ASUS Maximus Formula SE: X38 and DDR2 Unite!
by Rajinder Gill on November 9, 2007 7:00 AM EST- Posted in
- Motherboards
Overclocking with the ASUS Maximus Formula SE
For the vast majority of PC users, the announcement of a new CPU or chipset release generates a fair amount of interest, pegged with a realistic and thoughtful reaction towards the potential benefits (if any) over present hardware. Intel has forged ahead at such a pace this last year that it seems owning the performance crown for the foreseeable future is not even enough at this point. Releases of information from AMD about their future CPU and chipset plans have quickly been answered by Intel (with actual product releases in most cases) to ensure their continued market dominance in the enthusiast sector for the time being.
With this rapid release schedule comes improved fabrication methods that have brought many benefits: better speed scaling, lower transistor leakage, lower overall power consumption, and reduced production cost. These new CPU and chipset releases have also been coupled with competitive price cuts based on the competitive landscape that users have enjoyed the past 18 months. We have seen these improvements and strategies delivered with precision by Intel into the market place over the last year and a half while AMD has stumbled. If that's not enough, the upcoming release of the Penryn processor family features a CPU die shrink from the current 65nm process to 45nm along with additional cache sizes and several core improvements such as SSE4 support. This seems more than enough to guarantee an overall performance advantage over AMD for some time to come in the enthusiast sector.
The question from those steeped in conservatism is whether or not this industry will ever fully utilize the potential of current hardware before moving ahead to something new. From a corporate perspective, posing this question can be construed as completely naïve. Indeed, there's money to be had by making a series of small changes, while simultaneously holding a partially revealed trump card in current R&D, just in case there is the chance of a competitor counterstrike.
The Internet has become such a powerful marketing tool that a strategic selection of websites for a launch can create enough interest to guarantee a successful campaign, even when economically or performance wise a change might not be beneficial for the user. At this time, much of Intel's advertising success is accrued by placement of product in the hands of overclockers, who are enlisted to fulfill the role of presenting new hardware by showing gains in various synthetic benchmarks. This is a perfectly logical corporate decision - it is easier to sell new product to a group that is passionate about even the smallest gains than it is to satisfy the somewhat discerning real world PC user.
Since the launch of Core 2 Duo, we have shown that any company in support of Intel's CPU and chipset release strategy needs to have sufficient resources to co-exist with Intel's relentless marketing drive. This leads us to ASUS, who has managed to surf the Core 2 Duo wave to its full potential. They are usually among the first to release a motherboard with each Intel chipset launch, reaping the benefits of being one of the preferred launch partners for Intel chipsets. Most of the boards released by ASUS have attained substantial accolade for the most part, especially those priced in the sub $200 region. Further, there is ASUS's top of the line motherboards, the ROG (Republic Of Gamers) series. This particular product line is aimed to satisfy the needs of the upper tier gaming and overclocking enthusiasts.
Over the past year, we have seen several boards released bearing the ROG badge: the Striker Extreme, Commando, Blitz Formula, and Blitz Extreme. Most of these motherboards have been a popular choice for enthusiasts and gamers alike. Our own feelings about the ROG series is that ASUS has offered overall improvements in terms of overclocking friendly BIOS functions, component quality, features, unique on-board cooling designs, and in some cases improved compatibility - although not always. The ASUS Striker series was an especially finicky beast for non-QVL memory modules. Fortunately, the Intel chipset based boards are far friendlier in this department, a trend we hope continues with their latest products.
The latest addition to the "bells and whistles" ROG series is the ASUS Maximus Formula, based upon the Intel X38 chipset and featuring DDR2 memory support. To date, DDR2 performance using the X38 chipset has been a little underwhelming for us. Synthetic and real world benchmarks have shown little to separate X38 DDR2 scores from those using the P35 chipset. The motherboard suppliers should not be penalized for this situation as the X38 was certainly hyped to be the killer product for Intel's high-end aspirations. Our experiences with the X38 over the summer reminded us of a roller coaster ride with several highs but just as many lows as we tested each chipset revision and BIOS release. A month ago we were not excited about the X38, but with the BIOS releases maturing along with early performance results from Penryn we are ready to get back on the roller coaster.
Purchasing decisions often seem clear-cut, especially those based upon outright synthetic performance figures. Before we arrive at a conclusion today, let's look at some of the factors forcing/promoting use of the X38 chipset within selected markets (in this case the top-end enthusiast sector):
For the vast majority of PC users, the announcement of a new CPU or chipset release generates a fair amount of interest, pegged with a realistic and thoughtful reaction towards the potential benefits (if any) over present hardware. Intel has forged ahead at such a pace this last year that it seems owning the performance crown for the foreseeable future is not even enough at this point. Releases of information from AMD about their future CPU and chipset plans have quickly been answered by Intel (with actual product releases in most cases) to ensure their continued market dominance in the enthusiast sector for the time being.
