Simviation Forums

Well these SD3700 chips certainly clock well:



Stock voltage and stock cooling. And stable

Very pleased

Jon

Well these SD3700 chips certainly clock well:



Stock voltage and stock cooling. And stable

Very pleased

Jon

Yep... used that processor for almost a year with the same 25% overclock

That's why I raved on about the San Diego until I got sick of my own voice. Excellent CPU.

Ever wonder why we are always suggesting the 3700+ San Deago?

Cheers
Cameron

Ever wonder why we are always suggesting the 3700+ San Deago?

Now I know

Jon

I got mine right up there too (Congo was very patient with me (thanks again)).. It was cool and stable enough for everything.. Ran FS9 great.. EXCEPT.. Day of Defeat would crash until I slowed it down just a tad.

I think I could have worked with the memory a little (or got better memory).. but the difference in speed wasn't that much..

I'll have to check, but I think it ended up being a 20% O/C.. Very noticeable in FS9  

[quote]I got mine right up there too (Congo was very patient with me (thanks again)).. It was cool and stable enough for everything.. Ran FS9 got better memory).. but the difference in speed wasn't that much..

I'll have to check, but I think it ended up being a 20% O/C.. Very noticeable in FS9

It does rather depend on the stepping. Over on the OCUK forums some people have been hitting 3Ghz on air with the newest chips.

However, some of them can only manage 2.6Ghz @ 1.5v, so it really is luck of the draw.

Jon

Stability tests don't always work to guarantee stability at a particular clockspeed either. I'm beginning to believe that some gaming apps may have bad code that trips the CPU into a failure where it would normally test stable.

I get failures in IL2FB and NFS Most Wanted at high clockspeeds and intense game periods with maximum action, yet torture testing shows the CPU is stable. I'm also not ruling out memory as a possible culprit either, I just haven't got the knowledge to be absolutely certain of the cause.

In any event, Backing off the clock slightly usually cures the problem, though Need For Speed - Most Wanted requires that I go back to 2.6ghz or I get a crash after a fairly long session.

There is what is termed "errata" built into every CPU, these are documented design errors which have been identified. It's possible that all errata is not documented and therefore software developers may risk putting "hits" on those errata which could cause a crash.

Another explanation is just sloppy programming or perhaps a missed error when the program was compiled.

I'm not qualified in these areas, I'm simply explaining things the way I understand them.

Whether or not clockspeed could be a factor in these "software" crashes is something I'm not certain of, but I do suspect it's a possibility.