Date: July 3, 2007
Author(s): Rob Williams
Gigabyte, like others, released a slew of motherboards utilizing Intel’s P35 chipset when it was first announced. The P35C-DS3R is unique to their line-up though, since it will accept both DDR2 and DDR3 memory. We pit the boards performance using both types of memory against our recently reviewed ASUS P5K and P5K3.
By now, Intel’s P35 chipset needs no introduction. It has boosted performance over P965 and brings a few new things to the table, my favorite being the S-ATA disable. In our reviews of ASUS’ own P5K and P5K3, we were left impressed to say the least. Compared to our boards that utilize NVIDIA’s 680i and Intel’s 975x, the P35 cleaned house in near every test.
If you are in the market to build a new PC and don’t care for SLI support, then P35 boards are a good bet, unless you want to wait until X38 is launched later this year, which is supposed to be vastly superior to P35. Time will tell if that’s actually going to be the case or not, but so far it’s looking hopeful.
As mentioned in our intro, once P35’s NDA was lifted, motherboard manufacturers all over announced their new line-ups. ASUS themselves released a staggering seven P35-based boards and Gigabyte released five. MSI and abit also have great looking boards as well, a few of which we will be taking a look at in the weeks to come.
The board we are taking a look at today though delivers a newer concept, and one I still don’t completely understand. Users who purchase the board will have the ability to choose between DDR2 and DDR3 memory, which can’t be a bad thing, right? It means that consumers have a choice.
From a marketing perspective, I see this being beneficial for those who have DDR2 now, but want to upgrade to DDR3 down the road. But even then, why upgrade to DDR3 when DDR2 is not lacking in any way right now? Far down the road when 4GB of DDR3 doesn’t cost a limb, then perhaps it would be a great upgrade, but as it stands right now, there is no reason to set your sights on DDR3 unless you have found one of the few ways to utilize all of that excess bandwidth.
Though the performance benefits of DDR3 are yet to be seen, the fact of the matter is, it’s performance IS better than DDR2, primarily thanks to the faster frequencies. This is how it is, despite the fact that DDR3 has timings that make DDR2 fans weep. DDR3 is more efficient than DDR2, it’s just that most people don’t need to focus all of their attention to it. However, just because I can’t seem to see the sense in such a combo board, others may disagree and have personal reasons why a board like this is ideal. If so, please let us know your thoughts in our forums.
One might jump to the conclusion that because this is a combo board, it’s price is going to be extreme. It’s actually the exact opposite. This is more of a budget board than it is a high-end board, retailing for an average of ~$175. While not cheap per se, compared to the P5K and P5K3 boards we took a look at last month, which retail for closer to $230 – $250, it is.
One thing can be said about Gigabyte’s choice of box art… it’s the best I’ve seen. Clean, professional and delivers all of the information one would need when fighting with themselves over a purchase decision. The board is part of the Ultra Durable 2 series, which means it bounces off the floor when dropped.*
* Dropping board is a bad idea.
Besides the expected manual, software and case sticker, Gigabyte doesn’t go cheap on what you need. Included are four S-ATA cables, which include metal push tabs to secure it into the port and also for easy release. One IDE and one floppy drive cable is also included as well.
In the top-right of the image you will see the required components to get an E-SATA setup running. Though not tested for the review, the installation appears simple. You will need to clear open a port in the back of your computer and install it there. You will then plug a 4-Pin Molex connector from this port in to an available cable on your power supply. The two black cables are used externally, one for power and the other for the connection.
Next up, a look at the board and it’s layout.
Gigabyte is well known for making some of the most colorful motherboards on the planet, and to some this is a good thing. To others, it’s foolish. Me, I have no opinion, I just care about the performance. Overall the layout looks great, no immediate complaints. Everything is located in an accessible location, except for the hard drive IDE connector port. Though I don’t have an IDE drive, I assume a hulking cable as such is going to look ugly when plugged in there.
