Two 690G builds
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
Two 690G builds
I recently built two 690G based systems using different mobos and AM2 CPUs, and would like to share my experience with everyone. I took some power measurements with different configurations of mobo/CPU/PSU.
Components used:
* Biostar TA690G-AM2 (latest 605 BIOS)
* Gigabyte GA-MA69GM-S2H (latest F3a BIOS)
* AM2 Brisbane 4000+, 2.1ghz
* AM2 BE-2350, 2.1ghz
* 2 x 1GB DDR2 memory
* Seagate 750GB 7200rpm SATA drive
* Silverstone ML02 with built-in Enhance END 0512 power module; 120W brick
* 120W picoPSU with 80W brick
* generic ATX PSU
Base configuration:
* Biostar TA690G-AM2
* AM2 4000+
* picoPSU
Power consumption:
2.1ghz @ 1.056V (orthos stable 24hours): idle: 36W, load: 63W
1.0ghz @ 0.88V, 31W, idle (under CCPUID control)
1.8ghz @ 0.976V, 55W, load (under CCPUID control)
The Enhance PSU adds 2W to the numbers. The BE-2350 operating at the same Vcore and frequency reduces the numbers by 3W-4W. A generic ATX PSU adds 25W to the figures.
Using the Gigabyte GA-MA69GM-S2H board:
2.1ghz @ 1.056V (orthos stable 24hours): idle: 52W, load: 72W
1.0ghz @ 0.88V, 44W, idle (under CCPUID control)
1.8ghz @ 0.976V, 65W, load (under CCPUID control)
So the Gigabyte board consumes a full 8W-16W more power when compared to the Biostar!! My results show that not all the 690G boards are created equal.
It's clear that if you are building a cool running quiet system you'd want to go with the Biostar.
Here are other comparisons between the two boards:
* The Biostar board would not POST with the Enhance PSU. I'm not sure which to blame but I suspect both: The Biostar may require a brief, high power draw at power-on, and the Enhance's surge protection circuitry is not tolerant enough to handle that. In contrast the picoPSU has no problem whatsoever powering up this board, nor does the Enhance PSU have problem powering up the Gigabyte.
* The NB heatsink on Gigabyte runs HOT, HOT, HOT, even in idle! I suspect that's where the extra 8-16W of power is expended.
* The Gigabtye's heatsink mount is held together with plastic pins, whereas the Biostar's held together by metal screws. This is important if you are using an after market screwed on HSF like the Hyper HFC-20820-C1, a must if you are using a slim case like the Silverstone LC19 or ML02. The Biostar has the right backplate with the screw holes, whereas the Gigabyte does not.
* The Biostar has full undervoltage and underclock controls in BIOS, whereas the Gigabyte only has only underclock control. CrystalCPUID works well with both, though.
EDIT: * The Gigabyte board's standby indicator only works for S1+S3, but not for S3 only which presumably most of us reading SPCR use. So with the Gigabyte board you can't tell the difference between S3 and power off states.
Components used:
* Biostar TA690G-AM2 (latest 605 BIOS)
* Gigabyte GA-MA69GM-S2H (latest F3a BIOS)
* AM2 Brisbane 4000+, 2.1ghz
* AM2 BE-2350, 2.1ghz
* 2 x 1GB DDR2 memory
* Seagate 750GB 7200rpm SATA drive
* Silverstone ML02 with built-in Enhance END 0512 power module; 120W brick
* 120W picoPSU with 80W brick
* generic ATX PSU
Base configuration:
* Biostar TA690G-AM2
* AM2 4000+
* picoPSU
Power consumption:
2.1ghz @ 1.056V (orthos stable 24hours): idle: 36W, load: 63W
1.0ghz @ 0.88V, 31W, idle (under CCPUID control)
1.8ghz @ 0.976V, 55W, load (under CCPUID control)
The Enhance PSU adds 2W to the numbers. The BE-2350 operating at the same Vcore and frequency reduces the numbers by 3W-4W. A generic ATX PSU adds 25W to the figures.
Using the Gigabyte GA-MA69GM-S2H board:
2.1ghz @ 1.056V (orthos stable 24hours): idle: 52W, load: 72W
1.0ghz @ 0.88V, 44W, idle (under CCPUID control)
1.8ghz @ 0.976V, 65W, load (under CCPUID control)
So the Gigabyte board consumes a full 8W-16W more power when compared to the Biostar!! My results show that not all the 690G boards are created equal.
It's clear that if you are building a cool running quiet system you'd want to go with the Biostar.
