Sign in with
Sign up | Sign in
Your question
Closed

System Builder Marathon, Dec. 2009: System Value Compared

Tags:
  • System Builder
  • Performance
  • Systems
  • Product
Last response: in Reviews comments
Share
December 25, 2009 5:01:22 AM

Our mid-priced and high-end systems are ready to challenge the notion of value supremacy for low-cost parts. Will Intel’s latest architecture provide enough performance to break the $700 machine’s value winning monopoly? How do ATI's new GPUs factor in?

System Builder Marathon, Dec. 2009: System Value Compared : Read more

More about : system builder marathon dec 2009 system compared

December 25, 2009 5:34:00 AM

Great conclusion Thomas, good System Building Marathon overall. My only wish is to see something with an AMD processor next time, specially on the market segment where they shine.
Score
6
December 25, 2009 6:02:55 AM

I am really impressed by the performance of $1300 PC.

It came so close to the $2500 PC without breaking the bank.
Simply awesome!!
Score
8
Related resources
December 25, 2009 7:07:08 AM

typo on the last page, overclocked $650, not sure if that was the value used in the chart, might check that also.
Score
8
December 25, 2009 7:13:11 AM

Good conclusions..

This article can get more interesting if you add previous [September’s ] systems data to the charts.
Score
2
December 25, 2009 7:45:53 AM

Yes the previous systems would have been really nice to compare, espacially in the conclusion.
Score
6
December 25, 2009 8:04:12 AM

ibnsinaGood conclusions..This article can get more interesting if you add previous [September’s ] systems data to the charts.


Sorry, but that wouldn't be fair. First of all, September's systems used different benchmarks, settings, and OS. Second, September's systems used both AMD graphics that pre-date these, plus AMD processors, and people would have used the older graphics as an excuse to beat up on the CPU. AMD fans would have gone nuts as well, claiming the authors were trying to use the superior graphics of this SBM to skew readers against AMD. Tom's isn't interested in publishing invalid results or creating fake controversy.
Score
6
December 25, 2009 8:14:30 AM

psycho sykesA question..Does those come with Windows 7 installed? Or they won't be real 700-1300-2500 machines.. Right?!


Windows 7 was only installed for the benchmark analysis. For anyone who would like to copy one of the builds and still stay on budget, Ubuntu is suggested.
Score
5
December 25, 2009 9:30:47 AM

Crashman
[/quote]First of all, September's systems used different benchmarks. AMD fans would have gone nuts as well[/quote]

If you compare the benchmarks, is more or less identical with 1-2 minor differences. What’s wrong with idea of comparing different configurations? I don’t agree with the thought of AMD fans getting upset about it, to contrary they will be happy about gaining new knowledge.
Score
0
December 25, 2009 11:24:44 AM

First of all, Merry Christmas everyone!
Second of all, congratulations to Don. His tweaking contributed heavily to the superiority of the $1300 machine. The $700 machine pulled up lame when not gaming, and the $2500 machine was crippled by inadequate cooling.
If I win the big guy, I'll put it in my CM-RC690 and see how it does. The little guy will get my Q9450, but Don's build just needs a better cooler.
Nice series.
Score
0
December 25, 2009 4:18:20 PM

Merry xmas to the staff and thanks for writing these articles at an otherwise slow time of the year due to holidays. I've enjoyed reading them.
Score
3
December 25, 2009 4:23:33 PM

I enjoyed the whole series. Popular hardware in different configurations, always interesting to read the results.
I notice a lot of game players ask "will a cpu like the AMD 620 4x2.6 bottleneck a crossfire system of ie 5750,4850,5770". And the answer is evident in these benchmark results. YES, even the i5750 system with turbo liked o/c. All fps were very good except maybe the 700 dollar system at stock levels, but they all showed improvement with o/c. In these tests the o/c includes gpu o/c as well. It was bugging me, the 700 dollar system, using 2 4870's. I've been following prices diligently and I thought that was unrealistic, $250. I had suspicions that you snuck them in to make sure you beat the Sept system with 4850's. But a quick look this morning showed a Diamond 4870 for 129.99
http://www.zipzoomfly.com/jsp/ProductDetail.jsp?Product...
Thats a great deal to build on.
Score
2
December 25, 2009 5:08:40 PM

Love this series. Always a great read.
Score
2
December 25, 2009 5:42:29 PM

Crashman "Windows 7 was only installed for the benchmark analysis. For anyone who would like to copy one of the builds and still stay on budget, Ubuntu is suggested."

