- Joined
- Nov 30, 2004
Antec EarthWatts 380W
Vs.
Before we get started a special thanks goes to mcoleg who’s thread got this all started.
Purpose: To see if an EA 380W is up to the challenge of powering a single GPU gaming rig.
Specs:
e6600 @ 3.1Ghz - 1.4125v in bios
Sapphire x1900 XT 512MB – stock
2 x 1GB DDR2 @ 688 – 2.1v
4 HD + an optical
4 Case fans
X-fi
*Misgivings: While the EA 380 is a Seasonic built supply lets not pretend that it’s S12 quality. The EA380 is not the equal of an Etatis, Zippy, Win-tact or even an S12 quality Seasonic unit but with that said it should be more than adequate for the proposed job. I wouldn’t be comfortable running one at its max rated load (supplying 380W to a system) 24/7/365 but since neither my rig nor mcolegs draws that much it isn’t something we have to worry about. To be honest I’m more concerned about Oklahoma Wolf mistaking me for a deer and trying to eat me than I am worried for the safety of rig.
Testing methodology: For each test the voltage on the 12v rail is measured with a DMM and the power draw from the wall is recorded using a kill-a-watt meter. The peak values for each test are displayed in the results.
Test Points:
1. Windows Idle - … goes the CPU usage graph ; )
2. Full CPU load – Intel TAT, 100% both cores
3. CPU + GPU load – Intel TAT + 3dmark 03 (nature test)
4. Real world game test – Oblivion
5. Maximum power consumption – Intel TAT + 3dmark + all 4 hard drives reading & writing + optical read.
Results:
To estimate DC power usage I referenced the SPCR review of the EA430 and the jG review of the EA500. The highest observed efficiency was 83.5% so to be safe I used that value across the board (rather than try to plot points on the efficiency curve.) I'd rather overestimate than under estimate how much juice my system is using.
Analysis:
Looking at the voltage readings we see a drop of .13v when going from idle to the maximum achieved load. Not the best I’ve ever seen for sure but still solidly within ATX spec and acceptable by my standards (not quite as good as the Enermax it replaced.)
In terms of maximum power consumption it comes as no surprise that test 5 was the most demanding. What is surprising though was the very small delta between test 5 and the next most demanding test, test #3. To try and confirm these results I started with my system at idle and had all 4 hard drives read and write + the optical read at the same time which resulted in a 15W increase in wall draw. I believe this “discrepancy” in the increase in draw can be explained by the CPU cycles that are grabbed away from TAT are less taxing. To support this theory I loaded the CPU with Orthos and saw the system drew ~15 less watts from the wall when compared to TAT draw. Even if we aren’t losing many CPU cycles to the file copies, these operations have to be less “taxing” than Orthos so the result is a smaller increase in power usage than initially expected.
Another interesting point of test 5 is observed voltage. Despite being the greatest draw in watts the PSU is doing a slightly better job regulating the 12V rail in this situation. Being a group regulated design I would wager the increased load on the 5v rail from the hard drives and optical produced this result.
Overall I am very pleased with how my EA380 has performed in my main rig. Its been well over a week since I put it in and I’ve experienced no problems. Just incase anyone was wondering, this system has been stressed with TAT + looped 3dmark just about every night since it was installed so it has proven it can handle (at least in the short term) prolonged stress without a hitch. Only thing left to determine is long term reliability but given that at max I was only able to reach ~73% of this PSU's rated load I have no qualms about leaving this PSU in my main rig long term (at least until a GPU upgrade.) Its comforting to know that one can go to Circuit City and pick up a EA 380W for $50 and be able to power this kind of rig with it.
One last note - The power usage for my system was slightly less than I expected (290-300W range was my guess) so its nice to know that I over estimated my draw, rather than under.
To wrap things up how about some random equipment shots:
Kill a watt meter in action during test 3
DMM in action during test 3
Vs.
Before we get started a special thanks goes to mcoleg who’s thread got this all started.
Purpose: To see if an EA 380W is up to the challenge of powering a single GPU gaming rig.
Specs:
e6600 @ 3.1Ghz - 1.4125v in bios
Sapphire x1900 XT 512MB – stock
2 x 1GB DDR2 @ 688 – 2.1v
4 HD + an optical
4 Case fans
X-fi
*Misgivings: While the EA 380 is a Seasonic built supply lets not pretend that it’s S12 quality. The EA380 is not the equal of an Etatis, Zippy, Win-tact or even an S12 quality Seasonic unit but with that said it should be more than adequate for the proposed job. I wouldn’t be comfortable running one at its max rated load (supplying 380W to a system) 24/7/365 but since neither my rig nor mcolegs draws that much it isn’t something we have to worry about. To be honest I’m more concerned about Oklahoma Wolf mistaking me for a deer and trying to eat me than I am worried for the safety of rig.
Testing methodology: For each test the voltage on the 12v rail is measured with a DMM and the power draw from the wall is recorded using a kill-a-watt meter. The peak values for each test are displayed in the results.
Test Points:
1. Windows Idle - … goes the CPU usage graph ; )
2. Full CPU load – Intel TAT, 100% both cores
3. CPU + GPU load – Intel TAT + 3dmark 03 (nature test)
4. Real world game test – Oblivion
5. Maximum power consumption – Intel TAT + 3dmark + all 4 hard drives reading & writing + optical read.
Results:
To estimate DC power usage I referenced the SPCR review of the EA430 and the jG review of the EA500. The highest observed efficiency was 83.5% so to be safe I used that value across the board (rather than try to plot points on the efficiency curve.) I'd rather overestimate than under estimate how much juice my system is using.
Analysis:
Looking at the voltage readings we see a drop of .13v when going from idle to the maximum achieved load. Not the best I’ve ever seen for sure but still solidly within ATX spec and acceptable by my standards (not quite as good as the Enermax it replaced.)
In terms of maximum power consumption it comes as no surprise that test 5 was the most demanding. What is surprising though was the very small delta between test 5 and the next most demanding test, test #3. To try and confirm these results I started with my system at idle and had all 4 hard drives read and write + the optical read at the same time which resulted in a 15W increase in wall draw. I believe this “discrepancy” in the increase in draw can be explained by the CPU cycles that are grabbed away from TAT are less taxing. To support this theory I loaded the CPU with Orthos and saw the system drew ~15 less watts from the wall when compared to TAT draw. Even if we aren’t losing many CPU cycles to the file copies, these operations have to be less “taxing” than Orthos so the result is a smaller increase in power usage than initially expected.
Another interesting point of test 5 is observed voltage. Despite being the greatest draw in watts the PSU is doing a slightly better job regulating the 12V rail in this situation. Being a group regulated design I would wager the increased load on the 5v rail from the hard drives and optical produced this result.
Overall I am very pleased with how my EA380 has performed in my main rig. Its been well over a week since I put it in and I’ve experienced no problems. Just incase anyone was wondering, this system has been stressed with TAT + looped 3dmark just about every night since it was installed so it has proven it can handle (at least in the short term) prolonged stress without a hitch. Only thing left to determine is long term reliability but given that at max I was only able to reach ~73% of this PSU's rated load I have no qualms about leaving this PSU in my main rig long term (at least until a GPU upgrade.) Its comforting to know that one can go to Circuit City and pick up a EA 380W for $50 and be able to power this kind of rig with it.
One last note - The power usage for my system was slightly less than I expected (290-300W range was my guess) so its nice to know that I over estimated my draw, rather than under.
To wrap things up how about some random equipment shots:
Kill a watt meter in action during test 3
DMM in action during test 3