Here are other comparisons between the two boards:
* The Biostar board would not POST with the Enhance PSU. I'm not sure which to blame but I suspect both: The Biostar may require a brief, high power draw at power-on, and the Enhance's surge protection circuitry is not tolerant enough to handle that. In contrast the picoPSU has no problem whatsoever powering up this board, nor does the Enhance PSU have problem powering up the Gigabyte.
* The NB heatsink on Gigabyte runs HOT, HOT, HOT, even in idle! I suspect that's where the extra 8-16W of power is expended.
* The Gigabtye's heatsink mount is held together with plastic pins, whereas the Biostar's held together by metal screws. This is important if you are using an after market screwed on HSF like the Hyper HFC-20820-C1, a must if you are using a slim case like the Silverstone LC19 or ML02. The Biostar has the right backplate with the screw holes, whereas the Gigabyte does not.
* The Biostar has full undervoltage and underclock controls in BIOS, whereas the Gigabyte only has only underclock control. CrystalCPUID works well with both, though.
EDIT: * The Gigabyte board's standby indicator only works for S1+S3, but not for S3 only which presumably most of us reading SPCR use. So with the Gigabyte board you can't tell the difference between S3 and power off states.
Last edited by frank2003 on Sat Aug 11, 2007 6:25 am, edited 1 time in total.
Frank2003, this is just enormously useful, thanks. 2 followups.
1. Some of us on another thread have been speculatin' about whether the Gigabyte 690G boards will properly mount aftermarket NB coolers like the Thermalright HR-05. From your observation, would either or both of these boards have good mounts for a cooler like that?
2. Can you say a word more about the nature of the load you used in measurements, particularly in video-processing terms?
1. Some of us on another thread have been speculatin' about whether the Gigabyte 690G boards will properly mount aftermarket NB coolers like the Thermalright HR-05. From your observation, would either or both of these boards have good mounts for a cooler like that?
2. Can you say a word more about the nature of the load you used in measurements, particularly in video-processing terms?
On the gigabyte 690v and thus I suspect the 690g you can control the IGP mhz. On then 690v they overclock it by default, you can loose a few watts but matching to the default. Can the Biostar do that same ?
I found the 690v used more power than my asus 690g board initially. Once I changed the PSU to a 80% efficient Antec they dropped. The Asus uses about 43watts fully loaded using bios to undervolt. This was a drop of about 18watts. The Gigabyte dropped by 25 watts but it was starting at a higher watts. So now its using 39 to 38 which is pretty good. It seems strange that it used so much to start with with a standard Antec 350w PSU. I'm now running Antec Earthwatts 380 which is 80%+ efficient. Standby has dropped from about 7w each to 2 watts
I prefer the asus as you can underclock in bios but RMClock works great on the gigabyte. The gigabyte has the newer AMD Core bios features to change IGP clockspeed and the front side bus controls so you can very easily overclock. These figures were true for brisbane core Athlon 64 3600+ and 4000+
The Giga and Asus northbridge heatsinks run hot as per posting in this forum. Some people have replaced them but I've not done anything about it yet.
I found the 690v used more power than my asus 690g board initially. Once I changed the PSU to a 80% efficient Antec they dropped. The Asus uses about 43watts fully loaded using bios to undervolt. This was a drop of about 18watts. The Gigabyte dropped by 25 watts but it was starting at a higher watts. So now its using 39 to 38 which is pretty good. It seems strange that it used so much to start with with a standard Antec 350w PSU. I'm now running Antec Earthwatts 380 which is 80%+ efficient. Standby has dropped from about 7w each to 2 watts
I prefer the asus as you can underclock in bios but RMClock works great on the gigabyte. The gigabyte has the newer AMD Core bios features to change IGP clockspeed and the front side bus controls so you can very easily overclock. These figures were true for brisbane core Athlon 64 3600+ and 4000+
The Giga and Asus northbridge heatsinks run hot as per posting in this forum. Some people have replaced them but I've not done anything about it yet.
Holy cow !, and to think I thought seriously about the Gigabyte for a moment.So the Gigabyte board consumes a full 8W-16W more power when compared to the Biostar!! My results show that not all the 690G boards are created equal.
Thank you very much for this !!!
Btw, Frank, why did you go with the 80W brick for the picoPSU, and not with the 120W brick ?
