yosarianilives

Figured I'd get a score in while binning

I take back what I said about ath II zosma being crap. Seems only core 6 sucks on these chips.

Thanks to cbjaust for showing me the ideal batch and reduction size in his sub. Need to run this chip on ln2 again, this time with the correct background ;)

Up to 1.5-1.55v shouldn't damage it but you may lose scaling.

Unfortunately this is all too common for igp. People just select the first igp that comes up cause they're too lazy to select the right one. I've spent way too much time attempting to clean up some of the categories, and in the end it's only more work for the mods as I can only report subs.

Eh my cvf forgot how to mem. Was having issues running higher vdimm or running good mem settings for the past month then this last weekend it started refusing to post with any mem. It seems that boards just wear out after a while. Got many good SS sessions out of my cvf before that though, and my understanding is that sabertooth had done very well in a few ln2 sessions before it became a dice only board.

On single stage it's set and forget so I caught up on quite a few shows back when I took gold for this chip earlier in the year.

There is no love, only pure hatred. On the plus side it saved me some money as I have zero desire to own one now. I'll stick to the m7i, a much better board.

LN2 at -50 to -60c on sandy for a 45 minute run ? I could see it helping mem tho, so theoretically could be useful for timings. Did you use hypers? I had no luck with them on sandy myself. I couldn't set as tight of tertiary timings as on PSC so my score ended up taking a hit despite tighter primaries.

Also are you running hypers?

Nice! Do you have any advice for max bclk? I'll admit I'm kinda clueless and just use mostly auto voltages then put gpu in last slot on a board with plx chip.

Is this the chip I sent you? :D

Yes, wasted a lot of LN2 troubleshooting last weekend it before we determined that it was a board issue.

My experience with UD4P has been that it's a god tier board for am2 cpus (min did 420+ bclk ez on aio) but on AM3 cpus it isn't so good (struggled with 300 bclk on ss, my cvf is much better than that)

Just disable hwinfo, it won't fill in anything.

Perhaps sabertooth boards just have issues? This last weekend on sabertooth we were having issue across multiple cpus, am3 and am3+, where it had a CBB at -80c. Basically you could boot into bios, pull down, then boot into OS. However if it failed to run the settings or you needed to reboot from the OS you would need to pull up to around -80c to get it to POST at all.

Yeah seems that zosma Athlons are the binning rejects out of the the binning rejects. I wonder how much cache influences the weakness of the cores as normally there is no l3 cache active on the 640s however it gets unlocked when I unlock the cores up to 6. I still have a few more on the way and hopefully there is a decent on otherwise our third score for that stage may be on a quad.

Don't feel so bad. My binning of 840t says that only about 50% of zosma have 6 working cores (this is after 20+ chips), also 100% of my athlon II 640's were utter garbage. The 2 out of 3 that unlocked to 6 only clocked to 3.6 ghz at 1.65v on ss. Binning results here

D

Perhaps he realized he needed rig pic after tearing down and didn't want to risk the die with a quick and dirty mount?

Just found this article here, is it possible that there's more to unlocking multiplier than most people know? Seems like in this link it was all cpus on a certain mobo with a certain bios. If there's a way to reliably unlock multiplier on locked AM3 chips could mean a lot for things like 1 core records.

Interesting, I've always run my hypers at 2v on am3 and they seemed to like it. Was able to do 1760 6-6-5-22 1t that way with hypers on air, cpu on ss. Although lately my cvf has been unhappy with high vdimm so perhaps there's something to lower vdimm.

Similarly can we fix the unlocked cores feature to be correct to the cpu used? For example I can sub a ath II 651k as unlocked to 6 cores even though this is impossible as there are physically only 4 cores on the llano die.

Deleted and reuploaded outside of comp, good catch. Tbh this was just a testing score, expected to get more out of the chip before I broke my OS ?

Kinda hard to tell since I didn't even bench this chip properly. ie; no in OS OC, no pscheck, etc but probably would at least be a 7+ 32m and legacy chip in 7700k terms