User Article: [H]ard Economics of Video Card Upgrades

SpeedyVV · Sep 17, 2008

The [H]ard Economics
of
Video Card Upgrades

Author: SpeedyVV
Editor: SpeedyVV
Category: GPU / Video Cards
Date: Wednesday, August 27, 2008

Table of Contents

The [H]ard Economics of Video Card Upgrades.
Table of Contents.

Introduction.

Background.

The GeForce 8800 Series.

EVGA GeForce 8800 GTX..

The Competition.

Current Gameplay.

Current System Setup.

Video Card Evaluation Setup.

Evaluation Setup.

Game Play Summary.

Overclocking.

Temperature.

Power

Point A..

[H]ard Economics of Point A..

Upgrade Options.

The Goal

Value of Upgrade (Apples-to-Oranges)

The Bottom Line $.

Point B..

[H]ard Economics of Point B..

The Upgrade.

Upgrade Planning.

The Plans.

[H]ard Economics of The Plan.

Conclusion.

Decision.

[H]ard Economics of Video Card Upgrades.

Epilogue.

Appendix.

Game Benchmarks.

----------------------------------------------------------------------------------------------------------------------------

Do not go gentle into that good night,
Old age should burn and rave at close of day;
Rage, rage against the dying of the light.

Though fast video cards at their end know dark is right,
Because their ports had forked no lightning they
Do not go gentle into that good night.

Good cads, the last wave by, crying how bright
Their frail renders might have danced in a green bay,
Rage, rage against the dying of the light.

Wild cards who caught and painted the sun in flight,
And learn, too late, they grieved it on its way,
Do not go gentle into that good night.

Grave cards, near death, who see with blinding sprites
Blind eyes could blaze like meteors and be gay,
Rage, rage against the dying of the light.

And you, my 8800 GTX, there on the sad height,
Curse, bless, me now with your fierce shaders, I pray.
Do not go gentle into that good night.
Rage, rage against the dying of the light.

- Inspired by a poem by Dylan Thomas.

----------------------------------------------------------------------------------------------------------------------------

Introduction

Do I need to upgrade my graphics or not? How do I go about deciding that? How much will it cost me? What will I get in return? Will my wife kill me for buying another expensive video card? Will it be worth it?

If you ever pondered about these kinds of dilemmas, then I will try to answer them in this article using some concepts that I call the [H]ard Economics of Upgrades. In this case, of a video card upgrade.

While this article is based on my system and upgrade, the ideas and concepts are applicable to many other upgrade situations. Hopefully even yours.

But lets start with my potential upgrade.

----------------------------------------------------------------------------------------------------------------------------

Background

On September 15th, 2008, my EVGA 8800 GTX celebrated its 1st birthday. Is it time for it to go gentle into that good night?

On the day of purchase, the card cost me $549.99 plus tax which in Ontario, Canada equaled to $626.99. Today, on craigslist I see one listed for $225. I wonder if it will sell, even at that price. At least now I can feel the pain NVIDIA felt when they had to lower the price of their GTX 200 series cards. Something about walking a mile in someone elses shoes.

My gaming rig, which is also my work system and DAW, has 3 LCDs. One Dell 3007WFP-HC, flanked by 2 Dells 2007FPs, for a total resolution of 4960x1600 of pixel heaven.

Needless to say, this rig has given me countless hours of bliss. Gaming on a 30 monitor with a resolution of 2560x1600 is just awesome. However playing games at this high resolution is also very demanding on a graphics subsystem

After 1 year since I bought this card, there have been lots of new video cards that have appeared on the market. As a matter of fact, the price\performance landscape has changed considerably on graphic cards in this past year.

As most of you know, the latest top end cards, like the NVIDIA GTX 280, or ATI HD 4870x2, are insane performers. Many of you have already said that the only people that need cards like these are those that are running at an insane 2560x1600 resolution.

Hey, I am running at that so called insane resolution. Then, it only makes sense that I must upgrade my current graphics card to one of these latest, made for 2560x1600 video cards.

Sounds easy enough. I just do a quick search and find a deal on a nice Sapphire HD 4870x2 2G DDR5 PCI Express 2.0 Video Card for only $569.95 plus taxes, which will burn a $644.04 hole in my pants.

Uhmm, maybe not so fast then, hey? As a matter of fact, I recall [H] saying that my not so young 8800GTX single-GPU video card is providing the performance in todays games to make 2560x1600 a reality.*. Of course that was Brent Justice talking back in November 8th 2006.

* http://enthusiast.hardocp.com/article.html?art=MTIxOCwxNywsaHVudGh1c2lhc3Q=

Besides, ultimately I need to decide whether my system is fast enough to run the games I actually play, or plan on playing. And if not, then what I should replace it by.

If you will have to, or had to consider, or did a video card upgrade, then read on, and answer the poll when finished.
----------------------------------------------------------------------------------------------------------------------------

The GeForce 8800 Series

The 8800s like mine, aka G80, was launched on November 8th 2006. This release not only included the GeForce 8800 GTX but also the 8800 GTS

NVIDIA GeForce 8 series graphics processing units (GPUs) redefined the PC gaming experience. With a revolutionary unified architecture and full support for Microsoft DirectX 10 games, GeForce 8 series GPUs delivered unprecedented performance, extraordinarily detailed environments, and film quality game effects.

Features included the following:

· Full support for both Microsoft® DirectX 10 and DirectX 9 for unparalleled levels of graphics realism and performance

· NVIDIA PureVideo HD2 technology provides unsurpassed Blu-ray and HD DVD movie picture quality

· NVIDIA 8 Series graphics processors are essential for accelerating the best Microsoft® Windows Vista experience

For a thorough [H] review of the 8800 series on their release, take a look here: http://enthusiast.hardocp.com/article.html?art=MTIxOCwxLCxoZW50aHVzaWFzdA==

--------------------------------------------------------------------------------------------------------------------------

EVGA GeForce 8800 GTX

My 8800GTX is model 768-P2-N831, which is in essence a stock reference design 8800 GTX

The following is a list of its major specifications:

· Core Clock (MHz): 575
· Shader Clock (MHz): 1350
· Memory Clock (MHz): 900
· Memory Amount: 768MB
· Memory Interface: 384-bit
· Memory Bandwidth (GB/sec): 86.4
· Texture Fill Rate (billion/sec): 36.8

Some of the key features are:

· NVIDIA unified architecture with GigaThread technology
· Full MS DirectX 10 Shader Model 4.0 support
· NVIDIA SLI Ready
· 16x full-screen anti-aliasing
· True 128-bit floating point high dynamic-range (HDR) lighting
· NVIDIA Quantum Effects physics processing technology
· Two dual-link DVI outputs supports two 2560x1600 resolution displays
· NVIDIA PureVideo support
· OpenGL 2.0 support
· NVIDIA nView multi-display technology
· Built for MS Windows Vista

----------------------------------------------------------------------------------------------------------------------------

The Competition

When deciding on hardware upgrades, there are many approaches we can take.

There is the Super Enthusiast way, or aka heatlesssun way. Just buy the best there is, price be damned. Take a look at the thread GTX 280 3x SLI - 3 Teraflops and/or 3 terrible flops? (It is a very [H]ard article on the very best of the best.

Here at [H] you will find a lot of Enthusiasts. Their rigs are typically built using mostly the latest and the greatest. If there is a new component on the market, like a graphics card, they will typically buy it to update their systems.

The main difference between an Enthusiasts way and a Super Enthusiasts way is that the non-Super variant way, will typically based on some criteria. This could be because it offers significant improvement of gaming experience, or superior price\performance compared to current setup combined with a good ROI for sale of existing component. And off course they will typically be constrained buy some budget amount. These folks will typically have an 8800 Series in SLI, GTX 280, or a HD 4870x2.

BTW, there is a subgroup of the Enthusiast group, which is the Married Enthusiast subgroup. This is my group, and when a budget is calculated, it must then be multiplied by (# of children + 1+1) X (item cost). So a video card of $500 will require a budget of (3 + 2) X $500, or $2500 for a family with a 3 kids. If you dont understand why, then dont worry, you are not a Married Enthusiast. But think top of the line washer and drier, cause it is for the whole family, and each member deserves to spend as much in this house, and YOU do want your laundry done, right?

