Mad_Max: That's only true if your power is free or heavily subsidized. At typical rates a GPUless desktop in the US will use between $150 and $300/year in electricity. In western Europe twice that.
I am running 24/7 a SUN M2 Workstation with an Opteron 1210 CPU at 1.8 GHz, no GPU. My latest 3 months bill was 95 euro all appliances included, no AC, and heating and hot water provided by methane.
Tullio
On a Global Search unit my two core Opteron 1210 running Linux obtained the same timings of a 16 core Opteron 8356 running Windows. How come?
Tullio
I assume, you are compairing, Runtimes per Core, when you take a look
at the amounts of FLOPS (or Million Ops per second, Whetstone), you won't see
much of a difference, either.
Some calculations benefit quite a lot from a large L2 or L3 Cache or AMD's (now INTEL, too), QPI (Quick Path Interface?) INTEL's QPI. QPI Wikipedia.
The Opteron 8356 is a Quad Core but BOINC sees it as a 16 processor CPU. Its floating point speed (each core, I believe) is double that of mine. Einstein apps run on only one core on mine CPU, while AQUA, which multithreading, takes all two. Does the OS make the difference?
Tulio
Mad_Max: That's only true if your power is free or heavily subsidized. At typical rates a GPUless desktop in the US will use between $150 and $300/year in electricity. In western Europe twice that.
Yes, the cost of electricity is also desirable to consider if you plan to use the machines 24 / 7 and even if multiple machines simultaneously (even in Russia - energy prices here still relatively low, but growing at a faster rate of ~20% per year).
But where does the comparison Intel / AMD? If you mean that AMDs CPU will consume more energy, then you are wrong - about the same for the latest generations of chips (core 2 duo and later Athlon II / Phenom II and later). In addition, motherboards based on ATI's chipsets (the best choice for AMDs CPUs in desktop segment) consume less than similar Intel motherboards.
For example, my system (MSI 785GM + Athlon X2 250 + 2Ñ…1Gb DDR2-800 RAM + HDD Hitachi 1 Tb, with built-in MB video and off monitor) consumes less than 90 watts (including losses in the PSU) under full load, at the same time can produce up to 2200 RAC at work 24/7 on GW tasks.
Truth is not a very good result, more "green" will be installed Athlon II X4 on the same motherboard - consumption will increase somewhere up to 110-120 watts, and RAC up to ~ 4000@24/7 or lower-end models of the 6-cores Phenom line (i plan do it later, and use current MB+CPU+RAM to upgrade old second desktop witch run Athlon XP 2600 atm).
Most effective with respect to energy will be 12-cores Opterons - such as the above in original post - it just 80 watts with 12 cores!, but the processor itself is too expensive, that would nullify the entire economy.
So that the most economical crunch in AMD camp is Athlon II X3/X4 and Phenom II X6
P.S.
My main project is the Rosetta @ Home, so do not pay attention to my current low RAC - I connect 2 desktop machines to the E @ H only when the Rosetta have problems and cant supply work. During the rest time is only minor crunch on intel laptop.
Mark: Those really aren't the systems you should be looking at in a bang for buckish comparison. More enlightening would be the fastest i7-quad core is ranked 30 with 9.7k RAC. The #2 i7 quad is currently my 930 @ 4.1ghz - 8.9k RAC.
The lack of AMD systems in the chart is rather painfully obvious. The 2nd ranked AMD box is a 2xquad core at rank 111 6.6k RAC, the #3 AMD is also a 2xquad core. The 4th highest AMD system is a 6core CPU and is roughly in the same price bucket as an i7 quad. It's ranked at 194 and only has a 5.4k RAC. As the fastest system in its class I assume it's also heavily overclocked and that a direct comparison with the fastest i7 quads is reasonable. What it shows is amd 6 cores being only slightly more than half as fast as i7-quads.
AMD's entry level quads might be more competitive because they fall into a price range where intel's offerings are weak, but for E@H higher performing systems are a no-contest win for Intel.
As a side note, I'll point out that in general benchmarks AMD's chips are much closer so there might be some projects that favor them over Intel.
Using the Mk 1 Eyeball is appears performance is a function of the number of cores and the clock speed just as one would expect. That Intel is so highly represented appears to indicate nothing more than mother boards with multiple Xeon chips dominate the server market.
In general interrupt driven OSs like linux require less overhead than polling OSs like Microsoft. While this does matter (5-7% vice <1%) when boinc is a background task among other such as day to day usage it is not clear it will make much difference if it is running only boinc.
Interrupt driven responds faster to random and prioritized tasks such as the average user generates. This gives the user a more responsive machine.
But when the tasks are always the same without those random requirements to reallocate RAM, deal with swap space, and the rest when a new task is called there is not many extra overhead tasks to deal with.
It's comparing the power consumption of two computers to one. Unless you're comparing a high speed top of the line CPU to a much lower low power one you're not going to get a 2:1 delta in power consumption. And if you are doing that your performance delta becomes more like 3:1.
Hmm. Looks like Intel-fann wants to start a holy-war without arguments ...
I will not continue, because I do not like to participate in such things. Unsubscribe from this topic also.