How many of those popular games will run on Ubuntu?
Score
-3
December 25, 2009 5:45:12 PM

I'm still upset about the 2500$ build.

In addition to all the improper decisions you've mentioned in the conclusions page, the case was completely inadequate for this kind of high-end hardware (unless you mod it) and costs 2 times more (!!!) than the much more able antec 200.

That by itself could allow buying 2 intake fans and a better cpu cooler, thus increasing overclocking by a big margin, consequently improving overall system performance.
Score
-1
December 25, 2009 5:54:35 PM

doronI'm still upset about the 2500$ build.In addition to all the improper decisions you've mentioned in the conclusions page, the case was completely inadequate for this kind of high-end hardware (unless you mod it) and costs 2 times more (!!!) than the much more able antec 200.


It's nice when a case has enough room for both the graphics cards and the hard drives, not either/or.
Score
0
December 25, 2009 7:18:48 PM

The conclusion makes up for the Hi-end build, I guess next Hi-end build will rock. As for the 700$ build...please consider AMD for the next one.
Score
5
December 25, 2009 7:47:16 PM

@Crashman - Then how about Coolermaster 922 which costs 100$ on newegg?
I know that's 10$ more but your afoermentioned case has the same 100$ price tag on newegg.

Even better, for 90$ you could get a mighty Antec 900 (First gen) with all its 3x120mm + 1x200mm fans bundled.
Score
0
December 25, 2009 8:07:13 PM

doron@Crashman - Then how about Coolermaster 922 which costs 100$ on newegg?I know that's 10$ more but your afoermentioned case has the same 100$ price tag on newegg.Even better, for 90$ you could get a mighty Antec 900 (First gen) with all its 3x120mm + 1x200mm fans bundled.


Yeh, it's a good deal. Unfortunately it didn't fit within budget at order time, when the Lian-Li case was on sale for $80 and the total purchase was $2495.
Score
1
December 25, 2009 8:43:56 PM

hmm i have an idea. Give us, toms readers option to choose builds in each segment. Post few configurations for each budget for a survey, and build the winer.
Score
1
December 25, 2009 8:45:24 PM

i would have loved to see the athlon 620 overclocked to 3.2 gigahertz instead of that shitty dual core. i mean a quad core makes sense to drive those 4870's.
Score
5
December 25, 2009 10:00:37 PM

Well, as for the 5850 versus the 5870 in CF... after seeing much benchmarks of the 5970, it was obvious the extra money wasn't worth it.

As for cpus being a MAJOR drawback in a system... no... the AMD built with quad 4850 GPU actually won a few rounds over the CF 5850 and a core i5... which is really surprising...
Score
2
December 25, 2009 10:01:25 PM

I wonder how 3 4890 would perform over 2 5850... and even 2 5870.
Score
-2
December 25, 2009 10:04:36 PM

In canada, the Powercolor 4890 is selling for 150$CAD at NCIX... that means 140$US...

The 4890 is really now the sweetspot of gpu.
Score
2
December 26, 2009 1:55:19 AM

The cheapest HD4890 on Newegg (US) is $190. I'd take the slightly weaker HD5770 for $155 for the power savings (and resulting lower heat and fan noise), Eyefinity, and DirectX 11 support.
Score
1
December 26, 2009 6:07:39 AM

The 1200-1300 USD system, if built wisely, is almost-always the sweet spot. Its kind of unfair for the guys who build the other two systems!
Score
-1
December 26, 2009 7:57:42 AM

Quote:
Reaching only 1280x1024 at acceptable frame rates, we’d rather sacrifice a few details to gain higher resolutions on the overclocked $700 system.