Common of budget motherboards, the P35C-DS3R has a 4-Pin motherboard connector, instead of the usual 8-Pin. This proved an inconvenience to me, because I already had an 8-Pin routed through the back of my Antec P182 case properly. Normally this is not a problem, because you can plug an 8-Pin into a 4-Pin since both connectors have a top row of +12v and bottom row of ground. However that proved impossible here, due to a component getting in the way.
Another thing that also really inconvenienced me was the lack of 3-pin fan connectors. Besides the CPU fan connector, there are only two more, one beside the DIMM slots and the other beside the north bridge. I use a Corsair Nautilius 500 water-cooling system on the PC, and you are supposed to plug two connectors into fan ports on the motherboard. Here though, both fan connectors were out of reach, so it couldn’t be done. I am unsure why there is such a lack of connectors, though I feel it must part of the budget experience. It’s too bad though, as there is plenty of room for more.
Like here for example.
At the bottom of the board, a set of four USB connectors are clearly labelled. The P35 chipset introduced the ability for 12 USB ports off of it alone, and that’s possible right out of the box here, as long as you have that many ports available.
8 S-ATA ports are readily available at the right-side of the board, situated directly above the lone IDE connector. The board also supports RAID, if you want to take that route.
Although the P35 chipset supports Crossfire with ATI cards, it’s not supported on this board because of a lone PCI-E 16x slot. In addition to three PCI slots are three more PCI-E 1x slots. I have yet to own a single PCI-E 1x component, let alone three.
The southbridge has a modest passive heatsink, which will be enjoyed by those who like silent PC’s. During testing, I did not find this to get hot enough to warrant a fan.
Finally, the most interesting part of the motherboard. The red and yellow slots are for your DDR2, while the green are for DDR3. For DDR2 users, you will be able to use upwards of 8GB of ram, with 2GB modules. For DDR3, 8GB should be possible, but we have yet to see 2GB DDR3 modules, let alone 4GB. During installation, you have to choose between the two, neither can be used at the same time.
Compared to the south bridge, the north bridge has a massive cooler, appropriately so since the NB will get far warmer than the SB. For hardcore overclockers, a fan would be recommended.
There is sufficient space around the CPU socket for most coolers, I can’t see it being a problem to install even the largest on the market. Note the fact that the PWM doesn’t include any type of cooling whatsoever, not even a cheap passive heatsink.
The board uses the ITE IT8718F-S chipset used for I/O purposes, and though blurred, there is also a small Realtek ALC889A, used for the boards 8-channel audio.
On the back we find PS/2 ports for both Keyboard and Mouse, S/PDIF, parallel port, serial port, four USB ports, LAN port and finally, the audio jacks. Nothing is missing here per se, although on higher-end boards you would find an E-SATA and a second NIC port. As for the parallel port, the inclusion of that just confuses me, since it’s userbase would be so low. There would be a better chance of a PCI-E 1x port being used.
Here is a vertical view of the board to finish things off, since this is the way most people will be viewing it once it’s installed. Unless you turn your computer upside-down.
Before we jump into testing and performance reports, we will first take a look through the fairly robust BIOS.
On this page, we will take a quick walk through the BIOS to see what’s being offered. First up is the main page, which won’t be much of a stranger to anyone.
The first page offers many of the basic options you’d expect, such as time, date and hard drive configuration. Two options I normally disable here are the Floppy 3.5″ and the Halt On > No Errors. That way, it won’t stall at boot time because I don’t have a CPU fan plugged in (due to water cooling).
Under Advanced BIOS Features we can arrange our boot sequence, disable the full screen logo, disable EIST, enable SMART and more.
Integrated Peripherals offers USB-related options, audio and LAN enable/disable and also the RAID configuration.
The following screens represent the power and PnP/PCI options, as well as system health.
