned on, it will still look distinctly like a cobblestone road even as it approaches the horizon. AF is generally not very GPU intensive, even on its highest setting (which is 16x at the moment).

Antialiasing - Image quality method that reduces the “jaggies” or pixilation around the edges of in-game objects. Typically carried out using the multisampling method, but many other variants exist. Some forms sacrifice image quality for performance (such as MLAA), while others sacrifice performance for image quality (such as SSAA). Transparency antialiasing is antialiasing on transparent textures; things like the spaces in fences, trees or bushes. Forcing supersampling for transparent textures (SSTsAA) is generally recommended since you get much better image quality for relatively little cost.

Deferred Rendering - Rendering method that many modern games use which makes shaders and post-processing more efficient, but can affect the ability for the game to allow antialiasing.

DirectX - An application programming interface for 3D applications including games, usually distributed in runtime libraries. DirectX11 is the newest version of the 3D development platform, sporting features including tessellation, better multithreading support, Shader Model 5.0, and more optimized performance.

Frames per second (framerate) - The amount of frames that are being processed through hardware per second. Each frame is a fully realized and fully processed view of the game. Games start to feel significantly laggy below 40fps, and 60fps is currently the standard for optimal performance, as that's the refresh frequency of most displays. The frames per second in games is slightly different from the frames per second in television in movies in that it typically lacks motion blur, so high amounts are needed in order to compensate for smoothness. This is why TV and movies look smooth at 24fps, while games are basically unplayable at that framerate. In addition, you "feel" framerate in games whereas you only observe it in television or movies, which makes low framerate even more apparent in gaming. Your framerate cannot exceed the refresh rate of your monitor; 240fps on a 60hz monitor and 60fps on a 60hz monitor will look the same (in fact 240fps may even look worse due to screen tearing; see ‘vertical sync’).

Resolution – This will affect the overall detail of your game. Changes in resolution decrease performance as more pixels need to be processed per frame, but in return you get more detail. Currently, standard PC gaming resolutions are 1680x1050, 1920x1080, 1920x1200, and 2560x1600.

Vertical sync – Locks your framerate to that of your monitor (usually 60). The benefit of this is it prevents screen tearing, which appears as a thin horizontal line that stretches across your screen momentarily due to the frames exceeding that of your monitor and not matching up, and smooths the overall look of the game. The downside is it reduces performance and creates stuttering if you can’t achieve a consistent 60+ framerate. 


CF - Crossfire, ATI’s multi-GPU scaling solution. Crossfire generally refers to 2 cards, Tri-fire 3, and Quad-fire 4.

FRAPS - Handy program that measures framerates, records benchmarks and movies, and takes screenshots. Generally what reviewers use to measure framerates for their benchmarks.

GTX - Nvidia’s line of GPUs, the higher the number the better. Currently in the 5xx series.

Hyperthreading - Intel’s built in feature in some processors that allows a single physical core to be divided into 2 logical threads; enhances multitasking. Usually has little effect in games but functionality may increase as games become more multithreaded.

HyperTransport - Replacement for the front side bus, this system has a higher bandwidth and universalizes several different FSB frequencies.

LGA 1155 - "Sandy Bridge", Intel's newest CPU architecture as of 2011. SKUs range from Core i3 to Core i7, from dual cores to quad cores with hyperthreading.

PhysX – Nvidia’s onboard physics solution. Many of Nvidia’s newer cards come with an onboard PhysX chip, which enables exclusive PhysX effects for games that support it. Physx actually doesn’t require much hardware power, but it will take up an entire graphics card in a multi-GPU setup. ATI cards are unable to run Physx. Some people opt to buy a single, cheap Physx capable card such as the GT 220 and use that as a dedicated card to do Physx.

QPI - Quick Path Interconnect. The successor to the front side bus and originally seen on AMD chips in the form of HyperTransport, it allows for extremely fast transfer rates from the memory controller to the CPU.

Radeon HD - ATI’s line of GPUs, the higher the number the better. Currently in the 6xxx series.

SLI – Scalable Link Interface. Nvidia’s multi-GPU scaling solution. SLI generally refers to 2 cards, Tri-SLI 3, and Quad-SLI 4.

