Sign in with
Sign up | Sign in
Time To Upgrade, Part 2: Picking The Parts To Replace
By ,
1. Old Vs. New: Where To Spend Your Upgrade Dollars?

We recently looked at the effects of replacing the pieces in a 2007-era mainstream PC with modern components. We took a P35 platform with a Core 2 Quad, GeForce 8-series card, and a Western Digital Raptor hard drive, and then compared it to a brand new P55-based platform with a Core i7-870, a Radeon HD 5870, and a SandForce SF-1200-based SSD.

The results proved what most people expected: the differences are significant, but it only makes sense for an enthusiast or gamer to replace the entire configuration. What if we were to look at the component upgrades individually? Today we're taking the same components and analyzing three primary upgrade routes:

  • Motherboard/processor/RAM upgrade
  • Graphics upgrade
  • HDD/SSD upgrade


Sometimes you don’t need to go all the way and replace the entire PC. We're looking at which component provides the best upgrade value.Sometimes you don’t need to go all the way and replace the entire PC. We're looking at which component provides the best upgrade value.

Some readers criticize this type of article, saying it’s obvious that newer hardware will outperform older components, be more efficient, and both. We can’t argue with that. Our real question is to ask how large the differences are, and whether or not an upgrade actually makes sense for you. In more and more cases, the overabundance of performance available through current-gen hardware is almost irrelevant given that mainstream applications only use a fraction of the system's potential. Users replacing an older Porsche 911 with a new model indeed get more performance and more technology, but they probably won’t take advantage of those improvements unless they visit a race track, or at least drive during off-peak traffic hours.

The same applies to PCs. Ever since dual-core CPUs arrived, we've shed most multitasking bottlenecks. Applications that take advantage of multiple processing cores have realized greater performance gains than any clock speed increase could have delivered in the past. This results in more than enough performance for mainstream users, while enthusiasts and professionals remain on the hunt for state-of-the-art hardware.

Clearly, it doesn’t make much sense to outright dump a three-year-old PC, but there are ways to optimize performance for less cost than what a complete replacement would run. We decided to break down our upgrade package from that previous article and look at the benefits each component introduces separately.

Once again, the base system is a MSI P35 Platinum with a 2.66 GHz Core 2 Extreme QX6700 and 4 GB of DDR2 memory along with a GeForce 8800 GTS and WD’s 150 GB Raptor drive. You’ll find the upgrade components listed on the following pages.

2. Scenario 1: Motherboard/CPU/RAM

In this first set of test runs, we decided to replace the platform, meaning we exchanged the motherboard, processor, and main memory. We did not modify the graphics card or the hard drive, assuming that such an upgrade is most attractive to users needing lots of computing power for rendering, audio/video transcoding, or workstation-type workloads. The reason for swapping all three components is clear: the platform, CPU, and memory are all interdependent.

Motherboard: P55A Fuzion (Intel P55 Express)

We used this motherboard in the last article for its mainstream P55 platform. Most board makers include P55-based motherboards in their portfolios to address overclockers, because the LGA 1156 interface currently offers the best total value in Intel’s current lineup.

MSI’s interpretation of a modern P55 motherboard is based on the P55 Express and the Lucid LT22102 ("Hydra"), which is a PCI Express 2.0 SoC (system on a chip) switching solution with a total of 48 PCIe lanes. It can be used for PCIe load balancing, or enabling dual-card graphics setups. This includes pairing different GPUs from the same vendor and even cross-vendor setups in which an AMD card runs with an Nvidia card. We'll look at this feature soon in an upcoming article.

Thanks to the board's two PCI Express x16 slots, you can accommodate a pair of graphics cards or a graphics card and a high-bandwidth expansion card, such as a storage controller. Just remember that, even though Lucid's solution enables a pair of true x16 links, you're still limited to the bandwidth of Intel's 16-lane (total) integrated PCIe controller. Two more x1 PCIe slots and two 32-bit PCI 2.3 slots accept additional expansion cards. HD audio, gigabit networking, USB 3.0 ports, and a SATA 6 Gb/s controller are also included.

CPU: Intel Core i7-870 (2.93 GHz Quad Core)

We’re using the same quad-core, 45 nm Core i7-870 processor for this comparison that we used in our prior 2007 vs. 2010 story. With Turbo Boost, the 2.93 GHz CPU can crank up by as much as 667 MHz under single-core utilization. Even though Intel is gradually switching to 32 nm manufacturing, this Lynnfield-based chip remains both fast and efficient.

RAM: 2 x 2 GB DDR3-1600 OCZ Gold

Memory is less relevant in this shootout, since the performance difference between entry-level speeds (DDR3-1066) and hardcore overclocking memory (DDR3-2000 and faster) isn't very substantial. In fact, high-end memory only really makes sense in situations where lofty reference clock rates push memory data rates beyond the range of mid-range memory modules. In those cases, the extra headroom of a high-end kit might be the key to enabling an ambitious configuration. We grabbed some OCZ DDR3-1600 Gold series memory as a reasonable compromise.

