As I have promised, here is an article about my computer upgrading experience. I bought all the components on a Boxing Day sale and spent about two hours working on my computer to make it what it is today. Benchmarks and other test results are obviously included. Even thought I didn’t have a particularly pleasant experience doesn’t mean that you shouldn’t personally upgrade your computer.
If you read my first computer article, I recently built a machine for myself that was cheap but still pretty decent, but like a lot of computer users, I had the urge to add components to my rig, before the comp itself becomes too old to justify an upgrade. Before Christmas, I already made a “wish list” about what components I wanted to add to my relatively new rig, so when the Boxing Day Sales of electronic stores started online at 9PM PST, I quickly went online and bought all the parts. Here’s a list of the items.
Part | Description | Price (in CAN $) |
Optical Drive | BenQ DW1620 16X DVD+R\RW Burner | $69.98 |
Heatsink Fan | Thermaltake K7 Silent Boost | $23.98 |
Thermal Grease | OCZ Ultra 5+ | $8.19 |
RAM | 512MB Samsung DDR400 | $77.99 |
Total Cost plus Taxes | $205.36 |
Well, obviously the DVD burner is used to burn DVDs and CDs. I think this is a necesary upgrade to my computer since originally it only had a DVD ROM drive, so it couldn’t store any information on to optical media. Besides, my dad had tons of DVDs he wanted me make a copy of, so a DVD burner is “essential.” The stick of memory is used to complete dual channel on my motherboard, which supposely improves performance. Even if dual channel doesn’t do anything, in most cases, more memory is better. The other two items on the list are used to replace my stock heatsink fan. These items aren’t really “necessary;” I bought them because I wanted to overclock my processor and the stock fan isn’t doing well enough to keep the cpu stable at a higher speed. To tell you the truth, my parents probably don’t approve of overclocking, but since these items are basically my Christmas presents, they didn’t complain. They know I can take care of myself and my computer, so they bought this stuff for me. Thanks, Mom and Dad.
After I got all the parts I ordered, I went to my room and started working. I got everything I needed for the upgrade process: tools, instructions, lucky charm (kidding) etc. I proceeded to open up my computer and put it on its sides on a flat surface. I decided the first thing I should do was replace the hsf (abbreviation for heatsink fan), since that was probably the most difficult and delicate task for this upgrade. First, I got a screw driver and took off the stock cooler. This was pretty easy to do, but here ends the easy part. Now I was faced with the task of cleaning gooey thermal paste from the cpu core. Having visited several websites for tips, I followed their advice and used some Q tips to apply 99% isopropyl alcohol to dissolve the and get rid of the residue thermal paste. The leftover paste was quite sticky, so it took about a dozen minutes and the same number of Q tips to clean all the paste off. I was a little afraid at that point since there seems to be a lot of liquid on the processor, which probably wasn’t a good thing, so I took the liberty of using a lint free cloth to wipe the processor down. After I made sure the processor was clean and dry, I went on to the next step.
I bought some OCZ Ultra 5+ thermal paste, which is supposedly identical to the famous Arctic Silver 5 while being slightly cheaper. The paste came in a 3 gram syringe, and at first sight, it seemed pretty easy to use. I printed the application instructions off the OCZ website just to be safe. So the first thing I was told to do was to rub a bit of the thermal paste into the bottom of the heatsink. So I did that, and this may have ruined my lint free cloth. Then I was supposed to put a little bit of the paste on the cpu die and spread it into a flat, thin, and uniform layer. Easier said than done. The paste was very sticky and a little push on the syringe actually pushed out a lot of paste, so before the paste even got on the processor, I’ve already made a small mess on my work space. The spreading part wasn’t all that simple either, since the paste’s sticky nature made it hard to spread. I followed the instruction and used a plastic card, but the paste was more inclined to stick to the card then to the processor. After a bit of work, I finally spread the paste enough to cover all of the die. The layer was far from uniform, but it was the best that I could do.
So the final part of the hsf installation was installing the hsf itself. The hsf I chose was the Thermaltake Silent Boost, a cooler that is supposed to deliver good performance while operating at a “silent” 21 decibels. Most websites and regular buyers who’ve reviewed this cooler gave it a good rating, and it was on sale, so I decided to buy it. Compared to the stock cooler it replaces, the Silent Boost was huge and quite a bit heftier, due to its copper construction (instead of aluminum on the stock cooler) and bigger fan (80mm vs 60mm). The attachment method was the standard socket A three-pronged clip, which I’ve found to be a pain in the a$$ to install when I attached the stock cooler during the initial building of this computer. This time it actually turned out even worse than the first time, because it took major effort to get just one side of the clip on to the socket. I had to apply quite a bit of pressure to attach both sides and I was beginning to think that I was cracking the processor. When the heatsink finally got attached, I almost cheered out loud. “Hopefully, I didn’t damage the processor or smear the thermal paste around too much.” I thought at that point.
