Saturday, September 22, 2007

low cost system...utk mat salleh jer....kat Msia mahal giler laa

Introduction

In our previous system builder marathon, we built a budget PC for about $500. It performed really well compared to higher-end systems when we factored in the amount of money invested, but unfortunately, our $500 low-cost system didn't have a lot of overclocking potential to go up against the big boys.

In our overclocker marathon, we configured a new budget PC with an overclockable e4300 CPU and Geforce 8800 GTX video card. The final price was under $1,200, but it had what it takes to overclock and perform in the same league as PCs twice its price.

This time, our low-end machine will have a price cap of $1,000 and our medium grade system will be capped around $1,500, so we can really see what that extra $500 can get you. These higher price points will also make our systems better suited to go head-to-head in an overclocking marathon at a later date.

Now that we've doubled our budget from our original $500, what hardware will our $1,000 buy us? Let's look at the components we chose and discuss our rationale for selecting them.
CPU: Retail Intel Core 2 E6750



At just over $200, the new E6750 is an extremely easy choice for a low-cost system.

It seems it was only a few months ago that the E6700 was one of the most powerful CPUs available - and that's because it was. The E6750 offers a much lower price, coupled with a higher front side bus speed than its older sibling. The only bad thing you could say about the E6750 is that it's still a dual core part at a time when quad cores are becoming more popular. As far as it impacts the user experience, though, dual core still has a long life of usefulness ahead of it.

Thomas Soderstrom has chosen the new quad core Q6600 for the midrange system that Shelton Romhanyi will be building and testing. Watching what happens when the E6750 goes toe to toe with the Q6600 later in the marathon, should make for a very interesting read.






The new E6750 has a higher front side bus than the older Core2 Duos, which requires a newer motherboard than the Asus P5B we used in our last budget overclocking system.

Even though it's relatively new, the P5K already has a good reputation for overclocking, which will come in handy later in the series when we push these systems past their stock speeds. But for now, the P5K is a solid board for our low-cost system, at a decent price.

CPU Cooler: Cooler Master HyperTX 2




Cooler Master's HyperTX 2 is a capable cooler for a very attractive price. At under $30 including an active fan, the HyperTX 2 is a great choice for a budget system, especially if you plan to overclock a bit on the cheap.

Memory: 2 GB Of Wintec AMPO PC2-6400





We've used 2 GB of Wintec AMPO in our budget overclocking marathon system, and we're happy to do it again. This RAM's overclocking reputation remains, but its price has come down, so picking up two gigabytes of the stuff is even easier on our pocketbook. We found two 1 GB sticks of it for just $81.
Hard Drive: 320 GB Seagate Barracuda SATA II 7200 RPM

Our last budget system had a 250 GB drive, but hard drive space is always something that gets cheaper over time. Our choice for the current low-cost build is a Seagate Barracuda 320 GB SATA II drive. This is a fine amount of space for a budget machine and is a fairly quick performer for the price, to boot.

PC Case: Raidmax Smilodon




We've spent a little time with the Smilodon since our last budget marathon, and we like it even more now. There aren't a lot of cases out there that allow you the option of folding out the motherboard tray for ease of setting up. Our only beef with the case is that when it's used in conjunction with a large CPU cooler - like the Cooler Master HyperTX 2 - there isn't enough clearance to open the motherboard tray. Regardless, the Smilodon is a great case with lots of airflow at a very reasonable budget price of $80.
Power Supply: Fortron Source AX450-PN




Fortron has established a reliable reputation for itself, and we're pleased to once again select its budget-priced AX450-PN power supply for use in our low-cost system. At a low cost of $58, this PSU supplies a total 36 A of 12 V power split between two 12 V rails rated at 18 A each. This is a good companion to the 8800 GTS we've selected for our build.

Optical Drive: Sony NEC Optiarc 7170 SATA



At $35, the Optiarc 7170 is a very speedy DVD-RW drive. Add to this the convenience of SATA cabling, and you have a real winner.