3. Scenario 2: Graphics Upgrade

Our second scenario is probably most interesting for gamers, given that a fast quad-core CPU from 2007 typically still delivers enough performance (especially if you've overclocked it). Experience shows that you should maximize graphics horsepower before going after more CPU cores. Our step from the GeForce 8800 GTS to AMD's Radeon HD 5870 provides a significant jump in both performance and visual quality. The latest cards support DirectX 11, while the GeForce 8-series tops out at DirectX 10.

The pricey MSI R5870 Lightning is similar to most Radeon HD 5870-based cards, but the product arrives overclocked to a 900 MHz GPU speed, and its beefed-up cooling solution allows the product to run at higher clock rates and lower noise levels than AMD's reference cooler. You'll have to decide whether or not that last few extra percent in performance and the noise reduction are worth MSI's significant price premium.

4. Scenario 3: SSD Upgrade

This third option may not appear as significant as the previous two. However, SSDs have overtaken hard drives in terms of synthetic and real-world performance, so it makes sense to consider a simple hard drive replacement. A modern SSD is still rather expensive per gigabyte, but it delivers tangible performance gains, nevertheless.

The G.Skill Phoenix is based on the popular SandForce SF-1200 controller. SandForce solutions work without additional cache memory, unlike Indilinx, Intel, Samsung, or Toshiba architectures. Instead, it utilizes part of the flash memory to handle wear leveling and performance management features, such as algorithms that deal with operations to minimize write amplification.

This SSD, like many others based on a similar architecure, delivers more than 200 MB/s of read and write throughput plus excellent I/O and application performance. We examined how this would impact our 2007 system.

