Reply 30280 of 30283, by ubiq
tehsiggi wrote on Yesterday, 18:30:The 5V rail of your PSU is rated at 20A, which nowadays is sufficient, however that's only 100W. The 9700 will take around 20W of it's 60W from the 5V rail. That leaves you with 80W for the rest of the system. Given that socket 462 boards often rely on the 5V rail for CPU power, you might be a bit in a tight spot here under load. Not sure about your a7v8x-mx.
Awesome, thanks for bringing the numbers! Sounds like I'm right to suspect the PSU, even if I was completely off on the exact reasons. I have a X1950 Pro AGP that bluescreened but didn't trip the PSU off when I tested it in this system. I'm going to guess that means it pulls more on the 12V than the 9700, and that if (when) I build a system around it I'll probably be ok with a modern PSU.
tehsiggi wrote on Yesterday, 18:30:I'd leave the caps as is. Of all 9500s, 9700s and 9800s I had, none of them died / failed due to bad caps.
Cool - yeah, none of them look like they've wet themselves so I'm happy to leave them alone.
tehsiggi wrote on Yesterday, 18:30:I'd always grant the memory on 9700s and 9800s memory heatsinks. The memory gets quite warm and the stock cooler already has bad airflow over the memory. Given that you are cooling it passively, I'd highly recommend additional heatsinks on the memory.
Ok on it - I'll wait till I get that sorted to do any further kind of stress testing. Thanks!
Funny, last time I had a 9700 Pro (AIW), I was running it in a tiny Shuttle MiniPC with zero airflow (and a busted PSU in this pic):
Now I want to treat this one like precious cargo.