Compared to the replacement of the cpu hsf, installing the RAM and the DVD burner was a snap. The RAM, with a little push, snaps into the slot on the motherboard and the DVD burner just a few screws to attach to the case plus a few cables to connect. During RAM installation, I found out that the stick of RAM I originally have and the new stick weren’t identical. The two sticks are both 184 pin DDR400 RAM manufactured by Samsung and bought from the same store, but one of them had TCC4 chips while the other one had TCCC chips. It’s recommended that a person uses two identical sticks of RAM to run dual channel or there might be some problems. Well, I could have bought the cheap 1gig dual channel kit (in which both sticks of RAM are actually identical) that was on sale, but then the total cost of the upgrade would be $100 higher, so I thought “screw it” and put the new stick in anyways. This turned out to be a BIG mistake. After all the upgrading was done, I connected all the cables I pulled out and double checked everything. When I was satisfied about the conditions of my computer, I bought it back to my desk and plugged in all the back panel connectors and the power plug.
So, now it was the moment of truth. I first turned on the monitor and then pressed the power button on my rig. To my relief, the computer booted up just fine, updated its settings a little bit, and booted into Windows. All the fans were spinning and all the components got their power. I went into Windows and used an utility program, CPU-Z, to check status of the memory, and indeed the memory was running in dual channel. Great. CPU-Z also confirmed my observation earlier and identified the sticks of memory as being different. For instance, the newer stick have tighter memory timings than the old stick. As long as the thing works, it didn’t really matter to me.
Obviously, the next logical step was to run some benchmarks to test if dual channel did or did not improve performance. I ran 3DMark2001 SE, Aquamark 3, and Sisoft Sandra memory benchmark. Below are the results. Note that I didn’t use the results from my old computer article because I changed the driver of my graphics board and that may affect the score a little bit.
Benchmark | Score Before | Score After |
3DMark | 9290 | 9377 |
Aquamark | 18424 | 18453 |
Sandra Memory Bandwith | 2412/2237 | 2391/2247 |
As you can see, dual channel mode didn’t really do anything. The increases in 3DMark and Aquamark are both very small, and in the case of the Sandra test, the bandwidth was actually lower. The Athlon XP platform isn’t really known for its memory bandwidth, and the fact that I have two different sticks of memory might have something to do with this result.
The next thing to do was to check whether the new hsf and thermal compound did anything to imporove the temperature. I took a reading using the utility program that came with the motherboard before and after the installation of the new hsf. Here are the results with room temperature at around 18 degrees Celsius and case temperature at 28 degrees. The idle temperatures were taken when only the background programs are running. The load temperatures were taken during a gaming session.
Situation | With Stock Cooler | With New Cooler | Change in Temp |
Idle | 30 | 28 | -2 |
Load | 45 | 40 | -5 |
Idle/Load Difference (Load temp minus Idle temp) | 15 | 12 | -3 |
So indeed there was a decrease in temperature across the board. The load temperature dropped by 5 degrees, which means the new hsf and thermal compound can do a better job getting rid of heat from the cpu than the old combination. Mind you that these temperatures were taken just after the hsf was installed. On the OCZ website, they said that it takes the thermal compound about 120 hours to properly set in between the cpu and the heatsink, so it’s normal to see a 1-5 degrees reduction in temperature during that period of time. Hopefully, what they say is true. A load temperature at 35 degrees would be awesome for a traditional hsf such as the Silent Boost. It would give me quite a bit of room to work with when I try overclocking. I’m not keeping my hopes up too hight though, since I don’t think I did a very good job applying the thermal compound to the cpu die. Any reduction in temperature would be welcomed.
The last thing to check was whether or not my DVD burner worked. I bought this BenQ model because it was the cheapest burner I could find that still supported the new dual layer format. The customer reviews at the vendor where I bought this at were so-so for this burner, so I’m not extremely confident of the burner’s quality. To test the burner, I installed the software that was supplied and tried to burn an ISO image of a DVD movie onto a blank piece of DVD-R. The process took about 15 minutes and after the process had finished, I took the disc out and popped it into a DVD player for testing. It worked flawlessly. Needless to say, I’m pleased with the results and I’ll be burning more discs in the future.
Here are some photos of my newly upgraded computer.


The first photo is of the front of my computer. The second drive bay from the top has the DVD ROM drive and the bay below it is the DVD burner. Originally, I wanted to put the DVD burner on top and set it as master and the DVD ROM drive as slave. However, the power cable from the power supply wouldn’t reach the top slot and I was too lazy to mess with the jumpers on the DVD drive inside the case, so I installed the burner under the old drive and set it as slave on the IDE channel. In the second picture, you can clearly see the my new cpu hsf, the Thermaltake Silent Boost, doing its work.
