Bartons will undervolt and underclock quiet nicly.
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
Bartons will undervolt and underclock quiet nicly.
I have my 2500+ barton runn at 1.552 to 1.536 volts or so my usdm says.
I specified 1.525 but it overcolts. still it's running at 10x200 with 1.552. It's not a 1700+ but it's not too bad if you ask me. Maybe i can even overclock it further on this voltage i'll get back to you guys on that.
I specified 1.525 but it overcolts. still it's running at 10x200 with 1.552. It's not a 1700+ but it's not too bad if you ask me. Maybe i can even overclock it further on this voltage i'll get back to you guys on that.
-
- SPCR Reviewer
- Posts: 8636
- Joined: Sat Nov 23, 2002 6:33 am
- Location: Sunny SoCal
I'm at 1.3v and its rock solid. It would occassionally crash at 1.2v but I reckon its was OK at 1.25v, its just that I like the margin of safety the extra 0.05v gives.I undervolted my bartons to 1.35 volt (!) I think they'll go even lower.
I used to run my 1.6v Thoroughbred B at 1.2v, it was rock solid too.
AOpen AK79G Max. Undervolting and SilentBIOS are great, otherwise its at best an average board.What boards are you guys using?
Cheers!
undervolting does not neccesarilly mean underclocking; there is a bit of headroom, sometimes a lot. I was able to get an extra 350 mhz overclock out of stock voltage; I could have undervolted pretty welll at stock speeds.
Also, when we start underclocking, they usually don't go as low as we'd like; 1 ghz Athlon XP will have very low power consumption but they don't make any slower than 1533 mhz. Performance will be much better than with a Via too.
Also, when we start underclocking, they usually don't go as low as we'd like; 1 ghz Athlon XP will have very low power consumption but they don't make any slower than 1533 mhz. Performance will be much better than with a Via too.
Mine runs at stock speed but with a big power consumption reduction.What's the advantage of underclocking a processor over buying a processor with less power consumption to begin with? I am just curious, not trying to make a point or anything.
If you mean as compared to a mobile processor its much cheaper, if you mean compared to a low power comsumption processor like an Epia its much more computationally powerful. It also means I get by with a £7 Coolermaster HSF rather than a £50 Thermalright - and I'm cheap by nature so that suits me fine!
I'd love to know what the power consumption difference for a barton at 1.8 ghz versus 1.5 and undervolted (like, actual objective numbers).
Might make a really interesting article if someone had some processors, undervoltable MB's, and bench-marking tools. If you could get a barton down to around < 30 watts, it would be a good candidate for passive cooling with a high quality heat sink.
...If I'm not mistaken, the power consumption decreases linearly with the clock rate? What I don't know, however, is if this can only decrease so far before you can't run the processor with less power....combined with a voltage drop and a reduction in speed, I wonder what the lowest possible operating power consumption of a barton is. If you could get it less than 10 watts and still be running faster than 500 mhz (don't know if that's possible, or just a pipedream), screw Via--I'm all over that. Heh.
Any takers?
Might make a really interesting article if someone had some processors, undervoltable MB's, and bench-marking tools. If you could get a barton down to around < 30 watts, it would be a good candidate for passive cooling with a high quality heat sink.
...If I'm not mistaken, the power consumption decreases linearly with the clock rate? What I don't know, however, is if this can only decrease so far before you can't run the processor with less power....combined with a voltage drop and a reduction in speed, I wonder what the lowest possible operating power consumption of a barton is. If you could get it less than 10 watts and still be running faster than 500 mhz (don't know if that's possible, or just a pipedream), screw Via--I'm all over that. Heh.
Any takers?
-
- Patron of SPCR
- Posts: 946
- Joined: Sat Sep 06, 2003 9:57 pm
- Location: Berkeley, CA, USA
- Contact:
Definitely possible-- as long as you are willing to underclock as well as undervolt. At some point you are limited by the multiplier options, but you can always decrease the FSB to compensate. As always, the question is, what is the minimum level of performance you are willing to sacrifice for noise?
I think you should be able to calculate heat dissipation per clockrate mathematically based on the existing figures, shouldn't you?
I think you should be able to calculate heat dissipation per clockrate mathematically based on the existing figures, shouldn't you?
we've already got an article on undervolting/underclocking with mathematically calculated figures.
Tested numbers would be much harder to get.
http://www.silentpcreview.com/modules.p ... 102&page=1
Tested numbers would be much harder to get.
http://www.silentpcreview.com/modules.p ... 102&page=1
Ask and ye shall recieve:
http://forums.silentpcreview.com/viewtopic.php?t=7019
A Barton 2500 (@1833Mhz) defaults to a Vcore of 1.65v and puts out a max of 68.3 watts.
Underclocked to 1500Mhz and UV'd to 1.25v you're down to a max of 32 watts
You'd hit 10 watts at 600Mhz and 1.1 volts. So not quite as efficient as a C3, but in the same ballpark as a P-M
http://forums.silentpcreview.com/viewtopic.php?t=7019
A Barton 2500 (@1833Mhz) defaults to a Vcore of 1.65v and puts out a max of 68.3 watts.
Underclocked to 1500Mhz and UV'd to 1.25v you're down to a max of 32 watts
You'd hit 10 watts at 600Mhz and 1.1 volts. So not quite as efficient as a C3, but in the same ballpark as a P-M
There is no magic threshold.
