My Biostar TA690G Review: mATX Overclocking Gem

[Lawrence Lee]

I built the following rig for my cousins to replace their aging P3:

AMD X2 3600+ Brisbane (1.9Ghz, 65W, 1.300V)
Arctic Freezer 64 Pro PWM @ 7V
Biostar TA690G
1GB Corsair Value Select DDR2-667 (128MB allocated to video memory)
Maxtor DM9+ 120GB 8MB SATA1 HD
LG H54N 18x DL DVD+/-RW (PATA)
Sparkle SPI 400W PSU
Apex TX-319 Case
Fans: 100mm Scythe exhaust (came with the Mine Rev.B), 120mm Antec intake (specs unknown, about as loud as a D12SL-12), both at 7V
Acer AL1715S-8 LCD (1280x1024, connected through DVI)
Windows XP Pro SP2


WHAT'S INCLUDED
Motherboard accessories came in a nice nylon bag - very nice surprise. Very few accessories though, just an IDE cable, floppy cable, SATA cable, SATA power adapter and backplate. Also a manual and driver CD. Definitely a no-frills product.


LAYOUT
One of the best layouts I've seen in awhile. Power connections are at the edge of the board, IDE port is high and aligned vertically. They even put in a power and reset button right on the board, taking a page out of DFI's book. Only negatives are the northbridge heatsink looks awfully small and the area around the socket is a bit tight. The Arctic Freezer installed fine but the first DIMM is obstructed, so if you plan on using that combo, install the memory first (if you're going to use that first bank). Also, beware the 24-pin ATX connector... it's incredibly tight.


BIOS
The BIOS had a lot of options, especially when it came to memory timings and HT settings. However the BIOS it shipped with did not come Cool n' Quiet support, any kind of fan control or undervolting options (only overvolting). A BIOS update enabled C&Q and Smartfan. The flash utility inside the BIOS allows you to flash from a non-bootable floppy, but it still has to be a floppy. Updating to the April 27th BIOS improved things a lot and also all the overclocking options were moved to a single menu; very convenient. There's also a CMOS Reloaded section for you to save BIOS configurations, however any saved configurations disappear if you try to boot with unPOSTable settings. When this happens, cut power to the system and turn it back on again. It will POST with default settings, so you don't have to reset the CMOS.


SETUP/INSTALLATION
The board gave me zero problems until after I installed Windows. Welcome to driver hell. I could not find updated drivers for this board from any of the Biostar mirror sites. The Taiwan site has a review link for this board on the front page, but I could not find this model listed anywhere on the site. Funny, huh? Not really. I was lucky to even get a BIOS update - I Googled around and found a link to it on the Chinese site (http://www.biostar.cn/driver2005/BIOS/0 ... 9AM427.rar).

Eventually I found updated drivers from ATI/AMD and individual audio/ethernet drivers from Realtek. Unfortunately, I could not get Catalyst 7.4, 7.3, or 7.2 to work properly (7.4 is the worst, don't even bother trying it unless you like having the screen go black randomly). I had problems playing video until finally I gave up and used the drivers on the CD. They were at least 6 months old, but they worked.


ONBOARD VIDEO
The onboard video I found to be sufficient, though to be fair, I didn't exactly see the point in testing any games on it. I did put it through 3DMark2001 and 2003 and got 5000 and 1500 scores respectively if that counts for anything. It's quite obvious it's only suitable for older games or less demanding modern games at lower resolutions. It did play H.264 content fairly well without any stuttering. CPU usage comparisons (VLC Player):

H.264 720P
========
X2 3600+ @ 2.4Ghz/X1250: 25-38%
Opteron 165 @ 2.5Ghz/7950GT: 14-18%

H.264 1080P
========
X2 3600+ @ 2.4Ghz/X1250: 36-52%
Opteron 165 @ 2.5Ghz/7950GT: 18-31%


FAN CONTROL
Weird with a capital W. In the BIOS, the Smartfan values it claims you can use are 0-127. However, it won't let you input a 3 digit number so it's actually 0 to 99. At the "0" setting I estimate the lowest the fan speed was reduced to was about 9V during bootup, which was still too loud for my liking. The threshold temperatures seemed to have no effect on the speed. Once at the desktop level, the speed seemed to remain constant no matter what I did, so I tried the fan control utility off the Biostar driver disc.