Of most interest to enthusiasts will be the MB Intelligent Tweaker section, which has every overclocking related option you will need. Here’s where things became interesting for me though. It’s been quite a while since I’ve used a Gigabyte motherboard, so I had forgotten about the CTRL+F1 trick at the BIOS main menu. For those unaware, what that does is unlock more than the usual options in the tweaking section, such as memory timings. Without pressing CTRL+F1, the selection is lacking.
Everything here is pretty self-explanatory, except for the DDR2/DDR3 OverVoltage Control. The options within range between +0.1v up through +0.7v. Depending on whether you have DDR2 or DDR3 installed, the BIOS will set a different base voltage. For DDR2, it’s 1.8v and for DDR3, it’s 1.5v. So, by using the +0.7v option, you will be able to use 2.5v for DDR2 and 2.2v for DDR3, both of which are within the limits of sanity.
Overall the BIOS isn’t exploding at the seams with features, but everything important is there.
Throughout all of our benchmarking tests, regardless of what we are reviewing, testing is done in a clean and stand-alone version of Windows XP Professional with SP2. The exception is in our SYSmark 2007 tests, which uses Windows Vista Ultimate. No Windows Updates are applied for the sake of time, unless one is required for a piece of hardware on the computer.
We’ve recently updated our methodology, to reflect a more real-world experience. Instead of tweaking Windows to a minimalist state, we leave all initial processes running, as it is a realistic scenario. The primary form of computer latency is a network, so all network devices are disabled prior to any testing. This is done in addition to:
Games that are used for benchmarking have their average frames per second recorded with the help of FRAPS 2.82. Here is the specs of the machine used for todays testing:
|Processor||Intel E6600 (2.40GHz)|
|Memory||OCZ 2GB NVIDIA PC2-7200 (For DDR2 Boards)|
Kingston 2GB HyperX DDR3-1375 (For DDR3 Boards)
|Video||ASUS EN8800GTX 768MB (NVIDIA 158.22)|
|Sound||Onboard HD Audio|
|Storage||Seagate 7200.10 320GB|
LG 18x DVD+RW
|Etcetera||Windows XP Professional SP2|
Windows Vista Ultimate
Antec P182 Chassis
Corsair HX620W PSU
|Cooling||Corsair Nautilus 500|
|Compared Mobos||ASUS P5K|
Because we are using both DDR2 and DDR3 motherboards, it’s important to lay out the settings we used. These are based off what we feel to be the most common at the given frequency. For DDR2, we didn’t want to go beyond 1.9v, which is +0.1v above DDR2 standard. The same goes for DDR3, so we are using 1.6v with reasonable timings that should be possible on most DDR3-1066 kits. The memory speeds used are as follows:
SYSmark is an industry leading system benchmarking tool, which is completely automated but utilizes real-world tests. It installs common applications such as Microsoft Word and Excel, Photoshop CS2, 3DS Max, SketchUp! among others.
SYSmark grades the performance of the system with various methods, but most notably how well it could handle different operations. Systems with more than one core will benefit in the tests, since there is a lot of multi-tasking throughout. Once the test is completed, it will provide you with an overall score, in addition to showing areas where the computer excelled.
Intel’s XBX2 excelled in this test, for whatever reason, but ASUS’ P5K3 fell not too far behind. The DDR2 configuration with the P35C-DS3R exceeded the performance of the DDR3 by a small margin. Very, very small.
The individual test results can give us a deeper look at where each board failed, or performed well. Regarding 3D tests, all of the motherboards performed quite well to one another. When performing personal renders with 3DS Max, I found that this was always the case.
When looking at the individual scores, it’s easy to see that in the end, no board has an obvious advantage through all of the tests. Some excel in one test while others excel in another. Seen here again though, our DDR2 configuration performed better than the DDR3.
Futuremark has long offered benchmarking tools to enthusiasts that allow them to gage their systems worth. There is a lot of skepticism revolving around the importance of the overall scores, but we enjoy running them because it’s a quick fix to see differences between platforms. Real world benchmarks are by far more important, and we will cover those on the next few pages.