Video Card: EVGA Geforce 8800 GTS 320MB





There's really not much to say about the Geforce 8800 GTS 320MB except that it's incredibly powerful for the price.

Compared to its 8800 GTX and 8800 GTS 640MB brethren, the only notable performance difference appears at resolutions 1600x1200 and above. This does not mean that the 8800 GTS 320MB can't handle some 1920x1200 gaming mind you, the GTX will just be faster.

For $280, there's no better videocard you can put in your machine, period.

Test System & Benchmarks

System Builder Marathon: New Low-Cost PC Component Costs
CPU Intel Core 2 Duo e6750 $212
CPU Cooler Cooler Master HyperTX 2 $27
Motherboard Asus P5K $137
RAM Wintec Ampo DDR2 PC2-6400 - 2GB $81
Graphics GeForce 8800 GTS 320MB $280
Hard Drive Seagate Barracuda 320GB SATAII $80
Sound Onboard $0
Case Raidmax Smilodon $58
DVD-RW Sony NEC Optiarc 7170 SATA $35
Total Price $990

System Builder Marathon: Old Budget Overclocker PC Component Costs
CPU Intel Core 2 Duo e4300 $120
CPU Cooler Thermalright XP-90 & 120mm fan $55
Motherboard Asus P5B $115
RAM Wintec Ampo DDR2 PC2-64 00 - 2GB $81
Graphics GeForce 8800GTX $500
Hard Drive Western Digital Caviar 250GB $63
Sound Onboard $0
Case Raidmax Smilodon $80
Power Aerocool Zerodba 620w $125
DVD-RW Sony NEC Optiarc 7170 SATA $35
Total Price $1,174

We're going to compare our new low-cost system with the budget overclocker's system we created for our overclocking marathon. Our budget overclocking system is similarly priced to our new low-cost system: it cost $1,174, compared to the new low-cost system's $990 price tag.

The single item that raised the budget overclocker's system price above $1000 was the inclusion of a Geforce 8800 GTX. Had the old budget overclocker's build sported an 8800 GTS 320mb instead of the GTX, its price would have been almost identical to the new low-cost system: about $950. It's going to be interesting to see how the old e4300 CPU/8800 GTX combo will fare against the new e6750/8800 GTS 320MB combo in the gaming arena.

We won't be comparing the new system to the original budget system in the first system builder marathon. The original budget system was put together for $500 and is completely outclassed by our new $1000 budget champion.

System Hardware
Processor Intel Core Duo e6750, 2.67 GHz, 1333 FSB, 4MB Cache
Motherboard ASUS P5K, BIOS: 0507
RAM Wintec Ampo PC2-6400, 2x 1024 MB, CAS 5.0-5-5-16
Hard Drive Seagate Barracuda ST3320620AS 320 GB, 7200 RPM, 16 MB Cache, SATA 300
Networking PCIe Gb LAN controller featuring AI NET2
Graphics Cards EVGA GeForce 8800 GTS, 320MB RAM
Power Supply Fortron AX-450PN
System Software & Drivers
OS Microsoft Windows XP Pro 5.1.2600
DirectX Version 9.0c (4.09.0000.0904)
Graphics Driver nVIDIA ForceWare 162.18
Benchmarks and Settings
3D-Games
F.E.A.R Version: 1.0 Retail Video Mode: 1024x768 Computer: Maximum Graphics: Maximum-Custom-No Softshadows Test Path: Options/Performance/Test Settings
Doom 3 Version: 1.3 Video Mode: 1024x768 Graphics: High Test: Time Demo 1
Oblivion Version: 1.0 Retail Video Mode: 1024x768 Graphics: High, 4xAA Test: Inside & Outside save files
Audio
Lame MP3 Version 3.97 Beta 2 (12-22-2005) Audio CD "Terminator II SE", 53 min wave to mp3 160 kbps
OGG Version 1.1.2 (Intel P4 MOD) Version 1.1.2 (Intel AMD MOD) Audio CD "Terminator II SE", 53 min wave to ogg Quality: 5
Video
TMPEG 3.0 Express Version: 3.0.4.24 (no Audio) fist 5 Minutes DVD Terminator 2 SE (704x576) 16:9 Multithreading by rendering
DivX 6.1 Version: 6.1 Profile: High Definition Profile 1-pass, 3000 kbit/s Encoding mode: Insane Quality Enhanced multithreading no Audio
XviD 1.1.0 Version: 1.1.0 Beta 2 Target qantizer: 1.00
Applications
Autodesk 3D Studio Max Version: 8.0 Characters "Dragon_Charater_rig" rendering HTDV 1920x1080 & 1280x720
Synthetics
SPECviewperf 9 Version: 9.0.3
PCMark05 Pro Version: 1.1.0 System, CPU and Memory Tests Windows Media Player 10.00.00.3646 Windows Media Encoder 9.00.00.2980
3DMark05 Version: 1.2 System Test Only
SiSoftware Sandra 2005 Version 2005.7.10.60 CPU Test = Arithmetic, Multimedia Benchmarks Memory Test = Bandwidth Benchmark