5. Test Systems And Benchmarks
System Hardware
HardwareDetails
Performance Benchmarks
Base System
Motherboard
(Socket LGA 775)
MSI P35 Platinum (Rev. 1.0), Chipset: Intel P35, BIOS: 1.C (30/12/2009)
CPU IntelIntel Core 2 Extreme QX6700 (65 nm, 2.66 GHz, 8 MB L2 Cache, TDP 130 W)
RAM DDR2 (dual)4 x 1 GB DDR2-800 (Corsair CM2C1024-6400C4 5-5-5-15)
GraphicsZotac GeForce 8800 GTS, GPU: G80, Graphics RAM: 320 MB GDDR3 (900 MHz), Stream Processors: 96
Hard DriveWestern Digital Raptor, 150 GB (WD1500ADFD), 10 000 RPM, SATA 3Gb/s, 16 MB Cache
Power SupplyPC Power & Cooling, Silencer 750EPS12V 750 W
Upgrades
Upgrade I: Platform
Motherboard
(Socket LGA 1156)
MSI P55A Fuzion (Rev. 1.0), Chipset: Intel P55, BIOS: 1.0 (07/12/2010)
CPU IntelIntel Core i7-870 (45 nm, 2.93 GHz, 4 x 256 KB L2 and 8 MB L3 Cache, TDP 95 W, Rev. B1)
RAM DDR3 (dual)2 x 2 GB DDR-1333 (OCZ3G2000LV4GK)
Upgrade II: Graphics
Graphics CardMSI R5870 Lightning TwinFrozr II, GPU: Cypress (900 MHz), Graphics RAM: 1024 MB GDDR5 (2400 MHz), Stream Processors: 1600
Upgrade III: Hard Drive/SSD
SSDG.Skill Phoenix 100 GB SSD (FM-25S2S-100GBP1), SandForce SF-1200, SATA 3 Gb/s
System Software & Drivers
Operating SystemWindows 7 Ultimate x64
Updated on 2010-07-29
Drivers and Settings
ATI Radeon DriverATI Catalyst 10.7 Suite for Windows 7
Nvidia GeForce DriverGeForce Driver Release 256 258.96
Intel Chipset DriversChipset Installation Utility Ver. 9.1.1.1025
Intel Storage DriversMatrix Storage Drivers Ver. 8.​9.​0.​1023
3D Games Benchmarks and Settings
BenchmarkDetails
Grand Theft Auto IVVersion: 1.0.3
Video Mode:
- 1680x1050, - Aspect Ratio: Auto, - All options: Low, - View Distance: 10, - Detail Distance: 100, - Vehicle Density: 100, - Shadow Density: 16, - Definition: On, - Vsync: Off, In-game Benchmark
Left 4 DeadVersion: 1.0.0.5
Video Mode: 1680x1050
Game Settings
- Anti Aliasing: None
- Filtering: Trilinear
- Wait for vertical sync: Disabled
- Shader Detail: Medium
- Effect Detail: Medium
- Model/Texture Detail: Medium
Demo: THG Demo 1
Left 4 Dead
High Qualtity Settings
Version: 1.0.0.5
Video Mode: 1680x1050
Game Settings
- Anti Aliasing: 8x
- Filtering: Anisotropic 16x
- Wait for vertical sync: Disabled
- Shader Detail: Very High
- Effect Detail: High
- Model/Texture Detail: High
Demo: THG Demo 1
Left 4 Dead 2Version: 1.0.0.5
Video Mode: 1680x1050
Game Settings
- Anti Aliasing: 4x MSAA
- Filtering: Anisotropic 8x
- Wait for vertical sync: Disabled
- Shader Detail: High
- Effect Detail: High
- Model/Texture Detail: High
Demo: THG Demo 1
Audio Benchmarks and Settings
BenchmarkDetails
iTunesVersion: 9.0.3.15
Audio CD ("Terminator II" SE), 53 min.
Convert to AAC audio format
Lame MP3Version 3.98.3
Audio CD "Terminator II SE", 53 min.
Convert WAV to MP3 audio format
Command: -b 160 --nores (160 Kbps)
Video Benchmarks and Settings
BenchmarkDetails
HandBrake CLIVersion: 0.94
Video: Big Buck Bunny (720x480, 23.972 frames) 5 Min.
Audio: Dolby Digital, 48000 Hz, 6-Channel, English
to
Video: AVC1 Audio1: AC3 Audio2: AAC (High Profile)
MainConcept Reference v2Version: 2.0.0.1555
MPEG-2 to H.264
MainConcept H.264/AVC Codec
28 sec. HDTV 1920x1080 (MPEG-2)
Audio:
MPEG-2 (44.1 kHz, 2-Channel, 16-Bit, 224 Kbps)
Codec: H.264 Pro
Mode: PAL 50i (25 FPS)
Profile: H.264 BD HDMV
Application Benchmarks and Settings
BenchmarkDetails
7-ZipVersion 9.1 beta
LZMA2
Syntax "a -t7z -r -m0=LZMA2 -mx=5"
Benchmark: 2010-THG-Workload
WinRARVersion 3.92
RAR
Syntax "winrar a -r -m3"
Benchmark: 2010-THG-Workload
WinZip 14Version 14.0 Pro (8652)
WinZIP Commandline Version 3
ZIPX
Syntax "-a -ez -p -r"
Benchmark: 2010-THG-Workload
Autodesk 3ds Max 2010Version: 10 x64
Rendering Space Flyby Mentalray (SPECapc_3dsmax9)
Frame: 248
Resolution: 1440 x 1080
CineBench 11.5Version 11.5 Build CB25720DEMO
CPU Test single- and multi-threaded
Adobe Photoshop CS 4 (64-Bit)Version: 11
Filtering a 16 MB TIF (15000x7266)
Filters:
Radial Blur (Amount: 10; Method: zoom; Quality: good)
Shape Blur (Radius: 46 px; custom shape: Trademark sysmbol)
Median (Radius: 1px)
Polar Coordinates (Rectangular to Polar)
Adobe Acrobat 9 ProfessionalVersion: 9.0.0 (Extended)
== Printing Preferenced Menu ==
Default Settings: Standard
== Adobe PDF Security - Edit Menu ==
Encrypt all documents (128-bit RC4)
Open Password: 123
Permissions Password: 321
Microsoft PowerPoint 2007Version: 2007 SP2
PPT to PDF
Powerpoint Document (115 Pages)
Adobe PDF-Printer
FritzFritz Chess Benchmark Version 4.3.2
Synthetic Benchmarks and Settings
BenchmarkDetails
3DMark VantageVersion: 1.02 Patch 1901
Options: Performance
Graphics Test 1
Graphics Test 2
CPU Test 1
CPU Test 2
PCMark VantageVersion: 1.0.2.0 Patch 1901
PCMark Benchmark
Memories Benchmark
SYSmark 2007 PreviewVersion 1.05
SiSoftware Sandra 2010Version: 2010.1.16.10
Processor Arithmetic, Cryptography, Memory Bandwith
6. Scenario 1 Results: Synthetic Benchmarks

In Scenario 1, we replaced the motherboard, processor, and main memory, but keep the graphics card and hard drive.

All synthetic benchmarks that take advantage of the new platform’s additional computing power clearly benefit here. This applies to all benchmarks except 3DMark Vantage. Apparently, the Core i7 processor isn’t the best match for the old GeForce 8800 GTS.

7. Scenario 1 Results: Application Benchmarks

We saw these results in the first article, in which we executed a full component exchange. Since most application benchmarks are CPU-dependent, the results are similar. Depending on the application, the Core i7-based system from 2010 falls somewhere between a little faster and twice as fast.

