Re: [gentoo-user] Re: new machine : CPU : AMD FX-4100 ?
On Mon, Jul 30, 2012 at 2:16 AM, Nikos Chantziaras rea...@gmail.com wrote: On 30/07/12 07:28, Michael Mol wrote: On Sun, Jul 29, 2012 at 11:55 PM, Nikos Chantziaras rea...@gmail.com wrote: On 30/07/12 06:08, Michael Mol wrote: On Sun, Jul 29, 2012 at 10:50 PM, Nikos Chantziaras rea...@gmail.com wrote: On 30/07/12 05:23, Philip Webb wrote: i5-2550K FX-4100 both use 95 W (some of the more costly AMDs use 125 W ). Note that power savings are not important if you're not using a laptop. CPU power savings on a desktop don't translate to any relevant amount of money on your electricity bills. This is because neither of those CPUs really use 95W. That's just the thermal upper limit. To be fair, power savings are relevant if you're concerned about your electric bill, or if you're concerned about heat management in your system. Consider my dual E5345...leaving that on 24x7 appears to cost me about 90USD/mo. CPU power savings will transform that into a 89.9USD/mo ;-) That's what I mean. It's not worth much. It helps quite a bit with laptop battery life. But for desktops, it doesn't do anything too useful. If you really want the hard numbers, check out some place like Tom's Hardware or Phoronix. I forget which does the power consumption measurements. At some of the hardware review blogs, you can get numbers on idle vs full-load power consumption, as measured at the wall. The difference truly is striking. When you have full load, the CPU won't clock down. So nothing saved there. When you're considering full load, the TDP becomes a useful estimation of relative power consumption between different processors. If you don't have full load, the clock-down doesn't save much compared to max clocks while idle. This is where you're wrong. http://www.tomshardware.com/reviews/ivy-bridge-benchmark-core-i7-3770k,3181-23.html http://www.tomshardware.com/reviews/fx-power-consumption-efficiency,3060-11.html -- :wq
Re: [gentoo-user] Re: new machine : CPU : AMD FX-4100 ?
Am Montag, 30. Juli 2012, 10:08:24 schrieb Michael Mol: On Mon, Jul 30, 2012 at 2:16 AM, Nikos Chantziaras rea...@gmail.com wrote: On 30/07/12 07:28, Michael Mol wrote: On Sun, Jul 29, 2012 at 11:55 PM, Nikos Chantziaras rea...@gmail.com wrote: On 30/07/12 06:08, Michael Mol wrote: On Sun, Jul 29, 2012 at 10:50 PM, Nikos Chantziaras rea...@gmail.com wrote: On 30/07/12 05:23, Philip Webb wrote: i5-2550K FX-4100 both use 95 W (some of the more costly AMDs use 125 W ). Note that power savings are not important if you're not using a laptop. CPU power savings on a desktop don't translate to any relevant amount of money on your electricity bills. This is because neither of those CPUs really use 95W. That's just the thermal upper limit. To be fair, power savings are relevant if you're concerned about your electric bill, or if you're concerned about heat management in your system. Consider my dual E5345...leaving that on 24x7 appears to cost me about 90USD/mo. CPU power savings will transform that into a 89.9USD/mo ;-) That's what I mean. It's not worth much. It helps quite a bit with laptop battery life. But for desktops, it doesn't do anything too useful. If you really want the hard numbers, check out some place like Tom's Hardware or Phoronix. I forget which does the power consumption measurements. At some of the hardware review blogs, you can get numbers on idle vs full-load power consumption, as measured at the wall. The difference truly is striking. When you have full load, the CPU won't clock down. So nothing saved there. When you're considering full load, the TDP becomes a useful estimation of relative power consumption between different processors. If you don't have full load, the clock-down doesn't save much compared to max clocks while idle. This is where you're wrong. http://www.tomshardware.com/reviews/ivy-bridge-benchmark-core-i7-3770k,3181- 23.html http://www.tomshardware.com/reviews/fx-power-consumption-efficiency,3060-11. html I wouldn't trust anything Tom's publishes. That said, Intel's 'TDP' is not really a 'TDP' - for almost a decade Intel's 'TDP' is not the 'real' TDP but a 'usually you won't get higher than this' - until you run some really heavy stuff. Like compiling openoffice... AMD followed suit some time ago. So both numbers are misleading at best. That said, idlelow load consumption is fine with all CPU's. Mobos and PSUs influence that numbers a lot more. -- #163933
Re: [gentoo-user] Re: new machine : CPU : AMD FX-4100 ?