With this rapid release schedule comes improved fabrication methods that have brought many benefits: better speed scaling, lower transistor leakage, lower overall power consumption, and reduced production cost. These new CPU and chipset releases have also been coupled with competitive price cuts based on the competitive landscape that users have enjoyed the past 18 months. We have seen these improvements and strategies delivered with precision by Intel into the market place over the last year and a half while AMD has stumbled. If that's not enough, the upcoming release of the Penryn processor family features a CPU die shrink from the current 65nm process to 45nm along with additional cache sizes and several core improvements such as SSE4 support. This seems more than enough to guarantee an overall performance advantage over AMD for some time to come in the enthusiast sector.
The question from those steeped in conservatism is whether or not this industry will ever fully utilize the potential of current hardware before moving ahead to something new. From a corporate perspective, posing this question can be construed as completely naïve. Indeed, there's money to be had by making a series of small changes, while simultaneously holding a partially revealed trump card in current R&D, just in case there is the chance of a competitor counterstrike.
The Internet has become such a powerful marketing tool that a strategic selection of websites for a launch can create enough interest to guarantee a successful campaign, even when economically or performance wise a change might not be beneficial for the user. At this time, much of Intel's advertising success is accrued by placement of product in the hands of overclockers, who are enlisted to fulfill the role of presenting new hardware by showing gains in various synthetic benchmarks. This is a perfectly logical corporate decision - it is easier to sell new product to a group that is passionate about even the smallest gains than it is to satisfy the somewhat discerning real world PC user.
Since the launch of Core 2 Duo, we have shown that any company in support of Intel's CPU and chipset release strategy needs to have sufficient resources to co-exist with Intel's relentless marketing drive. This leads us to ASUS, who has managed to surf the Core 2 Duo wave to its full potential. They are usually among the first to release a motherboard with each Intel chipset launch, reaping the benefits of being one of the preferred launch partners for Intel chipsets. Most of the boards released by ASUS have attained substantial accolade for the most part, especially those priced in the sub $200 region. Further, there is ASUS's top of the line motherboards, the ROG (Republic Of Gamers) series. This particular product line is aimed to satisfy the needs of the upper tier gaming and overclocking enthusiasts.
Over the past year, we have seen several boards released bearing the ROG badge: the Striker Extreme, Commando, Blitz Formula, and Blitz Extreme. Most of these motherboards have been a popular choice for enthusiasts and gamers alike. Our own feelings about the ROG series is that ASUS has offered overall improvements in terms of overclocking friendly BIOS functions, component quality, features, unique on-board cooling designs, and in some cases improved compatibility - although not always. The ASUS Striker series was an especially finicky beast for non-QVL memory modules. Fortunately, the Intel chipset based boards are far friendlier in this department, a trend we hope continues with their latest products.
The latest addition to the "bells and whistles" ROG series is the ASUS Maximus Formula, based upon the Intel X38 chipset and featuring DDR2 memory support. To date, DDR2 performance using the X38 chipset has been a little underwhelming for us. Synthetic and real world benchmarks have shown little to separate X38 DDR2 scores from those using the P35 chipset. The motherboard suppliers should not be penalized for this situation as the X38 was certainly hyped to be the killer product for Intel's high-end aspirations. Our experiences with the X38 over the summer reminded us of a roller coaster ride with several highs but just as many lows as we tested each chipset revision and BIOS release. A month ago we were not excited about the X38, but with the BIOS releases maturing along with early performance results from Penryn we are ready to get back on the roller coaster.
Purchasing decisions often seem clear-cut, especially those based upon outright synthetic performance figures. Before we arrive at a conclusion today, let's look at some of the factors forcing/promoting use of the X38 chipset within selected markets (in this case the top-end enthusiast sector):
- Intel's CPU price/performance scaling strategy: Intel uses lower default multipliers for non-Extreme CPUs, demanding higher overall FSB potential from chipsets - especially for quad-core CPUs.
- Attaining the limits of economically viable engineering practices by using a separate memory controller hub: This creates the need to push towards DDR3 rather than trying to lower overall chipset latency and increasing memory bandwidth using current DDR2 technology. Using DDR3 and associated memory timing ranges can offer superior scaling potential; testing has shown increased synthetic benchmark scores and small gains in workstation and gaming software with the right memory. The obvious selling points of X38 in DDR2 format are suggested in points 1 and 3.
