System Builder Marathon, Q1 2013: System Value Compared

[pyro226]

I would like to see more low-end systems. Toms Hardware has already done a lot of articles with the I5's, I3's, and Phenom FX.

I know that spending less on the graphics card and changing to an i3 or a FX 4300 would cut costs; however, I am curious about where the balance lies between picking the recommended Athlon II X4 640 and the Phenom II X4 965 at the expense of graphics power.

The article "Gaming Shoot-Out: 18 CPUs And APUs Under $200, Benchmarked" contained the discontinued Phenom II X4 980 and Llano APU's. In terms of building for games, is it better to go with the AMD A8-5500 (Trinity) for $105 and possibly crossfire than Phenom II X4 965 at $100?

I think there's a lot left that can be reviewed and discussed with low-end hardware. It will also be interesting to see how steamroller and haswell change the recommendations at all price ranges.

 

[pauldh]

One reason is that the $1000 build is a "Performance PC" . That means everything is super snappy performance. Whenever you open any application or game, it should start in an instant.
Nobody is suggesting an SSD for the $600 or even for the $800 build.

And very strange that none of the 3 story writers have responded to my and Onus's (and others) suggestion of a different evaluation of the builds for 2 quarters in a row.

Apologies, ignoring you was unintentional. In fact, those comments shared within the last Day 4 article convinced me to re-weight the value equation within my own story, as it is something I have wanted to do anyway. Long ago I added a "Native Resolution Gaming" chart as it is such an important metric for this machine. As long as "Gaming PC" remains within my build's name, I'll strive to make it a solid gamer up to a realistic target resolution for the budget, even if that means dropping the CPU down and subsequently bowing out of the completion. I think 2/3 games, with added weight to 16x10 and 19x10 at the highest details, is a fair compromise for a $500-600 Gaming PC. I just can't drop the Application weight lower when ultimately the machine will be judged more on it's CPU than it's GPU. And when you think about it, performance within today's threaded applications may represent CPU demands in next year's games. We hope for photorealistic graphics, and accept replacing our graphics solution is par for the course along the way. As far as day 4, I'll touch on that in a bit, but overall while there have been many good suggestions, what you are asking for changes the series a lot, and may be impossible to implement to everyone's satisfaction.

IMO, drop "Gaming" and a $400-500 "Entry Level PC" MAY be able to top the value charts, but at the same time would not show up for the bulk of our game settings. Drop $100 off this $600 PC's graphics budget, and your left with i5-3350P at $500, hurting at 20-30 % of the tests. Phenom II X4 965 BE could possibly do the same for less money, maybe offer a bit better gaming budget. Shoot, a well-tuned Trinity build may be able to win at $400 using the iGPU. However none of these could be called a "Gaming PC." And is there a target audience here for this? How many Tom's readers could live with such a winning machine without making upgrades?

I didn't comment last time because it was in the day 4 story. That belongs to Thomas... his story, his time and energy, and his calculations. That said, we are all team players, and it's simple to express concerns if/when we have them. Thing is, the SBM series values overall performance, and I think all involved are quite satisfied with the value equation Thomas has pinned down. Real life needs of most folks seem to favor removing all platform bottlenecks, over just stuffing in the big graphics required for Ultra details with AA. (Comments wanting i3+7870 at $600 aside) And 10% storage scores representing the snappier user experience from an SSD, is fair IMO.

As gamers ourselves, we understand what you are saying. And we can also understand how the series falls short for the crowd who doesn't game (anymore?). A Gaming PC shouldn't be judged on productivity, and a beastly workstation or video editing machine shouldn't be judged on gaming. Everyone's needs are different, and it's impossible to agree on a purpose at each price point, without creating a separate series of competitions. Rather the SBM attempts to tackle all things well. After all, we just think our readership leans enthusiast, and hates any visible platform bottlenecks at any price point. We'll never all agree if CPU/GPU/SSD or even aesthetics and silence deserves the bulk of our funds, which is a big part of why we build our own machines. =)

The gaming community here is huge, and I took heat for not going with a pure gaming PC with more GPU. Shoot, I started the heat within my own conclusion, recognizing a potential winner isn't the best possible gamer. In fact, I think only Don pleased this demographic with his graphics budget allotment. So I will see if Thomas would consider adding the popular 1920x1080 res in his gaming chart just like he does with 25x16. Only the highest test settings should matter. That way, the SBM winner is still the most overall bang for the buck, but we should be able to also pin down a value winner rewarding builds focused ultimately on 19x10 and 25x16 gaming.

