Sign in with
Sign up | Sign in
Your question

System Builder Marathon: Price/Performance

Last response: in Systems
Share
September 21, 2007 6:47:32 PM

http://www.tomshardware.com/2007/09/21/sbm_price_performance/index.html

After a grueling four days of benchmarking and tweaking, it's time to see how our three systems compare in baseline performance, overclocked performance, and total value.
September 21, 2007 7:56:36 PM

I think there is a sequencing problem with the images. It seems to start on page 8 with an extra Sandra 2005 Memory Bandwidth table. From there on it all the images are off by one resulting in the final Price/Performance image not appearing in the article. The last image, % Total Performance, is chart29.png, but if you open that in your browser and change it to chart30.png you get the Performance Per Dollar chart.

I love these System Builder Marathons, including this one, but at the end I felt like the analysis/interpretation was cut short. Earlier on it was said that the results of these tests would influence the recommendation of the E6750 vs. Q6600 processors but there is no mention of this in either the Value Analysis or Conclusion. I would have liked to see more discussion of the difference the various components made and maybe a note about what might change in the near future with new releases.

Thanks for taking the effort to get through these marathons, they are a great resource.


September 21, 2007 8:11:44 PM

Nice job, but I agree with the above post.

You guys should do this again in November with some Penryns and an X38.

EDIT:
The comment on the high-end system being restricted by it's CPU speed in Oblivion makes me think that even with a 3.9GHz quad-core, a pair of top-end SLI cards will only really benefit newer games with brute CPU power. Perhaps it's an indication that the extra performance from the upcoming Penryn (when overclocked) will finally let demanding games (and SLI) do their thing. With titles like Crysis and FarCry2 on the way, they most definitely will be optimized for multi-core.

It seems the system of choice for the next two years will be a quad-core Penryn overclocked on either an X38 with CF or nVidia chipset with SLI.
Related resources
September 21, 2007 8:22:44 PM

I also love these reviews/comparison of systems. The only thing is that even though these systems seem to perfectly reflect the low,mid, and high range systems generally built, it doesn't take customization into consideration (which I know would be impossible). But some switching around of components would have been worth mentioning. For example what would the low end oc system have accomplished with a beefier psu and an additional video card? Or what if some extra money was spent on the GTX instead of the GTS? That I know would have brought the graphics performance much closer to the other two systems.

The other thing that I found interesting was the low oc results for the Q6600. They were so low I think it would be deffinitely worth testing out the system with the G0 stepping version. My current system (with little effort) gets 3.8Ghz stable overclock with the G0 Q6600. I am using the (I believe the same) Patriot xtreme ram, but on a gigabyte board. Temps are also very low (33 idle 50 load).

Well, I enjoyed this series very much and greatly look forward to more in the future!

animefanotaku
September 22, 2007 1:05:10 AM

I love how my system would have been high end 6 months ago, but now barely qualifies as low OC.

Mad props, Intel, always keepin the little man down, you bastard.
September 22, 2007 3:34:47 AM

The last Marathon series was what inspired me to do my 1st build.

Having done the build, this Marathon series was sooo much more interesting. I now appreciate the subtleties of the choices one has to make based on budget, requirements, personal taste, etc.,

Great job.

The one thing that I keep thinking about is, in Audio systems, they have a saying, "a system is only as good has the speakers". So what is the equivalent, for builds? I would argue that it is the input, speakers and monitors.

While those 2 components might have a longer "shelf life", it is really a big part of the price/value equation.

So how about in a future Marathon, we raise the budget to $1500, $3000, and $6000, but with keyboard/mouse/gaming input, speakers, and monitor?

Is $2K too much to add to the high end build? I dont think so. Why spend $4K to use a 20" monitor, and on-board audio, and $50 speakers? I say 30" @ 2560x1600, 5.1 Audio, etc,...

I think that would be interesting because, if one is to buy\build a new system, those pieces would also have to be priced in.

Of course, how the hell would you do a price/performance comparison????

That is your challenge THG :hello: 

Thanks again for the great work. Keep them coming!!
September 22, 2007 4:06:53 AM

Yea, we actually need the price/performance chart in there :lol: 
September 22, 2007 5:13:07 PM

Spelling Error: "Intel Core Duo e6750 " - missing a 2.
September 22, 2007 6:49:30 PM

Conceptuweasel said:
I think there is a sequencing problem with the images.


