by Ilya Gavrichenkov
06/26/2007 | 06:13 PM
A few days ago we posted a review of a new Asus mainboard on Intel P35 chipset – Asus P5K Deluxe. As we saw in our review, this mainboard offers advanced users a pretty attractive set of features and they will most likely love it. And it means that the sales volume of Asus P5K Deluxe should become quite significant and it might as well repeat the great success of P5B model.
This particular fact inspired us to continue investigating Asus P5K Deluxe, especially since there still were a lot of questions left that we didn’t have the chance to address in the previous article.
In particular, we are going to pay special attention to practical overclocking techniques on this mainboard, new BIOS Setup functions and perform another performance test session in order to estimate the efficiency of this solution in overclocked platforms.
Moreover, Asus engineers provided an additional intrigue by releasing a new BIOS version. This version 0404 added a few new interesting options to the BIOS Setup, which will definitely discuss also.
Note that this article is a continuation to our previous Asus P5K Deluxe Mainboard Review. If you haven’t yet read it you may find the today’s material a little hard to follow.
So, let’s get started. Please note that this time all the tests were performed with a newer BIOS version 0404 from June 21, 2007.
Until this day Asus mainboards on iP965 chipset were considered the best platform for Intel Core 2 Duo and Intel Core 2 Extreme overclocking. Only with these boards you could hit the highest overclocking results without any too sophisticated manipulations. So, we would certainly expect Asus P5K Deluxe to be a highly overclocking friendly solution as well. Especially, since there are a few reasons for that: BIOS Setup with rich set of options for overclocking and official support of 1333MHz bus.
For our tests we assembled a system on our mainboard with Intel Core 2 Extreme X6800 processor. Besides we also used 2GB of Corsair Dominator TWIN2X2048-10000C5D DDR2 SDRAM, OCZ GeForce 8800 GTX graphics card, Western Digital Raptor WD1500AHFD HDD and SilverStone SST-ST85ZF PSU. Thanks to the memory that can run at up to 1250MHz frequency, we didn’t have to worry about its overclocking potential when clocking it synchronously with the FSB bus and setting the timings to 5-5-5-15. Processor was cooled with Zalman CNPS-9700LED cooler. We didn’t install any additional fans on top of the North Bridge heatsink, because it didn’t warm up too much, even during overclocking. The processor stability during overclocking was checked with the well-known SP2004/ORTHOS utility based on Prime95 code. We also used S&M program to check memory stability during overclocking.
We will not bore you with the detailed description of our overclocking experiments, especially since there is not much to describe. We managed to find the maximum bus frequency when the system remained stable pretty quickly. This was easy mostly due to simple overclocking procedure on Asus P5K Deluxe. Although there are a lot of settings that can be adjusted for maximum overclocking success, most of them can be left untouched.
For example, to achieve the maximum FSB frequency when our processor could work stably and reliably (it equaled 505MHz), we only had to increase the processor Vcore and raise the Vmem a little bit. Other voltages were set at their minimums, however it was more than enough for stable system performance.
While we were working on this and previous article, our lab received a new Intel Core 2 Duo E6850 processor with 1333MHz bus based on the new G0 core stepping. Overclockers are a little bit concerned about this core stepping, because the CPUs built on it should have relatively low frequency multipliers and hence their overclocking potential may be limited by the mainboard’s ability to raise the FSB speed. Of course, we couldn’t resist the temptation to check out the overclocking potential of Asus P5K Deluxe with a processor like that. For our overclocking experiments we reduced the processor frequency multiplier to 7x and pushed its Vcore to 1.45V. The results are provided on the screenshots below:
It proved a smart idea to reduce the processor clock frequency multiplier. Asus P5K Deluxe mainboard worked perfectly stable at 540MHz FSB, which is just another piece of evidence that this solution boasts truly excellent overclocking potential. By the way, we didn’t have to touch the chipset voltage at all to hit this frequency.
So, Asus P5K Deluxe is a great choice for overclocking of any processors, with low as well as high frequency multipliers. The board retains stability at up to 540MHz FSB (and maybe even higher), and these superior results are achieved with minimal efforts, without going though the whole bunch of finest settings, without any voltmodding or additional cooling on the chipset and the CPU voltage regulator circuitry.