 

[tmk221]

[citation][nom]Crashman[/nom]That's an awesome idea too! We could get some old-fashioned Chieftech Dragon (or similarly-popular) cases, maybe some older 700W power supplies and hard drives, match everything and just change the platform. Anyone else think this is a good idea?[/citation]

Yes that's a great idea. I cary over case, dvd and hard drives, and sometimes PSU.

 

[TomFromBelgium]

Some suggestions:
- Is it possible to also include sound measurements (Decibel) for each build?
- As for the price ranges, I propose 700 - 1000 - 1300$.

- For me, you don't need to include an SSD in any build. The influence of SSD's is not on gaming or applicatenions performance, and pretty independent on the build. When building your own system, you just add the price of an SSD if you want one (to each built).

I'm looking for a +- 1000 Euro system (in Belgium), including SSD that I will not overclock (those days are over ). For me the noise it makes is also important

 

[pauldh]

[citation][nom]TomFromBelgium[/nom]Some suggestions:- Is it possible to also include sound measurements (Decibel) for each build?- ... I'm looking for a +- 1000 Euro system (in Belgium), including SSD that I will not overclock (those days are over ). For me the noise it makes is also important[/citation]
We fully understand noise is important, but accuracy for such measurements is a big issue. We can't toss these rigs in a sound chamber and test with an equal acoustic floor. Instead these are tested in different labs, and both acoustic and thermal baselines would vary to some degree or another. You can't compare dB readings from site to site, and even the same video card tested in Germany, California, and Canada for instance, would likely get different readings even when measured at the same distance and angle. We'd love to oblige and add acoustics to the SBMs, but only if the comparison could be done within reasonable accuracy and at a reasonable cost.

Most gamers seem to expect some GPU noise is par for the course, others seek out the quietest cooling solutions or at least the best noise to performance ratio. I've had plenty of those old overclocked gaming rigs sporting an abundance of 80mm fans, but those days are long gone, and for at least 5+ years now I just couldn't game on such an abrasive beast anymore. I can only put up with noise in a machine intended to play with, not anything I play on. My GPU's at home are mildly overclocked taking noise into consideration, and only pushed harder if/when a specific game requires a bit more performance. And I couldn't tell you the last time I used an AMD boxed cooler at home. Any Athlon II/ Phenom II got aftermarket cooling. Oh, and for non-gaming rigs or ones occupying living space (family room), well quietness is key and always gets funding. If you are not overclocking, and don't live in a hot climate, then Intel's boxed coolers usually do an acceptable job of balancing noise and temperatures.

 

[cypeq]

1000$ build would really need storage drive -_- in times where game can take up 30 gigs of you hdd 240 GB drive does not sound appealing.

 

[Strategist]

Forum seems to have eaten my reply.

Anyway, doing it again...

You guys are currently baselining a monitor, keyboard, and mouse as something everyone already has or is default... I think you guys should at the very least also baseline a NAS/storage server. In this day and age of tablets, smart TV's and cloud computing, having a NAS/storage server is simply a must have for every self respecting computer enthusiast. The notion that a PC must have room for storage is a very outdated one IMO.

I would also like to see a competition with baselined cases, 800W or so PSU's (IMO having the over capacity for a 2nd GPU is quite vital) and optical drive (because even if maybe it is too early to do without optical drive entirely, it is definitely too late to put money in a new one rather than use one you have laying around somewhere )... but those are a seperate story than the NAS.

Third, I personally think the price points you went for are too low. If I was to build a PC this year, I would be looking in the $800 to $1600 price range for the components. And most likely end up with something around $1200-$1300. So I would enjoy a $800/$1200/$1600 or $900/$1200/$1500 match up... or maybe $1000/$1200/$1400 if you want to keep the $200 gaps and the $1000 level.