1280x1024, is nothing to be ashamed of. It was a little barely a decade ago, when people were wetting their pants to play their games at highest settings on resolutions like 640x480 and 800x600 - difficult to stomach by today's standards. Personally, I usually prefer to sacrifice screen resolution over "Detail settings" whenever necessary in most cases as long at the resolution is at least a minimum of 1024x768.
Score
3
December 26, 2009 8:47:29 AM

UFO_WARVIPER1280x1024, is nothing to be ashamed of. It was a little barely a decade ago, when people were wetting their pants to play their games at highest settings on resolutions like 640x480 and 800x600 - difficult to stomach by today's standards. Personally, I usually prefer to sacrifice screen resolution over "Detail settings" whenever necessary in most cases as long at the resolution is at least a minimum of 1024x768.


Have you actually OBSERVED the difference in details between the lower and higher settings of that particular bench? The lower settings look great, the lower resolution looks terrible.
Score
-2
December 26, 2009 10:41:28 AM

I ran the Crysis bench the other day on my brother's system in fact. However, the highest resolution my monitor our monitors are capable of is 1280x1024, which I always thought looked spectacular, and is usually the resolution my family and I game on contently. Everyone has their own opinion I guess. So, to answer your question: No, I haven't observed the difference, nor can I. Everytime someone sets an old CRT next to the trash, I take it to storage (and repair it if neccessary), so my family and I can game on it when one of our other monitors fail beyond repair.

On a slightly different note, I'm a little concerned that the framerates were almost exactly the same when I added an extra Radeon 4850 to his system in crossfire. Our old CRT's don't offer the resolutions where crossfire and SLI can flex their muscles, and maybe the Phenom II x3 720 BE holds us back too. My system has the same CPU and a 4850x2 card with about the same results as his previous system with only one 4850 card. The GPU benchmark bundled with the game reports 29-32 average FPS runs on "Very High" and 1280x1024 No AA, and 0x to 4x AF on all 3 video card configurations. We use the trick to hack "Very High" settings on Windows XP, which supports only DX9 shaders, by editing the files in the CVarGroups folder.
Score
0
December 26, 2009 10:50:55 AM

CrashmanYeh, it's a good deal. Unfortunately it didn't fit within budget at order time, when the Lian-Li case was on sale for $80 and the total purchase was $2495.


Sorry for not letting this go but you miss the point.

The guy who built this rig should have used a bit more creativity. Heck, if he would opt for 2x1tb instead of 2x2tb he would save 400$(!!!). That alone (without addressing the other controversial parts) is enough to get 3xhd5850, an x58 asus rampage II extreme, a core i7 920 and will even leave 45$ to spare, be it for triple channel memory modules, better cooling, or both.

The Asus Rampage II, in addition to it being a high-end, feature-rich motherboard, achieved the best performance in your msi eclipse plus article in triple SLI testing. You were also using 3xgtx 285 for testing - roughly the same performance as the hd5850.

Note that the hd5850 is 15-20% slower than the hd5870 at worst but 30% cheaper.

It's not the build I'm upset about, it's the fact that I've learned almost all of what I just wrote from you guys, so all you had to do is open some articles from your archives and see this for yourself.
Score
1
December 26, 2009 11:13:50 AM

This is the most impressive $1300 build yet. The core i5 750 is currently a killer value, in fact one of the most impressive values I've seen from intel for a long time in the upper mid-range segment. I shrieked at the choice of the 4870 cards on the $700 build. Whenever I build my systems, I pay very close attention to articles like System Builder marathons and similar articles at Anandtech as guidelines to help me balance my PC, and getting the best value in my price range. I'm not complaining, and I know the video cards werea good deal and extremely lucky find and all. However, given that some hobbiests will likely refer to the $700 machine as a rough draft of their own future system I don't think it is practical to include impossible to find video cards in an article like this. Maybe, I'm in the wrong, but something about this, sort of strikes me a little oddly, and makes the $700 system article feel like the author did not take his audience seriously. These articles really should represent parts that enthusiast are reasonably likely to expect to find on the market at least modestly close in price as constructed here for a short time after the article is posted. This is partly what make's the content of Cleeve's monthly articles' recommendations straightfoward & useful
Score
2
December 26, 2009 11:20:01 AM

