The Bulldozer week/batch thread...

[pershoot]

I always use the "Ultra High" setting for LLC on the CPU. It results in a slight overshoot under load ~0.015v which actually helps clocks (but it does add a little heat)...

Try setting v's ~1.4375 and LLC to Ultra and see where that puts you under load.

with it usually on 1.455 - 1.468, using ultra high:
http://droidbasement.com/amd_zambezi/4.6_2.4ghz.jpg

core temps ~3c away (less) from rog cpu temp.

during building src code stress (what i primarily use pc for) in a VM, i see peak to ~1.475, low 40c, so it 'should' be ok that during occt it peaked at 1.508 (for a split second). would like a rock stable voltage, but i know that's not going to happen, hehe.

for some reason virtualbox sees 300MHZ (~4300) less (this happened on the previous amd system as well). not sure what the deal is with that.

will run it through its paces to ensure it holds up for 24/7.

[tbone8ty]

hmm FX-8120 is now $189 on newegg...

curious on the 2012 batches...

hey AMDnord how is yur new toy doing?

[AMDnord]

ii returned home really late and i just turned it on to see the VID :p

1.26v. Is it good for 8120s?

[Oese]

normal i'd say, not too bad..

[AMDnord]

i think i am gonna skip the air section and go straight for LN2 :P

[FlanK3r]

just got a brand new 8120!

1201PGN

gonna test it tonight (i hope) on air and on ln2 @ saturday.



wish me good luck :p

I am looking forward

[aRkangeL]

I have one 8120 1150PGT have about 1,285 stock vcore, im waiting for the mobo return to the RMA, and get some tests!

[tbone8ty]

wonder how the 2012 batches are doing...

any word on fx-8170 yet? cpu-world says march 2012 release.

want b3 stepping stat!

[BeepBeep2]

wonder how the 2012 batches are doing...

any word on fx-8170 yet? cpu-world says march 2012 release.

want b3 stepping stat!

There won't be a B3 stepping. Piledriver is already finished and I'm pretty sure Trinity is ramping up for a May launch. Latest roadmaps don't say a damned thing about FX-8170, they list 8150 as highest tier until FX-8350 Piledriver series.

[Daveburt714]

There won't be a B3 stepping. Piledriver is already finished and I'm pretty sure Trinity is ramping up for a May launch. Latest roadmaps don't say a damned thing about FX-8170, they list 8150 as highest tier until FX-8350 Piledriver series.

I hate to admit it, but I've been messing with mem so much lately I'm a little out of touch with new CPUs...

I thought Trinity (piledriver?) was an APU. Will there be a version sans GPU and will it still be AM3+ or require a new socket (FM1 or FM2)?

I know..... I should just Google it, and I will if I have to, but it sounds like some of you folks are pretty knowledgeable on the ropic.

Thanks.... Dave

[BeepBeep2]

I hate to admit it, but I've been messing with mem so much lately I'm a little out of touch with new CPUs...

I thought Trinity (piledriver?) was an APU. Will there be a version sans GPU and will it still be AM3+ or require a new socket (FM1 or FM2)?

I know..... I should just Google it, and I will if I have to, but it sounds like some of you folks are pretty knowledgeable on the ropic.

Thanks.... Dave

Bulldozer is the name of a micro-architecture...so is Piledriver. "Trinity" APU will be the first of the Piledriver CPU's to come on the market, in Q2. "Vishera" CPU will follow Trinity to the market, replacing "Zambezi".

Llano (enhanced "STARS") FM1 --> Trinity (enhanced "Bulldozer", "Piledriver") FM2
Zambezi ("Bulldozer") AM3+ --> Vishera (enhanced "Bulldozer", "Piledriver") AM3+

[l0ud_sil3nc3]

supposedly FM2 is launching in June now

I was hoping for a sooner launch as I have been dying to try a llano setup for mem clocking.

[FlanK3r]

maybe wil be one year after Llano launch exactly.