Another large group here at [H], is the Bang for the Buck Enthusiast. They will typically wait for enthusiast class hardware to come down in price, or buy 2nd hand, or just below top of the line hardware. Today you will see them with video cards like 7900 series, 8800 series, 9600 SLI, HD 3800 series, HD 4850; or even a GTX 260, or a HD 4870. Some might even have older hardware that was once enthusiast class and can still play their games at the resolutions of their displays. But most importantly, they are happy with it, and see no need to upgrade it.

And last but not the least, there is the non enthusiast. If you are one, and you are reading this, stop and go listen to iTunes, play Spider Solitaire or game at www.miniclip.com.

Since I am a married enthusiast, we can assume that we will want to look at graphic cards that will give us higher performance. So the initial list of candidates for this potential upgrade will look something like this:

· NVIDIA: GTX 260, GTX 280
· ATI: 4850, 4870, 4870X2

Let the games begin!

SpeedyVV · Sep 17, 2008

Current Gameplay

I like to think strategically, and the essence of a strategy, is to create a plan. This plan will take you from Point A to Point B. Point A is where you are today, and Point B is where you would like to be at the end of this plan.

So lets look at where I am today by benchmarking my system against the games I play. After that I can decide what Point B will look like, and how to get there.
----------------------------------------------------------------------------------------------------------------------------

Current System Setup

For my test system platform I am using my setup exactly like I have it on a day to day basis. I am using an EVGA 680i SLI motherboard with an Intel Core 2 Quad Q6600 processor at 3.15GHz (9x350MHz) and SpeedStepped to 2.1GHz (6x350MHz), an EVGA 8800 GTX at stock, a XFX 7600GT XXX at stock, and 2GB of Corsair Twinx DDR2 PC2-6400 also at stock.

While I have pre-defined low OC, med OC, and High OC configurations for both GPUs and CPU, most of the time I game with the low OC setup as described above.

I am using the 9.53 chipset drivers and the P30 BIOS for the evaluation. The above configuration, I have used for a year, and it is rock solid. The mobo has also been rock solid in my pursuit of CPU\GPU\Memory OC configurations. I do wish I had better cooling in my case, but dont we all?

----------------------------------------------------------------------------------------------------------------------------
Video Card Evaluation Setup

Both video cards in the setup were evaluated using the previously installed ForceWare version 175.16. While the system is a dual boot with Windows XP and Vista, the entire tests have been done on Windows XP Professional 32-bit with Service Pack 2. This configuration has proven to be stable over time for the games evaluated.

As a matter of fact, all these benchmarks we run without ever rebooting, restarting, or crashing the system. It took a few days, with XP taking the system to Sleep mode many times, the over clocked Intel Q6600 stepped down to 6x multiplier when not stressed.

I must be very lucky, because sometimes when I read comments on the [H]ard|Forum, you would think I am the only person with solid system built on a NVIDIA 680i SLI platform.

Along with the games being benchmarked, I will have my standard environment running. Things like Steam, Rivatuner, and M-Audio BlackBox control application.

Additionally, during benchmarking I will have, Intels TAT, Task manager, CPU-z, and Rivatuner Hardware Monitoring, and Microsoft Excel all available to support the benchmarking process.

----------------------------------------------------------------------------------------------------------------------------

Evaluation Setup

Please be aware that I tested my video card a bit differently from your typical video card review.

Since the tests are performed to confirm if my current video card is able to play my games on my main monitors resolution of 2560x1600 (high resolution) , 2 tests will be conducted: 2560x1600 at max eye candy, and at 2560x1600 with maximum eye candy that is still playable.

If the game has a built in canned benchmark, that will be used. For those that do not, 1 minute of game play will be performed with FRAPS capturing the Min., Max., and Avg. frames per second. At least 3 runs will be performed for each game to ensure that no anomalies are used, and a representative result will be posted.

The same tests will also be performed with full eye candy at the resolutions of 1920x1200 (medium resolution), and 1680x1050 (low resolution) for comparison purposes only since I only game at high resolution.

For consistency, Maximum eye candy is the maximum available through the games video options, and not using tricks like .ini changes, video driver settings, or any special download.

Remember, these tests are not about whether my 8800GTX is slower than the new cards. That is a given. It is to see if an upgrade is in order for my gaming needs.

I measured 8 games and other characteristics of my EVGA 8800 GTX and documented the details in the Appendix at the end of this article.

----------------------------------------------------------------------------------------------------------------------------

Game Play Summary

So how do we summarize game play? In essence, what is Point A in our upgrade strategy plan?

For my gaming needs, there are only 3 significant levels of performance:

1. It can not play my game at my displays native resolution

2. It can but without Max Settings

3. It can with Max Setting

So lets see how it adds up.

I need to mention that BioShocks evaluation is without AA. The Maximum setting is actually not the highest IQ the game can be played at. The game setting menu does not allow AA to be turned on. However you can use the NVIDIA Control Panel to force AA.

Considering that I game in high resolution on a 30 display, it is amazing that an almost 2 year old GPU can play all games I want to play at the displays native resolution of 2560x1600.

Also two of the games I am playing can be played at their maximum settings. One of which is my favorite game that I can play across 3 screens at a whopping resolution of 5260x1600. Nice.

----------------------------------------------------------------------------------------------------------------------------

Overclocking

I tend to not pay too much overclocking benchmarks I read in reviews. Every card has a different maximum overclock, so there is never any guarantees that yours is going to perform the same.

But even so, I do overclock my card when I game. It is basically what some call free performance upgrade. Usually nothing is free, and no exception here. Overclocking will cost in terms of heat generated, higher power consumption, and potentially stability and shorter MTBF. But most [H]enthusiasts will overclock.

The default clock speed is 576 MHz core and 900 MHz memory.
Using Rivatuner to overclock my 8800 GTX, I get a core speed of 625 MHz and memory at 1000 MHz.

This gives me about a 10% clock speed boost. While, this is not going to change game play drastically, it does take some games from being barely playable to a smoother game feel.

----------------------------------------------------------------------------------------------------------------------------

Temperature

If there is one thing I do not like about my system, is the heat it produces.

I want to strive for a system that is as cool as possible. What I mean is not only that has great cooling, but also that the components do not generate a lot of heat to start with.

Unfortunately, any enthusiasts rig is going to generate a lot of heat. There really is not much you can do about it.

One thing hardware manufacturers can to do better is reduce the amount of heat during idle or low load conditions. This is very important to me. I do not mind that a performance part generates heat when needed. But at idle? That is just wasted energy an off course extra heat.

According to Rivatuner, my GPU idles around 56C with fan at 80%, and during gaming anywhere between 75C-80C with the fan at 100%.

The video cards fan at 100% is noisy, but when I game, that is not the loudest fan in the system.

I would not call that a cool card.

----------------------------------------------------------------------------------------------------------------------------

Power

I have my rig and all 3 LCDs behind a APC Back-UPS XS 1500VA LCD.

One of the features is that it displays the power that is being output by the APC UPS.

The power reading for my system while idle is 400 watts. The equivalent of over 6 regular household light bulbs on for no purpose at all.

I guess I will not be winning any environmental awards.

The same system at full load consumes 490 watts. About 8 lights bulbs, but at least doing something.

To minimize wasted electricity, and heat generation I use the following power saving steps:

1. SpeedStep to 6x at low load

2. Turn of monitors at 10 minutes idle

3. Turn off hard disk at 15 minutes idle

4. Got to system standby at 45 minutes idle

Seeing that the system is drawing so much power, I am going to get more aggressive on those power saving settings.

But how much of that is the 8800 GTX?

To find out I am going to temporarily use only 2 monitors. 1 plugged in to the 7600 GT and 1 plugged in to the 8800 GTX. Measure idle and load watts, then remove the 8800 GTX and plug both monitors to the 7600 GT and re-measure.

At idle and during 3DMark0 Proxycon the 2 cards pulled between 333-439 Watts.

Using only the 7600 GT with 2 monitors the power drain was between 283-339 Watts.

So using this un-scientific measurement the EVGA 8800 GTX uses 50 Watts and 100 Watts in idle and load respectively. That is not too bad considering the age and performance of the card.

elleana · Sep 17, 2008

Absolutely awesome post.

SpeedyVV · Sep 17, 2008

Point A

We can now say where is our starting point, from a performance, stability, and power perspective.