Last thing:
You got 31W idle (1Ghz) with 750GB desktop HD and 2 ram sticks ?, this means that with a laptop hard drive and 1 ram stick, you could have gotten about 15W @ idle. I think this would have set the record for an Intel/AMD system.Components used:
* Biostar TA690G-AM2 (latest 605 BIOS)
* Gigabyte GA-MA69GM-S2H (latest F3a BIOS)
* AM2 Brisbane 4000+, 2.1ghz
* AM2 BE-2350, 2.1ghz
* 2 x 1GB DDR2 memory
* Seagate 750GB 7200rpm SATA drive
* Silverstone ML02 with built-in Enhance END 0512 power module; 120W brick
* 120W picoPSU with 80W brick
* generic ATX PSU
Base configuration:
* Biostar TA690G-AM2
* AM2 4000+
* picoPSU
Power consumption:
2.1ghz @ 1.056V (orthos stable 24hours): idle: 36W, load: 63W
1.0ghz @ 0.88V, 31W, idle (under CCPUID control)
1.8ghz @ 0.976V, 55W, load (under CCPUID control)
Judging from the picture of the HR-05, I really doubt there's room for it on both boards. On the Biostar, the stock NB heatsink is already too close to the CPU HSF; On the Gigabyte, it's too close to the PCI-e. The good news is that on the Biostar you won't need an aftermarket NB heatsink as the stock HS is only slight warm to the touch, even after playing 3 hours of 1080i HDTV recordings.colin2 wrote:Frank2003, this is just enormously useful, thanks. 2 followups.
1. Some of us on another thread have been speculatin' about whether the Gigabyte 690G boards will properly mount aftermarket NB coolers like the Thermalright HR-05. From your observation, would either or both of these boards have good mounts for a cooler like that?
The numbers I reported were obtained by running Orthos Blend test that stressed both the CPU and memory. I recorded the steady-state numbers after the initial transient jumps in power consumption.2. Can you say a word more about the nature of the load you used in measurements, particularly in video-processing terms?
In terms of practical video processing power consumptions: On the Biostar (4400+, picoPSU), I configured CCPUID for .88V @ 1ghz, 1.056 @ 1.8ghz. When playing 1920x1080p HD WMD movie trailers on a 1080p HDTV over HDMI, I'm seeing 51W of power usage at the wall, with average CPU utilization around 43% with clock running at 1.8ghz.
I just ran another test with 100% load at 1.8ghz. This time I was running orthos in the background while playing back 1080p WMV HD in the foreground. The power consumption recorded was 59-60W.
Both my Biostar and Gigabyte 690G boards default to 400Mhz for IGX clock.NX3 wrote:On the gigabyte 690v and thus I suspect the 690g you can control the IGP mhz. On then 690v they overclock it by default, you can loose a few watts but matching to the default. Can the Biostar do that same ?
The fact that Biostar's NB can run cool makes one wonder why other boards (ex asus and gigabyte) run so hot. Maybe they were made from different reference designs?The Giga and Asus northbridge heatsinks run hot as per posting in this forum. Some people have replaced them but I've not done anything about it yet.
Thanks! I noticed the BE-2350 didn't show up in Biostar's list of supported CPUs but it sounds like it ran fine for you. Do you have any advice on CPU selection? Oh and while I'm pestering you, what kind of memory did you use and does it matter?
It sounds like I'm trying to achieve the same thing you are: the lowest power draw and least noise possible within the constraints of a 2 megapixel monitor and video usefulness.
My only real worry with Biostar is sorting through all the driver and bios updates and whatnot -- if you compare their newegg feedback to Gigabyte they seem a little rougher in that department. If you have a minute, could you give a neophyte a quick roadmap to what you need to download from where in what order to get this board going?
It sounds like I'm trying to achieve the same thing you are: the lowest power draw and least noise possible within the constraints of a 2 megapixel monitor and video usefulness.
My only real worry with Biostar is sorting through all the driver and bios updates and whatnot -- if you compare their newegg feedback to Gigabyte they seem a little rougher in that department. If you have a minute, could you give a neophyte a quick roadmap to what you need to download from where in what order to get this board going?
If you search in Biostar's home page you'll find the 516 BIOS (released on May 16, 2007) and see that it supports the new BE-2350. But the latest BIOS is 605 (released on June 05, 2007). It can be found here: http://www.biostar.cn/bios2006/AM2/R69AM605.BST.RAR
I originally bought the 2.1ghz 4000+ before the release of BE-2350. But based on my measurments you can achieve similar results if you use the same vcore and frequency (the BE-2350 is a few watts lower). Other people also have good experience with other 65W parts (3600+, 3800). Which processor you need really depends on what you will be doing with your system. For playback of 1080i HDTV recordings, I had determined I really only needed 1.6ghz. But for many 1080p movie trailers, I needed to run at 1.8ghz in order to achieve smooth playback.
In addition to creating a quiet HTPC, I had another constraint which turned out not to be trivial to meet - I wanted my HTPC to fit inside a slim desktop case like the Silverstone ML02. My experience with this case is another story which I will share in the future.