Although old, 3D Mark 2001 proves a good benchmark to evaluate your systems overall performance. In 2000, this benchmark really stressed whatever GPU you owned, but today the GPU hardly comes into the equation. What does help you achieve a higher score is faster CPU and memory frequencies.
3D Mark 2006 tests your system in a similar manner that 2001 does, except this updated version actually does bottleneck on your GPU. The faster the GPU, the better the score. Multi-core processors also help greatly improve your scores here.
PC Mark is somewhat similar to SYSmark, which we discussed on the previous page. The difference is that PC Mark focuses more on synthetic benchmarking schemes, such as disk access and multi-tasking. Very little of the entire test will be seen by you though, as it all goes on behind the scenes.
With our synthetics out of the way, let’s move forward into our real-world tests, arguably the most important of any performance review.
Nothing can prove the performance of a PC better than real-world benchmarks. The only downside to real-world tests is that it’s difficult to compare to a friends computer, unless they happen to have the same software and media files that you are encoding/converting. We can show direct differences though, since we run the exact same tests on each board.
To test video encoding capabilities, we ripped our “Lamb of God – Killadelphia” concert DVD and then used Nero Recode to convert it into something that can be burned on a normal sized DVD. The direct DVD rip is 7.7GB, and Recode compresses it into a 4.5GB frame.
As humorous as it is to see the Intel XBX2 land in last place, despite being first in our SYSmark tests, the most important thing to take away is knowing that the P35C-DSR3 placed first, with both configurations.
Similar to our video encoding test, we originally ripped a solid FLAC file from our “Tiesto – Elements of Life” album. From there, we decompressed it using flac -d and then compressed it into a 320Kbits MP3 using lame -b 320. LAME 3.97b2 was used for testing.
Audio conversion, as we can see, may depend on a motherboard, but moreover it will depend on the actual frequencies used. Both of our P35C-DS3R and P5K DDR2 and DDR3 configurations posted the exact same times.
As mentioned earlier, I have performed numerous tests using 3DS Max 9 on multiple motherboards, to find that the end performance results hardly differed at all. My tests consisted of a 3200×2400 render, in addition to an export of 100 frames at 640×480. With both tests, the results were either exact or one second different.
Both the ASUS P5K and P5K3 had obvious benefits here, while the P35C fell a wee bit behind.
Different motherboards use different chipsets for SATA controllers, so these tests are a good way to see how one board will compare to another. The first test is synthetic, using HD Tach RW/3, while the File Compression is real world, which involves compressing a 4GB folder (4,809 files) using 7-Zip.
Intel’s XBX2 excels at very specific tasks, archiving being one of them. Behind it lies the P5K boards followed by our Gigabyte.
To pit these boards against some popular games, we chose to use Half-Life 2: Episode 1, S.T.A.L.K.E.R. and Need for Speed: Carbon. Each game offers its own flare to our benchmarking reviews for different reasons. HL2 is great simply because it’s one of the most popular games of all time, while STALKER has a wide open world to render and AI to churn. NFS: Carbon is included because racing games really enjoy powerful systems to push high FPS when you are driving at 200MPH.
As a reminder, we are running a 2.4GHz Intel E6600 along with an ASUS 8800GTX, which we choose because of its power and ability to rid out the GPU as being a bottleneck. All of the games were run on 1280×1024 using default options. The only options changed was to NFS: Carbon in order to select “High” detail. Results were tabulated with the help of FRAPS 2.8.2. Each play through lasted between three and five minutes, depending on the level chosen.
For Half-Life 2: Episode 1, I chose my favorite level, ep1_c17_02a. The level starts you off in a dimly lit hallway and you need to make it through to the roof where an airship is trying to gun you down. It’s a fun level, and really shows off HDR.