This is actually a surprise, as I thought when I first considered this benchmark, that some games would not be playable at high resolution.

On top of that, the benchmarking was completed over the course of a week, with absolutely no stability issues. Nice.

Strategically speaking, this is an important point in [H]ard Economics: Perceived Point A NOT EQUAL Actual Point A.

The only way to determine Point A is to measure. The more factual information we can find out or gather the better.

Objective data is better than subjective. But just because we cant always get objective data, does not mean subjective data is useless.

For example, I can just subjectively state that my system is not state of the art because is it built mainly of 2 year old components. While I could go into detail here and find out objectively by examining my system in detail, I am happy with the subjective conclusion.

Trends are better than absolute data points. For example, if we were to measure GPU temperature over time we could make better judgments as to whether the card is going to be stable over the long term. If I were to trend my benchmarks over the games release date, I could possibly extrapolate the expected date where my card would no longer play the latest titles at native resolution.

Marketing folks love stressing your perceived Point A:

· You are to fat right now, just look at all these Supermodels and how you dont look like any of them.
· You dont have a chance of winning Race 07 unless you buy a $250.00 Logitech G25 wheel. Of course only to be beaten online by drivers with $25.00 gamepads, because they are better at setting up the cars.

This is why so many of us like [H]s review approach. It is the great lengths the reviewers go to give us objective measures of actual game play.

While they might not benchmark 10 games, they provide actual valuable data, instead of canned and potentially marketing driven measures that dont necessarily equate to actual game play.

So lets list my Point A (in order of priority):

1. I can use all my 3 monitors simultaneously when not gaming(Objective)
2. My system is stable (Objective Trend)
3. I can play all my games at my monitors native resolution (Objective)
4. I can play all my games at my monitors native resolution smoothly with some minor exceptions (Subjective)
5. I can play my favorite game with max eye candy (Objective)
6. I can play some game across all three screens (Objective)
7. There are some games that I wish could be played at higher IQ (Objective)
8. My system uses too much power when idling (Objective)
9. My system gets hot (Objective)
10. The are games that I wish had higher FPSs (Objective)
11. My system is not state of the art (Subjective)
12. My system is 16 months old, and video card 12 months old (Objective)
13. My system uses 2 video cards (Objective)
14. I can dual boot with Windows XP and Vista (Objective)
15. New games will likely not be playable at high resolution (Subjective Trend)

---------------------------------------------------------------------------------------------
[H]ard Economics of Point A

Defining Point A was a pretty straightforward technical exercise in measurement and observation.

In benchmarking, measurement seemed like a simple exercise, but in reality it only looks that way because we taken certain liberties.

We have taken many factors, such as accuracy, units of measure, and measurement tools for granted.

For example, lets just look at FRAPS. How accurate is it? When it says that the minimum frame rate was 35 frames per second, could it actually have been 36 instead? And what did it use to measure time? Is that clock accurate?

So for something that we marked as Objective there sure were a lot of assumptions.

What about the subjective measure in point number 4 of our list? That is an important one for us [H]enthusiast. How did we measure that? Actual game play off course. But just how long was my sample? What if it was just a rare slow spot in the game? And what is smooth for me might not be smooth for you.

Another thing to remember is that Point A can change even if we do nothing. Meaning that at any point in time, Point A is fixed, but because of external forces (or even internal) Point A can change over time. For example the system is aging even as I write this article. New products coming out will make my system less state of the art. New games that I want to play but cant at full resolution, will become available.

What about the person doing the measurement? Well, I am speaking of myself here. Why would I lie to myself? Did I not say I was a married enthusiast? Do I really want to tell my wife that I can play all my games right now?

Point A will have a huge impact on the ultimate decision of whether I upgrade my video card or not, and with a wife and 3 kids, $500.00+ is not something I want to take for granted.

This is actually a very interesting and profitable field. Yes, you heard it right. Measuring Point A as part of a strategic plan is something business, military, and science will spend a lot of money on.

So how [H]ard can Point A be? Think of this question, Are we ready to send our troops into war? Very hard core indeed.

Valset · Sep 17, 2008

elleana said:
Absolutely awesome post.

+1, And you need to upgrade to the 4870X2 stat !

SpeedyVV · Sep 17, 2008

Upgrade Options

Defining Point A was a simple straightforward technical exercise in measurement. Point B will be a much more interesting exercise.

This is where creativity, imagination, experience and knowledge will be very important.

While at any instance in time there is only one actual Point A, there are endless possibilities for Point B.

As pleasant as endless possibilities sound, it is not very manageable. So lets create some boundaries, or constraints. Things like: off the shelf technology, reviewed by [H]ardOCP, within a certain budget and provides significant value.

Lets further organize characteristics of our Point B options (or solutions) into the following 3 groups: Must Haves, Should Haves, and Nice to Haves.

If you think this is starting to sound like our requirements, then you thought right.

Lets look at our last paragraph in the Introduction:

So the list of candidates for this potential upgrade is:

· NVIDIA: GTX 260, GTX 280

· ATI: 4850, 4870, 4870X2

Now, how did I come up with that short list without thinking about the requirements?

I could even say that I am going to pick the 4870x2 because I read all the reviews that say that this card is the fastest single card out there, and I have enough cash flow to buy it.

Sure I could, but that would be not according to the concepts of [H]ard Economics.

I can almost hear the ATI fanboys screaming that I and am a NVIDIA fanboy and that [H]ard Economics is just another tool to make ATI look bad.

Ok fine. I could even say that I am going to pick the GTX 280 because I read all the reviews that say that the GTX 280 is the fastest single GPU solution out there, and I have enough cash on hand to buy it.

Again, I would not because of [H]ard Economics.

I can almost hear the NVIDIA fanboys screaming that I and am an ATI fanboy and you get the picture.

As a matter of fact, you are not going to find out if this is going to be a ATI or NVIDIA solution until the very end of the article.

One thing we do know is what Point A is, so lets define what Point B should look like.

But how do we do that?
---------------------------------------------------------------------------------------------------
The Goal

In the Current Gameplay section we mentioned that a strategy is in essence a plan that takes us from where we are to where we want to go.

The problem with that definition is that it does not talk about what makes a good versus bad strategy.

For example, in a RTS type game, we could choose to just nuke the enemy as our strategy to defeat them. Well that is fine and dandy in a RTS game if we have the nukes.

In real life, if the U.S.A. had chosen that strategy in the 1st Gulf War, things would have been a bit more complicated. Sure, it would have gotten the Iraqis out of Kuwait in a jiffy, but the consequences would have been disastrous.

It would have been a strategy, albeit a really bad one.

But lets use that well documented example, to try to understand some ideas behind what makes a good strategy.

In the United States Air Force (USAF) paper Battlefield of the Future: 21st Century Warfare Issues, chapters 4-5 provide 2 articles on the Future Airpower and Strategy Issues , including discussion on strategic principles.

To paraphrase, strategy is the art and science of translating objectives into practical plans and operations.

So it is important to accept that we are talking not only about science, but also art.

We need to not only plan, but also operationally execute that plan, driven by our desire to achieve and objective. This is an essential concept of [H]ard Economics. We need to define Point A and B in the context of a Goal! It tells us why we want to go to Point B.

Col John A. Warden III, who was a key planner and organizer of the allied air campaign that gave the Coalition air superiority over Saddam Husseins air force in the 1991 Gulf War, further states:

· Military operations must be conducted so as to give reasonable probability of accomplishing desired political goals at an acceptable price

· Military commanders, with the exception of a few really stupid ones, have always weighed costs as they were planning or conducting an operation

This leads us to 2 more concepts of [H]ard Economics:

· We must weigh Risk and the probability of success of achieving our Goal

· We must carefully manage cost not only during planning, but also during the execution of the plan

As an example, if our goal is to build a future proof system for gaming all new games over the next 5 years, what is the risk of not achieving our goal? Pretty high I would say.

Lastly, the cost of our upgrade is not only the cost of the Video Cards, but also the cost of additional energy consumption over the life of the card. Not to mention the potential cost for damage control with the wife. Remember, I am a Married Enthusiast.

In other words, our ability to act or not to act must be based on some kind of cost-benefit ratio.

Now that we have learned what we needed from the USAF, lets leave them to plan wars, and lets get back to planning our video card upgrade.

