780's with Samsung/Hynix memory AND the B1 stepping?

[Ralfies]

I searched and couldn't find any good info about this. I'm going to be upgrading soon, and if the non-reference R9 290/290X's turn out to be poor overclockers I will almost certainly go with a 780.

My first question is which 780's are known to have the B1 stepping, and does it really make a difference? I'm hearing mixed things about whether or not it has an impact on overclockability/heat/power.

My second question is which 780's are known to have Samsung/Hynix memory? The 780 Classified was my first choice for a 780 until I found out they shipped with Elpida memory. I'm not going pay a >$50 premium for a card built for overclocking with memory that doesn't OC.

My third, and most important question is are there any 780's that have BOTH?

On a side note, I was thinking of a 780Ti if the 290's don't pan out. I have the money for it, and wouldn't mind paying ~$750 IF the card had more ram. If I spend that kind of dough on a card it would have to last me 2-3+ years. When I got my GTX 670 a year and a half ago everybody was saying how 2GB Vram was enough even for 1440p. It didn't take long for me to regret not getting a 7970 or a 4GB 670. I'm sure 3GB of ram will be plenty in the mean time, but it didn't take long for 2GB to go from plenty to not enough.

 

[radrok]

I'd still get the 780 classified, doesn't matter if your memory is a poor clocker, you still have up to 1.5v on the core which will make up for the lack of memory headroom.

Some elpida chips have reported very good overclocks, also.

Well, to be honest you can get 1.5v+ on reference too, but to run those voltages on a reference design you'll need an EK block with very low water temps to keep your VRM from frying

 

[LAN_deRf_HA]

Is 1.5v ok on B1? Cause it sure isn't on the old chips. All I've seen is very rapid degradation and dead chips at those volts.

 

[radrok]

?

Could you please give some data on that?

I'm running 1.35v-1.4v on my A1 for gaming and it's all fine and dandy.

Don't see why B1 would be ruined by voltage, provided temperatures are in check.

 

[LAN_deRf_HA]

Over in the OCN threads right after they figured out how to unlock those volts on the Titan and 780 people reported severe degradation and dead chips. If you've been running those volts for a prolonged period I'm a bit surprised there's been no problem. Maybe some chips are more fragile than others.

 

[radrok]

The only thing I noticed is an increased coil whine, proportional to voltage increase.

At 1.5v the whine is just unbearable

Anyway with clocks above 1400 MHz it all becomes an artifact fest so gaming wise I'd say 1300-1350 MHz is the sweet spot.

Maybe those you've seen were running high core temperatures on air, I can believe that people are that stupid.

Never gone over 42C on the core here, even at 1.5v, got a massive loop that doesn't go over 4c delta even with 1500W heatload


Thank you, though

 

[LAN_deRf_HA]

I tried to find the posts I had remembered about people saying they damaged stuff, but couldn't find a single post I remembered so idk what threads I was reading. Haven't looked into it in awhile. This post was the only recent thing I saw and he was indeed on air cooling. http://www.overclock.net/t/1421221/...ard-vdroop-fix-solid-1-325v/330#post_21029216 Seems people are saying no one has had an issue on water, though I swear someone said 1.5v killed their Titan but I can't find the post so I'm starting to feel like I'm imagining things haha.

 

[radrok]

I can honestly believe that, there aren't many phases on reference GK110 board so with 1.5v the circuitry has to push an insane amount of amps through itself.

Lots of voltage + heat = dead cards.

That dude pushed almost 1.3v on air, way to go lol....

I remember before installing waterblocks the cards ran at a very high temperature with just 1.2v, something like 85c... If the core was 85c then the VRM were probably 10-15c more.

I'm pretty sure that dude ran his VRM at over 100c for a prolonged amount of time, doesn't take much to draw conclusions...

Anyway, people have been benching cards at 1.6v with waterblocks on reference design and it's all fine (not that I would ever recommend using that kind of voltage 24/7).

Overvolting takes just a bit of brain and common sense, that's why Nvidia probably locks down these cards.

 

[the54thvoid]

I was an early Titan adopter and i've read literally every post the Titan Owners forum at OCN. There are actually very few Titans with reported failure (technically none from OCN people). The recommendations are that air cooling shouldn't above 1.212v and water should safely stick to 1.3v for regular use (Radrok take note!). People at OCN generally only use 1.35v+ for benchmarks and 1.4v+ for suicide benchmark runs.
And the coil whine does kick in as soon as you start to over volt.

And Radrok mate, you need to bench Firestrike at close to 1400mhz. It's your duty!

 

[radrok]

I can do better, 1550 mhz at 1.5v