Orthos Error

Spudgun88 · 3 Jan 2007 at 09:41

Hello all,
I recieve the following error while running Orthos -

Code:

FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 3 minutes 45 seconds - 1 errors, 0 warnings.
Execution halted.

My specs are:
DS3 Mobo
6300 CPU
Sapphire X1950 Pro 256mb
2Gb of Geil 6400

I recieve the same error a lot quicker if I set the RAM to run at 2.1v, and then I tried running the test with 2.0v going through the RAM, and the test managed to run for about 3 minutes longer before the error.

Anyone got any idea what's up?

Thanks a lot,
Spudgun

Jokester · 3 Jan 2007 at 10:05

Run Memtest to make sure your RAM is stable first.

What Orthos test are you using?

Jokester

Spudgun88 · 3 Jan 2007 at 10:17

Jokester said:
Run Memtest to make sure your RAM is stable first.

What Orthos test are you using?

Jokester

Okey dokey, I'll let Memtest run for a while.
I was using the 'Blend' test.

Defcon5 · 3 Jan 2007 at 10:28

Jokester said:
Run Memtest to make sure your RAM is stable first.

I could run memtest for hours, but orthos failed in 7 seconds

blknoel · 3 Jan 2007 at 10:42

Which means the failure is not Ram related

, elimination method

Next time you have problems with OCing/stability testing, would be useful to provide some essential infomation:

- Vcore and Temp and Clocks
- Vdimm
- VMCH (northbridge voltage)

Stability is based on at least those factors, if you say you have a 6300 with DS3 setup and running 2.1v vdimm, how are we to know how high CPU clocks or temps are? Unless you don't consider them crucial to stability

.

mmj_uk · 3 Jan 2007 at 11:21

blknoel said:
Which means the failure is not Ram related , elimination method

whilst memtest is a good indication of faults in memory, i find that it is no where near as sensitive/stressful to memory as prime95/orthos is.

it could be chipset overheating or memory timings related, if you're overclocking the first thing you should do is set everything to stock and just take it slowly, run the computer for 1-2 days at a time at any given overclock and just monitor for any strange crashes/errors as you do your day to day stuff, if you get a crash/error then back off.

james32 · 3 Jan 2007 at 17:40

blknoel said:
Which means the failure is not Ram related , elimination method .

wrong i had a matched pair of ram that could pass memtest all day long
but blend failed in under 5 mins

i checked and indeed my ram had 2 diffrent part numbers
with cpuz so that meant i had 2 diffrent trc numbers

replaced ram and blend will pass all day long

1st always use orthos stress test for 8hrs if it passes, your cpu is good
2nd then do orthos blend 4hrs max.

if it fails blend then start to look at Ram

Bomag · 3 Jan 2007 at 19:20

james32 said:
wrong i had a matched pair of ram that could pass memtest all day long
but blend failed in under 5 mins

i checked and indeed my ram had 2 diffrent part numbers
with cpuz so that meant i had 2 diffrent trc numbers

replaced ram and blend will pass all day long

1st always use orthos stress test for 8hrs if it passes, your cpu is good
2nd then do orthos blend 4hrs max.

if it fails blend then start to look at Ram

If they had different TRCs then they were not a matched pair

The problem was that the memory controller could not stably run the memory at the chosen settings rather than dodgy mem. The OPs problem will probably be the same, raising VMCH and FSB voltages should help.