If you are interested about strategy, I do encourage your to read at least Chapter 4 of the USAF paper in its entirety.

With all we learned, lets map out our next steps.

1. We define our Goal for the upgrade.

2. We describe what Point B should look like using the metrics we used to measure and define Point A, that way we know we can measure if we successfully reached Point B once we set forth our plan, because we know that we can measure them.

3. Use additional requirements to ensure that Point B supports the achievement of our Goal.

Also, interestingly enough we then can define the value we get from the upgrade as the difference between Point A and Point B less the cost of executing our plan to get to Point B.

Lets go ahead and define our goal for the upgrade:

To upgrade the Video subsystem with state of the art Video Card(s), to support gaming at full 2560x1600 resolution, and at least with 4xAA/16AF and average 30 frames per second for all the games that I own, or that are currently released and I plan on purchasing before the Christmas Season, while maintaining all the functionality of the current setup and the cost is reasonable.

Notice that it is a SMART goal: Specific, Measurable, Attainable, Realistic, and Timely. That helps us minimize the risk of failure, which is one of our principles.

By specifying a goal, we now have an objective for Point B.

Now we need to describe Point B with our predefined metric in the context of our goal.

1. I can use all my 3 monitors simultaneously when not gaming(Must Have)

2. My system is stable (NH)

3. I can play all my games at my monitors native resolution (MH)

4. I can play all my games at my monitors native resolution smoothly (MH)

5. I can play all my games with max Quality Setting (MH)

6. I can play some game across all three screens (Nice to Have)

7. I can play all my games at 4xAA/16AF (SH)

8. My system consumes no more than 400 Watts while idling (SH)

9. My system should reduce heat production (NH)

10. I can play all my games with average of at least 30fps (SH)

11. The Video Cards subsystem is state of the art (SH)

12. The Video Cards should be no older than 6 months (SH)

13. My system uses 2 video cards or less (NH)

14. I can dual boot with Windows XP and Vista (SH)

15. My system should be able to play new games at 2560x1600 (NH)

Must Have are binary, meaning either it complies or it does not.

Should Have, are important but can have variable degrees of compliance, or measurability.

Nice to Have, are optional, of lesser importance, or difficult to be objectively measured.

Lets add some constrains and features that are important to me:

a) Solution must not cost more that $700.00 with Tax (Cash flow)

b) Solution must include only video cards and drivers (Scope management)

c) Solution must be Commercially Off The Shelf (COTS) (Time, Risk Management)

d) Solution must have been reviewed in [H]ardOCP (Risk Management)

e) Solution must be available at time of evaluation (Scope Management)

f) NVIDIA or Intel solutions are desired ($5.00, Risk Management)

g) Double Lifetime warranty is preferred ($10.00, Resale)

h) Lifetime warranty is desired ($5.00, Risk Management)

i) Step-Up Programs are desired ($5.00, Upgradability)

j) Single card solutions are preferred ($20.00, Supportability, Scalability)

k) Dual slot design are desired ($5.00, Performance)

l) Dual Double Linked DVI desired ($5.00, Upgradability)

m) Integrated HDMI preferred ($10.00, Upgradability)

n) Scalable solution preferred ($10.00, Scalability)

o) Nice looking desired ($5.00, Desirability)

p) Includes new game preferred ($20.00, Functionality)

q) Includes game I own desired ($5.00, Profit)

r) Tier-1 Brand preferred ($10.00, Risk Management)

s) Solutions with at least 768MB preferred ($15.00, Performance)

t) Fastest solution in the market preferred ($10.00, Performance)

The Must items are constrains, and will help us manage scope of solutions, and the other features will help us evaluate our options.

Kil4Thril · Sep 17, 2008

Holy shit, nice article, dude. You did your homework. Thank you. It was also very nicely written.

SpeedyVV · Sep 17, 2008

Value of Upgrade (Apples-to-Oranges)

In the last 10 years of working with sales executives (sales people), I have learned that the best ones, and also most successful ones, are the ones that best understand value for the client.

Think about it, they are able to understand the customers problems, positioned a solutions of great benefit, implementable with minimal cost and risk. That way the customer is more likely to pay a higher price, but will be satisfied because they will have a high probability of getting the benefit that is higher than the cost.

Sounds easy, but it is not. If you think you have that capability, I would suggest you change jobs, because great sales execs are one of the best paid people in most organizations.

Unfortunately I am not one of those great sales execs. So I have to rely on methods and processes to have a reasonably good probability of finding a valuable solution.

Lets just look at the Wikipedia definition of Value in marketing (http://en.wikipedia.org/wiki/Value_(marketing)):

Value of a product within the context of marketing means the
relationship between the consumer's expectations of product quality
to the actual amount paid for it. It is often expressed as the equation:

Value = Benefits / Price

Or:

Value = Quality received / Expectations

Or even:

Value = Benefits - Price

This value concept is very tricky. As we mentioned before, marketing folks are always trying to trick us into thinking something is more valuable than it is really is.

Even without marketing things can be complicated because of our biases.

In order to defend ourselves from making the wrong choices, we can apply another concept of [H]ard Economics: We must clearly define the evaluation criteria before we identify our potential solutions.

1. Define evaluation criteria in terms Must Have scoping and Value in Dollars from our metrics and priorities

2. Define candidate solutions for Point B

3. Rate candidate solutions against the Must Haves to scope out non qualifying solutions

4. Rate candidate solutions for Point B against the value criteria defined

Here are the two tables defining our Must Haves, and Value.

Notice how, in the Value table, I have made the arbitrary decision to assign $600.00 to the Benefit that a perfect card would give me while using it for 2 years, and that such a card would give me a net TOTAL cost of $600.00.

Also, based on my limited experience eBay, I have decided that a video card looses half its value during a year. Therefore at the end of two years, my $600.00 can be sold on eBay for $150.00.

Using the formula:

Value = Benefits Price

=> Value = ($600.00+$150.00) - $600.00

=> Value = $150.00

By the way, this $150.00 is the entertainment plus resale Value this card will give me at the end of 2 years less its original cost.

Now that we have this criterion, lets get a set of candidate solutions.

Our original set was:

· NVIDIA: GTX 260, GTX 280

· ATI: 4850, 4870, 4870X2

Considering our Goal, this is a pretty good list of state of the art cards that have a good probability of giving us the gaming performance we were looking at.

I have also decided to add to this evaluation 2 other cards that could have very large price performance advantage. I can buy another 8800 GTX for $200.00, or I can sell my 8800 GTX and 7600 GT for $250.00, and buy 2 8800 GTs for $219.98.

Also added the rumored 260 die shrink that has been rumored around the [H]ardForum.

What I would like to do now of course is to take my 2 tables and get samples of all solutions, and do an evaluation using an Apples to Apples style benchmarking.

This is exactly what most organizations would do. As an alternative they might send a formal benchmarking criterion and ask the various vendors to respond.

Well, I am just buying a simple video card. And while $500+ is a lot of money for me, vendors are not likely to lend me samples.

So I will have to rely on sites that I trust and make some judgment calls. This is what I call my Apples to Oranges comparison. Sometimes even a single site like [H]ardOCP will have some variance between reviews because of the nuances and approach of different reviewers.

From this scoping exercise the following candidates have been eliminated:

· our current setup and the single ATI 4850 solutions due to not being able to play at Max IQ

· all CrossFire solutions due to needing a new motherboard

· 280 SLI and Tri SLI because cost exceeds available cash

· 260b die shrink because it is not available for purchase

----------------------------------------------------------------------------------------------------------------------------

The Bottom Line $

Now that we narrowed our list, lets see which of the solutions does best?

So lets look at our table and some of its formulas:

TOTAL Benefit = ∑Features over 2 years

TOTAL COST = (Cost of Cards) (Sale of existing Cards no longer needed)

ReSale after 2 Years = Cost of Current Solution ÷4

TOTAL Value = (TOTAL Benefit + ReSale after 2 years) TOTAL Cost

Net Value = (TOTAL Value of Point B) (TOTAL Value of Point A)

----------------------------------------------------------------------------------------------------------------------------

Point B

So now we have our short list of Point B solutions and the dollar Net Value calculations, it is a simple exercise of filling the numbers and looking at the bottom line.

Unfortunately there is technically one more little step.