[tbone8ty]

anybody lap their FX chip and see any substantial temp drops? or are you better off lapping the heatink/waterblock

[Daveburt714]

I told myself I wasn't going to buy another FX chip until there was a new revision, but I'm weak...
Anyway, heres another chip for the list:

Daveburt714 | FX8120 | 1152 PGT | 0052 | 4613 @ 1.4625v | custom water | Max untested


My old 8120 is an 1135, so I thought I'd try another one to see if they'd gotten any better.
I was a little disapointed when I saw that this ones VID was 1.350v
My week 35 was 1.2375 and it actually clocks pretty well...

Anyway, this new one is a voltage hog (compared to the week 35), but it runs MUCH cooler!
Anything over 1.375 on the cores (on the old chip) resulted in temps well over 65c, but this one seems more voltage/temp tolerant...

Guess I'm going to have to buy a Kill-A-Watt so I can see how much power each one is actually drawing.
Overall though, this new one comes pretty close to the clocks I got from the week 35
I get the feeling this new one may actually use a little less electricity. BTW, the IMC on this 52 defaults @ 1.2375v which is also very high compared to the week 35 (1.1175v).

Ignore the MEM clocks in this SS, I'm still working on those, but the Cores and IMC are pretty darn stable using Prime Blend/Max Mem...


Even though the vid on this chip is high it's not too shaby (would probably do really well on air).
Maybe they are getting a little better...

[FlanK3r]

wow, IMC looks good. Can u tets max CPU/NB for superpi test? Because I seen in Taiwan forum aircooled FX with superpi32M and NB over 3000 MHz!

[BeepBeep2]

wow, IMC looks good. Can u tets max CPU/NB for superpi test? Because I seen in Taiwan forum aircooled FX with superpi32M and NB over 3000 MHz!

2500 NB @ 1.25v isn't very impressive at all...SuperPi will probably be about 2600 max

[FlanK3r]

but not bad...about 1.35V is still safe for 24/7 to CPu/NB...

[tbone8ty]

nice dave,

was itching to pick up a 8120 as well but i wanna see how the 2012 versions are doing....

question...you have 36gigs of memory? what sticks are you running for the specific SS

[Daveburt714]

nice dave,

was itching to pick up a 8120 as well but i wanna see how the 2012 versions are doing....

question...you have 36gigs of memory? what sticks are you running for the specific SS

Yeah, I was looking for a 2012 chip too, week 52 of 11 was the best they had though....

For mem, I'm using my RipJawX 2133 9-11-10 kit 8Gb (2x4).
The CPU-Z mem tab gets messed up sometimes while using Turbo Evo to overclock.
To be honest I didn't even notice the 36Gb , it usually just shows wrong timings.
Those settings are good for ~30 minute blend from cold boot though (without the CPU-z screwup)

FlanK3r: I haven't really pushed the IMC that far yet, but thats about all it has in it @ those V's.
I left voltage on "Auto" in bios and it booted up @ 1.40v with 2400 Mhz IMC...

I'll see if I can more out of it this weekend, but I don't really feel comfortable with anything over 1.275.

[BeepBeep2]

but not bad...about 1.35V is still safe for 24/7 to CPu/NB...

I don't think so at all. First of all you add 5-10c in heat and scaling drops off significantly at 1.2-1.25v. I thought Zeus said he degraded his CPUNB pretty quick with about 1.4v too.

[FlanK3r]

Limitation for FX are very simillary to AMD Thuban. Look at technical documents....

[BeepBeep2]

Limitation for FX are very simillary to AMD Thuban. Look at technical documents....

Skimmed through both the Software optimization guide and BKDG guide. They say nothing about voltage tolerance...what tech doc are you talking about?

What I'm saying is that
A. They don't scale much past 1.25v
B. Raising CPUNB voltage creates a lot of unwanted heat...

[FlanK3r]

heat is problem every time...with Phenoms, Bloomfileds etc. IMC improves soemtimes memory stability and clocks (+performance of CPU with higher uncore and a bit CPU stability at high clocks), but chips are hotter and harder to cooling.

[FlanK3r]

heat is problem every time...with Phenoms, Bloomfileds etc. IMC improves soemtimes memory stability and clocks (+performance of CPU with higher uncore and a bit CPU stability at high clocks), but chips are hotter and harder to cooling.

Guys, tomorow Il try my FX-8120 again under LN2, hope, chip will be stil live after that