On Mon, Jul 30, 2012 at 12:36 PM, James wirel...@tampabay.rr.com wrote: When you have full load, the CPU won't clock down. So nothing saved there. When you're considering full load, the TDP becomes a useful estimation of relative power consumption between different processors. If you don't have full load, the clock-down doesn't save much compared to max clocks while idle. This is where you're wrong. OK. both sides are well stated. Amp meters are less than $50 USD. They clamp around the power cord, or any wires inside the computer you can fit the clamp around. So make your choice, based on actual measurements? That's how an EE would make a decision on how a given processor or software setup actually effects the power consumption. REAL DATA. Note, some of the fancier meter's have an integrator function where measurements are taken frequently over a time period to get an even more realistic picture of power consumption... http://www.allaboutcircuits.com/vol_6/chpt_2/4.html I've got a Kill-o-Watt which I use for these kinds of purposes. The problem with obtaining first-hand data for a purchasing decision is getting the parts to perform the tests on. -- :wq
Re: [gentoo-user] Re: new machine : CPU : AMD FX-4100 ?
On Mon, Jul 30, 2012 at 9:36 AM, James wirel...@tampabay.rr.com wrote: SNIP Amp meters are less than $50 USD. They clamp around the power cord, or any wires inside the computer you can fit the clamp around. SNIP hth, James I haven't read this thread but I do use one of these which costs less than $20: http://www.amazon.com/P3-International-P4400-Electricity-Monitor/dp/B9MDBU/ref=pd_sim_hi_1 Personally I think CPU power consumption is a red herring without including the power consumed by the rest of the box: MB power? Hard drive power? Hard disk power GPU power? DRAM power? The 5 above can easily become the dominant power hogs. I use an Intel i7 980X 6-core hyper-threaded CPU, so that's 12 CPUs in top, which burns _lots_ of power, but I suspect it's not the biggest power consumer when compared to the total of the 6 500GB 7200 RPM hard drives I have in the box. WRT to money spent to run a machine I hope someone stated earlier than this that it's the whole system that matters and not just the CPU. Cheers, Mark
Re: [gentoo-user] Re: new machine : CPU : AMD FX-4100 ?
On Mon, Jul 30, 2012 at 1:04 PM, Mark Knecht markkne...@gmail.com wrote: On Mon, Jul 30, 2012 at 9:36 AM, James wirel...@tampabay.rr.com wrote: SNIP Amp meters are less than $50 USD. They clamp around the power cord, or any wires inside the computer you can fit the clamp around. SNIP hth, James I haven't read this thread but I do use one of these which costs less than $20: http://www.amazon.com/P3-International-P4400-Electricity-Monitor/dp/B9MDBU/ref=pd_sim_hi_1 Personally I think CPU power consumption is a red herring without including the power consumed by the rest of the box: MB power? Hard drive power? Hard disk power GPU power? DRAM power? The 5 above can easily become the dominant power hogs. I use an Intel i7 980X 6-core hyper-threaded CPU, so that's 12 CPUs in top, which burns _lots_ of power, but I suspect it's not the biggest power consumer when compared to the total of the 6 500GB 7200 RPM hard drives I have in the box. Spinning disks consume surprisingly little power once they're up to speed. My GPU, by comparison, doesn't seem to reduce heat generation very much when relatively idle. WRT to money spent to run a machine I hope someone stated earlier than this that it's the whole system that matters and not just the CPU. I didn't state so explicitly, no, but I believe I mentioned the two machines had been otherwise comparable in their equipment loadout. If I missed that, my bad. -- :wq
Re: [gentoo-user] Re: new machine : CPU : AMD FX-4100 ?