As we mentioned before, to go from Point A to Point B, we need to also to plan it out to insure we have a good chance of implementing our solution successfully.

Sometimes the cost of implementing a solution is greater than the actual capital cost of the solution, which could derail our whole initial value proposition.

So lets revisit our list of possible Point Bs after we look at planning.

[H]ard Economics of Point B

Maybe you just have not thought of it this way, but us enthusiasts are all about Point B. We push the technology so that tomorrow is better than today.

We think of crazy ways to use the technology. We mod. We fold, We bench. Hell, we even dump liquid nitro to make Point B just a little better.

And do you know what else is cool about a great Point B? When you nail it, we make friends, family, others, and even corporations act. We take their goals and objectives and we sell them our vision of what is possible.

When you understand goals, have knowledge and imagination, and some discipline, it can make things very interesting.

Obviously this approach is overkill for the purchase of a video card. But if you take away the formalism and use the ideas and concepts, youll be surprised how much of this you already do.

The point is, when you are going to do any upgrade, do it for a reason. Make sure you quantify and qualify that reason, so that you or others can find good options, and then make the right decision and not be swayed but tricky marketing tricks.

SpeedyVV · Sep 17, 2008

The Upgrade

Now that we have our list of Point Bs and a criteria to select one as the one, lets just cross our T-s and dot our I-s.

We do that by thinking about could happen when we go from A to B:

1. Draft plans to execute the journey from Point A to all Point Bs.

2. Ensure we have a risk mitigation strategy to increase the probability of success

3. Manage all possible cost from planning to execution

4. Make final decision

This is nothing more than treating the planning and execution of the upgrade from decision to implementation like a project.

If you are a project manager by profession, you already know how to do this. If you are not then this is beyond the scope of this article to explain it in detail.

But lets make things simple and just think this project through.

---------------------------------------------------------------------------------------------------
Upgrade Planning

A project has by definition a beginning and an end. For us that is Point A and Point B.

A project plan has a series of tasks and required resources (people, equipment, money) to take us from A to B.

For our upgrade, I am going to be the only people resource, and I will perform all the tasks for free so I wont worry about that cost.

But be careful with those kinds of assumptions. Things can go wrong, and they usually will at the worst possible time. Hence the need for a Risk Management approach.

What is a Risk you ask? Well it is anything that stands in the way of our success. For example, what if you when you install your shiny new video card, your Power Supply chokes because it is no longer capable of supplying the required amps on the one rail?

Well, you buy a new one right? Does that then change the cost benefit equation of your Point B? It does, but you are not sure, so it is not fair that you add that cost ahead of time, either.

And once we have the plan, what happens if things dont go as planned? Then you can add, remove or modify tasks to bring the project on track. Another approach is to change Point B taking into account the new reality. In essence we can steer the project to a new direction to try to minimize variance on time and budget.

One thing to remember is that no matter how much planning you do, you cant predict the future which takes us to our last two concept of [H]ard Economics:

1. Plan carefully so you know how to react when things go wrong, and not only when they go right. That part is easy.

2. Whenever possible have a contingency for when things go wrong (time, money, resources)

----------------------------------------------------------------------------------------------
The Plans

I am actually not going to show all the plans and their details in this article, but I am going to draft out a simplified plan to illustrate the key concepts, using the example plan for the 8800 GT SLI solution:

As you can see, when we map it out like this, it becomes easier to list the resources, risks, and potential hidden costs which could derail our upgrade.

You may even think that this level of planning is too much, but look around the forums and you will find lots of upgrade horror stories that could have been prevented with some very simple planning.

---------------------------------------------------------------------------------------------------
[H]ard Economics of The Plan

As a great project manager once told me, project planning is not about the project plan, but about the thought process that went into creating it.

Therefore the [H]ard Economics of the plan is about thinking before doing. Some will say measure twice, cut once.

Remember, all the best reviews and articles cannot think for you. It will inform you, but ultimately only you know the specifics of your Point A and Point B.

---------------------------------------------------------------------------------------------------

Conclusion

So we come to the point where we can use our process to execute this video card upgrade.

Lets summarize the steps:

· Determine where we are today by measuring and benchmarking Point A

· Determine where we can go for Point B:

o We define our Goal for the upgrade.

o We describe what Point B should look like using the metrics we used to measure and define Point A, that way we know we can measure if we successfully reached Point B once we set forth our plan, because we know that we can measure them.

o Use additional requirements to ensure that Point B supports the achievement of our Goal.

o Define evaluation criteria in terms Must Have scoping and Value in Dollars from our metrics and priorities

o Define candidate solutions for Point B

o Rate candidate solutions against the Must Haves to scope out non qualifying solutions

o Rate candidate solutions for Point B against the value criteria defined

· Plan the Upgrade:

o Draft plans to execute the journey from Point A to all Point Bs.

o Ensure we have a risk mitigation strategy to increase the probability of success

o Manage all possible cost from planning to execution

o Make final decision

Remember, what is important is not to formalize the steps into a strict process. What is important is that you understand the goal of each step, and have the discipline approach to you upgrade projects.

Each time you do an upgrade, try this out. See what works. Improve it each time you do it.

It will not guarantee you 100% success, but it sure will improve your chances of an upgrade that gets you your moneys worth and with less unpleasant surprises.
---------------------------------------------------------------------------------------------------

Decision

I promised you a decision and here it is.

A completely filled out decision table, and the bottom row will tell us which solution will give me the highest net value.

The winner by a slim margin is a NVIDIA GTX 280 coupled with the existing 7600 GT.

Here are the numbers for the winning solution:

The TOTAL Benefit of using the solution for the next 2 years is $455.00 out of a possible $600.00.

The TOTAL Cost is the cost of buying a GTX 280 less the sale value of the 8800 GTX which is $469.99 - $200.00 = $269.99.

The ReSale after 2 Years is the current cost of the cards divided by 4 or $469.99 + $50.00 ÷ 4 = 130.00.

The Total Value of our Point B is the (TOTAL Benefit + ReSale after 2 Years) TOTAL COST or ($455.00 + $130.00) - $269.99 = $315.01

The table also shows the TOTAL Value of the status quo or our Point A if we use it for another 2 years. It is $32.50.

This shows us that the Net Value of upgrading our EVGA 8800 GTX + XFX 7600 GT to a GTX 280 + 7600 GT is $315.01 - $32.50 or $282.51.

Now that I have talked the talk, will I be willing to walk the walk?

There you go:

---------------------------------------------------------------------------------------------------

[H]ard Economics of Video Card Upgrades

As you have seen, video card upgrades can potentially be about hundreds of dollars.

There are some strategic concepts that dictate how that money can be and should be spent. We called these the [H]ard Economics of Video Card Upgrades:

1. Perceived Point A NOT EQUAL Actual Point A. Measure Point A!

2. We need to define Point A and B in the context of a Goal.

3. We must weigh Risk and the probability of success of achieving our Goal

4. We must carefully manage cost not only during planning, but also during the execution of the plan

5. We must clearly define the evaluation criteria before we identify our potential solutions.

6. Plan carefully so you know how to react when things go wrong, and not only when they go right.

7. Whenever possible have a contingency for when things go wrong

Thank you for reading this far.

This article took you trough a very long journey of an upgrade to highlight many techniques required in the acquisition of computer equipment:

· Measurement

· Value

· Return on Investment

· Requirements Management

· Strategic Thinking

· Selling

· Project Management and Planning

· Risk Management

Each of those subject areas were only touched on this article, but hopefully you got to see their application in something we enthusiast do all the time

This article is a give back to [H]ardForum after my 1st year here. If you enjoyed it let me know, and I will try to contribute some other stuff that I have in mind.

SpeedyVV · Sep 17, 2008

Epilogue

Point A

Benchmarking Point A

Selling Point A

Interim 2560x1600 with XFX 7600 GT XXX

Back up arrives

Point B

psyshack · Sep 17, 2008

WOW!!!! All that brain power to buy a video card.

Cost of card $269.99. Cost of post pricelss, I think, maybe, hell I don't know.

Enjoy the new card! WOW,, who knows.

Atech · Sep 17, 2008

TOO much information...it's a videocard, not a house.