When the fan control utility loads up, it's set to AUTO, which is quite aggressive. Switch it off AUTO and you can control the CPU fan and one case fan manually, from a claimed 0-100% (in actuality probably 5V to 12V). However when you exit the utility and load it up again, it sets itself back to AUTO. You can't save the settings and you can't configure it to change dynamically. If you load the utility, set the fan speeds manually, then exit, it will keep the fans at those speeds until you reboot or the utility is run again. Not a completely worthless program, but extremely inconvenient. FYI, you have to connect a 4-pin fan to the CPU fan header to have any control over it.

Unimpressed, I of course installed Speedfan. Unfortunately, Speedfan does not work, you can't control any fans and the temperature is not even reported correctly. Of course it's also reported incorrectly in the Biostar utility as well, always hovering around 50C. Disappointed, I 7V modded all the fans and left it at that. Hopefully Alfredo will be able to add support for this board in the future because two fan headers CAN be controlled, it's just a matter of better implementation.


OVERCLOCKING
I prefer overclocking without adding voltage, so I dropped the HT multiplier, set the memory setting to lowest (400Mhz, which is 1:1), loosened the memory timings to 5-5-5-15 and started cranking up the HT frequency in increments of 5Mhz. POSTed at 230Mhz (CPU/RAM), wouldn't POST at 235Mhz. After reading a tip on another forum, I increased the memory setting to 533Mhz (I know that goes against all overclocking sense and reason, but for some reason it helped). POSTed up to 265, but not at 270. Increased vCore by 0.1V just to see if it was the motherboard or the CPU holding me back out of curiousity. It POSTed up to 275 with the additional voltage so I guess it was the CPU's fault. I like even numbers so I settled at 253Mhz to give me 2.4Ghz. I would've gone for 258Mhz to give me 2.5Ghz, but I wasn't confident the cooling setup would hold up all summer in a mATX case.


UNDERVOLTING & POWER CONSUMPTION
Using CrystalCPUID, I found the lowest multiplier I could use was 5. At 4 or 4.5 the system would freeze up. I went with multipliers of 9.5, 7 and 5. Here are the lowest stable voltages I could get along with along with system power consumption figures (directly from the wall socket) at idle and with Orthos running (small FFT setting):

5.0 x 200Mhz = 1.00Ghz, 0.800V, no power figures
5.0 x 253Mhz = 1.27Ghz, 0.900V, 62W, 71W
7.0 x 253Mhz = 1.77Ghz, 1.100V, 64W, 83W
9.5 x 200Mhz = 1.90Ghz, 1.000V, no power figures
9.5 x 253Mhz = 2.40Ghz, 1.275V, 70W, 108W


FINAL THOUGHTS
Mixed opinions, but generally positive. The layout is excellent, the overclocking ability is top-notch (for a mATX board anyway), and it's got a very good set of features (HDMI, DVI, 8 channel HD audio). Drivers are poor as is the fan control, but I feel both will be remedied in time. I found the overclocking memory quirk to be baffling, but then again a year ago I had never heard of FSB holes either, so no big deal. The northbridge heatsink does get quite hot, but it never caused any problems as far as I could tell. If I had to describe this board in one sentence I would say "Quirky but formidable mATX overclocking HTPC board."

[jojo4u]

Nice review, thanks. About Speedfan: With my Gigabyte, I had to change the speed's option in the advanced options tab from Smartguardian to the other option.

[Lawrence Lee]

Nice review, thanks. About Speedfan: With my Gigabyte, I had to change the speed's option in the advanced options tab from Smartguardian to the other option.

There was no such option for this board. Speedfan recognized 3 chips and there was no PWM option on any of them.

In addition, on some boards it would show fan controls (nothing would happen when you change the speed), but in this case ithe controls weren't even there - it was just blank.