 

[mayankleoboy1]

[citation][nom]pauldh[/nom]Apologies, ignoring you was unintentional. In fact, those comments shared within the last Day 4 article convinced me to re-weight the value equation within my own story, as it is something I have wanted to do anyway. Long ago I added a "Native Resolution Gaming" chart as it is such an important metric for this machine. As long as "Gaming PC" remains within my build's name, I'll strive to make it a solid gamer up to a realistic target resolution for the budget, even if that means dropping the CPU down and subsequently bowing out of the completion. I think 2/3 games, with added weight to 16x10 and 19x10 at the highest details, is a fair compromise for a $500-600 Gaming PC. I just can't drop the Application weight lower when ultimately the machine will be judged more on it's CPU than it's GPU. And when you think about it, performance within today's threaded applications may represent CPU demands in next year's games. We hope for photorealistic graphics, and accept replacing our graphics solution is par for the course along the way. As far as day 4, I'll touch on that in a bit, but overall while there have been many good suggestions, what you are asking for changes the series a lot, and may be impossible to implement to everyone's satisfaction. IMO, drop "Gaming" and a $400-500 "Entry Level PC" MAY be able to top the value charts, but at the same time would not show up for the bulk of our game settings. Drop $100 off this $600 PC's graphics budget, and your left with i5-3350P at $500, hurting at 20-30 % of the tests. Phenom II X4 965 BE could possibly do the same for less money, maybe offer a bit better gaming budget. Shoot, a well-tuned Trinity build may be able to win at $400 using the iGPU. However none of these could be called a "Gaming PC." And is there a target audience here for this? How many Tom's readers could live with such a winning machine without making upgrades?I didn't comment last time because it was in the day 4 story. That belongs to Thomas... his story, his time and energy, and his calculations. That said, we are all team players, and it's simple to express concerns if/when we have them. Thing is, the SBM series values overall performance, and I think all involved are quite satisfied with the value equation Thomas has pinned down. Real life needs of most folks seem to favor removing all platform bottlenecks, over just stuffing in the big graphics required for Ultra details with AA. (Comments wanting i3+7870 at $600 aside) And 10% storage scores representing the snappier user experience from an SSD, is fair IMO. As gamers ourselves, we understand what you are saying. And we can also understand how the series falls short for the crowd who doesn't game (anymore?). A Gaming PC shouldn't be judged on productivity, and a beastly workstation or video editing machine shouldn't be judged on gaming. Everyone's needs are different, and it's impossible to agree on a purpose at each price point, without creating a separate series of competitions. Rather the SBM attempts to tackle all things well. After all, we just think our readership leans enthusiast, and hates any visible platform bottlenecks at any price point. We'll never all agree if CPU/GPU/SSD or even aesthetics and silence deserves the bulk of our funds, which is a big part of why we build our own machines. =)The gaming community here is huge, and I took heat for not going with a pure gaming PC with more GPU. Shoot, I started the heat within my own conclusion, recognizing a potential winner isn't the best possible gamer. In fact, I think only Don pleased this demographic with his graphics budget allotment. So I will see if Thomas would consider adding the popular 1920x1080 res in his gaming chart just like he does with 25x16. Only the highest test settings should matter. That way, the SBM winner is still the most overall bang for the buck, but we should be able to also pin down a value winner rewarding builds focused ultimately on 19x10 and 25x16 gaming.[/citation]

Glad to see someone reply

Basically saying that if the name is "Gaming PC", its overall weightage should depend more on games.
For a "general purpose PC", it should do games and apps equally good.

If you want to make a gaming heavy build , why dont you name all the builds as "$600/800/1000 gaming build" and so forth, instead of "general purpose" and "enthu build" ?

I would be happy if there is an "alternate analysis" of data,(on a separate page) according to (or similar to) the method i suggested. The comments by readers can tell you if all of us like it or not.