3D Games

When we benchmarked the old budget overclocker's system, we were forced to cap our testing resolution at 1600x1200, because at the time there wasn't a high-end monitor available in that particular lab.

This situation has been remedied for the new system tests, so we will show you results up to 1920x1200 in our final comparison articles between the new low-end, midrange, and high-end systems. For this budget article, however, we will only use 1600x1200 and lower




We see that Doom3 is bottlenecked by the e4300 processor at about 110 fps across the board in the old system. The new e6750 unleashes a lot more performance at lower resolutions.

At 1600x1200 with 4xAA and 8xAF, however, the 8800 GTX in the e4300 system stretches its muscles a bit, and we see it take a small lead from the e6750/8800 GTS 320MB combo. At higher resolutions, this trend would likely have continued and the gap would have widened.




F.E.A.R. shows a very similar story to what we saw with Doom3, with slightly more exaggerated results. With 4xAA and 8xAF enabled, the e4300/8800 GTX combo takes a small lead at 1280x1024, and a large lead at 1600x1200.














resolutions minimize the CPU impact, and maximize the graphics card impact on game performance.

Once again we see that high-resolution gaming can usually be done quite acceptably with a slightly slower CPU paired with a fast video card. Our conclusion article later this week will probably show an even larger spread between the 8800 GTS 320MB and the 8800 GTX at 1920x1200.

Audio




Unlike the gaming benchmarks, we see here the true difference between the e4300 and e6750 CPUs, with no graphics card influence. The e6750 in the new low-cost machine pummels the e4300 with about 150% of the latter's speed




Once again, we see an incredible 50% speed increase from the e6750.

Let's see if these results hold up in the video arena.

Video




It looks like it doesn't matter whether you're encoding audio or video - the e6750 will be about 50% faster than the e4300.




Different codec, but the same story. The percentage performance difference between DivX and Xvid encoding is negligible.




It's beginning to sound like a broken record, but the magic 150% number keeps showing up. It's no surprise that the e6750 truly looks impressive compared to the e4300

Application




Our first synthetic benchmark is 3dMark05, and once again we see what happens when the e4300 is combined with the powerful 8800 GTX. The e4300 bottlenecks the 8800 GTX's graphics performance so consistently that the results hardly change regardless of the resolution or image quality enhancements.

On the other hand, the e6750 allows the 8800 GTS to work to its full potential. At low resolutions, the e6750/8800 GTS 320MB combo easily outpaces the e4300/8800 GTX combo. Just like our game benchmarks showed, though, at higher resolutions the GTX flexes its muscles and doesn't budge, while the 8800 GTS 320MB starts to show its limitations.