[Max Slowik]

I found that with more than a few smaller-form factor motherboards that SpeedFan doesn't work because Cool'N'Quiet is forced on in the BIOS, and that there's no option to disable it.

With the AMD driver installed, does that appear to be the case here? A layer of idiot-proofing hampering the tweaking? Not that it would be a good thing, but it would be a 'splanation.

[Chocolinx]

Nice review, thanks. About Speedfan: With my Gigabyte, I had to change the speed's option in the advanced options tab from Smartguardian to the other option.

There was no such option for this board. Speedfan recognized 3 chips and there was no PWM option on any of them.

In addition, on some boards it would show fan controls (nothing would happen when you change the speed), but in this case ithe controls weren't even there - it was just blank.

I have the ASUS version of this board M2A-VM. I got the same problem with SpeedFan not giving any speed control. It's probably just because SpeedFan doesn't support the A690G chipset yet.

[Squirrel]

ONBOARD VIDEO
The onboard video I found to be sufficient, though to be fair, I didn't exactly see the point in testing any games on it. I did put it through 3DMark2001 and 2003 and got 5000 and 1500 scores respectively if that counts for anything. It's quite obvious it's only suitable for older games or less demanding modern games at lower resolutions. It did play H.264 content fairly well without any stuttering. CPU usage comparisons (VLC Player):

H.264 720P
========
X2 3600+ @ 2.4Ghz/X1250: 25-38%
Opteron 165 @ 2.5Ghz/7950GT: 14-18%

H.264 1080P
========
X2 3600+ @ 2.4Ghz/X1250: 36-52%
Opteron 165 @ 2.5Ghz/7950GT: 18-31%

Thank god! Finally some numbers on what I can expect. Every site I've visited has totally different requirements for HD video. But your build is almost exactly what I was planning for mine, and it seems to work well. Been wondering if I should go for a E4300 to be able to clock it if I would need to. (It doesn't help that some label the X2 idle power at half the Core 2, some label it as equal, some give the advantage to Core 2...)

But a question: is VLC multi-threaded? Wondering if that 52% is one core maxed and another for background processes.

I'm upgrading from the budget variant of the Geforce 3, so I think I will enjoy the upgrade either way.

[Lawrence Lee]

Been wondering if I should go for a E4300 to be able to clock it if I would need to. (It doesn't help that some label the X2 idle power at half the Core 2, some label it as equal, some give the advantage to Core 2...)

But a question: is VLC multi-threaded? Wondering if that 52% is one core maxed and another for background processes.

For a mATX build it seems next to impossible to get a good overclock out of a E4300 unless you use the Abit Fatality board.

VLC is indeed multithreaded as far as I can tell (i.e. Task Manager).

[Mats]

FYI, this mobo is a nice overclocker that can go higher than 300 MHz, see link.

[swissguy77]

Thanks for your review Amourek!

I'm interested in this biostar board as well (because of HDMI plus DVI). I want to pack it into the small Silverstone LC19B case with its passive power supply. My "wish" would be to be able to cool the cpu passively. Perhaps by cutting of the stock heatsink so that it is just touching the case, making the case part of the passive cooling. However, I don't know if this is realistic or not...

Which AM2 CPU would you recommend? The older 90nm EE-Versions or the new 65nm Brisbane?
How hot does your 3600+ Brisbane get? Would you feel comfortable to run it at stock speed, perhaps slightly undervolted, passively?

If I understood your review correctly, this biostar motherboard is a good undervolter. Is AMD's cool and quiet still working, if you change, the CPU's voltage?

Thanks for your comments!

[Lawrence Lee]

Which AM2 CPU would you recommend? The older 90nm EE-Versions or the new 65nm Brisbane?
How hot does your 3600+ Brisbane get? Would you feel comfortable to run it at stock speed, perhaps slightly undervolted, passively?

If I understood your review correctly, this biostar motherboard is a good undervolter. Is AMD's cool and quiet still working, if you change, the CPU's voltage?