At the same time we decided to check out the ability of our Asus P5K Deluxe mainboard to overclock memory. The thing is that mainboards based on previous generation Intel chipsets didn’t cope with DDR2 SDRAM overclocking as well as the solutions based on Nvidia core logic sets. Unfortunately, our tests showed that Intel P35 didn’t improve the situation that much. Once you’ve hit 1200MHz frequency on the memory bus, further overclocking is a real pain, independent of the selected timings or other settings. We managed to get our Corsair Dominator TWIN2X2048-10000C5D test kit to work only at 1213MHz in Asus P5K Deluxe mainboard, while in NVIDIA nForce 680i SLI based platforms it ran just fine at 1250MHz and higher.
However, I can hardly complain about this memory bus frequency being not high enough. Although Asus P5K Deluxe loses to some selected solutions when it comes to maximum memory bus speeds, its potential is more than enough for most overclocker DDR2 SDRAM kits out there, with very few exceptions.
The recent poll we held on our site showed that Asus mainboards are the most popular out there. About 44% of our readers prefer mainboards from Asus. And it means that Asus P5K Deluxe based on the new Intel P35 chipset has all the chances to become a pretty demanded platform for systems with Core 2 Duo processors. At least, it should be as popular as the P5B Deluxe that has because the choice of many computer enthusiasts. That is why we decided to pay special attention to some fine tuning tricks you could use on Asus P5K Deluxe in order to achieve maximum performance. Especially since the BIOS Setup of this solution offers you a lot of new unique features.
The latest BIOS versions for this mainboard allow enabling 1T Command Rate mode. As you already know, all Intel chipsets used to work only with 2T Command Rate and didn’t allow the user to adjust this memory controller setting. By the way, this parameter allows boosting the performance pretty impressively in platforms on Nvidia nForce 600i core logic.
Unfortunately, the adjustment of Command Rate parameter on Asus P5K Deluxe is not always beneficial. Its operation depends on the selected FSB:Mem divider. To be more exact, we only confirmed that 1T Command Rate can be set for 5:6 or 5:8 memory frequency dividers. With all other dividers, 1T Command Rate results in system instability and hence cannot be used.
As far as the efficiency of this function is concerned, we decided to check it out by testing the system performance with different timings settings and different Command Rate settings. the tests were run on an Asus P5K Deluxe based platform with the FSB frequency set to 400MHz. Core 2 Extreme X6800 processor we used in this testbed was configured as 8 x 400MHz, i.e. it was overclocked to 3.2GHz. Other testbed components were the same as during our overclocking experiments.
The tests showed that by setting Command Rate to 1T we managed to slightly increase the performance of the memory controller. The bandwidth increases and latency drops. However, the actual performance gain we get is not that significant. In real applications the system speeds up by 0.5% at most. So, pushing the memory frequency one step up turns out more efficient than fighting for a more aggressive Command Rate setting. And taking into account that 1T Command Rate lowers the memory frequency potential quite noticeably, we cannot state that this setting is a useful one. It will hardly help you achieve maximum performance.
Among the Asus P5K Deluxe BIOS Setup settings we also found a new option that we haven’t seen before called Transaction Booster. You can set it to Auto, Enabled or Disabled. By selecting Enabled or Disabled, you can access another additional parameter: Boost/Relax Level.
We decided to see how this new option affects the system performance. The tests were performed in the same conditions, with the CPU configured as 8 x 400MHz. the memory was working as DDR2-800 with 3-3-3-10-2T timings. Unfortunately, we couldn’t check out Transaction Booster operation with the Boost Level set to 1, because the board wasn’t stable enough in this case.
The obtained results showed that enabling Transaction Booster helps increase the system performance. In real application this improvement hits 1% over the performance level you get with this parameter set to Auto. If you disable Transaction Booster altogether, the performance will drop with any Relax Level settings other than 0.
So, Asus engineers provided their mainboard with another option for performance improvement. Although we have to admit that Transaction Booster function has very limited application field. When it is enabled, the memory frequency potential lowers, so it actually works only with low memory frequency settings. For example, Corsair Dominator TWIN2X2048-10000C5D memory we used in our test platform allowed enabling Transaction Booster function only when its frequency was set below 900MHz. Enabling Transaction Booster at higher memory frequencies led to system instability.
Our tests show how Transaction Booster affects the performance of the memory subsystem and the whole platform in general. However, it still doesn’t answer the question on how Asus engineers managed to speed up the performance of the Intel P35 chipset. MemSet utility can cast some light on that matter, as its latest versions run quite OK with Intel P35 based mainboards.
