- Joined
- Nov 10, 2003
- Location
- Canberra, Down Under
My aging generic PSU was finally starting to give me trouble (Hyena 500W, ~2 years old), particularily with cold-booting. Probably a capacitor issue or something. Anyhow, I decided "stuff it" and went out and bought an Antec TruePower 550 EPS12V plus a 24-pin to 20-pin converter cable. The results were almost comical.
The system is a dual Barton at 2GHz, 1.50V VCore, 9800SE softmodded + OC'd to 9800XT, WD800JB and a few other things (USB card, etc).
With the old PSU, the rails were within a percent or two of being right. I've posted here before about it, and it had minor variation on the rails between load and no load. Nothing out of spec.
Upon first sight, the results were a little disappointing. The 12V rail was at 11.73V (2.25% low), the 5V at 4.84 (3.20% low) and the 3.3V at 3.20V (3.03% low). This is all from the motherboard readings; I can't remember the differences between motherboard and calibrated multimeter, so I might go borrow one from the lab sometime to check.
The comical part is ... they don't change. At all. There is absolutely no change in the read values between idle and any combination of Prime95, BurnK7, 3DMark2001, ATITool, or anything. I was starting to wonder if the monitors had broken or something, until there was a 1-reading blip down to 4.81V on the 5V line, then back up to 4.84V. I got bored after watching the values for a few minutes, reset the high/low values in MBM5, and went off to dinner.
Currently, after about an hour of Prime95, the High/Average/Low values are
{} 5V = 4.84/4.84/4.81
{} 12V = 11.73/11.73/11.73
{} 3.3V = 3.20/3.20/3.18
Compared to the useless voltage regulators on the K7D board itself:
{} VCore = 1.55/1.47/1.33
{} VMem = 2.56/2.48/2.38
I've always been a fan of overkill, but I think this is a new record for me
It'd be more stessed on a dual Xeon/Opteron system, but I think it's safe to say that a dual Athlon system is a piece of cake for this PSU. It's also very quiet (fan currently running at 1400RPM apparently ... it goes slow enough to break the monitoring hardware sometimes), much quieter than the old one. Now, I've just gotta decide whether it's worth blowing the same amount again to get a PSU for the A7M266-D (which incidentally cost less than half of what I spent on the PSU) that's in storage before the safety officers get to Antec and stop them putting out 36A on a single rail
The only problem is that the "external" 4-pin connector interferes with my case. I'm going to do a bit of surgery with a hacksaw (on the case of course) which should fix it. Err, that and the fact that the -5V is at a rather out-of-spec 3.55V, but nothing seems to be broken so I assume it's not being used.
The system is a dual Barton at 2GHz, 1.50V VCore, 9800SE softmodded + OC'd to 9800XT, WD800JB and a few other things (USB card, etc).
With the old PSU, the rails were within a percent or two of being right. I've posted here before about it, and it had minor variation on the rails between load and no load. Nothing out of spec.
Upon first sight, the results were a little disappointing. The 12V rail was at 11.73V (2.25% low), the 5V at 4.84 (3.20% low) and the 3.3V at 3.20V (3.03% low). This is all from the motherboard readings; I can't remember the differences between motherboard and calibrated multimeter, so I might go borrow one from the lab sometime to check.
The comical part is ... they don't change. At all. There is absolutely no change in the read values between idle and any combination of Prime95, BurnK7, 3DMark2001, ATITool, or anything. I was starting to wonder if the monitors had broken or something, until there was a 1-reading blip down to 4.81V on the 5V line, then back up to 4.84V. I got bored after watching the values for a few minutes, reset the high/low values in MBM5, and went off to dinner.
Currently, after about an hour of Prime95, the High/Average/Low values are
{} 5V = 4.84/4.84/4.81
{} 12V = 11.73/11.73/11.73
{} 3.3V = 3.20/3.20/3.18
Compared to the useless voltage regulators on the K7D board itself:
{} VCore = 1.55/1.47/1.33
{} VMem = 2.56/2.48/2.38
I've always been a fan of overkill, but I think this is a new record for me
It'd be more stessed on a dual Xeon/Opteron system, but I think it's safe to say that a dual Athlon system is a piece of cake for this PSU. It's also very quiet (fan currently running at 1400RPM apparently ... it goes slow enough to break the monitoring hardware sometimes), much quieter than the old one. Now, I've just gotta decide whether it's worth blowing the same amount again to get a PSU for the A7M266-D (which incidentally cost less than half of what I spent on the PSU) that's in storage before the safety officers get to Antec and stop them putting out 36A on a single rail The only problem is that the "external" 4-pin connector interferes with my case. I'm going to do a bit of surgery with a hacksaw (on the case of course) which should fix it. Err, that and the fact that the -5V is at a rather out-of-spec 3.55V, but nothing seems to be broken so I assume it's not being used.