RE: Mad_Max: That's only
)
I am running 24/7 a SUN M2 Workstation with an Opteron 1210 CPU at 1.8 GHz, no GPU. My latest 3 months bill was 95 euro all appliances included, no AC, and heating and hot water provided by methane.
Tullio
I never actually checked to
)
I never actually checked to see how much these add to my light bill and I have at least 3 machines running 24/7 and for 10 years (October 30th)
But then I always have the monitors turned off unless I am on one.
You can look at my pc's since they aren't hidden.
I started running a cheap ($400) AMD Phenom 9150e Quad-Core Processor a little over a year ago and so far so good.
A couple years ago a Pentium D CPU 2.80GHz which I don't check all that often but it is just ok.
But the last few months I added 3 more machines at a nice price ($1,500 total with free shipping)
I do watch those 3 all day and the fastest of the 3 is the AMD Athlon II X4 630
It will do 4 Global's every 5:45 and Arecibo's in about 2.5hrs
Now the funny thing is the other 2 are exactly the same.
AMD Phenom II X3 720 but one does 3 Globals every 6:12 ave.
And the one right next to it does 3 Globals at a 6hr ave. with many less than 6hrs and the other never does less than 6hrs
And it zips through the Arecibo's doing over 25 per day.
My RAC is definitely higher than I have ever had in all these years.
Like most of us I have a couple in my parts stash that I could bring back to life if I get the time.
On a Global Search unit my
)
On a Global Search unit my two core Opteron 1210 running Linux obtained the same timings of a 16 core Opteron 8356 running Windows. How come?
Tullio
RE: On a Global Search unit
)
I assume, you are compairing, Runtimes per Core, when you take a look
at the amounts of FLOPS (or Million Ops per second, Whetstone), you won't see
much of a difference, either.
Some calculations benefit quite a lot from a large L2 or L3 Cache or AMD's (now INTEL, too), QPI (Quick Path Interface?)
INTEL's QPI.
QPI Wikipedia.
The Opteron 8356 is a Quad
)
The Opteron 8356 is a Quad Core but BOINC sees it as a 16 processor CPU. Its floating point speed (each core, I believe) is double that of mine. Einstein apps run on only one core on mine CPU, while AQUA, which multithreading, takes all two. Does the OS make the difference?
Tulio
RE: Mad_Max: That's only
)
Yes, the cost of electricity is also desirable to consider if you plan to use the machines 24 / 7 and even if multiple machines simultaneously (even in Russia - energy prices here still relatively low, but growing at a faster rate of ~20% per year).
But where does the comparison Intel / AMD? If you mean that AMDs CPU will consume more energy, then you are wrong - about the same for the latest generations of chips (core 2 duo and later Athlon II / Phenom II and later). In addition, motherboards based on ATI's chipsets (the best choice for AMDs CPUs in desktop segment) consume less than similar Intel motherboards.
For example, my system (MSI 785GM + Athlon X2 250 + 2Ñ…1Gb DDR2-800 RAM + HDD Hitachi 1 Tb, with built-in MB video and off monitor) consumes less than 90 watts (including losses in the PSU) under full load, at the same time can produce up to 2200 RAC at work 24/7 on GW tasks.
Truth is not a very good result, more "green" will be installed Athlon II X4 on the same motherboard - consumption will increase somewhere up to 110-120 watts, and RAC up to ~ 4000@24/7 or lower-end models of the 6-cores Phenom line (i plan do it later, and use current MB+CPU+RAM to upgrade old second desktop witch run Athlon XP 2600 atm).
Most effective with respect to energy will be 12-cores Opterons - such as the above in original post - it just 80 watts with 12 cores!, but the processor itself is too expensive, that would nullify the entire economy.
So that the most economical crunch in AMD camp is Athlon II X3/X4 and Phenom II X6
P.S.
My main project is the Rosetta @ Home, so do not pay attention to my current low RAC - I connect 2 desktop machines to the E @ H only when the Rosetta have problems and cant supply work. During the rest time is only minor crunch on intel laptop.
RE: Mark: Those really
)
Using the Mk 1 Eyeball is appears performance is a function of the number of cores and the clock speed just as one would expect. That Intel is so highly represented appears to indicate nothing more than mother boards with multiple Xeon chips dominate the server market.
RE: ... Does the OS make
)
In general interrupt driven OSs like linux require less overhead than polling OSs like Microsoft. While this does matter (5-7% vice <1%) when boinc is a background task among other such as day to day usage it is not clear it will make much difference if it is running only boinc.
Interrupt driven responds faster to random and prioritized tasks such as the average user generates. This gives the user a more responsive machine.
But when the tasks are always the same without those random requirements to reallocate RAM, deal with swap space, and the rest when a new task is called there is not many extra overhead tasks to deal with.
RE: But where does the
)
It's comparing the power consumption of two computers to one. Unless you're comparing a high speed top of the line CPU to a much lower low power one you're not going to get a 2:1 delta in power consumption. And if you are doing that your performance delta becomes more like 3:1.
Hmm. Looks like Intel-fann
)
Hmm. Looks like Intel-fann wants to start a holy-war without arguments ...
I will not continue, because I do not like to participate in such things. Unsubscribe from this topic also.