VDNKh wrote on 2022-12-11, 18:03:
hellslinger wrote on 2018-06-28, 02:16:
The first 2 processors I tried were a Pentium Dual Core E5200 (SLB9T) and a Core 2 Duo E7200 (SLAPC). Both worked the first time I turned the power on, and they would reboot without problems, but after I powered off, they would not POST again. I know this board takes a bit of time to POST, but nothing would happen after minutes so I knew something was wrong.
I have been having this same issue with my R3.0 board and Wolfdale E6800. The CPU is officially supported and I exceed ASRocks requirement with some Corsair, heat spreader, DDR400 CL2 spec'd DIMMs, as well a VGA card. I'm fairly certain it just doesn't like my memory. I checked CPU-Z after a CMOS reset and it downclocks the RAM to 177MHz and goes into single channel mode. That's why it works after a CMOS reset, it always defaults to that 2:3 ratio as a fail safe. But after a cold boot it will try to use the applied BIOS RAM settings and fail to POST. My best guess at this point anyway. At one point I thought that my actual BIOS chip was dying but re-flashing it didn't throw any errors or fix the issue. Setting the CL to 2.5 or configure by SPD do not fix the issue either.
Anyone have any better recommendations for DIMMs that are know to work with this board?
So I installed an E5800 (3.2 GHz, 800MHz FSB) and all is well now, sort of. I found that the CAS 2 latency option is a lie. CPU-Z reports CAS 2.5 no matter how it's set in BIOS, and there's no difference in performance between CAS 2 and CAS 2.5. Maybe a lower clocking CPUs can utilize CAS 2, but the E5800 can't.
I think I discovered why this board hates my RAM so much. I was looking up reviews of RAM similar to mine and I read:
Before you ask, there are no CAS 3.0 results because Winbond BH-5 memory chips do not support CAS 3.
I'd never heard that before, but that would explain the motherboard's weird behavior with this RAM. If, after a CMOS reset, it kept the timing at CAS 2.5 and downclocked the CPU to achieve stability at 1066MHz FSB, that would explain why it worked then. But after a cold start, if it tried using CAS 3 to boot up, it wouldn't work because the RAM doesn't support that timing. I don't think I have Winbond memory though. According to this thread, I have "Promos -5 Rev A" (XMS3203v6.3) which I can't find any information on. But maybe it has the same limitation? Has anyone ever heard of this?
Edit: The CAS 2 option does work. But if you're overclocking it's really not worth it. Loosening the CAS to 2.5 lets me OC the FSB to 233 MHz and maintain 1:1 RAM to FSB (stable at 5 hours with P95), while CAS 2 only lets me push it to around 210 MHz. Plus the extra clock speed trumps any performance gains on CAS 2.
I was able to push the FSB to 245 MHz and get 3.92 GHz on the E5800, successfully ran P95 for 2.5 hours on a single core in Windows 98, if I set the RAM:FSB to 4:5 and made the timings super loose. Kinda pissed I couldn't get to 250 MHz so I could get an even 400 MTs with a 4:5 RAM clock rate. I want to see how far this can be pushed if I had some DDR 500 RAM. Maybe I'll retry the E6800 if I ever get my hands on some fancy DDR 500 RAM.
Exploring the BIOS, the failsafe option is "03" which I think is [Auto] if it goes by the menu order, who knows what it does. Perhaps setting failsafe and default to option "00", which going by menu order is CAS 2.5, would solve my problem with the E6800. More investigation is needed.
Spoiler
The attachment cas.PNG is no longer available