Specviewperf is an interesting benchmark, because it stresses both the CPU and video card. You can really see which benchmarks are CPU dependent (where the new system has a large lead) and which benchmarks are graphics card dependant (where the old system is tied or has a slight lead). On average, the new system wins by a notable margin.




Applications, Continued

We experienced some difficulty getting PCMark05 to complete the graphics benchmarks, so we couldn't get a total system score to compare. The WMV encoding test apparently has some problems on certain systems, but we'll see if we can get things running for our conclusion article.

The system and memory tests show results close to the 150% speed increase we've seen before. The hard drive test shows a 133% increase, which is certainly noteworthy.




Once again, pretty much every CPU benchmark shows the same 50% performance increase. Memory speed is even higher - around 160% of the previous system's performance - which might be due to the higher bus speed, and memory optimizations on the Asus P5K board.

Performance Summary




When we look at the game performance difference, we have to keep in mind that this chart represents an average of game performance at 1600x1200 and below.

It looks like the new e6750/8800 GTS 320MB combo enjoys an advantage over the e4300/8800 GTX combo, but this isn't really true: at 1600x1200 and above, the 8800 GTX will perform better, even when paired with a slower CPU. This will become more apparent in our summary article, when our e6750/8800 GTS 320MB goes up against the midrange system, which sports an 8800 GTX.




There's no special interpretation to be made here: the e6750 CPU performs software applications at about one-and-a-half times the speed of the e4300. This is an incredible speed difference, one that really matters in real-world applications.




The synthetic benchmark summary is a little fudged, because we don't have proper results for PcMark 2005 - as we stated, we couldn't get a total system score due to a benchmark glitch. We brought down the new system's average score by assigning it an arbitrary 100% compared to the old system, even though its actual score would likely have been much higher. Even with this artificial handicap, though, the PcMark results were on average 133% faster than the old system.

3dMark 2005 is a synthetic graphics benchmark, and tells pretty much the same story as actual gaming benchmarks did. Looking to Specviewperf and Sisoft Sandra, we see the same 150% performance numbers from the new system that we saw in the application benchmarks.




When we average everything out for a figure to represent the overall performance of the new system vs. the old system, we get a figure of 136%. The interesting part is that the new system costs less than the old one! This will make for an interesting value analysis indeed.




With a lower total cost and higher performance at lower resolutions, the new budget system's overall gaming value is much higher than just the performance delta would suggest. We see a number of 140% representing its gaming value compared the old budget overclocker's system.

Once again, we need to take these figures with a grain of salt, as the performance of the slower CPU with faster video card would have shone at higher resolutions we weren't able to test in this review, specifically above 1600x1200.




The total system value looks very impressive. When the new budget system's lower cost is factored in along with its superior performance, it looks to be an incredible value: about 160% of the value of the old budget overclocker's system overall.
Conclusion

The conclusion is obvious: $1,000 buys a very formidable PC in today's market, thanks in no small part to Intel's new e6750 CPU. While it is true that many parts of the budget PC have upgraded in speed since our last marathon, it's the e6750 that looks to have the most performance increase for the buck.

This might be a 'budget' machine, but there's nothing this low-cost PC couldn't eat up and spit out. I don't think the $1,000 price point has ever seen such fine performing hardware.

The real question is: how will this $1,000 dual-core system compare to a $1,500 quad-core system when we calculate value per dollar? Stay tuned this week to see the results!
Editor's Opinion

The results I'm personally looking forward to the most will be the comparison between this budget e6750 machine and the midrange q6600 machine brought to life by Shelton Romhanyi and Thomas Soderstrom. With a $500 price difference, will the extra two cores in the Q6600 CPU allow the midrange system to surge ahead in applications? Or are the applications we use in our test suite not yet sufficiently able to take advantage of quad core optimizations? We'll have to see.


No comments: