Biostar TPower I55 Review!

Introduction :
Biostar were founded in 1986, with their original focus being on XL motherboards. Since then, like any good business, they've developed and expanded their range and product quality. As far as enthusiasts go, back in the Socket A/ 478 days, their star (no pun intended) was pretty dim and awareness of their products fell behind those of companies like Asus, Abit, Gigabyte et al. They made some headway with AMD Socket 939 but it wasn't until Intel socket 775 developed and gave us P45 that Biostar went supernova. Their P45 boards filled global leaderboards until X48 came along, but Biostar were already much more in the enthusiasts field of vision. Biostar have been fairly quiet on the Intel side since Nehalem came along. They released X58 boards in the first wave of I7-fever, but were spending more time with AM2 and AM3 boards, with some notable success. The release of Intel P55 has seen Biostar release new boards, against strong competition from all the usual names. We have the TPower I55 here for review... lets see how it stacks up.


Box, Bits, and Board :




Very eye-catching! The gold is a head-turner... and is also damned hard to take a good picture of! Same criticism as the TH55.... the selling points in the corner are.... not selling points.










Flipping over the box, we get details on the more interesting features of the board. The "G.P.U. Power Indicator, sadly, has nothing to do with GPUs or GPU power saving. It stands for "Green Power Utility" and the naming is heading towards being mis-leading. It is basically a set of LEDs that follow PWM utilisation.


The Bits:






Opening up the box, we get a very clean presentation. The contents are split into two layers, with the top layer being split into Left-Right packaging for protection.

Accessories: Five single S-ATA power leads, five S-ATA data cables, IDE and Floppy connectors, I/O plate, SLI and CF bridges and a bridge retention bracket. The cabling bundles are gathered together in branded velcro ties. Useful!

Also included are the usual manual and driver installation CD..




Removing the top layer to get to the board, I was impressed to see the board sat in its own foam cutout. I'm not used to seeing this and does a very good job of protecting the board in transit. +1 here, no doubt about it.


The Board:











Onto the board... its covered in feature-promoting stickers. I would like to see some of these... or all of them.... shunted to the front of the box to help tempt buyers. They're no use inside the box. A user has already bought the board before they see this... the advertising bit is finished!







On-board voltage readpoints. They're reasonably easy to access and accurate enough for my liking. Ideally, I'd like to see them presented as "pits" or holes, so its harder to slip whilst probing and short something out.





This was meant to show the vDIMM PWM.. but look! an IDE port!





On-board power & reset buttons and a debug display.





Nice beefy heatsink 





On P55/ H55, heatsinks in this position are essentially waste as they're not covering a heat source. I guess it can dissipate excess heat from the SB and/ or PWM (all connected via heatpipes) but... the board runs cool enough that its not a problem.

The Board, continued





Speaking of PWM heatsinks....... These are well designed and offer high surface area with a good transmission area from the heat source to the fins. Some boards i've seen recently did not feature well-designed heatsinks, so + 1 here.









I/O is well-equipped. Connections are nice and tight

BIOS notes

BIOS options include:

CPU clock Swing

CPU + PCH Clock delays. There are only delay options, no clock advance options.

CPU voltage is, like the H55 board, measured relative to it's base value, which i'm guessing (hoping) is VID. The CPU can have up to 0.7875v added to it, which will give a maximum of ~1.9v depending on the CPU VID. There are limited undervolting options, allowing up to 0.1v less than VID.


vTT can be set up to 2.03v


CPU PLL can be set as high as 2.73. I really doubt anyone will go anywhere near that high though....


vDIMM ranges from 1.60v to 2.53v. Like the TH55, low vDIMM options appear to be an afterthought and are listed after the ridiculously high vDIMM options and are also shown in red. Users should not be put off by this though, the values are accurate.


PCH voltage can be set between 1.1 and 2.03v, as can the PCH PLL voltage and PCH ME Voltage. I have no idea what this voltage is!


Clockgen voltage can be set between 0.015v and 0.6v higher than standard. 



There are also phase frequency options for some PWMs, which is a nice touch.

CPU phase freq. can be set: -20, +20 or +40%

DIMM phase Freq. can be set +30, +60 or +90%

vTT phase Freq can be set between +20 and +80%

While I suspect not many users will dabble heavily with these settings, an increased PWM Frequency can help stabilise relevant clocks, but at the expense of power consumption and heat. It would be nice to know the actual frequency being supplied though.

Available RAM timings are fairly standard, but we have access to R-R, W-R and R-W timings for same and different rank. One slip-up here.... There are two settings for R-W delay, same rank. I don't know which one is the real one... and I have no idea what the duplicate is controlling.


Another BIOS problem..... again, like on the H55, Using the + and - keys to change the DRAM Timing Control options results in a BIOS hang. Biostar engineers were unable to replicate this however. It might be a one-off linked to my keyboard.





Setup, testing, assumptions, drawbacks.