Finally, the DDR2 memory I use are 2 sticks of G. Skill F2-6400CL4D-2GBPK. I had no particular reason for selecting them other than the fact they were on sale.
I originally bought the 2.1ghz 4000+ before the release of BE-2350. But based on my measurments you can achieve similar results if you use the same vcore and frequency (the BE-2350 is a few watts lower). Other people also have good experience with other 65W parts (3600+, 3800). Which processor you need really depends on what you will be doing with your system. For playback of 1080i HDTV recordings, I had determined I really only needed 1.6ghz. But for many 1080p movie trailers, I needed to run at 1.8ghz in order to achieve smooth playback.
In addition to creating a quiet HTPC, I had another constraint which turned out not to be trivial to meet - I wanted my HTPC to fit inside a slim desktop case like the Silverstone ML02. My experience with this case is another story which I will share in the future.
Finally, the DDR2 memory I use are 2 sticks of G. Skill F2-6400CL4D-2GBPK. I had no particular reason for selecting them other than the fact they were on sale.
I clocked the gigabyte IGP to 200 but it doesn't run cooler. It saves a about 2-3 watts though. It seems odd the Biostar runs cool. The biostar northbridge chip looks a good size, bigger than the Gigabyte. The asus is tall but not that big overall foot print. I doubt the size though would be enough to explain why it runs cool. It would be interesting if you could get some temps for the northbridge heatsink and I'll try to get some for the giga and asus.
39W is the total system consumption or just mobo+cpu?NX3 wrote:Once I changed the PSU to a 80% efficient Antec they dropped. The Asus uses about 43watts fully loaded using bios to undervolt. This was a drop of about 18watts. The Gigabyte dropped by 25 watts but it was starting at a higher watts. So now its using 39 to 38 which is pretty good. It seems strange that it used so much to start with with a standard Antec 350w PSU. I'm now running Antec Earthwatts 380 which is 80%+ efficient. Standby has dropped from about 7w each to 2 watts
What else is on your system if 39 is total?
Frank, you said,
2. Silentpcreview review of the picoPSU said that the 120W brick's fan only kicked in at 90W and above, not something you'll ever reach with your system. Still, having a higher wattage brick gives you that peace of mind that whatever you'll do, your system won't need more power than the brick can deliver. For $10 more, it seems like with the 120W brick you are loosing nothing in comparisson to the 80W brick, only gaining.
1. So what is the advantage of going with the 120W picoPSU with a 80W brick ?, why didn't you go with the 80W picoPSU ?I shied away from the 120W brick after I read that it contains an internal fan that could make a lot of noise when it kicks in.
2. Silentpcreview review of the picoPSU said that the 120W brick's fan only kicked in at 90W and above, not something you'll ever reach with your system. Still, having a higher wattage brick gives you that peace of mind that whatever you'll do, your system won't need more power than the brick can deliver. For $10 more, it seems like with the 120W brick you are loosing nothing in comparisson to the 80W brick, only gaining.
From my research using a standard ATX PSU, I knew I was able to get by with ~75W of power at the wall. So I knew with a more efficient PSU like the picoPSU, 80W brick would be plenty powerful.jones_r wrote: 1. So what is the advantage of going with the 120W picoPSU with a 80W brick ?, why didn't you go with the 80W picoPSU ?
However, 80W max at the wall does not mean all the 80W power will go to the rails that require it. So to play it safe, I went with a 120W which would allow the 12V rail (max 7Amps) to draw say, 75W of power and other rails to draw 1W of power on all other rails (these are made up numbers to illustrate a point). An 80W picoPSU would not have enough juice on the 12V (3 Amps only) rail to power this hypothetical PC.
I must have missed this point. I'll keep that in mind on my next picoPSU order. Thanks!2. Silentpcreview review of the picoPSU said that the 120W brick's fan only kicked in at 90W and above, not something you'll ever reach with your system. Still, having a higher wattage brick gives you that peace of mind that whatever you'll do, your system won't need more power than the brick can deliver. For $10 more, it seems like with the 120W brick you are loosing nothing in comparisson to the 80W brick, only gaining.
The Gigabyte comes with the Easy Tune 5 utility that shows the system (presumably NB) temperature. It pretty much stays at 49C during idle (compared to 20-30's for CPU). This is with a fan flowing on it, too!NX3 wrote:I clocked the gigabyte IGP to 200 but it doesn't run cooler. It saves a about 2-3 watts though. It seems odd the Biostar runs cool. The biostar northbridge chip looks a good size, bigger than the Gigabyte. The asus is tall but not that big overall foot print. I doubt the size though would be enough to explain why it runs cool. It would be interesting if you could get some temps for the northbridge heatsink and I'll try to get some for the giga and asus.