8. Scenario 1 Results: Efficiency And Power Consumption

Idle power consumption decreases considerably just from the platform upgrade, while peak power doesn’t drop as much. If you recall the results from our prior full system upgrade, you'll realize that the new Radeon HD 5870 graphics card helps to save additional power, especially in idle.

Translated into efficiency results, this means that the 2010 system delivers roughly twice the performance per watt.


9. Scenario 2 Results: PCMark Vantage

Scenario II represents the choice in which we keep the platform and hard drive, but exchange the Zotac GeForce 8800 GTS graphics card for the MSI R5870 Lightning.

10. Scenario 2 Results: 3DMark Vantage

These results are more impressive. The Radeon HD 5870 ranges between slightly and many times faster than the aged GeForce 8800 GTS in 3DMark. We’ll look at a few games on the next page to verify these results under real-world conditions.

11. Scenario 2 Results: Games And Power Consumption

The benefits in Grand Theft Auto IV are small because our test run was limited to 1680x1050 resolution and low quality settings. Left 4 Dead also uses this resolution, along with low quality settings. At these levels, the games run smooth enough to justify not buying a new graphics card. However, once you start cranking up quality settings, the Radeon HD 5870 graphics card outperforms the old GeForce 8800 GTS card manyfold. We also tried Left 4 Dead 2 to confirm the results. Clearly, gamers wanting maximum visual quality should pick a current-gen graphics card.

Power Consumption

Simply replacing the older graphics card with a newer one resulted in an idle power consumption drop of 22 watts.


 

12. Scenario 3 Results: PCMark Vantage

Finally, we replaced the Western Digital Raptor hard drive with G.Skill's Phoenix SSD. Despite its 100 GB capacity, the drive delivers more than twice WD's maximum throughput, blistering minimum transfer rates, and many dozen times the hard drive's I/O performance. In some tests, the SSD is hundreds of times faster than the hard drive.

The HDD benchmark shows the SSD being 7x faster, and all other tests show the SSD as superior.


13. Scenario 3 Results: SYSmark 2007 Results

Next, we used SYSmark 2007 to look at performance benefits in real-world office applications. Replacing the CPU with a faster one typically yields a slight gain, but using our SSD instead of the older hard drive results in 5% to 15% performance boost. This is more than a CPU update could achieve, for the most part.

14. Scenario 3 Results: Startup, Shutdown

Lastly, we examined load times on Windows startup, an offline version of Tom’s Hardware browsed with IE8, Photoshop CS4, and Windows shutdown. The SSD's advantages are not only desirable for enthusiasts, but for average users. After all, why wait if it's not necessary?

15. Conclusion

In this article, we took the components used to upgrade a three-year-old PC in the article Time to Upgrade: Should You Dump Your 2007 PC? and tested them individually. We replaced the P35 motherboard with a Core 2 quad-core and 4 GB of DDR2 RAM with a new MSI P55A Fuzion, Core i7-870, and 4 GB of DDR3 RAM. The older Zotac GeForce 8800 GTS made way for MSI's R5870 Lightning, and WD’s 150 GB Raptor hard drive downsized to G.Skill's Phoenix 100 GB SSD.

Scenario 1: Motherboard/CPU/RAM Replacement

Replacing platform components heavily impacts power consumption while significantly increasing performance. However, even a three-year-old quad-core CPU still has lots of firepower for most modern applications. Our conclusion for the full upgrade in the prior article remains valid here: save your money if you don’t use applications that necessitate a platform upgrade.

Scenario 2: Graphics Upgrade

Spending a few hundred dollars on a graphics card is a trickier proposition. True gamers will most likely not have waited three years for a graphics card replacement, as newer cards are faster, provide more advanced functionality, and in the case of AMD's Radeon HD 5000-series, deliver lower power consumption. Occasional gamers should consider purchasing a mainstream graphics card instead of the Radeon HD 5870 we chose, since even a mid-range card like Nvidia's GeForce GTX 460 will improve performance, features, and power savings.

Scenario 3: HDD/SSD Upgrade

This upgrade option is the real surprise. Although decent SSDs start at $200 ($300+ in the case of our 100 GB drive), the benefits are much more noticeable than in Scenarios 1 and 2. The SSD-enabled system boots up much quicker, shuts down a bit faster, and launches applications in only a fraction of the time required by even today's top hard drives. SSD benefits will be glaring to all user types.

We would prioritize these upgrade options in the following order:

Gamers

  1. Graphics Card
  2. SSD
  3. Motherboard/CPU/RAM


Power Users

  1. SSD
  2. Motherboard/CPU/RAM
  3. Graphics Card


Mainstream Users

  1. SSD
  2. Motherboard/CPU/RAM (if at all)
  3. Graphics Card (if at all, or select a mainstream card)