Test rig:

Biostar T-Power I55 .
Intel I7 860. Quad-core w/HT
2x2GB G.Skill EO 800MHz 7-8-7 @ 1.35
PSU: Xigmatek 1200W NRP.
GPU: PCI 5200 for power consumption tests, nVidia 8800GT single & SLI @ 800/1100 for 3D performance
HDD: Samsung S-ATA II
Watercooling for the CPU + 8800GT GPUs

* The PCI 5200 was used because of its low power draw. I wanted the focus to be on the CPU and board. The card is about as close to "no power drawn" as a discrete GPU can be.
* The PSU is my bench unit. Given that the power consumption with the 8800GT SLI was comfortably under 500W, the unit is total overkill and power efficiency is unlikely to be at its maximum, given the ~40-45% loading. Users with high quality 6-700W PSUs will probably see lower power consumption numbers.
* The CPU is the second-most power hungry s1156 CPU. May as well give the board a workout!
* The PWM got quite warm if no airflow was present, but if I put two low-speed 92mm fans over the RAM, the airflow overspill left the PWM cool to the touch. Very impressed!


C1E was enabled for all the power consumption tests, which explains the very low idle voltages. The feature was disabled for the 3DMark testing as the load-idle cycling between tests played havoc with stability.



Power consumption/ power efficiency


Running stock MHz- 133x21 @ 1.12v (-0.1v in BIOS.) vDIMM @ 1.45v, vTT @ 1.15v, we get the following:

Power-saving off:

Idle: 0.80-0.81v, 96-99w

SPi32M: 1.11v, 113-114w

wP 1024: 1.12v, 155W

Power-saving enabled:


Idle:0.76v, 92W

SPi32M: 1.04v, 106-108w

wP 1024: 1.12w, 155W









V12 Auto-OC. After setting the RAM timings to 7-8-7 manually (a trick learned from the TH55) I enabled the V12 OC and was offered:


145BCLK. x26 idle, x25 for SPi32M and x22 for wP1024.


Voltages and power consumption:


Idle: 0.9v, 89-91W

SPi32M: 1.28v, 116-117W

wP1024: 1.29v, 193W


Again like the TH55, the RAM performance was disappointing, setting 1.60v and using a 1:5 ratio to give 725MHz.







Pushing to 200BCLK... very simple.


200x19, 1.28v.RAM @ x5, timings: 7-9-7, 1.70v. vTT @ 1.35v


Power saving off:


idle: 1.00v, 114-115W

SPi32M: 1.25v, 146-148W

wP 1024: 1.28v, 221-222W









Power-saving enabled:


Idle: 0.96v, 104W

SPi32M: 1.19v, 126-128W

wP1024: 1.28v, 222W







Comparing all these numbers to something.


Enter the MSI P55 GD-80. Same chipset, aimed at enthusiasts, 20%+ more expensive.....


Matching the voltages and settings of the Biostar board, the MSI offered us the following:


Power-saving turned off:

Idle: 1.15v, 98-101w

SPi32M: 1.16v, 107-109w

wP 1024: 1.17v, 152-154w


Power saving enabled: 

Idle: 1.15v, 79-82w

SPi32M: 1.15v, 99-102w

wP1024: 1.17v 153-154w


 




Only a few watts in it, in MSIs favour. Consider this a success for the Biostar board, given the price difference!






BCLK testing:


Setting vTT to 1.42v, 213 was stable.





Not quite as good as the H55, but on a par with what the MSI GD-80 offered me with this particular CPU

Discussion of results.

At stock, the CPU will be the bottleneck, especially in SLI conditions. This will allow us to see the CPU-side efficiency for 3D benching. Tests run @ 3.8GHz should give a good guage of how the scores scale. 3.8GHz with a single GPU should shunt the bottleneck back towards to the GPU, which will indicate the GPU-side efficiency of the board. These numbers will then all be compared to a rival of this board... the MSI GD-80.


Plan A was to compare the Biostar board to the MSI GD-80- an ideal apples- apples comparison. Sadly... the MSI board was totally unwilling to work in SLI mode. I'm still looking into *why* but as a stop-gap, I brought out the Gigabyte X58A-UD3R. A higher-end chipset, higher performing, more expensive (generally) and uses more power.

With those points in mind, the numbers are surprising. There is less than 2% in it....and thats picking the most X58-friendly results.

Comparing the P55 single-card results, the Biostar generally holds a small lead. 

I have all the screenshots available for those who realllllyyyyy wish to pore over sub-test breakdowns. Theres not much to see though.

Im really pleased with how the Biostar held up. Its overshadowed by Asus, eVGA, MSI and Gigabyte but its pulling its weight, performance wise.



Conclusion.

Well.... its up against strong competition. Asus, Gigabyte and MSI have strong offerings in the same price bracket. The BIOS niggles continue to annoy me. I'd like to see tweaks in the vCore department in both the presentation and range. I hope Biostar will act on my comments  

The colour scheme of the board might make some colour-themed rig builds a challenge.

Performance is good... and that makes up for a lot of other small indiscretions.

Availability is... good enough. Scan have it in stock.

Power efficiency is totally acceptable and the board runs cool with even slight airflow.

Hmmm..... conclusion.... cautious recommendation. Sort out the vCore, tweak the colour scheme and its all good.

My thanks to Biostar for the review sample