As far as day 4, I'll touch on that in a bit, but overall while there have been many good suggestions, what you are asking for changes the series a lot, and may be impossible to implement to everyone's satisfaction.

pretty poor reason of not trying something new. (no offense intended, please)

 

[Stevemeister]

I think it would be interesting to simply set a budget for the Motherboard, RAM, CPU and GPU and look at the relative value performance of these items. IMHO a problem with the value charts is value is calculated based on the cost of the "total system cost" which includes costs for components like the Case, PSU, DVD/BR drive etc. all of which can have a significant effect on the total cost but no effect on the value. I'm not sure if you look at the percentage of cost these items represent for the different priced builds but the bigger the budget the fancier the cases tended to be and while some may offer better cooling than others I'm not sure if that is being evaluated accurately. The long and short is that you could in theory spend twice as much on these components and not increase the performance one iota while dropping the system value . . . . so in my opinion the value analysis would be more useful if these "overhead costs" were removed from the value analysis but also simply treated as an "add price" to arrive at the cost of a complete system for budgeting purposes.

I think it would also be interesting to run a value analysis on systems with and without an SSD in both gaming and productivity applications. I have an SSD because subjectively I like the speed at which the computer boots up I'm not sure for most of what I do whether it actually makes that much difference BUT even for gaming I have 370 GB of gaming programs and I don't store theses on my 120 GB SSD - they are parsed off to my higher capacity 2 TB storage discs. I don't know how many people are in similar situations . . . once 500 GB SSD's hit the $250 level then I may splurge but for me at least its just not worth that much me to pay a lot to store gaming programs.


Finally the value impact of different cooling systems would make an interesting evaluation - how much more performance can you wring from overclocking by spending more on a good cooling solution i.e. what is the incremental performance gained for extra $ spent.

A lot of people who read these forums are enthusiasts and a lot of us already have Cases PSU's DVD-BR's and Coolers that get re-used in upgraded builds (buy well here and the items can last a long time). . . . . I have a Silverstone Temjin TJ07 with a PC Power and Cooling PSU and a big triple radiator water cooling loop that cools both the CPU and Graphics cards . . . when I upgrade I only change MB, CPU, RAM, & GPU and even then not always at once. . . . so these are the only bits I'm really interested in.

 

[Aegean BM]

The sweet spot is interesting precisely because there is some elasticity around the sweet spot and elasticity around most people's budget. I propose two different approaches to finding the sweet spot(s).

One, why put out spending tiers and then after the fact try to interpolate the sweet spot, a rather indirect approach. Instead of giving the three amigos three tiers, take the direct approach and give them all the same assignment: build the sweet spot. We'll find out tons from where they agree and where they differ.

Two, the three amigos agree on the sweet spot on a component by component basis. The sweet spot can be a range, where each dollar less is leaving performance, but not that much; or each dollar is buying diminishing returns, but again, not that bad. In some cases the sweet spot is fairly linear. Dollar per gigabyte of memory is close from 4 GB to 8 GB to 16 GB. The three amigos still build out high, medium, and low systems based on the agreed high, medium, and low sweet spot components. There won't be a surprise for the three amigos because they made the choices together, but I'm sure there will still be some surprise in the results. The interesting result for me will be the natural cost points of high, middle, and low around the sweet spot.

Competition has some strengths; so does collaboration. I would really like the three amigos to join together to design and build the sweet spot machine together.

I agree with the decision to exclude mice, monitor, keyboard, etc. I also agree with excluding software including the OS. I disagree with the required optical drive. It may be cheap, but its existence annoys me as much as a floppy drive. I disagree with the optional SSD and waving it away as merely "user experience". Back in the day, hard drives were labeled optional for the same reason. Same with high speed internet over dial up. Have you ever known someone who made the jump to ever revert back?

I am intensely interested in the sweet spot machine because I plan on building two systems this year. I'd like to build on the higher end of the sweet spot for myself, and the lower end for my sister in law.

 

[bwcbwc]

[citation][nom]agnickolov[/nom]I'd suggest stepping the budget up by $200 for the next SBM - $800, $1000, $1200.[/citation]
I think the differences between budgets need to be widened without shifting the bottom end. Based on the way everyone chose a low-end case, it was obviously hard for the reviewers to differentiate their builds based on perfomance. So maybe $600/900/1200 or $600/1000/1400? A budget range that puts a small SSD in the midrange build, 4/8/16 GB or 8/8/16 GB of ram, allows for a 3770k on the high end would be about right.