From what I've read 65nm Brisbanes use slightly less power than 90mm EEs. I'd go with a Brisbane.

I don't know how hot the CPU actually got because I could not get a good temperature reading from either Speedfan or Biostar's utility. It read in the low 30s in the BIOS though. My Brisbane undervolted fairly well at stock speed (Orthos stable for 4 hours at 1.000V). The heatsink itself didn't get very hot, and I would be comfortable running it passive but only with a better tower heatsink like the Ninja.

The board doesn't have any undervolting options in the BIOS so I had to do with CrystalCPUID. I didn't want it interfering with C&Q so I changed the power setting so it wouldn't activate. I don't really like C&Q... doesn't undervolt low enough and it takes too much CPU activity for it to increase to stock speed.

[noee]

The board doesn't have any undervolting options in the BIOS

Not true, at least in the newer BIOS version (4.27 and above). Do a search and you will find that there are a few of us who have undervolted the TA690G with a Brisbane quite effectively, using the BIOS settings only.

You can do it with CnQ enabled or disabled, but I prefer to disable CnQ as well.

HTH.

[Lawrence Lee]

The board doesn't have any undervolting options in the BIOS

Not true, at least in the newer BIOS version (4.27 and above). Do a search and you will find that there are a few of us who have undervolted the TA690G with a Brisbane quite effectively, using the BIOS settings only.

Well I don't know what to say... I was using the 4.27 BIOS and the voltage adjustments only went higher, not lower. *shrug*

[ppl4golf]

Great review, covered a lot of stuff folks really want to know.

Do I understand right : a 3-pin cpu fan would NOT be controlled in this mobo ??

[sun4384]

There are a few issues to be resolved with this board...

1. As pointed out in the review, no speedfan support yet. Smartfan feature in the BIOS which can be enabled with bios update may work, but at least it doesn't with 3-pin fans.

2. Also as pointed out, CPU temperature reading is incorrect. Exactly speaking, different monitoring programs give different results. I'm not sure which one to believe.

3. Random black screen, either with or without the control center. See the user review from newegg, and the articles in vr-zone http://forums.vr-zone.com/showthread.php?t=134098

4. Though it may not be a big issue, the Vdimm starts from 1.95v, not 1.8v. Anyone know some reasonable explanations?

The included T-xxx utilities are very low-quality ones. Quite pathetic... RMClock seem to work very fine with this board. Now I wish there is something that can control FSP in Windows. Does ClockGen work with this board?

[Lawrence Lee]

4. Though it may not be a big issue, the Vdimm starts from 1.95v, not 1.8v. Anyone know some reasonable explanations?

Yeah I noticed that, forgot to comment on it. I believe it's for compatiblity reasons, i.e. memory that won't POST with 1.8V.

[sun4384]

One more thing to add:

5. Though there are numerous memory tweaking options, they seem to be ignored and the SPD values are used instead according to the DDR setting. Could be solved in the next BIOS update.

By the way, I'm using the latest 0601 bios with catalyst 7.3 driver now. So far the black screen thing didn't occur. In case of 7.5, it happened within 30 minutes of operation. So, for the moment stay away from the catalyst 7.5 version.

[noee]

Yes, I can verify what you describe with any ATI drivers after 7.3, regardless of the BIOS rev. I too am on the 0601 rev.

My guess is that we'd have to wait for Biostar to release a driver pkg instead of using the driver right from ATI.

And just to be clear, their are two voltage options for vcore in the BIOS, one is for explicitly setting the vcore (this is how I undervolt to 1.152) and the other is a stepping option from startup vcore that only goes up.

I tested the SPD bug with the 0516 BIOS and it was fixed, but I haven't tried yet with the 0601 rev.

[sun4384]

This memory timing thing drives me mad... Probably I'll have to try the 0516 BIOS. BTW, what was the improvement from 0516 to 0601 BIOS? I cannot notice at least any changes in the setup menu.