doronSorry for not letting this go but you miss the point.The guy who built this rig should have used a bit more creativity. Heck, if he would opt for 2x1tb instead of 2x2tb he would save 400$(!!!). That alone (without addressing the other controversial parts) is enough to get 3xhd5850, an x58 asus rampage II extreme, a core i7 920 and will even leave 45$ to spare, be it for triple channel memory modules, better cooling, or both.The Asus Rampage II, in addition to it being a high-end, feature-rich motherboard, achieved the best performance in your msi eclipse plus article in triple SLI testing. You were also using 3xgtx 285 for testing - roughly the same performance as the hd5850.Note that the hd5850 is 15-20% slower than the hd5870 at worst but 30% cheaper.It's not the build I'm upset about, it's the fact that I've learned almost all of what I just wrote from you guys, so all you had to do is open some articles from your archives and see this for yourself.


AHAH! The reason 1280x1024 looks so bad on Tom's Hardware monitors is that all of them are 1920x1200 or 2560x1600 widescreens and 4:3 resolutions don't scale properly. 1680x1050 does.
Score
0
December 26, 2009 12:32:33 PM

1280x1024 is 5:4 not 4:3 which scales even worse on 16:10 and 16:9 monitors, except in few games where you can manually adjust the aspect ratio regardless of the resolution (e.g Burnout Paradise, Lost planet..)
Score
0
December 26, 2009 12:49:12 PM

arkadihmm i have an idea. Give us, toms readers option to choose builds in each segment. Post few configurations for each budget for a survey, and build the winer.


Good idea. I think it would be simpler to realize if the survey only asked for which Cpu+Gpu's should be baught and the rest of the system would then be matched to the budget left
Score
0
December 26, 2009 5:14:18 PM

2 x XFX HD 5850 with an Intel cooler? Then why am I wasting $50 plus on buying Prolimatech or Thermalright for a single Sapphire 5770 card running Core i7-860 processor? Clarify on this confusing thing.

Next, what about a good LCD Monitor, OS (Windows) and accessories like Keyboard/Mouse, Speakers etc costs? Aren't part of any system? They won't come less than $400 for sure!
Score
-1
December 26, 2009 5:15:56 PM

UFO_WARVIPERThese articles really should represent parts that enthusiast are reasonably likely to expect to find on the market at least modestly close in price as constructed here for a short time after the article is posted.

UFO, we certainly can't help the discontinuation or price increase of a component after we purchase them. The 4870 was in and out of stock at $125 for weeks, and impossible to beat for the price. As pointed out by a reader ... HD 4870 for $130 shipped and in stock right now: http://www.zipzoomfly.com/jsp/ProductDetail.jsp?Product...
Score
0
December 26, 2009 5:18:12 PM

one of my giffs with the high end system is the use of a solid stat drive. I know they can dramatically increase load times and desktop responsiveness, but these builds are primarily about gaming, otherwise they wouldn't dump so much money into video cards. So if they would have just left out the solid state, that would have probably been another 5870 and a aftermarket hsf, which I think would have pushed the 2500 alot higher over the 1300 and 700 builds
Score
-2
December 26, 2009 5:27:08 PM

Um...I don't think the last post made ANY sense...what did I miss?
Score
2
December 26, 2009 5:31:20 PM

To clarify, an SSD will decrease load times, but this quarter's $2500 build didn't use one (i.e. it WAS left out). And, an aftermarket cooler WAS used. The poor OC results I think can be attributed pretty clearly to a case with insufficient airflow. Jonnyboy, did you read the same SBM article I did?
Score
2
December 26, 2009 6:27:08 PM

Still promoting EOL parts - End of Life parts?
(o yeh, invest in discontinued junk and rationalize the low resolution thing, and keep the idiot public in the dark).

Still promoting spintel?

Still misguiding your readers? ... or perhaps the only readers still here are spintel fanboys.