Socket AM3 - Socket for AMD’s current generation of processors including the Athlon II and Phenom II lines. The Phenom II series consist of quad core CPUs while the Athlon II series consists of dual and triple core CPUs. These currently span the mid-end.

#2 Posted by s7evn (1072 posts) -

Wow, nice guide. Possibly could deserve a sticky over at the Tested site, hell even a stick here for the work that went into this.

#3 Posted by Mikemcn (6875 posts) -

Radeon HD 4850, never stop being beautiful.

#4 Posted by fwylo (3556 posts) -

These posts are great.  I built one around December without looking at one of these and it turned out to be sitting exactly the same as your first extreme build you posted. Quite the powerhouse.

#5 Posted by Geno (6479 posts) -

I see that the hyperlinks in the spoiler tags are hard to see, do any of you have a problem with this? 

#6 Posted by s7evn (1072 posts) -
@Geno: Yeah, they are bright yellow on a white background--very difficult to read.
#7 Posted by ISuperGamerI (1880 posts) -

Great idea and good work bro. :)

#8 Posted by defaulttag (890 posts) -

I love my core i7 920 overclocked and GTX 275

#9 Posted by Geno (6479 posts) -
@s7evn said:
" @Geno: Yeah, they are bright yellow on a white background--very difficult to read. "
I might've found a solution, just a sec. 
#10 Edited by atomic_dumpling (2442 posts) -

I just glanced over it, but the different tiers seem to be pretty accurate. Well done!
The prices for VGA-cards are completely whack at the moment. They simply refuse to drop. Also availability is a real problem these days. The GTX 260 through GTX 295 are EOL and disappearing from the market with the quickness - the only way to get a hand on those old (but still powerful) cards is the used market. It's probably easier to get a HD5770 or HD 5830 at this point. The problem is that between GTX 285/HD5770/HD5830 and HD5850/GTX470 there is currently a huge gap worth about 100 bucks. The upcoming GTX 465 was supposed to fill that gap, but as the first reviews show this not going to happen - the GTX 465 is going to be closer to the old GTX 275 and GTX 280 with the same high power consumption and heat generation as the GTX 470/480.

#11 Posted by Dipstick (547 posts) -
@defaulttag said:
" I love my core i7 920 overclocked and GTX 275 "
<3 you bro,<3 you 
We have similar builds. Let's be friends :)
#12 Edited by Geno (6479 posts) -
@atomic_dumpling: Thanks. I tried to be fairly scrutinizing with how the cards were ranked so hopefully it paid off.  
I think newer cards will almost always be better value for money simply by the fact that manufacturers and retailers actually update their prices according to market demands as opposed to EOL cards that don't. 
From first glance it looks like the GTX 465 is smack dab between HD 5830 and HD 5850 both in price and performance; this may or may not be a good thing since 5830 is kind of bad value, and I think most people would just spend the extra $20 for a 5850.  
With regards to the hyperlink color in the spoilers, I've tried every text formatting option that GB has, but nothing appears to fix it. Perhaps I'll be forced to ditch the spoiler tags, but I don't like how it'll be an epic wall of text either. 
#13 Posted by defaulttag (890 posts) -
@Dipstick:  haha fosho
#14 Posted by mrhankey (695 posts) -
@Geno:  Do you think you could link your sources regarding the new Nvidia graphics cards and their 15%-20% performance difference....
#15 Edited by Geno (6479 posts) -
@mrhankey said:

" @Geno:  Do you think you could link your sources regarding the new Nvidia graphics cards and their 15%-20% performance difference.... "

Of course, sorry about that. Here is how I derived the perf advantage; taking the overall average between 4 established hardware review sites that tested over a large quantity of games. I excluded 3DMark scores as they are synthetic, and I also excluded Unigine Heaven tesselation benchmarks as that would provide an unfair advantage to the GTX 400 cards, since they are exceedingly strong at tesselation but no game really uses it to much effect yet.  
 The new Nvidia drivers also push performance up by at least another 5% 
Which is independently confirmed here, as well as other places.  
Meanwhile Catalyst 10.5 has no similar improvements, so for the time being I think it's fair to say that GTX 480 has a distinct performance advantage over the 5870, in total about 20% or so, probably even a bit higher. 
#16 Edited by lamegame621 (975 posts) -

Edit: Great Job.