The board keeps trying to "optimize" the memory timings and the CPU:MEM divider on the fly according to the change of the CPU multiplier. If done correctly it would be very useful when using programs like RMClock. However, on each DDR setting, no matter it is DDR400, 533, 667, or 800, there is a "dead zone" in the CPU multiplier setting which momentarily overshots the memory setting which will cause the system to fail.

My guess is that when the CPU multiplier is increased from low to high, the memory divider does not increase instantly along with, leaving the memory too overclocked to work stably for a brief moment. Then the system crashes... This usually happens in 7 ~ 8 CPU multiplier range.

By the way ClockGen "works" with this board. I had to set the clock generator to xxxx16 (can't remember correctly) to make it work, not being sure if that was the correct choice. There are comments about making ClockGen work with hotkeys, but I can't find how to do so, Any idess?

[noee]

Seems like I read that the 0601 rev fixes some CPU support issues introduced in 0516, I can't remember. I think I read it on VR-zone?? I don't really notice anything else so far with it.

Yeah, that dead zone on my setup appears very often with 533 divider, 400 works well though and my mem is SuperTalent 667. I think you're probably right re: the "automatic" stuff that it's trying to do.

Sorry, can't help with ClockGen. If I find anything else with 0601 rev, I'll post it back here.

[Prodigy Freak]

hi to all from serbia, the land of kalash

ive purchased ta690g a week ago with brisbane 3600+ and 3x1gb apacer ddr2 800mhz.

flashed to bios 601 and tryed to overclock but only got stable to 235htt :s

flashed back to 427 and managed to god this, stable.



this board rocks my ass :hat1:

[frank2003]

Thanks for your review Amourek!

I'm interested in this biostar board as well (because of HDMI plus DVI). I want to pack it into the small Silverstone LC19B case with its passive power supply. My "wish" would be to be able to cool the cpu passively. Perhaps by cutting of the stock heatsink so that it is just touching the case, making the case part of the passive cooling. However, I don't know if this is realistic or not...

Which AM2 CPU would you recommend? The older 90nm EE-Versions or the new 65nm Brisbane?
How hot does your 3600+ Brisbane get? Would you feel comfortable to run it at stock speed, perhaps slightly undervolted, passively?

If I understood your review correctly, this biostar motherboard is a good undervolter. Is AMD's cool and quiet still working, if you change, the CPU's voltage?

Thanks for your comments!

I'm very interested in how you are doing in your quest of putting everything in the LC19 box, as I'm trying to do the exact same thing.

In my search for the perfect HDMI mobo I came across this Biostar board, but hit a brick wall in terms of getting product info. It appeared they didn't have any docs on this board, as if it didn't even exist. They did have docs on their 7050 board, though. Maybe they are trying to phase this 690G board out?

At the moment I'm considering the Sapphire 690G board, which has confirmed undervolting support in the BIOS. But the only source for it (zipzoomfly.com) does not have it in stock, and there are reports that no retailers are getting it.

So does this Biostart board have any undervolting support in the latest BIOS?

[Mariner]

Thanks for your review Amourek!

I'm interested in this biostar board as well (because of HDMI plus DVI). I want to pack it into the small Silverstone LC19B case with its passive power supply. My "wish" would be to be able to cool the cpu passively. Perhaps by cutting of the stock heatsink so that it is just touching the case, making the case part of the passive cooling. However, I don't know if this is realistic or not...

Just seen your post for the first time. I'd have to say that I don't really think that you'll be able to passively cool even a low-power CPU in the LC19 case, at least not with currently available heatsinks.

I built my system around the start of this year and also hoped to have a fully-passive system (my motherboard is an ASRock ALiveNF4G-DVI, incidentally). My CPU is an AM2 Sempron64 3200+ and I also thought of using the standard heatsink to passively cool. Unfortunately, even after undervolting and underclocking to 1.4GHz @ 0.9V the core reached over 50 degrees C after running Prime torture test overnight. This was also during the winter months when ambient temperature my house is much cooler than in the summer. The stock AM2 Sempron heatsink was also too tall to fit in my case as it stood - I would have needed to remove a few millimetres off the top of the heatsink if I was to put the cover on the case.