 

[Aegean BM]

I agree with making a weighted balance between gaming and productivity. However, as we all acknowledge, needs vary, so your balance is not my balance. I'd like to see the results on a simple web app with weights I can change.

 

[shrapnel_indie]

Maybe I am crazy... but... showing off a $400-450 system in this could be done just to prove a point at least. IF it follows general wisdom, it's poor performance will be quite clear. IF it breaks that general wisdom, well... we've learned something new. It probably wouldn't be much good beyond 1920x1080 though. Not everyone runs with triple monitor+ gaming or even dual monitor gaming. Monitors beyond 1920x1080 start getting pricey fairly quick too.

Also on a realistic note, we buyers still usually spring for the OS, and probably a new monitor once in a while in addition to mouse and keyboard.

 

[Aegean BM]

In justifying SSD, it drives me crazy to wait area loads on Borderlands.

In my nongaming life, I'm a programmer/consultant who has to use VMs to mimic my client's environment. Shutting down 2 VMs took 30 min to flush to disk, longer if you get disk thrashing happening. Changing to SSD brought it down consistently to 5 min, and no disk thrashing.

 

[ashon]

In the next edition I would like to see $600, $900, and $1200 for the three price points.

 

[stu2k2]

Tom's,
I greatly enjoy these articles, but there is something that this specific article proves that needs some more attention.

What is the difference between the $800 and $1k computers? An SSD. By the addition of an SSD to the computer, you increase the storage numbers by a significant amount (10% weight of overall score) but do literally nothing to enhance the gaming capability of the computer. Now, do I want an SSD on my computers? Yes, I hate long boot times. Do I want to see two of the same computer, with the SSD being the only difference, compared? Not really.

I think the best SSD for the money, when viewed in concert with these comparison, is a great way to understand what you should want to purchase for a computer. Most of the faithful readers of these articles understand that a decent SSD will run around $150-200 and will make a huge improvement in the feel of the computer, but not the gaming performance.

If you want to look at how the guts of a computer change across the $400 range, then make certain items fixed. Case, DVD, and Hard Drive (not SSD), all the same. That would give you a fairer comparison of what the extra $400 buys. At this price point you aren't going for a 4TB hard drive, or a $250 top of the line case. Too much is given up in other areas. Something like an Antec 300 and a decent 1TB hard drive will make the differences a result of the other choices made.

I want to know what the difference is from going AMD to Intel, i3 to i5 to i7, DDR3-1600 to 2133, or 7850 to 7870. Plus, I already have a case, storage, and a DVD. I am looking for what goes into my gaming rig so that I can migrate some of my older stuff into my HTPC/Server.

As a faithful reader, I am not criticizing the choices made in this quarter's articles, but asking for you to draw on the lessons learned. An SSD costs $200 which improves storage benchmarks and improves the overall feel of a computer. We already knew that...

 

[Crashman]

[citation][nom]stu2k2[/nom]Tom's,I greatly enjoy these articles, but there is something that this specific article proves that needs some more attention...Do I want to see two of the same computer, with the SSD being the only difference, compared? Not really. [/citation]False. False premise too. You start by pretending that these builds were coordinated, but that wouldn't be a competition.

Since SSD affects only 10% of the score, something must have went awfully right with the $1000 machine for it to beat the $800 machine in overall value.

 

[Jalmag]

As a starting point for SBM you could use the question: "What will the machine be used for?" Say, make up a few archetype users of toms and then try to build the machine for them. For example one guy just wants to game and his priority is strictly on budget while another wants to enable all the bells and whistles. Some guy rarely plays games but does heavy video editing everyday. Everyone has their own needs but you could bundle them together to cover most bases.

 

[stu2k2]

False. False premise too. You start by pretending that these builds were coordinated, but that wouldn't be a competition.

Since SSD affects only 10% of the score, something must have went awfully right with the $1000 machine for it to beat the $800 machine in overall value.

Funny thing is, I didn't have a premise when I read your article. I guess you do...

I don't see it as a competition between builders. I see it as a competition against the numbers. I could care less which builder gets bragging rights at the water cooler. Honestly, I would expect the builds to be coordinated. Who wouldn't? Otherwise you end up doing a comparison of two identical machines...