#17 Posted by Geno (6479 posts) -

I decided to take the lesser of two evils and toss the spoiler tags so that the hyperlinks are now legible. Also, I added resellerratings links to each of the recommended system builders. More updates to come next month, or when needed. 

#18 Posted by Computerplayer1 (991 posts) -

I'm still doing very well with my E8500 @ 3.8ghz and SLI 9800GT SC cards. Aside from the GPUs being able to heat a small home, the system still runs most everything maxed out or nearly maxed out at 1920x1200.

#19 Posted by warxsnake (2626 posts) -

Amazing post gj

#20 Posted by paudle (85 posts) -

Nice info that you are giving. I appreciate the effort. 
I think that I am looking for something around the price/power range of a core i5 750 and ati 5850. Something that will last me awhile but isn't quite top of the line. Would you recommend this or something from amd and nvidia?

#21 Edited by Geno (6479 posts) -
@paudle: The 5850 should be an above average card for another ~2 years (at which it will drop to mid-end), the Core i5 750 for about 3-4 (processors age slower than graphics cards, and the Core i5 750 is higher amongst its peers than the 5850 as well). It depends ultimately on what quality settings you will want to run at. If you refer the graphics card rankings sections, you should find a description that fits your needs. If you need benchmarks for any individual card I will be happy to provide them as well (I'm thinking of adding individual benchmarks to the graphics card rankings anyway), or you can search them up from one of the recommended sites.
#22 Posted by IncredibleBulk92 (936 posts) -

I love your mid end build.  That thing will run anything at a great fps and the 480 looks like it'll hold up well for the next two years at least. 
@Paulie:  I'm using a Core i5 and 5850 right now and I can tell you that it runs any game with no problem.  I haven't had any problems running any games so far.  The only slight stoppers have been Crysis, Metro 2033 and Stalker using certain settings (SSAO + DX11).  None of these games have problems at very high settings but whenever I start pulling the anti-aliasing up the frame rates can begin to drop but to be fair that's pretty normal.  I'm slightly drunk/tired right now so I'm not certain this makes sense but if you want to ask more questions reply to me or PM or something.  I also have some screenshots posted in the PC forum if you want games look like on my machine.

#23 Posted by paudle (85 posts) -

thanks guys and thoughts about the amd/nvidia stuff for a similar price?

#24 Posted by SlasherMan (1725 posts) -
@paudle: For a similar price on the AMD/Nvidia side, you'd probably be going with a Phenom II X4 and a GTX 470. Although considering the GTX 470 is very close to an HD 5870 in price ($350 vs $390, lowest prices for both on Newegg), I would recommend going for a Phenom II X4 + 5870 rig instead.
As far as the i5 vs the Phenom II, in gaming you won't notice much of a difference so going for the Phenom is probably a better idea. However, if you do need the extra power for stuff outside of gaming, then an i5 would be the better choice. It all depends on what you need, really.
#25 Posted by paudle (85 posts) -

How is the Phenom II X4 different than the Phenom II X6 in terms of game performance? Which X4 would you recommend?
I see that the  AMD Phenom II X6 1055T is basically the same price as core i5 750
For video cards (in canada) I see prices like this:

#26 Edited by SlasherMan (1725 posts) -
I'd recommend going with the Phenom II X4 965. You could just go with the 955 and save a little money though, since they're pretty much the same aside from the frequency, which of course isn't an issue if you're overclocking. So, if you're overclocking go with the 955, if you're not then the 965 (although they're both Black Edition CPUs dying to be unleashed).
Also, the Phenom II X6 is a hexacore CPU (hence the X6 in its name, signifying 6 cores), and I honestly don't think it's worth it as far as gaming is concerned. You can read this article for more info, comparisons and such (just to clarify, this compares the Athlon II X3 440, which is a low end triple core CPU to a high-end i7 920. You should be able to see why I don't think the X6 is worth it when compared to the X4). 
  This article has a direct comparison between the X4 965 and the X6 among other CPUs, which on second thought, would have probably been a more appropriate one to link to in the first place. Just skip to the gaming benchmarks and look at the ones that relate to your monitor's resolution. You can look for more benchmarks on other hardware sites, too, to get a broader view of the picture.
And yeah, those prices seem consistent with the ones I saw on Newegg as far as the differences between the cards go. $40 difference between the cheapest GTX 470 and HD 5870.
#27 Posted by Geno (6479 posts) -

Slasherman is right, there is no benefit to the X6 processors at the moment. In fact, in most benchmarks that I've seen it actually performs worse than similarly clocked quad cores. Really not a gaming CPU. 

#28 Posted by paudle (85 posts) -

thanks guys, this helps clear things out

#29 Edited by teh_pwnzorer (1482 posts) -
@Geno: Meta-analyses are at the boundary of science and pseudoscience.   Your computation of the average performance difference is comical to say the least.
#30 Edited by Geno (6479 posts) -
@teh_pwnzorer said:

" @Geno: Meta-analyses are at the boundary of science and pseudoscience.   Your computation of the average performance difference is comical to say the least. "

Yes, I didn't go through it with all of the proper statistical trimmings, it was just a quick 10-15 min job, but I think adequate for the purpose at hand. If you go to hardware forums such as xtremesystems, [H]forum or techpowerup you should find that a 15% performance edge at release is generally accepted. If you can provide a better one I would be glad to adopt it and incorporate it into the next edition. 
#31 Edited by teh_pwnzorer (1482 posts) -
@Geno said:

" @teh_pwnzorer said:

" @Geno: Meta-analyses are at the boundary of science and pseudoscience.   Your computation of the average performance difference is comical to say the least. "

Yes, I didn't go through it with all of the proper statistical trimmings, it was just a quick 10-15 min job, but I think adequate for the purpose at hand. If you go to hardware forums such as xtremesystems, [H]forum or techpowerup you should find that a 15% performance edge at release is generally accepted. If you can provide a better one I would be glad to adopt it and incorporate it into the next edition.  "
If you actually look at the numbers, the same game  (i.e., Bad Company 2) can produce opposite results on different sites -- because their setup is different.  Your "15%" is meaningless.   EDIT: Listen, if you have a dozen crappy experiments, you don't get a "good" experiment by taking the average.  You don't get signal from noise by taking the average of the noise.
BTW, I've mostly been going to tom's hardware for the last 14 years.  I never heard of the sites you mention.
#32 Edited by Geno (6479 posts) -
@teh_pwnzorer: Results will differ going from site to site, different sites use different testing methodologies and different hardware. That's the whole purpose of an average; everyone in a country doesn't necessarily have to have the average height, or even close to it, in fact there could be not a single person in a country with the average height, the average is still equally valid. It is difficult to decide the way in which to take averages for benchmarks for varying websites, which is why I used multiple methods. As you can see, they all came in to about the same value. Perhaps I could use median, but that doesn't work as well for smaller samples. Just look through some benchmarks yourself (if you don't prefer the given ones) and consult some enthusiast forums, I think you'll find that I'm about right on. I'm not really willing to discuss this further with you until you can provide some numbers and evaluations of your own, and I also hold your general knowledge of hardware suspect if you've "never heard of" sites such as [H]ardOCP or Xtremesystems, some of the most popular sites amongst the enthusiast community. Xtremesystems had a 300-page thread series about Fermi prior to its release for instance.  
As for Tom's, it is a good site, that is why I named it as a recommended site for news. For reviews however, I find that they test too few games with too few cards most of the time. Tom's does fine for news, but you can find better reviews elsewhere. If you disagree I am not forcing you to look elsewhere, but do some research before making your statements. 
#33 Posted by Mariwana (5 posts) -

what's the noticeable difference btw a 128bit graphics card and a 256bit graphics card

#34 Posted by SlasherMan (1725 posts) -
@Mariwana: There really isn't an answer to your question, as It's not as simple as just comparing memory buses like that. You also have to look at memory speed, to work out the memory bandwidth, and even then, that really doesn't mean much on its own.
For example, an HD 5770 can have around 75 GB/s of bandwidth with a 128-bit bus because it also uses GDDR5, whereas on the other hand, an HD 4850 for example only has around 65 GB/s of bandwidth despite it having a 256-bit bus because of the slower GDDR3 memory.
Generally, though, the best way to compare cards is by looking at benchmarks. You can always try to estimate theoretical performance by looking at specs and comparing, but not everything that would make sense on paper would necessarily translate the same way in real life. 

A big testament to that could clearly be seen a few years back when the 2900 XT was going to be released. On paper, the card seemed like it would beat the pants off the 8800 GTX, which was king at the time. However, the 8800 GTX performed better in every way, except perhaps in synthetic benchmarks (which are hardly something a card's strength is strictly based on), IIRC. 
For reference, in relation to your question about memory buses, the 2900 XT had a 512-bit bus coupled with GDDR3 memory that was capable of delivering about 105 GB/s, whereas the 8800 GTX had a 384-bit bus with GDDR 3 memory, which resulted in a memory bandwidth of 85 GB/s.
Everything on paper would suggest that the 2900 XT was actually the better card, but in reality it was the exact opposite, for whatever reason. Some might say the 2900 XT features were never really used or tapped into, it had potential, etc.. but that doesn't really mean much in the end when the competition delivered much better performance.
My point in writing this long-winded post is, don't judge cards based on specs alone but look at benchmarks instead. There are many hardware sites that put every card in existence to the test and provide a lot of extensive information and benchmarks so that you can see how cards perform in real life.
#35 Posted by Mariwana (5 posts) -

wow.... this is some big lecture ma bro, lol.......... well of this whole reply i could just understand a few lines, thats because am a medical student and i dont have the time to check out all those stuffs, but i play games a lot. 
512MB GDDR 3 256bits Nvidia card..... i think its geforce 9600 
Asus M4A77TD mother board, with AMD phenom II black edition 3.2Ghz
3GB DDR3 ram 
320 GB hard cost me about $700 
when i max out all graphics options of assassins creed , it seems the game gets slower or a bit shaky, and the graphics still dosnt come out nearly as good as the one i have seen on a ps3, so right now i'm thinking of some upgrade.......the only upgrade i'm certain about now is the hard disk......i want to get an additional 750GB, my engineer says its not really necessary to upgrade the ram video card and processor, what do you think bro......

#36 Edited by SlasherMan (1725 posts) -
@Mariwana: I would probably agree with your engineer about the upgrades not being necessary. The 9600GT is a pretty decent budget card and relatively pretty capable. In any case, if you do happen to disagree and still think an upgrade is necessary (after all, only you can decide if you're content with the general performance of your machine), it is the weakest part in your set up so that's what you should be looking at.
As far as Assassin's Creed (the first one, I'm assuming) is concerned you shouldn't have problems running it. The "shaky" thing is called suttering, and this is most likely a driver problem as I've also had that same issue with the game. I'd suggest updating your drivers first of all, and if that doesn't help, open your Nvidia Control Panel and look for a setting called "Maximum Pre-rendered Frames". This usually defaults to 3, but try changing it to 1 and see if it helps any with the game.
Also, about the game not looking good, do you make sure to run the game at your native resolution? Most games will default to 1024*768 or something similar, so be sure to bump up the resolution to whatever your screen's native res is.
#37 Posted by Mariwana (5 posts) -

you didnt say anything about the ram 

#38 Posted by SlasherMan (1725 posts) -
@Mariwana: 3GBs are generally enough. Never hurts to have more, but the question is do you actually need it? Do you ever notice that your computer runs out of memory while running games or doing stuff? If the answer to that question is yes, then sure get some more RAM. If it's a no, then it's probably not needed.
#39 Posted by Mariwana (5 posts) -

thanks bro....... you deserve some weed from meee.... i swear

#40 Posted by Mariwana (5 posts) -

bro my win7 is complaining about not being genuine, how can i tackle this?

#41 Posted by Geno (6479 posts) -
@Mariwana: If you go to My Computer --> Properties, you should see a small section near the bottom where you can validate your Windows, you will need the product key. 
#42 Posted by HitmanAgent47 (8576 posts) -

Hey I was building a pc for alwaysangry. I thought I post my opinion on my $600 pc. It has a better cpu, bigger hardrive, 50 watt lower psu. I didn't read your build when I made the decision, however I see we have a few similar components.

cpu $99.00 
videocards for gaming $159

this one is cheap (yet the hd5770 has DX11, it's newer)      
Cheap AM3 mobos  (still has crossfire at least if you want that option, you need DDR3 ram though)     $104     
ram   $109 
hardrive (7200 rpm which is fast and a full TB) $80 
case $39 
psu  $44.99 
Reuse your old dvd burner drive, might need some cheap sata cables for the hardrive, reuse mouse and keyboard. Geno didn't include windows 7, I am not going to either.

#43 Posted by Geno (6479 posts) -
@HitmanAgent47: Thanks for the build, I will take it into consideration. 
#44 Posted by SolemnOaf (535 posts) -

Hey, I wanted to make a follow-up post to my thread for some semblance of closure I guess.  I actually did just that, detailing how awesome owning a new computer feels and how happy I am with my purchase.  When I tried to post, however, Giantbomb chose that moment to temporarity disable it's servers and I basically said "Argh! There's no way I'm writing that thing again!".  Plus I figured nobody would really care that much and all that I really meant to say is; thanks for making this thread.
#45 Posted by AlwaysAngry (2924 posts) -
@Geno: Is there a way you could recommend a build between the $600.00 and the 1K? If my rez would be 720p, is there a way to make an $800.00 or so that could have the same performance as the 1K one except running on Windows XP and a slightly lower rez such as 720p?
#46 Posted by Geno (6479 posts) -
@AlwaysAngry said:
" @Geno: Is there a way you could recommend a build between the $600.00 and the 1K? If my rez would be 720p, is there a way to make an $800.00 or so that could have the same performance as the 1K one except running on Windows XP and a slightly lower rez such as 720p? "
The $600 build plays games at 720p perfectly. Games such as Stalker and Crysis will be outliers however, but they always are. 
#47 Posted by paudle (85 posts) -

Is there some guide out there for how to pick a motherboard?
I think I want usb3 and crossfire/sli (maybe not both necessarily) still looking at amd/intel

#48 Edited by FirePrince (1762 posts) -

ATi 5770 HD 
2 Gigs Of Ram 
Amd Athlon 5600+ X2 
I like playing PC games.Recently,I have been losing my faith in PC games,mainly because people don't do ports well.After I bought my 5770 I could not play PC games because of a ton of problems. 
But I still have faith.I am considering a upgrade.After seeing Crysis 2 in action-I have to have it on PC.If there's one company that can do PC games right,it's Crytek.Crysis ran great on 1980x1080(the res of my monitor) with all on high. 
What do you PC masters recommend?What should I upgrade?Also,I'm considering buying said upgrade around christmas.Thanks.

#49 Edited by atomic_dumpling (2442 posts) -

Well, when you are trying to upgrade from a HD5770 you are pretty much boned, as in "nothing good for less than 300 bucks" boned. The cheapest decent upgrade would be HD5850 or GTX465. 

#50 Posted by Ben_H (3200 posts) -
@atomic_dumpling said:

" Well, when you are trying to upgrade from a HD5770 you are pretty much boned, as in "nothing good for less than 300 bucks" boned. The cheapest decent upgrade would be HD5850 or GTX465.  "

Essentially.  The 5830 is only marginally better than the 5770 so the options are to buy a 5850 or higher to straight up replace the 5770   or to buy a second 5770 (Only if the motherboard can handle it at full bandwidth in both slots.  It wouldn't be worth it if you can only run them at x16/x4.  At x16/x16 they are about as powerful as a 5870.)
Oh hey, I just noticed I was credited in the OP for being useful.  No problem!

This edit will also create new pages on Giant Bomb for:

Beware, you are proposing to add brand new pages to the wiki along with your edits. Make sure this is what you intended. This will likely increase the time it takes for your changes to go live.

Comment and Save

Until you earn 1000 points all your submissions need to be vetted by other Giant Bomb users. This process takes no more than a few hours and we'll send you an email once approved.