In the end, the only suitable heatsink I could find which would fit in the exceedingly low-profile LC-19 was this Hipermedia cooler. I currently run my Sempron at 1.8GHz @ 0.975V with the stock 80x15mm fan running at 5V. This is inaudible more than 1 metre away from the case and runs around 35 degrees C under load.

Incidentally, I have just received delivery of a Brisbane X2 3600+ which I intend to use in my PC - the Sempron was just a little too underpowered. Hopefully, I'll have the opportunity to install this tonight or tomorrow.

I'm expecting to undervolt the Brisbane to the lowest level I can and possibly underclock a little also. If I could get it down to say 1.5GHz @ 0.9V I think this could also run very cool whilst providing quite a bit more processing power than my old Sempron. I'll report back once I know how successfully this works with my current fan/heatsink combo.

[frank2003]

Great info, Mariner!

I was looking at the Zalman CNPS8000 low profile HSF with 62.5mm height. Do you think it would fit in the LC19 (I read somewhere it has 68mm headroom)?

Silverstone has a new case, the ML02. It looks like it may have more breathing room than the LC19.

[Mariner]

Great info, Mariner!

I was looking at the Zalman CNPS8000 low profile HSF with 62.5mm height. Do you think it would fit in the LC19 (I read somewhere it has 68mm headroom)?

Silverstone has a new case, the ML02. It looks like it may have more breathing room than the LC19.

Regarding the Zalman: NO NO NO NO NO!!!

Before I built my system, I spent quite a while looking for a heatsink which would fit in the 68mm height of the LC19 as specified on the Silverstone web site. Imagine my surprise when the case arrived and I discovered that the height quoted actually includes the feet on the case as well. The height of the enclosure alone is just 50mm which makes it practically impossible to source heatsinks!

The Hiper heatsink I linked to is the only AM2 one I've found anywhere which will fit in this case and that only has a 15mm high fan attached. At a push a 20mm high fan might just fit but, of course, all the best 80mm fans are 25mm high and there would not be enough room to put the top on the LC19 if a 25mm high fan was used. I suppose if somebody produced a heatsink around 38-40mm in height with some heatpipes, it may be possible to passively cool in this case but, if such a beast exists, I've never seen one.

As another word of warning, if you look at the pictures of the ML02 on the Silverstone web site...

http://www.silverstonetek.com/products/ ... 2&area=usa

...you'll see that they appear to be providing misleading specs once again. It appears to me that the enclosure itself is 50mm and it's just the 'feet' and fascia on the front and back which take it to 82mm. Check out the side and rear views especially.

I'll report back once I've had time to install and test the Brisbane in my system.

[Mariner]

I actually posted my initial report in another thread so here it is copied across verbatim:

Just to confirm that my initial testing of a Brisbane X2 3600+ in my LC19 seems to be going quite well. The enclosure of the LC19 looks to be almost exactly the same dimensions as the ML02 though there are differences in the vents at the top. I also note on the Silverstone web site that a new version of the case, LC19R includes a PCI-x riser and a card reader as standard.

I've undervolted the Brisbane to 1.9GHz @ 1.0V and, running 2 instances of Prime95 overnight, Speedfan reports CPU temperature at 45 degrees Celsius and System temp around 41 degrees C. The only cooling in the system is the 80x15mm fan on my low-profile heatsink and this is undervolted to 5V - pushes about as much air as a mouse's fart!

Hopefully my Brisbane will also run stock speeds at 0.975 or even 0.95V - more testing required. I can see myself having a play with RMClock if I get the opportunity as well since I'll rarely need more than half the power this system can provide.

Very impressed with the Brisbane - I'd imagine an undervolted single-core Lima would use a miniscule amount of power.

[frank2003]

Just want to add another data point for those thinking of building a quiet HTPC to play back HDTV recordings using this board.

I was not sure what processor to get that would be fast enough to play back HDTV 1080i recordings, and I knew I certainly did not want to overclock. So I went with the 65W 2.1ghz 4000+.