The Biostar comes with a fan/CPU monitor that shows only CPU temperature but not chipset temperature.
When will SpeedFan support 690G chipset?
It's nice to know that BE-2350 indeed draws less power than normal 4000+, even in undervoting situation. I've the desire to get one 45W BE to compare my 35W 3800+ now.
Imo, for double the price of the brisbane, the BE-2350 simply ain't worth it.The BE-2350 operating at the same Vcore and frequency reduces the numbers by 3W-4W
An explenation for the delta between the Biostar and Gigabyte could a difference in the number of on board features. Would it be possible to take measurments after disabling some in bios, creating a match between the 2 boards? It would be interesting to see what impact they have on power consumption.
Already done that - I turned off pretty much all the features in BIOS but it did not do much to the power consumption or the heat on the NB.ntavlas wrote:An explenation for the delta between the Biostar and Gigabyte could a difference in the number of on board features. Would it be possible to take measurments after disabling some in bios, creating a match between the 2 boards? It would be interesting to see what impact they have on power consumption.
I would be anxious to hear more about the ML02 as that is certainly an interesting little HTPC case. In particular I would be interested in what type of cooling solution you might have used since the case is marketed as a MoDT case and only 82mm high per the specs and there are very few true low profile coolers out there. Would be nice if someone were to begin to sell a low profile cooler for the new 45W AMD CPUs going forward. The biggest drawback seems to be the lack of expansion slots. Silverstone was demoing the ML02 with an unreleased DTX motherboard at one the trade shows if I remember. OK, back on topic......frank2003 wrote: In addition to creating a quiet HTPC, I had another constraint which turned out not to be trivial to meet - I wanted my HTPC to fit inside a slim desktop case like the Silverstone ML02. My experience with this case is another story which I will share in the future.
I'll get one when my friend upgrade his computer. Borrow it from friends, I'll lose nothing and gain the fun. Lowest power draw topic interests me a lot. What a pity, I can't get Biostar boards here, and Gigabyte popularized around my area. Hard decisions for my friend.jones_r wrote: Imo, for double the price of the brisbane, the BE-2350 simply ain't worth it.
-
- Posts: 524
- Joined: Sun Oct 22, 2006 7:39 pm
- Location: Denver, Colorado USA
- Contact:
-
- Posts: 98
- Joined: Fri Jan 26, 2007 1:59 pm
- Location: Battlefield, MO
-
- Posts: 524
- Joined: Sun Oct 22, 2006 7:39 pm
- Location: Denver, Colorado USA
- Contact:
38w @ desktop idle via CnQ and with CnQ off 45wpapakoo wrote:39W is the total system consumption or just mobo+cpu?
What else is on your system if 39 is total?
Seasonic S12 380w, 2 x 512mb 800mhz memory, AMD 64 x2 4000, Gigabyte 690v mobo, Western Digital 160gb SATA hard drive, stock AMD HSF, Pioneer DVD writer.
GnatGoSplat, if you read the reviews of the gigabyte board they make reference to the new features and more or less tie up. Only the vcore adjustments in bios aren't available but thats gigabytes choice to make them available. I did ask them about it and they replied to say they didn't include them as this board isn't aimed at that type of user / market.
Using a Antec Earthwatt 380 the same setup 38 again but top end is 46/47w.
Both operate at 2w on standby.
Feature wise thee 690g boards are more or less the same. If you look at the boards the Asus, Gigabyte they have quite a large number of capacitors. The biostar below it seems to have 3 by the 4 pin 12v connector and the rest solid caps ? Are solid caps much more efficient ? The Asus and gigabyte has some solid caps but not as many.
http://blog46.fc2.com/k/kemx2pcanime/file/ta690g1.jpg
-
- Posts: 98
- Joined: Fri Jan 26, 2007 1:59 pm
- Location: Battlefield, MO
Wow, Gigabyte is fast. I already updated to the new BIOS last week, but I didn't know they updated anything other than what was stated, "Improve HT bus overclocking". The review says, "The new BIOS features also gave us the ability to overclock the board's HyperTransport link to over 1.5GHz, which was an increase of approximately 50% over "stock" which is very good for an inexpensive chipset with an IGP."
I wonder how they pulled that off? I'll have to give that a shot, previously even overclocking the HT link to 1040MHz was enough to make my system spontaneously reset.
I wonder how they pulled that off? I'll have to give that a shot, previously even overclocking the HT link to 1040MHz was enough to make my system spontaneously reset.