Just like there is no such thing as passive cooling. It all depends on the airflow. Even without a fan the heat produces its own convective flow. But that flow is so low and variable that its impossible to predict really, it all depends on the individual environment.
If your case has good ventilation 10watts is pretty safe for passive. But if its a sealed box 10W will cook, eventually.
If only I hadn't killed all those brain cells in college.....
Just like there is no such thing as passive cooling. It all depends on the airflow. Even without a fan the heat produces its own convective flow. But that flow is so low and variable that its impossible to predict really, it all depends on the individual environment.
If your case has good ventilation 10watts is pretty safe for passive. But if its a sealed box 10W will cook, eventually.
If only I hadn't killed all those brain cells in college.....
-
- Patron of SPCR
- Posts: 946
- Joined: Sat Sep 06, 2003 9:57 pm
- Location: Berkeley, CA, USA
- Contact:
Right, even my slowest EPIA box (533mhz?) needs a small fan. Completely passive cooling is so hard to achieve that it's almost not worth it. You can get very, very close-- really, what difference is one nearly inaudible fan between friends? It's a psychological hurdle more than anything else.
Totally passive is like the holy grail-- so close, yet so far away. I do have high hopes for these new "giant heatsink" case designs from Zalman, though.
Totally passive is like the holy grail-- so close, yet so far away. I do have high hopes for these new "giant heatsink" case designs from Zalman, though.
Rusty075 wrote:But if its a sealed box 10W will cook, eventually.
No, it won't neccessarily cook. It depends upon the heat transfer of the case to the outside environment--granted, when the case is closed this transfer is very small, but you have to take into account the temperature of the room the case is sitting in, and the ability of the case material (for example, a case built of wood and sealed would "cook" much faster than a sealed case of aluminum) to conduct heat away...
You're just making "air" the medium through which the heat transfer occurs, which is a hidious means of transfering heat--but still one way of doing it. It's no different than using a heat pipe or water or any of the other means people use to transfer heat--just less efficient.
Anyways, seems there's already lots of info on power dissipation versus clock speed and core voltage.....heh. Anyone have info on unlocking bartons?
-
- SPCR Reviewer
- Posts: 8636
- Joined: Sat Nov 23, 2002 6:33 am
- Location: Sunny SoCal
I'm currently running a TBred XP 1800+ underclocked to 133x5 = 666Mhz, vcore = 1.1v. Should be around 11w max dissipation. Running passive with a TT Volcano 7 heatsink, only fan in the system is from the PSU (Nexus NX-3500 that I'm reviewing, it's a Fortron FSP350-60PN (PF) painted in black, they didn't even remove the FSP logo from the fan controler).
50°c under CPUBurn. A Barton 2500+ underclocked and undervolted to the same level should perform the same (it's around 11w too). This setup won't run passive with no airflow, I tried it outside of the case and the temperature would rise and rise and rise under CPUBurn, I had to stop the test.
This test was part of an experiment I did in order to compare an undervolted XP and a Nehemiah 1Ghz. The Nehemiah will run passive too in the same case and with the same airflow, but it gets toasty under CPUBurn. As Zentar said, an undervolted XP has a better performance/thermal dissipation ratio, and will be a better choice if you don't care for the microATX form factor : a microATX board that can undervolt to 1.1v is hard to find. And the Nehemiah is also cheaper.
50°c under CPUBurn. A Barton 2500+ underclocked and undervolted to the same level should perform the same (it's around 11w too). This setup won't run passive with no airflow, I tried it outside of the case and the temperature would rise and rise and rise under CPUBurn, I had to stop the test.
This test was part of an experiment I did in order to compare an undervolted XP and a Nehemiah 1Ghz. The Nehemiah will run passive too in the same case and with the same airflow, but it gets toasty under CPUBurn. As Zentar said, an undervolted XP has a better performance/thermal dissipation ratio, and will be a better choice if you don't care for the microATX form factor : a microATX board that can undervolt to 1.1v is hard to find. And the Nehemiah is also cheaper.
-
- Patron of SPCR
- Posts: 946
- Joined: Sat Sep 06, 2003 9:57 pm
- Location: Berkeley, CA, USA
- Contact:
I'm currently running a TBred XP 1800+ underclocked to 133x5 = 666Mhz, vcore = 1.1v. Should be around 11w max dissipation. Running passive with a TT Volcano 7 heatsink
Wow, that's pretty nifty. And 50c is actually extremely cool for cpuburn with no fan other than the PSU!50°c under CPUBurn.
Of course, that's more way performance than I am willing to sacrifice in the name of (semi) passive cooling, but it's a neat experiment nonetheless..
Rusty075 wrote:True Beyonder, I was thinking of sealed in a calorimeter sort of way.
And all Bartons are unlock automatically.
Damn. I'm up on the theory and down with the fact.
Doh!
So.......does anyone know any good motherboards for undervolting the Barton? Methinks it's about time for GOLDFISHY to get some bigger guns, since she's about to become a HTPC in addition to web/file server....I was thinking Barton 2400 undervolted/clocked to about 1000 mhz, with Seasonic Tornado 300w, m-audio audiophile ( [shameless plug] need the sound quality for my "dynaco st-35 based tube amp/Bottlehead Straight-8" rig, w00t!! Real amps glow in the dark.... [/shameless plug] ), and some ati card with no fan.....maybe a tele tuner. I dunno.
I guess I just want a really basic MB for undervolting/underclocking, preferably cheap and stable. Don't really care all that much about overall performance....