After struggling with different versions of the BIOS (boy, were they hard to find) I was able to arrive at 1.6ghz @ 0.9v that gave acceptable playback performance of 1080i recordings in VLC and WMP10. The total power consumption was 67W idle, 72W with VLC and Orthos running (CPU @ 100%). The CPU temperature stayed around 42-44C using stock HSF with fan running around 1850rpm.

I tried to determine if I could run the processor fanless. At 1.6ghz @ 0.9V, the core temperature slowly crept up from 37C to 58C when I stopped the experiment. So fanless operation was out of the question for HDTV playback.

I used the HDMI output to connect the board to a 1080p HDTV. The audio came through the HDMI cable; that was very nice as it eliminated the need to run a separate audio cable. I had to struggle with getting the ATI driver to output 1080p without underscan (black border all around the TV). Apparently this bug affects other ATI GPUs, not just the 1250 used in this board, and has been around for at least 10 months without an official fix from ATI. A simply registry hack fixed that.

With the full load power consumption at a mere 72W, I now feel more comfortable about fitting the board in a slim case like the Silverstone LC19 or ML02. I'll report back once I get the final system set up.

Thanks to all for an excellent thread.

[frank2003]

Some more thoughts:

I originally had my mind set on the 45W, 2.1ghz BE-2350, but it was nowhere to be found. You may argue that by undervolting the 4000+ 65W you could achieve the same or better power consumption objectives. That's partly true, but there's a subtle difference. I've found that changing the vcore means you have to turn off Cool 'n Quiet. Whereas the BE-2350 would run in 1, 1.8 and 2.1ghz under CnQ, thus giving you a system idle consumption of around 52W; the undervolted 4000+ @ .9v, 1.6ghz on the other hand draws 67W at idle.

I also tried undervolting using CrystalCPUID, but found two major issues:
1. CCPUID did not kick in until the system booted up. Between power-up and system boot up, the power consumption surged and peaked at over 100W. With undervolting in the BIOS, the consumption never went above 70W. While this may not be a big deal for most PSUs, it was too close for comfort for the case I was considering which only had a 120W DC-DC PSU powered by an external brick.
2. When the system woke up from sleep or hibernation, CCPUID did not detect this condition, and the CPU settings went back to stock 1.325V, 2.1Ghz and the system idled at 87W. Not good.

[Steve_Y]

I also tried undervolting using CrystalCPUID, but found two major issues:
1. CCPUID did not kick in until the system booted up. Between power-up and system boot up, the power consumption surged and peaked at over 100W. With undervolting in the BIOS, the consumption never went above 70W. While this may not be a big deal for most PSUs, it was too close for comfort for the case I was considering which only had a 120W DC-DC PSU powered by an external brick.

Have you tried undervolting in the BIOS and still using CrystalCPUID? It's not something I've tried, but I don't see why CrystalCPUID couldn't override an already undervolted configuration when used.

2. When the system woke up from sleep or hibernation, CCPUID did not detect this condition, and the CPU settings went back to stock 1.325V, 2.1Ghz and the system idled at 87W. Not good.

That's strange. I've just tried putting my system to sleep and CrystalCPUID's multiplier management is still working fine now that it has woken up. It no longer seems to be detecting my external hard drive correctly, and explorer seems to be freaking out, but the power management seems fine.

[frank2003]

I saw the wake from sleep problem when I was experimenting with the "AMD K6/K7/K8/LX Multiplier" settings. I didn't spent much time working with the multiplier management, but will try it.

I just found that the wake from sleep problem might not be entirely the fault of CCPUID at all. Without CCPUID running, the BIOS resets to stock frequency and voltage when the PC awakes from sleep or hibernation! So it's clearly a BIOS bug. But I think CCPUID should still have reset the lower frequency and voltage as configured.

[Max Slowik]

In theory, CrystalCPUID requires that C'N'Q be enabled in the BIOS settings to operate; many motherboards automatically disable C'N'Q when under-volting.

However, I do both on my HTPC, and it works flawlessly (except when it doesn't--I didn't spend enough time stress-testing the different voltages).