- Dual x16 PCI-E 2.0 support for upcoming AMD graphics solutions (RV670 to start) to improve CrossFire performance.
24 Comments
View All Comments
mbf - Friday, November 9, 2007 - link
...handle DDR2 ECC memory? I for one would like to know. ASUS seems to be of two (or possibly more) minds on the matter stating conflicting information all around the product pages for their respective X38 board offerings. Then again, the P5W DH Deluxe still seems like a smart choice, considering the very small performance delta between the i975x and later chipsets. Also, ASUS claims Penryn support for several of their "mature" offerings, including the P5W DH Deluxe.AnnihilatorX - Friday, November 9, 2007 - link
Personally I would recommend everyone including enthusiasts to not to buy over-priced performance RAMs.And of course looking at price at the moment to choose DDR2 over DDR3
A low latency low frequency RAM are potentially *much* cheaper than a high frequency one. The performance discrepancy is at most 5% which relates to perhaps 2-3FPS in a game. This has probably the lowest cost-to-performance ratio of a system component.
steve4717 - Sunday, May 16, 2010 - link
when will the new bios be ready, and i exspect it, this time to make it possible so it can see, ddr2 1066 at long last.nleksan - Wednesday, November 28, 2012 - link
I have been looking for a motherboard for a recently acquired (given to me free of charge) set of somewhat older but almost entirely unused (i.e. no more than 100hrs use on anything, most have around 20-40hrs; came from 9 different full-or-partial PC's) hardware including:- Core2Duo E8600 (under 20hrs use, known to run stable at 4.5Ghz on air, 4.9Ghz on water)
- 4x2GB G.Skill DDR2-1066/1150 4-4-4-9
- 4x1GB OCZ DDR2-1066 4-4-4-12
- 3x 250GB Samsung Spinpoint SATA3Gbps HDD's
- 2x 150GB WD VR SATA3Gbps 10krpm HDD's
- 1x WD WD800BB 80GB SATA3Gbps HDD
- 4x WD2500JB 250GB SATA HDD's
- 3x WD Caviar Blue 320GB (AAKS) SATA3Gbps HDD's (repurposed for new X79 build)
- 2x Seagate 7200.7 160GB E-IDE HDD's
- 3x Hitachi Deskstar 320GB SATA3Gbps HDD's
- Enermax 690W High-Efficiency PSU
- Antec SOHO Server Case with 8x3.5" bays/5x5.25" bays (fits a Xigmatek 4x3.5-in-3x5.25 with 120x25mm fan converter nicely, for a total of 12xHDD's) and ripe for some heavy modifications
OR
- Thermaltake XASER V Limited Edition with 5x3.5" bays + 6x5.25" bays
OR
- Buy a new sub-$100 case for this (Rosewill ThorV2 would be nice for price, Antec 1100/1200/P283
- DD Maze6 CPU Block
- 2x DD Maze6 GPU Blocks
- 3x Swiftech MCW82 GPU Blocks
- Laing DDC3.25 + 2x Laing D5 Vario Pumps
- Swiftech MCRES-Rev2
- HWL Black Ice GTX 360 rad
- HWL Black Ice GTS 280 rad
- 4x Misc 120-240 Rads
- 11x Delta Fans (7x 120x38mm 2200-4500rpm up to 133cfm 14.25mmH2O, 4x 120x25 2400-4800rpm up to 155cfm 15.2mmH2O)
- 5x NIDEC Fans (120x38mm 0.98amps 11.5-13.2V, up to 4250rpm 151cfm 22.32mmH2O)
- >50x Misc 80x15/25/38mm, 92x25/38mm, 120x12/25/38mm, 140x25mm Fans
I have been looking for two things: a Motherboard and a GPU (or pair of GPU's), and while this will be a Home Server/Media Server, it will also function as a F@H box. I am thinking that 2x 9800GTX+'s or 2x GTX260 216core's in SLI would suffice, but perhaps not? I don't know much about the C2D/C2Q era MB's/GPU's....
I have been looking at the following boards:
- Asus P5Q Premium (huge amount of connections, would allow 4x GPU's for F@H or 3xGPU + 1x RAID Card)
- Asus Maximus Extreme
- Asus Rampage Extreme
For GPU's, I've been really considering the following, from lowest cost to highest:
- 2-3x 8800GTS 512MB (G92)
- 3x 8800GTX's
- 2-3x 9800GT's
- 2-3x 9800GTX+'s
- 2x 9800GTX2's
- 2-3x GTX260(216core)-to-GTX295's
- 2-3x GTX460's-to-GTX480's
Anyone remember enough about this older hardware to help me out?