On Mon, Jul 30, 2012 at 10:14 AM, Michael Mol mike...@gmail.com wrote: On Mon, Jul 30, 2012 at 1:04 PM, Mark Knecht markkne...@gmail.com wrote: SNIP MB power? Hard drive power? Hard disk power GPU power? DRAM power? The 5 above can easily become the dominant power hogs. I use an Intel i7 980X 6-core hyper-threaded CPU, so that's 12 CPUs in top, which burns _lots_ of power, but I suspect it's not the biggest power consumer when compared to the total of the 6 500GB 7200 RPM hard drives I have in the box. Spinning disks consume surprisingly little power once they're up to speed. My GPU, by comparison, doesn't seem to reduce heat generation very much when relatively idle. OK, point taken about the disks at least when talking about a single disk. I see we measure the same way with a Kill-a-Watt so that's at least consistent. And I think we're in more or less violent agreement, but reducing the CPU power in the end won't save all that much on the electric bill, or so I think. I ran around really quickly to find some spec values for the machine I'm on right now. No idea if the numbers are right. They are just what I found quickly: CPU - i7-980x - 130W vs Asus Rampage II Extreme - Can't find so far, so let's guess 25W for the chipset, NICS, audio devices, etc. WD RAID Edition 500GB - 8W * 6 = 48W NVidia GTX 465 class card = 80W 24GB DRAM - Total guess, but about 1W/DIMM looks safe, so 6W total So a __really__ rough guess is my machine should use about (130 + 25 + 48 + 80 + 6) Watts, or 289W assuming I added it up correctly. That's under full load though. My UPS has a power meter in it. The UPS is driving this machine, 3 monitors, a small switch, a wireless access point and maybe one or two other small things I've forgotten or have plugged in somewhere. When the machine is essentially idle that mater reads 330W. When I start a VM that uses 6 processor cores and runs for 30 minutes at full tilt the power consumption is 385W. I no longer remember what I have set up in terms of the CPU clocking stuff. It's on the machine and let's the box go full tilt, but I'm not sure what it does when the machine is idle. Anyway, if you assume that the 55W jump was the difference between the 980x idling, and then using 3 cores full tilt, then 6 cores (12 threads) might be more like a 100W jump which seems about right according to Intel's spec. In terms of the electric bill, don't forget the PC power supply is only 80-90% efficient, so 10-20% is thrown away there also. Now, assume you get a CPU that draws half the power. This setup would still likely draw something close to 330W when it's idling, and might only jump up by 60W when running full tilt. That would save maybe (330+60)/(330+100) or only about 10% on the whole system power consumption. For that reason I don't think skimping on the CPU makes much sense to me. I'll happily turn the box off 2 hours a day vs go slower all the time, but that's just me. WRT to money spent to run a machine I hope someone stated earlier than this that it's the whole system that matters and not just the CPU. I didn't state so explicitly, no, but I believe I mentioned the two machines had been otherwise comparable in their equipment loadout. If I missed that, my bad. And I don't know that you did as I haven't read the thread, but part of my argument is that you have to know the WHOLE system and not just the CPU to decide if changing the CPU costs or saves much power. Cheers, Mark
Re: [gentoo-user] Re: new machine : CPU : AMD FX-4100 ?
Am 30.07.2012 19:14, schrieb Michael Mol: On Mon, Jul 30, 2012 at 1:04 PM, Mark Knecht markkne...@gmail.com wrote: On Mon, Jul 30, 2012 at 9:36 AM, James wirel...@tampabay.rr.com wrote: SNIP Amp meters are less than $50 USD. They clamp around the power cord, or any wires inside the computer you can fit the clamp around. SNIP hth, James I haven't read this thread but I do use one of these which costs less than $20: http://www.amazon.com/P3-International-P4400-Electricity-Monitor/dp/B9MDBU/ref=pd_sim_hi_1 Personally I think CPU power consumption is a red herring without including the power consumed by the rest of the box: MB power? Hard drive power? Hard disk power GPU power? DRAM power? The 5 above can easily become the dominant power hogs. I use an Intel i7 980X 6-core hyper-threaded CPU, so that's 12 CPUs in top, which burns _lots_ of power, but I suspect it's not the biggest power consumer when compared to the total of the 6 500GB 7200 RPM hard drives I have in the box. Spinning disks consume surprisingly little power once they're up to speed. My GPU, by comparison, doesn't seem to reduce heat generation very much when relatively idle. When I built my NAS box I did meassure the power consumption of my box, first with one HD, then two, three and so one. And I figured that one of my (Samsung) HDs uses about 5 Watts when running idle.