Bigbacon · Sep 17, 2008

wow, it equates to what a lot of people say all the time. Buy the best single card/core solution you can. (that may have changed with the X2 though, not sure)

AthlonXP · Sep 17, 2008

Lol it looked like an econ project.. Anyhow the GTX280 seemed to be your best overall option as it was a pretty much straight forward upgrade.

SpeedyVV · Sep 17, 2008

Appendix

Game Benchmarks

Bioshock

BioShock, according to 2kgames:

BioShock is the "genetically enhanced" first person shooter that lets you do things never before possible in the genre: turn everything into a weapon, biologically mod your body with plasmids, hack devices and systems, upgrade your weapons and craft new ammo variants, and experiment with different battle techniques in an incredible and unique underwater city.

I played BioShock in the middle of the Apollo Square Level. I run around using all sorts of weapons and shooting at anything that moves for 1 minute.

All eye candy that is possibly turned on in the games graphics option panel is turned on

The game was totally playable at the native resolution of the monitor. However, there were times that a little bit of choppiness would be encountered, but nothing that would stop the game from being playable.

At the medium and low resolutions (1920x1200 and 1680x1050), the game felt smoother, so obviously the game could have been better at high.

For a detail review of BioShock DX10 and IQ, including benchmarks with the 8800 Ultra and 8800 GTX, look at the September 05, 2007 article on [H] here

----------------------------------------------------------------------------------------------------------------------------

Image Quality

At the high resolution, with all eye candy turned on, BioShock looks amazing. While I did not enjoy this game as much as I thought I would, the visual quality of this game just kept making me come back.

I think I will have to play this game again.

----------------------------------------------------------------------------------------------------------------------------

Call of Duty 4: Modern Warfare

Call of Duty 4 is a pretty cool 1st person shooter game. I specially like being on a gunship raining hell on a bunch of poor souls.

My 1 minute of game play took place in Game Over, Day 6 in the Altay Mountains, Russia. Our jeep is pursued buy a bunch of trucks with occupied by gunners and rocket launchers. The benchmark is taken as I dispose of the nuances.

So CoD 4 will not run at high resolution and be playable with 4xAA. As a matter of fact to be honestly playable I had to turn off AA all together. Doing so made the game completely smooth, with just the odd slow down.

For a comparison of a 9600GT OC SLI vs. a 8800 GTX to play CoD 4 take a look at this article in [H]

----------------------------------------------------------------------------------------------------------------------------

Image Quality

At 2560x1600, sometimes you can get away with no AA. Some will go as far as saying that it is not needed, but I dont agree with that general statement.

However in my opinion CoD does look ok with no AA compared to other games. Maybe it is the pace of the game, but I am not sure.
----------------------------------------------------------------------------------------------------------------------------

Race Driver: GRiD

GRiD is a follow up to DiRT, but instead of being a off road based racing game, it is instead based on road courses. Being an Arcade style game, it is a lot of fun, but of course you are not getting the realism of a racing sim game.

Unlike what you might expect from an arcade style game though, GRiD looks amazing. I will say that it is without a doubt the best looking game I own, period.

The clouds, asphalt, 3D crowd, red hot glowing brakes, haze, and smoking tires, all add up to give a mind blowing visual gaming experience. If you have not played GRiD, you have not seen what the best looking graphics in games today. I have not seen Crysis, so I guess I could be wrong.

I keep wishing that [H] use this game as part of their video card benchmarking suite because it has a whole different dimension for game play feel. With a steering wheel like the Logitech G25, nuances on the properties of your gpu will affect playability.

Unfortunately, as per Kyle, the game just is not demanding enough for the current performance capabilities of top end video cards. Unfortunately, I dont have one of them to confirm that he is right.

Playing a single race with a full field of cars, I wait for the red lights to come on, then I press F11 to start FRAPS benchmarking. The only difference is that after the 1 minute of game play, I never stopped until the end of the race because it is so much fun.

Unfortunately the benchmarks for the high resolution are not the using the max eye candy. With the Overall Detail at ULTRA and max AA at 16XQCSAA (whatever that means) the game would just crash. So for GRiD, I used ULTRA setting with 8XCSAA (again, whatever that means) instead for all resolutions.

At 2560x1600 the game was just unplayable. It is a shame too, because while the cars were all lined up on the grid, it was just stunning. Oh well. To be playable the game was set at MEDIUM detail and 0xAA.

Even at the medium resolution, while completely playable, the steering did not feel 100% responsive, but it was playable.

When played at the low resolution, the game was very responsive, but under the sounds of the engines screaming, I could hear the voices of the developers saying You are running our beautiful game! Please stop, please!.
----------------------------------------------------------------------------------------------------------------------------

Image Quality

GRiD is supposed to have fast game play, damage deformation physics, realistic and aggressive driver AI, and off course stunning visuals.

Using the Dells full 30 potential with AA turned on is just out of sight. Unfortunately for me, that is not possible while keeping the other characteristics of the game.

While I have to set the game to MEDIUM with 0AA for the game to be playable, I actually will play it with Multisampling set to 2 and HIGH quality setting. In racing, one of the big things is the racing line, and if that line is full of jaggies then it will suck. So I will put up with a sluggish response with 31.5 fps. I just deal with the fishtailing using my superior driving skill.
----------------------------------------------------------------------------------------------------------------------------

Race 07: Official WTCC Game

Speaking of driving skill. If you have any, Race 07 is one of the games to try. There are other good racing sims, but Race 07 has better graphics and less of a learning curve.

Race 07 according to SimBin:

REAL CARS, REAL RACING;

That's what its all about in RACE 07. Enjoy 9 different racing classes, each with its own distinctive character. From the prestigious World Touring Car Championship to the insane speed of the F3000 class. These are just a few of the many cars and classes featured in this ultimate racing sensation.

Hit the road on one of the 32 real-life racing circuits or city tracks and see the track through the unique Helmet Camera or with the immersive Action Camera. Enjoy RACE 07 and experience virtual racing like you've never experienced it before.

The French street circuit of Pau, with its tight turns and hard walls make for some really exciting and skillful racing. Again I benchmark from the red light for 1 minute of racing.

This game is obviously not very demanding on the GPU for todays standards, so why did I benchmark it?

The 1st reason is that this is my favorite game bar none. So in order to decide if my card is a keeper, this game had to perform flawlessly. And for a simulation game that means at least 60fps. So the 8800 GTX does not disappoint.

The second reason is on the 4th line of the benchmark table. Did you notice? Yes, thats right. 5260x1600!

That is an insane 8.4 megapixels that the 8800 GTX has to render, then the system has to split the image into 3 parts, discard 150 horizontal pixels per bezel, and send the 2 side images across the PCIe bus to the 7600 GT so that it can simultaneously display the images on the 2 side panels, and the 8800 GTX the centre panel.

If you have never seen this in action, take a look at my video here
----------------------------------------------------------------------------------------------------------------------------

Image Quality

Race 07 is about realism and less about visual appeal. Yes GRiD is much more better visually, but Race 07 with high resolution, max settings, and 6AA is no ugly duckling by any means. It is just that you will be too busy driving to notice.

Besides, since this game is less demanding on the GPU, I can run it Triple Head Surround Gaming mode with softTH, http://www.kegetys.net/SoftTH/ for free. Just make sure you dont try that online in multiplayer mode, because the other drivers will have you for lunch. 1/10th of a second slower per lap and you are toast.

evilcartman · Sep 17, 2008

Damn nice article. I never took any business courses in college (^ Education major) so reading your article was very informative, for me at least. Is it something I'd go through when I upgrade my video card? Probably not, but it was interesting to read about the nuances of choosing a new card and weighing risks/benefits/etc. I know in past upgrades I have gone through something similar to what you described (through my research on new parts), but it was just for the purpose of finding the best bang for the buck.

magoo · Sep 17, 2008

Interesting process.

But the one flaw is selling the 8800 GTX for 200 dollars.
Given: the new 260 GTX is selling for near 200 dollars now, and is a better card; and the HD 4850 is selling for under 150 dollars...............

I think you're hosed.

I had trouble selling two of my 8800 GTX OC versions for 150 each BEFORE these new cards were released.

Anyway, good article with alot of thought into it.