It's fixed now, all the charts are there.
September 22, 2007 8:40:24 PM

I know there must be a comfort feeling with using familiar games as benchmarks, along with being able to compare rates over many years, but you really need to put in some new, pc punishing games, like Company of Heroes, BioShock, STALKER, Call of Juarez, etc. I bet your low cost build would choke on the games most pople are playing today (as Oblivion nearly did), esp at the resolutions they play at.
September 22, 2007 10:22:23 PM

Prices vary so quickly that it is difficult to judge true price/performance. I just purchased two new machines from Dell about a month ago. $609 got me a Q6600 machine with 2GB Ram and all of the trimmings. An additional $250 got me an 8800 GTS and power supply upgrade. $859 for a nice quad core system with OS. Most of what I do is math oriented so the quad core crushes my old machine.

Funny thing is Dell stopped offering the system (I guess it was eating into their high end systems). My wife and I are set for another 4 years. Maybe the TerraScale project will be done by then :) 
September 23, 2007 3:21:47 AM

netsez said:
I know there must be a comfort feeling with using familiar games as benchmarks, along with being able to compare rates over many years, but you really need to put in some new, pc punishing games, like Company of Heroes, BioShock, STALKER, Call of Juarez, etc. I bet your low cost build would choke on the games most pople are playing today (as Oblivion nearly did), esp at the resolutions they play at.

Cleeve has said several times that they are updating the benchmarking suite soon. Oblivion is still relavent IMO because it still brings even the fastest computers to their knees.
September 23, 2007 6:15:38 AM

Yep, a new benchmark suite is in the works. Sorry it's taking so long fellows but we're trying to come up with something we can synchronize across all of our sites. Some red tape to cut through, but it IS coming.

In the meantime, for articles I personally write I'll include Bioshock andenemy Territory: Quake Wars, OK? :) 
September 23, 2007 6:21:25 AM

Although the price/performance ratio of the low(er) cost system is excellent you also have to figure how often you change out components and what the lifespan of any of these components are. For example the low(er) cost system is pretty good, but in 2 years it'll pretty much suck. However the high(er) cost system will probably still be on the good end in a few years, although maybe not the best. I'd probably still get the lower cost system because even if it's not the best in 2 years it'll still be alot cheaper for right now even if you can see an occasional stutter, also in a few years things will be so different that I'd probably "have to" change things anyway

On another note, I don't really think that the e6750 is a low end CPU, it's not high end but why not choose something like an e2180, or an e4300 as a low(er) end option. I think the e6750 is pretty good for the price. It even beats the e6600.
September 23, 2007 6:43:21 AM

Quote:

In the meantime, for articles I personally write I'll include Bioshock andenemy Territory: Quake Wars, OK?

Nice!

Now include Crysis and I will be satisfied ;) 
September 24, 2007 5:08:36 PM

It seems to me that these performance reviews are somewhat misleading. When benchmarking in terms of gaming performance there will be a threshold of fps above which improved performance isn't discernible. I would find it interesting to know what that fps threshold is and see value based discussions based up that, a 100% value should correspond to that threshold value.

Also, as a scientist, I would like to see better FP analysis done. I've seen many discussions where the poster claims that even now the FP performance of the AMD chips exceeds that of the newest Intel chips. If this is true that would be a major deciding factor for those who are doing scientific computations.
September 24, 2007 5:54:16 PM

hjruf1 said:
It seems to me that these performance reviews are somewhat misleading. When benchmarking in terms of gaming performance there will be a threshold of fps above which improved performance isn't discernible. I would find it interesting to know what that fps threshold is and see value based discussions based up that, a 100% value should correspond to that threshold value.

Also, as a scientist, I would like to see better FP analysis done. I've seen many discussions where the poster claims that even now the FP performance of the AMD chips exceeds that of the newest Intel chips. If this is true that would be a major deciding factor for those who are doing scientific computations.


As a scientist you should know then that something like a discernible threshold in graphics is subjective at best. As each human is different and thus their perceived visual capacity is different then there is no way you could say "above this line no-one can see the difference" without making the line so high that it is irrelevant for testing.

Just my 2 bits.

I do agree that FP numbers for a cpu test would be nice as it is probably useful for many tom's readers... but who knows if they will get that in there... it tends to polarize the fanboi readers. ;) 
September 24, 2007 8:32:05 PM

Quote:
In DivX, the high-performance system was able to convert DVD video to MPG4 nearly three times as fast, a boon to anyone who's looking to back up his or her entire movie collection to hard drives


Not really. You still have to wait the same amount of time to shovel the data off the DVD drive. Then you still take the same amount of time bumbling with software settings. Then you set it to run overnight, no? So why would it matter at all?
September 25, 2007 4:57:14 AM

shadowmaster625 said:
Quote:
In DivX, the high-performance system was able to convert DVD video to MPG4 nearly three times as fast, a boon to anyone who's looking to back up his or her entire movie collection to hard drives