But that wasn't the point to my post. By leveling the playing field, removing the costs that are shared between builds (DVD, HDD, Case), you truly look at the efficiency of the components. You already don't take into account the operating system, monitor, keyboard, mouse. They are integral to the feel of a gaming system, and they vary wildly in price. Cheap keyboard or slow monitor, and you may game horribly. No SSD, and your load/boot times will suck et al...

By using the same case, HDD, etc., you can remove that cost from the efficiency calcs and come up with a better number.

Without going too far into it, the cost of the case, power supply, DVD, and HDD is around $200. That means the $600 machine has one third of its cost spent on components that do nothing to improve performance. The $1k machine has 20% of its cost tied up in similar components. That's a significant difference in percentage. Remove those costs and rerun the numbers. Heck, I will do it after I get some sleep because you guys do such a detailed job in your write ups.

Ultimately, this month was an experiment in changing the brackets that I don't see as working out well because two intelligent builders who know the products better than nearly anyone ended up picking nearly identical core components; one with an SSD and one without. Why? Because they know the parts, and they know the rules. Take a step back, read the article and you may see it.

Now, as I stated, I am a faithful reader to your site. I am not someone who regularly flames a post, and hope that you don't take my observations that way. Truthfully, this is the place that I go to improve my technical knowledge, and ensure that I am buying a component that is worth something. Want a Z77 board for my new gaming rig? Head to Tom's and use the exact board (AsRock Extreme 4) that they used in their build. (I did that last quarter) Why? Because I know it works. It's just some feedback on how the article felt to me.

 

[Onus]

[citation][nom]Fulgurant[/nom]Yes, exactly. The assumption that more expensive hardware will last longer relies on a faulty premise -- namely, that there's a universal standard of acceptable performance. Realistically, someone who buys a $2,000 rig will tend to have much higher performance standards than someone who buys a $600 rig. Thus, the $2,000 rig isn't likely to offer its user any more longevity. The above-described principle is further reinforced by the native-resolution restrictions of LCDs. You can effectively shoe-horn yourself into high-end computer hardware if you get accustomed to gaming at high resolutions -- and if you buy an expensive, high-res monitor (or monitors), you may feel compelled to buy high-end hardware to justify that initial investment. By contrast, someone who only has a middling-res monitor, and is satisfied with it, can get an awful lot of mileage out of even low-end computer hardware.Ultimately, and perhaps ironically, the most important things to consider when buying computer hardware are unquantifiable: the user's personal needs and expectations. Technical knowledge is obviously important too, but it's useless without context.[/citation]
The bold is mine. I understand what this post says (and the one to which it replied), but I specifically disagree with the bolded point; it depends on where the extra money was spent.
Even the briefest view of performance charts will clearly show that if you want to play the latest games at "good" resolutions of the day (I remember when it was 1024x768), a graphics card upgrade every couple of years (maybe more often) is essentially unavoidable. Buying budget parts invites early replacement, as they won't keep up with upgraded graphics cards, or latest interfaces, or maintain cooling, or enough power, or... And there's the key. After a decade of thinking I was being frugal buying just what I needed (maybe a hair over), I realized it's what "justified" frequent upgrades. Now, my current systems both contain premium motherboards, top-tier PSUs, SSDs, and plenty of drive space. Omega is almost two years old, and I've felt no desire to rebuild it, even though a GPU upgrade from time to time is highly likely. Phoenix isn't as old, but the same goes there. By significantly exceeding my needs when I built them, and using quality parts, I have banished the upgrade itch, and will no doubt save a LOT of money in the long run.

 

[Onus]

[citation][nom]stu2k2[/nom]...I don't see it as a competition between builders. I see it as a competition against the numbers. I could care less which builder gets bragging rights at the water cooler. Honestly, I would expect the builds to be coordinated. Who wouldn't? Otherwise you end up doing a comparison of two identical machines...[/citation]
I disagree strongly. The fact that skilled, non-collaborating system builders selected so many of the same parts shows a convergence of value that other builders would do well to consider. There's fun in the competition, but the overall goal is to find value, if only by adding data points regarding price and performance. The discussions that follow do a good job of fleshing out the details; calling out fitness for purpose, or issues with a cheap case, or longevity. Together, they provide a pretty good outline for someone contemplating a new build.