Shameless pro-spintel spinside lies, supported by lying "benchmarketing" tests, backed by spintel "sabotage compilers".

- this is old news to those that know - you must be desparate to harvest in a hurry.

... cos the big storm is coming as the ftc tears spintel apart - or perhaps you don't read the knews.

- search ftc antitrust and get edgy-kated.

yes, this happening now. and spintel can't buy it's way past the gov.
Score
-6
December 26, 2009 7:57:47 PM

sighQ2Still promoting EOL parts...and spintel can't buy it's way past the gov.


But what about the moon landing? AND JFK!?!?!
Score
2
December 27, 2009 6:36:59 AM

I think we should have 2 systems. One being Intel built and one being an AMD one. At all price ranges. What ever happened to the $500 builds?? Those almost always favored AMD, so they upped the $ to $550, $600, now to $700. When will we be back to a pure budget build?? Times are tough and a $200 price difference is quite a bit now. Here's what I'd consider:
Low end Budget: $500
1 CPU, 1 GPU, 1 HD

Mid Lower End: $750
1 CPU, 1 GPU, 1 or 2 HD's (maybe the extra drive is a storage drive), Aftermarket CPU HSF??

Mid: $1000
1 CPU, 1 or 2 GPU's, 1-2 HD's

Mid Upper: $1500
1 CPU, 1 or 2 GPU's, 1-2 HD's

Upper: $???? Whatever you like

To me that is a better break down on systems. Yeah I know there are more systems here, but in all reality most people won't have more than $500-$1500 to spend. I for one miss the true Budget builds that you founded these SBM's by. I think your losing the original SBM ideas. I could be wrong here, but I believe most will agree with me that the $500 SBM is missed.
Score
1
December 27, 2009 7:16:54 AM

lunyoneI think your losing the original SBM ideas. I could be wrong here, but I believe most will agree with me that the $500 SBM is missed.


I think that's a GREAT idea, but my vote doesn't count yet. Anyone want to second the vote for a $500 PC?

Personally, I thought the $500-$1500-$4500 SBM's were good because they included a super-budget PC and a dream machine. Would anyone like to vote for that?
Score
0
December 27, 2009 2:59:39 PM

psycho sykesMe.


That's one vote to include both the $500 and $4500 systems next time. Anyone else? Either or both?
Score
0
December 27, 2009 3:04:21 PM

I'd like to see $500, $1000, and $1500-$2000 price points. The high end is least important to me; everyone knows you can just throw in a SSD, or add more storage, or another GPU. The most interesting and creative solutions will be where it's not so simple, where trade-offs are involved. Also, the "winner" may not be just the highest benchmarks, it may be the system where I think "hmmm, another $75 gets me [this], and it will REALLY kick butt."
Score
1
December 27, 2009 4:06:32 PM

About the resolutions and picture quality. Something apparent to all of us in 2d is the image from our monitors always looks the best/sharpest in its native resolution. The same thing probably goes on in 3d. Running a game in your monitors native res maybe looks it has higher quality settings.
Grabbed this explanation,
"Native Resolutions

All LCD screens can actually display only a single given resolution referred to as the native resolution. This is the physical number of horizontal and vertical pixels that make up the LCD matrix of the display. Setting a computer display to a resolution lower than this resolution will either cause the monitor to use a reduced visible area of the screen or it will have to do extrapolation. This extrapolation attempts to blend multiple pixels together to produce a similar image to what you would see if the monitor were to display it at the given resolution but it can result in fuzzy images."

Is this true in 3d mode?
Score
0
December 27, 2009 7:14:51 PM

-1 for a $4500 PC. No one would put that together, and it would only show how your performance per dollar goes to zero after $2k or so. We all know that.

$650 seems to be the sweet spot between the $500 and $750 that people can gauge for themselves if they want to add/remove from a $650 build to reach a 500/750 price point. So I'd leave that there. Besides, adding 2 more build configurations to you would just make more work for you.