Re: [gentoo-user] Re: new machine : CPU : AMD FX-4100 ?
On Mon, Jul 30, 2012 at 11:12 AM, Michael Hampicke gentoo-u...@hadt.biz wrote: SNIP Spinning disks consume surprisingly little power once they're up to speed. My GPU, by comparison, doesn't seem to reduce heat generation very much when relatively idle. Idle on a GPU (in Linux) might be more when the screen is black. I don't even know how to drive my GPU hard. It's just not part of my life here. Maybe certain games or something? When I built my NAS box I did meassure the power consumption of my box, first with one HD, then two, three and so one. And I figured that one of my (Samsung) HDs uses about 5 Watts when running idle. That's very consistent with the WD numbers I found. Something like 6W idle, 8W max, etc. for my RAID Edition drives which are not WD Green drives which likely have lower numbers. Point is that there's little power saved in a box with 6 drives going from max use to idle, etc. But I know you know that! :-) Anyway, we're getting similar results is what it sounds like to me. - Mark
Re: [gentoo-user] Re: new machine : CPU : AMD FX-4100 ?
On Sun, Jul 29, 2012 at 10:50 PM, Nikos Chantziaras rea...@gmail.com wrote: On 30/07/12 05:23, Philip Webb wrote: i5-2550K FX-4100 both use 95 W (some of the more costly AMDs use 125 W ). Note that power savings are not important if you're not using a laptop. CPU power savings on a desktop don't translate to any relevant amount of money on your electricity bills. This is because neither of those CPUs really use 95W. That's just the thermal upper limit. To be fair, power savings are relevant if you're concerned about your electric bill, or if you're concerned about heat management in your system. Consider my dual E5345...leaving that on 24x7 appears to cost me about 90USD/mo. -- :wq
Re: [gentoo-user] Re: new machine : CPU : AMD FX-4100 ?
On Sun, Jul 29, 2012 at 11:55 PM, Nikos Chantziaras rea...@gmail.com wrote: On 30/07/12 06:08, Michael Mol wrote: On Sun, Jul 29, 2012 at 10:50 PM, Nikos Chantziaras rea...@gmail.com wrote: On 30/07/12 05:23, Philip Webb wrote: i5-2550K FX-4100 both use 95 W (some of the more costly AMDs use 125 W ). Note that power savings are not important if you're not using a laptop. CPU power savings on a desktop don't translate to any relevant amount of money on your electricity bills. This is because neither of those CPUs really use 95W. That's just the thermal upper limit. To be fair, power savings are relevant if you're concerned about your electric bill, or if you're concerned about heat management in your system. Consider my dual E5345...leaving that on 24x7 appears to cost me about 90USD/mo. CPU power savings will transform that into a 89.9USD/mo ;-) That's what I mean. It's not worth much. It helps quite a bit with laptop battery life. But for desktops, it doesn't do anything too useful. If you really want the hard numbers, check out some place like Tom's Hardware or Phoronix. I forget which does the power consumption measurements. At some of the hardware review blogs, you can get numbers on idle vs full-load power consumption, as measured at the wall. The difference truly is striking. Now, at least part of the problem with my E5345 setup is that I'm running two high-performance Xeon processors that only have operational clock speeds: 2.33 GHz and 2.00GHz. Desktop-targeted CPUs often will clock down to just a hair over 1GHz, if not a hair under, if you have proper power management daemons running. -- :wq