SpeedyVV · Sep 17, 2008

Frontlines: Fuel War

Frontlines: Fuel of War is an open-world First Person Shooter set in the next Great War. In a world ravaged by a global energy crisis, environmental decay, and economic depression, players assume the role of an elite soldier in the Stray Dogs strike force in the Western Coalition Army on an epic crusade against the Red Star Alliance to control the last of the worlds oil reserves in the Caspian. Frontlines: Fuel of War combines intensely cinematic, non-linear game play, with next generation firepower and a revolutionary frontline combat system to deliver the most thrilling and engaging FPS to date.

That is what THQ, the publisher of Frontlines had to say about the game here

A bit boring is what I have to say about it. The graphics are ok, but I really have nothing else to say about this game, other than I am benchmarking it because I have not finished playing it yet, and I was too dumb to pay for it before reading a review. I was at the store and liked the box, so now I guess I will finish it.

Anvil, in chapter 3, where you have to Hold The Town, is where I go to the roof of the HQ, wait for incoming forces to show up, get on the machine gun and rain hell on the enemy for 1 minute.

So this game is not playable at the highest possible Overall Quality settings. At Low setting the old 8800 GTX can still play this newer game at the highest resolution.
Honestly, not that I care.
----------------------------------------------------------------------------------------------------------------------------

Image Quality

It looked fine. Frontlines proves that good visuals is not enough to provide a good gaming experience.

Race 07 proved the reverse, a great game does not require the greatest visuals.

Note to self: Never spend more than $9.99 on a game without 1st reading a review.

----------------------------------------------------------------------------------------------------------------------------

Assassins Creed

Jerusalem, 1191 AD.

The Third Crusade is tearing the Holy Land apart. You are an elite Assassin sent to stop the hostilities by suppressing the powers on both the Crusader and Saracen sides. But as you carry out your missions, a conspiracy begins to unfold. You find yourself tangled up in a conflict that threatens not only the Holy Land, but the entire world.

"Nothing is true, everything is permitted

Ubisoft Montreal created a pretty stunning game. Mind you, it took me a bit to get used to this game. I kept waiting for something to happen. Maybe a cyber dude from space to attack me, or a portal to open and drop me into some shooting mayhem, but it did not.

Once I got used to the pace of the game, it became much more enjoyable. As a matter of fact, having time to look around and enjoy the graphics while slicing a guys throat every now and then was rather therapeutic.

I entered the Animus DNA Memory, then went into a crowded area, used my sword to kill some innocent folks, which caused a lot of guards to show up. That is when the benchmarking started. Just me taking on a whole bunch of guards at the same time.

The performance of this game is kind of interesting. 1st off, when you set the resolution of 2560x1600 you cannot change the default Multisampling setting of 1. While it is possible to mess with the .ini files to force a change, I did not bother. The game was barely playable at this setting.

From the numbers, it would seem that an Average of 32.8 would be unplayable. However, due to the nature of the games pace, anything above 25 is acceptable, yet I still tried out the high resolution with Shadows: 1/3, Graphic Quality: 2/4, and Level of Detail: 2/4. That made the game much smoother, instead of barely playable.

----------------------------------------------------------------------------------------------------------------------------

Image Quality

This is one of the games that makes me get a more powerful video card.

Not being able to turn on AA while slowly walking through the games gorgeous scenes, then all off a sudden you see these columns full of jaggies, is just criminal.

If someone tells you that AA is not needed at 2560xx1600, well show then this screen shot. It just kills the hard work the developers and artists went through to provide us with all this beauty.

Setting the resolution at 1680x1050 did allow me to set Multisampling to 3/3, and made the Aliasing go away and still provided a smooth average frame rate of 48fps. But no, I am not going to play this game at this resolution. I refuse. I will store the DVD away until I get a card that makes the game playable at my monitors native resolution with AA.

----------------------------------------------------------------------------------------------------------------------------

F.E.A.R.: First Encounter Assault Recon

Activisions description:

You aren't a soldier. You are a weapon. A paramilitary force infiltrates a multi billion dollar aerospace compound taking hostages, but issuing no demands. The government responds by sending in its best special operations teams, only to have them obliterated. Live footage of the massacre is cut short by an unexpected wave of destruction that leaves military leaders stunned and in disbelief.

My description: The best first person shooter I ever played. Period, end of sentence.

Someone please tell me there is something better out there. I like FPSs. And while I am saving this game so I can play it again, along with both expansion packs, I am hoping, like all gamers, for that next game that provides us with that thrill of an awesome game.

This is still the only game that made me literally jump off my seat. Twice. That Alma, man.

I was not going to benchmark this game, but decided that while it is an old game, I recalled that I was not able to play it at high resolution with full eye candy.
All I did was run the canned benchmark and play a little of the game to confirm the results.

Such an old game, yet it is amazing that the 8800 GTX cannot play at Maximum Graphics setting and high resolution without any slow downs or stuttering. It was playable most of the time, but in really demanding scenes you could still be killed because of the performance.

With 2xAA and Soft Shadows off, the game was completely playable, but during multiplayer action, I would still need to lower some other setting in order to stay alive.

----------------------------------------------------------------------------------------------------------------------------

Image Quality

Why does this old game look so good to me? Even with only 2AA and Soft Shadows off, the game still looked amazing.

In single player mode the 8800 GTX can perform up to par and look good at the same time. In multiplayer, it can almost perform but not 100%.

Yet, I dont think I would play this game online anymore. But I will definitely play it single player.

So no need to upgrade my card for F.E.A.R..
----------------------------------------------------------------------------------------------------------------------------

World in Conflict

World in Conflict is the only game that I got excited about buying it. I got the Collectors edition that comes with a piece of the Berlin Wall, a documentary DVD, "Modern Marvels: The Berlin Wall" by The History Channel, and packaging featuring the Stars and Stripes, and The Sickle and Hammer.

I am old enough to remember the Cold War. I remember seeing the Berlin wall coming down in total disbelief. So I get a kick out of the Red box sitting on my shelf, and seing my kids watching the DVD is total awe. That alone was worth the cost of the game.

But lets see if my system will play it using the built-in benchmark.

My system just is not able to play this game with Graphics Detail set at Very High, This is DX9, because if I want DX10 enabled to see those beautiful God Rays and shadows from clouds it would likely bring the game to a total crawl.

If I lower the detail to Medium, then the game is playable. At this level, AA is turned off.
I think I am starting to see a pattern here.

----------------------------------------------------------------------------------------------------------------------------

Image Quality

The built-in benchmark has a gorgeous nuke going off, and a fantastic carpet bombing scene. This accounts for a lot of the lowest frame rates in the benchmark. Mind you, in real game play those two scenarios can be playable because you basically wait and stare at the beautiful destruction.

I bought the game about the same time I bought my 8800 GTX, which is to say about a year ago. But believe it or not, I have yet to finish the game.

I find it too confusing to play it in bits, like I play most games. So I will wait till I have some time without distraction to give it the attention it deserves, and the video card to be able to play it with full detail, 4xAA, and in DX10 to see those shadows.

This is one of the few games that I have installed in Vista (32-bit). I wanted to see if the DX10 features would make a difference. While not drastic, I think the game should be played in DX10.
Maybe my next video card will let me do that.

SpeedyVV · Sep 17, 2008

magoo said:
Interesting process.

But the one flaw is selling the 8800 GTX for 200 dollars.
Given: the new 260 GTX is selling for near 200 dollars now, and is a better card; and the HD 4850 is selling for under 150 dollars...............

I think you're hosed.

I had trouble selling two of my 8800 GTX OC versions for 150 each BEFORE these new cards were released.

Anyway, good article with alot of thought into it.

Actually I just sold my 8800GTX last week for US$190.00 + US$20.0 S&H.

So that equates to $204 Canadian plus shipping.

I actually had the 8800GTX @ $150.00 but updated to the new amount based on actual sale on eBay.

gmkmay · Sep 17, 2008

Amazing post, great read. Thanks for sharing

Valset · Sep 17, 2008

SpeedyVV said:
Actually I just sold my 8800GTX last week for US$190.00 + US$20.0 S&H.

So that equates to $204 Canadian plus shipping.

I actually had the 8800GTX @ $150.00 but updated to the new amount based on actual sale on eBay.

I just listed one for 150 to buy now, I may have to up that

SpeedyVV · Sep 17, 2008

Valset said:
I just listed one for 150 to buy now, I may have to up that

Yes you should.

Now you know why I started the article with the 8800GTX raging against the dying of the light.