So, I thought that I’m done, right? Not quite. Shortly after, the computer started experiencing some problems. At first there were some small software errors and some programs wouldn’t start. I thought these were just little “hiccups” the computer was having and ignored them, but soon after that, the computer started rebooting mysteriously and finally a serious error in the operating system was found, and I couldn’t boot into Windows any more. I didn’t panic and booted into safe mode, where I performed system restore and got back to an earlier configuration. The computer was able to boot into Windows now but when I tried to run the Prime95 torture test, the computer failed in 20 seconds. Same thing with PCMark 2004. I know then I had a major problem on my hands because Windows XP is pretty stable as Windows go and these events mean something important has screwed up. Windows will probably have to be reinstalled. I moved all my personal files to my mom’s computer, erased the hard drive, and started from scratch.
Windows XP installed without a problem and I applied all the updates and was about the install service pack 1, so I can use USB2 and the full capacity of my 160GB hard drive. However, during service pack deployment, the computer screwed up again and I was locked out of Windows for the second time. By that time, I know it was probably a hardware problem that’s messing up my computer. I reinstalled Windows again and download the Prime95 torture test as soon as possible. It confirmed my suspicions as the computer failed in a very short time. Seeing this result, I immediately turned off the computer and took out the most likely source of the problem: the new stick of RAM. I turned my rig on and used torture test again and this time the computer ran the test for several minutes without any problems. So the troublemaker was the RAM and I finally can get my computer back. I had no problem whatsoever after taking out the RAM, and reinstalled all the software and got back all the files.
I went back to vendor and gave the technician there the newer stick to test. The tech couldn’t find anything wrong with that stick and told me to bring back both the old and new sticks and he’ll find a matching pair for me. So I did that and exchanged my old TCC4 with a new TCCC RAM stick. I then went to test the two sticks that I have and… Prime95 failed within a minute. I was thinking: “Not again!!” This time I was smarter and tested the RAM sticks one by one. Through testing, I found out that the stick given during the exchange was bad, so I had to go back the vendor again to ask for an exchange and this time the tech said he gave us a brand new one. I was about to ask:” You mean you gave us an old one before?!” However, the tech was busy and went away again so I couldn’t ask him. I took the RAM home again and tested the brand new one first. No problem. Then I put the other RAM in and tested. To my relief, the computer finally didn’t fail Prime95 within the first minute and ran for about 10 minutes before I stopped the test. So after several trips to and back from the vendor, I finally got a pair of RAM that can actually work in dual channel. Yippeee! (I’m being sarcastic)
Well, now that the computer works, I had to run some benchmarks again. I don’t have any Aquamark results yet since I didn’t download it yet. My SANDRA memory bandwidth is still about the same, but my 3DMark jumped to 9657, which is a noticeable improvement (about 4%) over the previous 9290. It might have been a fluke, but it’s probably due to the dual channel (hopefully, anyways).
After all that trouble, here is the present components list for my rig.
Part | Description | Price (in CAN $) |
CPU | AMD Athlon XP 2500+ retail with heatsink | $109.99 |
Heatsink Fan | Thermaltake K7 Silent Boost | $23.98 |
Thermal Grease | OCZ Ultra 5+ | $8.19 |
Motherboard | Soltek 75FRN2-L Nforce2 Ultra Socket A | $79.98 |
RAM | 2x 512MB Samsung DDR400 | $187.97 |
Hard disk | Western Digital Special Edition 160GB | Got it for free |
Video Card | Mad Dog Nvidia GeforceFX 5600 128MB AGP 8x | $113.99 with $40 mail-in rebate |
Optical Drive | BenQ 1650S 16x DVD ROM | $29.99 |
Optical Drive | BenQ DW1620 16X DVD+R\RW Burner | $69.98 |
Case + Power | Powmax 212-4HL 17 inch case with 400W power supply and side window | $46 |
Floppy | Floppy drive | $11 |
Case Fan | Coolermaster TLF-R82 Blue LED fan | $6.99 |
Keyboard | Ione Scorpius M6 USB Keyboard | $9.99 |
Mouse | Logitech Optical Mouse | $9.99 |
OS | Windows XP Professional | Already have it |
Total Cost + Taxes | $764.29 |
So the total cost of my computer is up to over $750, so I can’t brag about how inexpensive it is. Upgrading is never really economical anyways, but I felt that the money was worth it, minus the “having to reinstall your operating system twice” problem of course. I think the problem I experienced was caused by trying to run two different sticks in dual channel. Conclusion: this is probably my own fault, even so bad luck was also involved. Even though this experience was a lot worse than I expected, I did get something out of it: don’t try to run two nonidentical sticks in dual channel. Hey, two out of three isn’t bad. The new heatsink is performing to its specs and the DVD burner works fine. Oh yeah, I also got two new sticks of memory instead of just one, which is a plus to this otherwise troublesome experience. Don’t let my bad luck discourage you from upgrading your own computer. Chances are, you’ll probably do fine and no problem will occur.
This is probably the last upgrade this computer will ever get. Well, maybe I’ll add a TV tuner when I get cable TV again so I can watch TV on the computer or if I win a top of the line graphic card or something else, I’ll add my prize in. I haven’t tried overclocking the processor again yet, because I’m waiting for that “1-5 degrees decrease” in temperature that the OCZ site reported. My overclocking ventures may become another article, so stay tuned, and thanks for reading my article.