In S.T.A.L.K.E.R., I chose a run-through with the thumb drive mission, which occurs near the beginning of the game. Through it, there are many people who die and you get to leave with a thumb drive. Does it get much better?
In our NFS: Carbon test, we played through the first normal race when choosing one through the Quick Race mode. Two choices of car are given, an upcoming Chevrolet Camaro and a Koenigsegg CCX. I think it’s obvious which one I chose.
With those out of the way, let’s head straight into our final thoughts.
One of the most important features of any motherboard is overclocking ability, and this is where the P35C-DS3R falls a little short. After a long overclocking session, I found 450FSB to be the highest the board would possibly go before becoming unstable. Compare this to the P5K boards that hit 475FSB, and even that is not high by todays standards.
What’s interesting though, is that pumping more voltage into the FSB didn’t seem to greatly help the overclock. Both the SB and NB have voltage options of +0.1v – +0.3v, and 450FSB was stable with only +0.2v. So what good is the +0.3v? There’s no good at all in it, it seems. 455FSB delivered errors at both +0.2v and +0.3v, but 450FSB was completely stable at +0.2v. Therefore, +0.3v will likely never have to be touched, unless you receive a board that wants to overclock a bit further.
While the board won’t be for hardcore overclockers, it still offers lots of OC’ing room for those who want high CPU frequencies without the benefit of a major FSB overclock. In truth, 450FSB may have been the limit thanks to our fanless NB/SB, but I didn’t proceed with extensive testing to see if that was the case. Most people won’t want to ghetto-mod fans inside their case to keep these two cool. The fact is handled 450FSB with modest heatsinks is good.
Besides overclocking ability, there isn’t too much to complain about. The lack of a second PCI-E 16x slot wipes out Crossfire capabilities, so if you are interested in that route, this board is a non-option. During installation, I didn’t run into any clearance problems at all. Even the 8800GTX installed just fine, without anything blocking it’s way.
Despite the wider DIMM lanes, you will still be able to use your RAM fan, because regardless of whether you are using DDR2 or DDR3, your fan will cover all the modules just fine thanks to how they are laid out. The combo DDR2/DDR3 feature is one I don’t quite understand, but if you have a specific need for a board like this, then so be it. Personally, I don’t find there to be a large difference between DDR2 and DDR3, as was backed up with our tests. At much higher overclocked frequencies, we will begin to see differences, but are they needed? That’s the question. There will be a point where beneficial returns are nil.
This board currently retails for ~$175, which to me is a reasonable price, if you plan on using the boards features to their full potential. This would include the E-SATA port, which will be appreciated by those who require it. It’s a little different than most E-SATA ports, since they include a cable for power. If you have an E-SATA enclosure, you won’t need the power cable, but if you have a spare S-ATA drive lying around, you will be able to hook it up to your PC without opening your tower.
Although Gigabyte’s color-schemes are sometimes questioned, one thing that excels is their packaging and presentation. It’s clean, professional, and informative. It’s easily some of the best looking packaging I’ve seen from any motherboard manufacturer. They include all the cables you will need, and a fully detailed installation sheet for new computer builders. Although the board itself isn’t exploding with features, the overall package feels like a good value.
When all said and done, I am going to rate the P35C-DS3R an 8 out of 10 on the Techgage scale. With what we’ve been delivered, I feel the board is about $20 more expensive than it should be, but I believe that’s because of the combo RAM functionality, but functionality I can’t see many people using. Even still, for the price, this is a stable board for regular consumers and enthusiasts alike. It doesn’t offer explosive overclocking ability, but even at 450FSB, the board appeared completely stable. The ASUS P5K boards performed better in almost all of the tests, but they cost upwards of $60 more, but also include WiFi and slightly better overclocking ability. With those facts in mind, the ultimate decision is yours.
If you have a comment you wish to make on this review, feel free to head on into our forums! There is no need to register in order to reply to such threads.
Copyright © 2005-2019 Techgage Networks Inc. - All Rights Reserved.