We used MemSet 3.3 to compare secondary timings with enabled and disabled Transaction Booster. It revealed that this setting affects Performance Level parameter that proved to have serious influence on the memory controller performance.
The smaller is the Performance Level value, the higher is the actual performance. By enabling Transaction Booster, you can reduce the value of the Performance Level parameter.
By the way, Transaction Booster is not the only BIOS Setup parameter that affects the Performance Level. It can also be lowered by disabling DRAM Static Read Control option.
Enabling DRAM Static Read Control increases the memory controller performance; disabling – decreases.
Besides the influence Command rate and Transaction Booster settings have on the system performance, we also paid special attention to the FSB frequency when Asus P5K Deluxe changed Strap and the resulting performance drop. For instance, Asus mainboards on iP965 chipset start changing strap at 401MHz FSB, which leads to noticeable performance drop when we hit the corresponding frequency during overclocking.
In order to find out what the situation looks like on Asus P5K Deluxe mainboard, we tested the memory subsystem performance with the FSB frequency increment of 10MHz. the memory frequency in this session was set with the 4:5 FSB:Mem divider (this is the minimal divider available in the entire FSB frequency range). The CPU clock frequency multiplier was set to 8x. The memory timings were set manually at 5-5-5-15-11-42-10-10-10, so that the memory could work stably at maximum frequencies. Transaction Booster was set to Auto, DRAM Static Read Control – to Disabled. The obtained results are given in the table and on the diagrams below:
Up to 460MHz FSB the performance is growing more or less gradually. If we disregard the unexplainable speed upsurge at 320MHz FSB, we can state that the memory subsystem performance is growing almost linearly within 266MHz-460MHz FSB frequency range. In other words, there is simply no Strap changing in this interval, or it occurs without any visible changes in the mainboard performance.
To check out that there are no performance drops during further increase in the FSB speed, we continued the performance test of the memory controller. Of course, in order to get to higher FSB frequencies we had to reduce the processor clock frequency multiplier down to 7x and set the front side bus and memory bus to synchronous clocking mode. Other settings remained unchanged.
We don’t see any performance drops here either. The memory controller performance results form an almost ideal straight line. So, Asus P5K Deluxe appears the first mainboard that doesn’t suffer from negative effects of the FSB Strap changing on performance.
We could have stopped right here, if it hadn’t been for one thing. The new BIOS version 0404 provided the ability to manually adjust Strap frequency that can now be set to Auto, 200, 266 and 333MHz.
However even this function doesn’t show us for sure if the performance remains unchanged during Strap changing. The thing is that the list of available memory frequency dividers depends on the selected FSB Strap value. Moreover, the ranges of available dividers at different Strap settings do not overlap.
FSB Strap frequency | Available FSB:Mem dividers |
200 MHz | 3:5, 1:2 |
266 MHz | 4:5, 2:3, 1:2 |
333 MHz | 1:1, 5:6, 5:8 |
This allows us to conclude that FSB Strap frequency on Asus P5K Deluxe is connected to the set divider forming the memory frequency. In other words, Strap is changed only when the memory frequency divider changes. That is why we didn’t notice Strap changing in our tests, because it occurred when we corrected the testbed parameters – selected a different memory frequency divider and reduced the processor clock frequency multiplier.
So, it turns out that Asus P5K Deluxe doesn’t have any set boundaries for Strap changing, placing full responsibility for it on the user who can either set it directly or through memory frequency. Unfortunately, for the same very reason we cannot compare the mainboard performance with different Strap settings.
However, we shouldn’t forget that FSB Strap frequency is not really that interesting as is, but bears huge auxiliary importance. Strap is the FSB frequency limit after which overclocking doesn't lead to a significant performance boost. Therefore we decided to carry out one more test: to analyze the performance of systems with the CPUs overclocked to the same speeds but using different multipliers and FSB frequencies. By comparing these results we will be able to find the most “optimal” FSB frequencies for successful overclocking.
The obtained results are given in the table below. The CPU in these tests worked at 3.2GHz. The memory frequency was set as close to 1066MHz as possible. The timings were locked at 4-4-4-12-8-30-8-8-8. Transaction Booster was set to Auto, and DRAM Static Read Control – to Enabled.