Not really. You still have to wait the same amount of time to shovel the data off the DVD drive. Then you still take the same amount of time bumbling with software settings. Then you set it to run overnight, no? So why would it matter at all?


wow... I can't remember the last time I set anything to run overnight. If I am doing something that takes that long (outside of massive pro-level 3d rendering) then it is time to upgrade. ;) 

seriously though, I have been looking and experimenting with options to get my entire collection (currently at about 450 dvds) onto disk and if I had to set it to run only one overnight that would take FOREVER... believe me when I say that more cores is where I have been looking as my dual core just does not cut it.

also, you don't bumble w/ software settings. Once you have things set up and your ripping software set correctly there is nothing to bumble with.

I am not flaming you, just pointing out that a 3x increase in speed on that alone truly IS a draw.
September 26, 2007 7:40:57 PM

I thought the entire project was very well thought out and presented. Thanx again for taking the time to do this. I'd imagine that the building and benchmarking were fun, but converting it all to data can be a PITA.

For me time is money. So the best bang for the buck for me is still a high end system, which this project reinforced quite well.

I think there are a few components that don't necessarily need to be the most expensive (EG: I'm quite happy with the performance of the E6750@3.6GHz on air, and knowing it'll do 4.0GHz if needed.) If/when quad core is taken advantage of by apps, my opinion on that may change. But multi-tasking, ripping, burning, encoding, and gaming haven't suffered in the least for now. I was quite happy to sell my Q6600 with GO stepping and keep the E6750.

And it's hard to argue with the price & performance of Crucial Ballistix PC2-800. Awesome product. Spending 3x as much on Corsair or Super Talent just doesn't make any sence.
September 26, 2007 8:04:10 PM

Dark41

hear here.

Scientists love numbers, it's in their genes. Statistics have a tendancy, however,to be manipulated by the statisticians

We strive to be accurate. <smile>
October 1, 2007 5:28:59 PM

I love numbers too. I'm fortunate that as system builders we test a lot of components, so I'm rarely baffled by BS. :) 
October 5, 2007 9:50:36 PM

I have been baffled by BS, and I am not afraid to admit it.

Fortunately, real world pereformance requires nothing but observation.
November 28, 2007 12:26:52 AM

I just wanted to say that I'm running an Opty 185 @ it's stock 2.6ghz with an EVGA 8800GTS 640mb card, and Oblivion crawls at parts with high detail on. Pretty sure that some good overclocking would help my system but it's obvious to me that my processor is the main bottleneck here. Oblivion also makes frequent HDD accesses, and even with raptors they become a bottleneck as well. Considering that the GTX still runs over $400, I think a lot of people are out there spending money on graphics cards when they ought to be putting more cash into upgrading their current platforms instead. Hopefully the future will SOON bring us multi-core applications and flash hard drives, or we're going to continue to dribble about with small increases just to get a slight edge. We get new techs like DDR3 and faster bus speeds but is it really that big of a gain when your graphics card is still waiting on your CPU and your stupid mechanical hard drive?
November 28, 2007 8:32:20 AM

THIS REVIEW IS TOTALLY POINTLESS BECAUSE OF THE GRAPHIC CARD OPTIONS: Who the hell wants to know about 3 system that have been tested 1000times in 1000 other reviews. Show me 8800GT or 3870, show me something i don't already know. And ya somebody pointed out earlier who the hell picks a e6750 for a Low end system. And the high end system has pc6400 cl4 PLZZZZZ Show me some DDR3 or something. Oh ya RAID docent make any kind of impact so thats just waste of test...

Oh ya PLZ explain the choices of PSU

OLD NEWS ARE SO EXITING !!!!!!!!
November 28, 2007 9:08:03 AM

genored said:
THIS REVIEW IS TOTALLY POINTLESS BECAUSE OF THE GRAPHIC CARD OPTIONS: Who the hell wants to know about 3 system that have been tested 1000times in 1000 other reviews. Show me 8800GT or 3870, show me something i don't already know. And ya somebody pointed out earlier who the hell picks a e6750 for a Low end system. And the high end system has pc6400 cl4 PLZZZZZ Show me some DDR3 or something. Oh ya RAID docent make any kind of impact so thats just waste of test...

Oh ya PLZ explain the choices of PSU

OLD NEWS ARE SO EXITING !!!!!!!!


Wow you're an idiot! I can't remember the last time I've seen such a stupid post. The 8800GT and 3870 didn't even exist in September. And the high-end system used some of the best memory available, it does 1200MHz quite easily. Nearly everyone here agrees DDR3 is a waste of money right now.
!