 

[Aegean BM]

I've read a lot of crap about $800 and $1000 machines ending up the same. I'll paraphrase it as "the only difference was an SSD and we already knew what that would do." This is hindsight talking. No one knew that's what would happen. I love that the experiment was done, and the result was the same GPU and CPU. It gives me MORE confidence in those two choices for my next build. That is EXACTLY what I'm looking for. Great job!

Concerning case, PSU, fans, acoustics, etc., the competition rules tend to obliterate lots of real concerns, even if they are subjective. I would like to know more what the builders would do if it were their machine. Therefore, I propose a quarter where the budget is rather vague. Build a cheap machine. Build a realistic enthusiast machine (don't SLI 4 cards). Build a sweet spot machine between the two. Let the goal determine the budget.

 

[Fulgurant]

The bold is mine. I understand what this post says (and the one to which it replied), but I specifically disagree with the bolded point; it depends on where the extra money was spent.
Even the briefest view of performance charts will clearly show that if you want to play the latest games at "good" resolutions of the day (I remember when it was 1024x768), a graphics card upgrade every couple of years (maybe more often) is essentially unavoidable. Buying budget parts invites early replacement, as they won't keep up with upgraded graphics cards, or latest interfaces, or maintain cooling, or enough power, or... And there's the key. After a decade of thinking I was being frugal buying just what I needed (maybe a hair over), I realized it's what "justified" frequent upgrades. Now, my current systems both contain premium motherboards, top-tier PSUs, SSDs, and plenty of drive space. Omega is almost two years old, and I've felt no desire to rebuild it, even though a GPU upgrade from time to time is highly likely. Phoenix isn't as old, but the same goes there. By significantly exceeding my needs when I built them, and using quality parts, I have banished the upgrade itch, and will no doubt save a LOT of money in the long run.

I humbly submit that your standards are high, then. The fact that you name your builds is a bit of a tell, IMO.

Most people don't upgrade piecemeal at all. Personally, I think the entire concept of upgrade-ability is overrated, for some of the very reasons you touch on. Every platform has limitations, and it's hard to know exactly when those limitations will become stifling.

Someone who bought (for example) an AGP motherboard in ~2003 would be pretty much screwed a couple of years later no matter how expensive or well-made that motherboard was at the time of purchase. Someone who bought an RDRAM platform back when Intel seemed intent to move the market in that direction would be likewise screwed vis-a-vis later upgrades. (I did both of those things, actually.)

The only components you can really count on, with respect to their long-term usefulness, are the case and to a lesser extent the power supply. (External peripherals and hard drives also have some legs, but their long-term value is heavily subjective.) Everything else is up in the air; chances are that if you play your cards right at the outset, you'll be better off simply waiting to rebuild the whole thing more-or-less from scratch 3-4 years down the road.

And that's where budget, value, and most importantly, the user's needs and expectations come into play. If you know what you're doing and if your expectations aren't too high, you can get 3-4 years of useful service time out of low-to-mid-range components. And if, during that time, this-or-that game needs to have this-or-that setting tuned down to play at a given resolution? It ain't the end of the world. Frankly, you'll encounter that problem on rare occasions even with the highest of high-end hardware. (Hardware is only as good as the software you expect to run on it.)

 

[crzyone01]

Where are the AMD chips in these builds? For the price of the i5 used you can get an 8350 or just gone with a 6300 and had another $100 +/- o add to the video card. The change in price points was great but there is no variety here really. The only difference in the top 2 is the SSD.

 

[Onus]

Every system I build gets a name; I think I've always done that.
By gaming standards, they're nothing special; my most powerful GPU is "only" a HD7870. It is indeed likely that socket changes will dead-end my systems' CPU upgrade potential sooner rather than later, but hopefully in terms of overall lifespan I'll get ten or more years out of them. There's nothing about my RAID-1 arrays that would appear in a typical value analysis, or having a Seasonic X-560 vs. a Crappermaster Extreme Power, but by eliminating most "ravages of time" or other quality issues, they make both systems extremely future-resistant.