Actually, we didn’t expect anything else. Higher FSB frequency on Asus P5K Deluxe always leads to higher performance, independent of the memory frequency dividers and FSB Strap. So, this mainboard has no hidden issues such as unpleasant performance drops occurring once you get beyond certain FSB frequencies.
In conclusion to our extensive discussion I would like to say a few words to those Asus P5K Deluxe owners who for some reason believe that it changes FSB Strap at 500MHz. In reality there is no Strap changing whatsoever and cannot be. All tests that allegedly prove this statement haven’t been performed correctly. The thing is that if the DRAM Static Read Control parameter is set to Auto, the mainboard changes its value at this frequency. That is why the performance drops, which is erroneously considered to be caused by Strap changing. To eliminate this unpleasant effect, you should set DRAM Static Read Control to Enabled in the BIOS Setup.
In our previous article devoted to Asus P5K Deluxe mainboard we studied its performance in nominal mode. However, there are a lot of experienced computer users who work on overclocked systems. For some strange reason many reviews out there do not touch upon the performance of overclocked systems. We decided to make up for this omission, so today we are going to discuss the performance of our mainboard against its competitors in overclocked systems.
Like in the previous article, we will compare Asus P5K Deluxe not only against Gigabyte GA-P35-DS3R mainboard, but also against a few solutions from the same price range but based on the older chipsets. Namely we will take Asus P5B Deluxe on Intel P965 chipset and Asus P5N32-E SLI on Nvidia nForce 680i SLI chipset.
We will compare the performance of our testing participants with the FSB overclocked to 400MHz. We used a Core 2 Extreme X6800 processor overclocked to 3.2GHz that was configured as 8 x 400MHz. The memory was working at 1000MHz with 4-4-4-12 timings.
Our testbeds were configured as follows:
Notes:
When we raised the system bus frequency above its nominal level, the mainboards that we believed performed almost identically started to demonstrate very different results. The main reason for that – different Strap changing on these mainboards. As a result, Gigabyte GA-P35-DS3R turns into a real outsider compared with Asus products. However, will the same thing happen in more complex tests and real application benchmarks? Let’s find out!
If you are an overclocker, then forget about the performance tests in the nominal work mode that we discussed in our previous article. Overclocking changes the whole picture dramatically and shows how greatly your system performance may actually depend on the right choice of mainboard.
The fastest mainboard in our test session with the FSB overclocked to 400MHz appears the good old Asus P5B Deluxe on Intel P965 chipset. However, you should bear in mind that this solution can only boast high performance during FSB overclocking to 400MHz. High FSB frequency will most likely change the situation, because of Strap changing leading to slowing down of the memory controller.
The newcomer, our today’s hero – Asus P5K Deluxe – is close behind the leader. However, Gigabyte GA-P35-DS3R on the same Intel P35 chipset cannot boast the same great performance. It yields to Asus mainboards on Intel chipsets in all tests.
The slowest of all during overclocking is the solution on Nvidia nForce 680i SLI. Asus P5N32-E SLI cannot compete with any Intel P35 or Intel P965 based solutions during overclocking.
Summing up we can say once again that we were very pleased with Asus P5K Deluxe mainboard. This solution offers excellent opportunities not only to mainstream computer users but also to overclocking fans, who would be delighted to check out the new functions in their experiments. The mainboard’s BIOS Setup contains a lot of options for system fine tuning and performance maximization. Moreover, the board performs stably and predictably, just like any high-quality solution should.
The results of our second test session proved once again that Asus P5K Deluxe boasts excellent performance and rich set of features that is why we award it with our Editor’s Choice as the today’s best mainboard for Core 2 processors.
Frankly speaking, I would like to say that in my opinion Asus P5K Deluxe is close to ideal at least if we were choosing from the solutions available in the market today. Great build quality, rich functionality, flexible settings – all this is true for Asus P5K Deluxe. Of course, it has a few drawbacks that we have already pointed out in our previous article, but they will hardly be critical for most users anyway.
However, Asus engineers shouldn’t relax at this time, they still have room for improvement of their Intel P35 based product family. So, we hope that the next mainboard from the Republic of Gamers series will not only inherit all the advantages from the award-wining Asus P5K Deluxe, but will lose its biggest drawbacks as well. BY the way, these new solutions have already been presented in Japan and the new mainboard on Intel P35 with DDR2 SDRAM support should be named Asus Blitz Formula. We look forward to getting this mainboard in our labs very soon.