There a lot of demand for that card. my was sold in Buy It Now and had 12 Watchers.

Good luck.

lowteckh · Sep 17, 2008

I had a bit of thinking to do in relation to this as well when I was upgrading from a 8600GT. Had the choice of keeping it but lose the value later, or purchasing a 9600GT($140) and selling the 8600GT ($80), then selling the 9600GT later (estimated at $80 at later date) which brought the cost of ownership to acceptably low. Girlfriend was against it at first but then (kinda) saw how logical it was (or to try and shut me up). But then.. after the 9600GT I couldn't stand not having something new so now I've got 2 4850s, throwing all that before me into the trash. Impulse always triumphs

magoo · Sep 17, 2008

I still can't believe anybody in their right mind would buy a two year old 8800 GTX for 150 dollars when there are cards two generations newer for nearly the same price????

And..........I forgot...........TigerDirect has the 8800 Ultra new for 199.

Man, I sold my two on this forum, probably got a lower price because "we" know the value and look at the cutting edge stuff all the time.

PT Barnum still lives on ebay........you know, a sucker born every minute

and I wanted to sell mine before the bottom fell out.

flegg · Sep 17, 2008

all those nice electronics and you have your keyboard and mouse on a pull-out.....

SpeedyVV · Sep 17, 2008

psyshack said:
WOW!!!! All that brain power to buy a video card.

Cost of card $269.99. Cost of post pricelss, I think, maybe, hell I don't know.

Enjoy the new card! WOW,, who knows.

Thank you. Just started checking if the system is stable. If so, going to play some GRiD finally with AA.

Atech said:
TOO much information...it's a videocard, not a house.

Sorry you did not get something out of the article.

Bigbacon said:
wow, it equates to what a lot of people say all the time. Buy the best single card/core solution you can. (that may have changed with the X2 though, not sure)

True, but that single GTX 260 came within less than $5 of Net Value from being on top. Which means that based on my choice of feature value that is an awsome bang for the buck rig.

Last but not least for those on a budget the 8800 GT in SLI is an awsome bargain for those of us with SLI mobos.

AthlonXP said:
Lol it looked like an econ project.. Anyhow the GTX280 seemed to be your best overall option as it was a pretty much straight forward upgrade.

The thing is, I had already been sold on the 4870x2. I "wanted" the fasted card so I was actually surprised when the GTX 280 came on top.

flegg said:
all those nice electronics and you have your keyboard and mouse on a pull-out.....

You got a point

What would you suggest?

knob · Sep 17, 2008

Very impressive [H]ard work there. When you work out the scientific method like that, you really can crush fanboyism without mercy.

Once again, daggone fine work - way to back up your purchasing decision. Don't listen to those who suggest you've done too much work. Poor decision-making leads to all kinds of issues, including wife-aggro

flegg · Sep 17, 2008

SpeedyVV said:
You got a point What would you suggest?

I feel very handicapped when using those things. If your desk is deep enough you can just put the keyboard and mouse on top, given you can raise the armrests on your chair to that level. Doesn't look like it's deep enough though.

I looked forever for a desk and ended up just getting one of these:

http://www.1stfoldingchairs.com/plastic_folding_tables.htm

They are friggin awesome for the money. I even have a 100 pound CRT on mine.

SoAndSo · Sep 17, 2008

seems like way too much to get into.

Spyhawk · Sep 18, 2008

The thing is, I had already been sold on the 4870x2. I "wanted" the fasted card so I was actually surprised when the GTX 280 came on top.

Hope you have fun with your new card. Excellant article and well thought out purchase decision. Personnally even with all that info, I would of still gotten the 4870X2. Its simply too hot a deal. Its the fastest card on the planet for well under what other high end cards in the past sold for like the 8800GTX, 8800Ultra, and the GTX280(when it launched)

QUOTE]

Syntax Error · Sep 18, 2008

A very interesting read, especially through a fundamental economic analysis of determining if upgrading your computer's marginal cost is outweighed by the marginal benefit of upgrading.

Love the subtle humor in your charts, BTW.

alg7_munif · Sep 18, 2008

G25 awesome wheel.

SpeedyVV · Sep 18, 2008

Spyhawk said:
Hope you have fun with your new card. Excellant article and well thought out purchase decision. Personnally even with all that info, I would of still gotten the 4870X2. Its simply too hot a deal. Its the fastest card on the planet for well under what other high end cards in the past sold for like the 8800GTX, 8800Ultra, and the GTX280(when it launched)

Funny you said that the 4870x2 "Its simply too hot a deal".

The reason I ended up not buying it was because the card runs too hot. If you look at the value table thats is where it lost most of the points.

I absolutely think it is the most awsome card, except I think I would have regreted once it would start heating my basement office. Maybe I can make it like tires. Summer = GTX 280; Winter = ATI 4870x2

FadedSpark · Sep 18, 2008

Just so you know, EVERY SINGLE SLI SOLUTION CAN NOT RUN 3 MONITORS (Not shouting, just hoping you update an otherwise FAN FUCKING TASTIC article

)

SpeedyVV · Sep 18, 2008

0mega said:
Just so you know, EVERY SINGLE SLI SOLUTION CAN NOT RUN 3 MONITORS (Not shouting, just hoping you update an otherwise FAN FUCKING TASTIC article )

thanks for the compliment

yes I do know that SLI is single monitor. And the triple monitor criteria as stated in the article is when not gaming (ie multi gpu disabled).

But I do agre that the table is misleading. I shortened most of the text for tables, but this one should not be shortened. I will fix it.

EDIT: Fixed. I hope it is clear now.

Tekkaman · Sep 22, 2008

I think something that would actually complete this article here would be an Excel spreadsheet that has fixed values for each card's points throughout the article under a standard system setup. The reason for this is because then we can have both an ongoing list as well as one that is constantly updated by the document holder(s) so that the information is never useless.

What I mean is that by doing an Excel sheet through Google Docs we could come up with a great way to show how cards will perform under most setups. People would be able to run a benchmark that is standardized with instructions on what resolution and all the other jazz and post their results to help harbor a fan-base, real world benchmark that would allow people to get a better idea of what their card is capable of. With these ratings, values, or even check boxes we can deduce what card is good for a setup should someone not truly know what they're looking for.

SpeedyVV · Sep 24, 2008

Tekkaman said:
I think something that would actually complete this article here would be an Excel spreadsheet that has fixed values for each card's points throughout the article under a standard system setup. The reason for this is because then we can have both an ongoing list as well as one that is constantly updated by the document holder(s) so that the information is never useless.

What I mean is that by doing an Excel sheet through Google Docs we could come up with a great way to show how cards will perform under most setups. People would be able to run a benchmark that is standardized with instructions on what resolution and all the other jazz and post their results to help harbor a fan-base, real world benchmark that would allow people to get a better idea of what their card is capable of. With these ratings, values, or even check boxes we can deduce what card is good for a setup should someone not truly know what they're looking for.

I like the idea, the only challenge is that there is really no such thing as a "Standard System Setup".

Also things like single slot vs double slot would be of different benefit to different people. If you have great case ventilation single slot might be good because it saves space and give you an extra potentially usefull slot. But the reverse could be true if you have bad case ventilation.

The main point of this article is that ultimately, no matter how many [H] articles and even more reviews all over the net, you cannot make a decision without using your "Noodle".

In essense Noodle > Google everytime!

User Article: [H]ard Economics of Video Card Upgrades

Supreme [H]ardness

Supreme [H]ardness

2[H]4U

Supreme [H]ardness

Supreme [H]ardness

Supreme [H]ardness

2[H]4U

Supreme [H]ardness

Supreme [H]ardness

Supreme [H]ardness

Limp Gawd

2[H]4U

Fully [H]

Fully [H]

Supreme [H]ardness

[H]ard|Gawd

[H]F Junkie

Supreme [H]ardness

Supreme [H]ardness

Gawd

Supreme [H]ardness

Supreme [H]ardness

Supreme [H]ardness

[H]F Junkie

[H]ard|Gawd

Supreme [H]ardness

Limp Gawd

[H]ard|Gawd

[H]ard|Gawd

Limp Gawd

2[H]4U

Supreme [H]ardness

Supreme [H]ardness

Supreme [H]ardness

Supreme [H]ardness

Limp Gawd

Supreme [H]ardness