broken 1.6A

PageFault · 2003-10-14, 22:12

Known bad box ... what to do?

So, a feature was added to our stats system ... bad tests category. At some point last fall, some doublechecks egan needing a third test. This went by me while I did about four months of 33M. Having seen the problem, some corrections were made and 10 dc's used to test have failed to match.

Can it be trusted to trial factor? I suppose I could test with low bit known factors and a few in the SSE2 range. Box might work at 1.6 (stock) but I'll never trust it again, besides that's too slow for me. Trial factoring would be a short term solution - replacement is inevitable.

battlemaxx · 2003-10-15, 21:44

Bad box...cut it loose now...

outlnder · 2003-10-15, 22:56

After my 1.6A continued to give errors after going back to stock speeds with the slowest memory settings, I sold it on Agora for $35.

PageFault · 2003-10-15, 23:15

I know of your problems, they couldn't be resolved. It is starting to look that way with this box too.

I'm usually very thorough in my research. Last year I solved a problem that baffled a mill, and the community, for forty years. I nearly pulled my hair out but in the end I persevered and found what went wrong.

The latest insanity is a drift in iteration times. I did fix the AM radio interference - that was due to the asynchronous ram / fsb. Never an error flag, never a burp in torture. vcore is stock, so it's not a case of NSDS - rather, a gradual drift in the ability to produce correct results. No other incidents were observed with this box - it runs for a long time, only going down when mains power does - zero crashes since May last year.

I have a feeling its hammer time. If I can be bothered, I could put it in a new boart. The new boart will have a 2.4C so I doubt I can be bothered.

PageFault · 2003-10-15, 23:18

Just out of interest, did that, or any other suspect cpu, ever go in a new boart? I'm loathe to think a cpu can croak but the boart is another thing - never a cheapie for me again.

I recall checking your stats and you generated many error flags - that difference bothers me - zero flags and zero matching dc's

Quote:

Originally posted by outlnder
After my 1.6A continued to give errors after going back to stock speeds with the slowest memory settings, I sold it on Agora for $35.

outlnder · 2003-10-16, 01:40

If you look at outlnder02 or outlnder52, you will see a lot of zero errors but incorrect results. These 2 are the same machine, just named differently.

As to the boart question, yes I put the 1.6 in a different boart and it still produced errors. I think somerthing on the chip got fried with the overclocking.

outlnder02 was on a cheapo ECS boart. I hate to admit it, but Xyzzy was right about the Intel chipsets. I just wish they didn't cost so much.

2003-10-14, 22:12	#1
PageFault Aug 2002 Dawn of the Dead 5×47 Posts	broken 1.6A Known bad box ... what to do? So, a feature was added to our stats system ... bad tests category. At some point last fall, some doublechecks egan needing a third test. This went by me while I did about four months of 33M. Having seen the problem, some corrections were made and 10 dc's used to test have failed to match. Can it be trusted to trial factor? I suppose I could test with low bit known factors and a few in the SSE2 range. Box might work at 1.6 (stock) but I'll never trust it again, besides that's too slow for me. Trial factoring would be a short term solution - replacement is inevitable.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
broken CPU bracket	wildrabbitt	Hardware	2	2017-05-19 20:17
N-1 Broken?	wblipp	FactorDB	4	2012-11-30 22:54
Something broken?	schickel	Forum Feedback	9	2011-08-16 04:43
84822 is broken	fivemack	Aliquot Sequences	3	2011-01-18 17:59
Prime95's backups broken?	abstractius	Software	4	2007-12-18 02:31

2003-10-15, 21:44	#2
battlemaxx Nov 2002 43 Posts	Bad box...cut it loose now...

2003-10-15, 22:56	#3
outlnder Aug 2002 13E₁₆ Posts	After my 1.6A continued to give errors after going back to stock speeds with the slowest memory settings, I sold it on Agora for $35.

2003-10-15, 23:15	#4
PageFault Aug 2002 Dawn of the Dead 11101011₂ Posts	I know of your problems, they couldn't be resolved. It is starting to look that way with this box too. I'm usually very thorough in my research. Last year I solved a problem that baffled a mill, and the community, for forty years. I nearly pulled my hair out but in the end I persevered and found what went wrong. The latest insanity is a drift in iteration times. I did fix the AM radio interference - that was due to the asynchronous ram / fsb. Never an error flag, never a burp in torture. vcore is stock, so it's not a case of NSDS - rather, a gradual drift in the ability to produce correct results. No other incidents were observed with this box - it runs for a long time, only going down when mains power does - zero crashes since May last year. I have a feeling its hammer time. If I can be bothered, I could put it in a new boart. The new boart will have a 2.4C so I doubt I can be bothered.

2003-10-16, 01:40	#6
outlnder Aug 2002 476₈ Posts	If you look at outlnder02 or outlnder52, you will see a lot of zero errors but incorrect results. These 2 are the same machine, just named differently. As to the boart question, yes I put the 1.6 in a different boart and it still produced errors. I think somerthing on the chip got fried with the overclocking. outlnder02 was on a cheapo ECS boart. I hate to admit it, but Xyzzy was right about the Intel chipsets. I just wish they didn't cost so much.