+1 with posting configurations (even if it is just CPU/GPU/RAM) that people can vote on to see put into the SBM. It would even be nice to see an AMD put up against a Intel in each price bracket, because it's really impossible to take a current SMB and compare it to a Previous SMB (using different architectures) and see an equivalent technology/price point comparison.

I'd also like to know if you guys can post a 'balance' rating on these builds. was the CPU limiting the GPU you configured? Were they well ballanced? I've read the 'build your own' articles numerous times, and wihtout doing oodles and oodles of benchmarking it's really hard to measure the balance of a PC without running the same benchmark against so many different configs. If you guys had some kinda tool that would be able to make that determination...or perhaps would just running the benchmark at 60%/80%/100% give enough MhZ data to know if the game would benefit from more CPU or less cpu?

I know that would mean a lot more data (ie a lot more benchmarking) but god...I hate WASTE, and i'd really like to know the optimal configuration of hardware configurations...of course, we're learning about that in the "build balanced pc" articles, but that info will be out of date in a few months.
Score
2
December 28, 2009 12:06:39 AM

CnoxI hate WASTE, and i'd really like to know the optimal configuration of hardware configurations...of course, we're learning about that in the "build balanced pc" articles, but that info will be out of date in a few months.


OK, who wants to vote that we pull the price caps off completely, have the low-end build target 1680x1050 games and "best value" application performance, the mainstream build target 1920x1200 gaming and "better" application performance, and the high end build target 2560x1600 gaming and "top" application performance?
Score
-1
December 28, 2009 12:07:09 AM

Nice find Mr. Henningsen $130. There second cheapest 4870 cards are $155, which is still cheaper than newegg's where they have scyrocketed back up to $180 on newegg. zipzoomfly.com - I'll have to remember that site. I want you to know that I've greatly enjoyed many of your articles, by the way.

Take any of my suggestions, with a grain of salt, especially since you're a professional journalist and have alot more 1st-hand experience publicly making hardware reccomendations to the masses, and probably are considering some things I may fail to recognize. To me, Reccomending an almost disappeared product like 4800 series cards still seems dicey for people like you and me. It can offer a compelling value if you're lucky enough to still find one near its lowest still-in-production price. But, when that one fleeting deal disappears, the people you help might come back and bitch at you and me when they can't find it.

When I help a friend or family or customer build or upgrade a computer, I take in account, how many boards with a particular GPU/chipset are available (in-stock) near the price of the cheapest one I can find. If all or nearly all of the other offerings, I can find are far removed from the price I found, I personally feel uneasy making a blanket recommendation for that item.

To a large extent, the SBM articles also double as "Best Gaming System for the money" article somewhat. The components chosen for each system in each price category really do serve as implicit reccomendations (or at least suggestions). Projected Market availability is still important but not as crucial for monthly editorials like Don's Best Gaming CPU and GPU For the Money articles.If a product suddenly falls into complete absence or unexpectedly skyrockets $30 higher 2 weeks after the article is written, there will will be a reference available in only 2 weeks to reflect that change. In a quarterly article like SBM, projected market availability plays much larger role during that term's reccomenations. Since readers will likely refer to this article until about a month before the next System builder marathon article, it would seem that we would want to recommend products whose supplies that won't dry up in a couple of weeks. We both know the supply of 4800 cards are about to be extinct because of AMD's previous announcement of ending production in favor of 57xx and 58xxx cards and rapidly increasing prices over the past few weeks confirm there is indeed a rapid reduction in supply, which has been reflected in rapid increases in price from most supply sources. Thus, it is predictable that the price is expected to rapidly increase relative to alternative products on the market. Admittedly, your recommendation is a perfectly valid and acceptable one if you view the lifespan of SBM articles as having a lifespan on the order of a couple of weeks. It all depends on what kind of role and purpose you intend the SBM articles to fulfill. I am sure other readers don't quite see these articles as fulling the same role as I see them, and that's ok.
Score
0
December 28, 2009 12:22:50 AM

Good, interesting, and like always say. Not always the more expensive is the best and this article check my theory :) 
Score
0
      • 1 / 2
      • 2
      • Newest
!