AMD and priceless moment!

That "125W TDP*" Phenom 9850 processor just went to around 240W* at 3.5GHz and 1.52v. The watercooling wasn't optional. More like a volcano in silicon instead of a wolf. :p

* 125W rating, real life and comparison to other processors use shows this probably is an artificially low value. Overclocked/over-volted power is probably also underestimated above under 100% load.
 
Nice OC For a phenom but its still junk. You wouldn't of been able to achieve that OC w/o it being a black edition with unlocked multis. And burns about double the power to get an extra 1000mhz.
 
Simpson5774 said:
Nice OC For a phenom but its still junk. You wouldn't of been able to achieve that OC w/o it being a black edition with unlocked multis. And burns about double the power to get an extra 1000mhz.
Click to expand...


LOL, who cares thats a 1000MHz OC sure it's going to take power and stuff.
 
Considering that it is a "native" quad and AMD's CPUs normally have a lower clock limit, the OC is really impressive.
 
Great OC, but how long will that Phenom last being run like that is the real question...?
 
I would not think 1.52v is gonna cause a heavy decline in proc longevity, or at least anything anyone here would worry about:D Either way, even if the average OC is at least 3.0 with minor bumps in voltage that still is not too bad and would likely be high enough to prevent any CPU bottlenecks in gaming especially with the growing number of people with 22 and 24" monitors with settings turned up
 
APOLLO said:
Great OC, but how long will that Phenom last being run like that is the real question...?
Click to expand...
That's a good question, but not one easily answered because 6 months after K10 "launch" the technical data sheets still aren't on AMD's web site. It would normally be trivial to look up the maximum voltage to see if 1.52v exceeds it.

The crystal ball says the lifespan will probably be significantly shortened, but it probably doesn't matter to that overclocker.
 
Wow thats great. And How can you say this is junk it is not like they are charging an arm and a leg for black edition, really who cares if the htt won't overclock worth shit, when they can do it with just uping the multipliers. Really overclocking is overclocking does it really matter how you get your final overclock. But yeah it looks like these phenoms got potential, it seems he got 3.2ghz stable which is really a nice overclock.
 
pxc said:
That "125W TDP*" Phenom 9850 processor just went to around 240W* at 3.5GHz and 1.52v. The watercooling wasn't optional. More like a volcano in silicon instead of a wolf. :p

* 125W rating, real life and comparison to other processors use shows this probably is an artificially low value. Overclocked/over-volted power is probably also underestimated above under 100% load.
Click to expand...

you have any idea how many W a Q6600 @ 3.6 takes? just about 200 0.0 so it's not a surprise for that phenom
 
Digital Viper-X- said:
you have any idea how many W a Q6600 @ 3.6 takes? just about 200 0.0 so it's not a surprise for that phenom
Click to expand...
It depends on the voltage. A G0 Q6600 @ 3.6GHz, 1.40v would use an estimated 175W at 100% load. If you need more than 1.40v, i'm not sure it's worth it.

My Q6600 does 3GHz @ stock voltage. 3.1GHz requires a 0.05v bump. The silent CPU cooler limits overclocking and that's fine to me considering the steep power and noise tradeoffs to get higher.

The power I'm estimating here and above is at the CPU socket, not counting VRM or PSU losses. 240W on the 9850 for the rather large voltage bump and 1GHz overclock is not something to take lightly. It wouldn't be surprising that many Phenom boards could not supply that much power reliably to the CPU.
 
pxc said:
It depends on the voltage. A G0 Q6600 @ 3.6GHz, 1.40v would use an estimated 175W at 100% load. If you need more than 1.40v, i'm not sure it's worth it.

My Q6600 does 3GHz @ stock voltage. 3.1GHz requires a 0.05v bump. The silent CPU cooler limits overclocking and that's fine to me considering the steep power and noise tradeoffs to get higher.

The power I'm estimating here and above is at the CPU socket, not counting VRM or PSU losses. 240W on the 9850 for the rather large voltage bump and 1GHz overclock is not something to take lightly. It wouldn't be surprising that many Phenom boards could not supply that much power reliably to the CPU.
Click to expand...

on average people using 1.4-1.5v for 3.6 ^^ thats still alot of wattage, where did you get the 240w from
 
Don't forget that when you OC a Q6600, you need to increase the northbridge clock spped which sometime needs a voltage bump too and Q6600 also doesn't have an imc like the Phenom. So looking at the heatsink for X38 and 790FX chipset, I'm not suprised why Intel has a lower CPU power consumption at the socket.
 
Digital Viper-X- said:
on average people using 1.4-1.5v for 3.6 ^^ thats still alot of wattage, where did you get the 240w from
Click to expand...
OK, i'll humor you with 1.5v since you want to bicker: ~205W for a 1.2GHz overclock on the Q6600 G0.

There's a relationship between power used by switching transistors that's proportional to the speed (S1/S2) and the square of voltage (V1^2/V2^2). Plug in the numbers and you'll get an estimate. :p I label it an estimate because not all parts of the chip are overclocked or utilized the same way, and power consumption can also vary with the temperature (higher when overclocked).

alg7_munif: it's not as much as you think, possibly in the 10-15W range at most. Intel does need to start moving chipsets into better processes though, but Nehalem is going to make that moot within a year.
 
pxc said:
OK, i'll humor you with 1.5v since you want to bicker: ~205W for a 1.2GHz overclock on the Q6600 G0.

There's a relationship between power used by switching transistors that's proportional to the speed (S1/S2) and the square of voltage (V1^2/V2^2). Plug in the numbers and you'll get an estimate. :p I label it an estimate because not all parts of the chip are overclocked or utilized the same way, and power consumption can also vary with the temperature (higher when overclocked).

alg7_munif: it's not as much as you think, possibly in the 10-15W range at most. Intel does need to start moving chipsets into better processes though, but Nehalem is going to make that moot within a year.
Click to expand...

the point is, it's not that big of a difference. get over it. 3.5ghz @ 1.5v is not bad for a phenom.
 
Digital Viper-X- said:
the point is, it's not that big of a difference. get over it. 3.5ghz @ 1.5v is not bad for a phenom.
Click to expand...
LOL, of course it's not that big of a difference when you consider:

1) that CPU @ 3.5GHz CPU hasn't been stability tested yet (ask me about my 3.3GHz Q6600 tests @ stock voltage!) and
2) you only accepting a higher overclock (1.2GHz vs 1GHz for the Phenom) at the higher edge of the voltage range (1.4-1.5v) for the Q6600 isn't reasonable.

If you're going to cherry pick one result as typical for AMD, why not accept Q6600 @ 3.6GHz, 1.4v results too? :rolleyes:

Anyways "Tony" already said 3.5GHz wasn't stable and noted this about the incredible amount of heat generated by that CPU:
Tony said:
I only test with water man, 3 test benches create a lot of heat so I have water to water heatexchanger with 220L of rainwater as the cooling side. 2 20Mcoils of copper submerged in this create my loop.

So to say you need water...I have no idea...I ONLY have water here.
Click to expand...
He has already backed down to 3.2GHz and finally managed to run some benchmarks. That's only a 700MHz semi-stability tested overclock, and there's probably nothing so special about his chip (or abilities) that other reviews already noted: other sites got 2.8-2.9GHz (possibly 3GHz) stable overclocks with B3 using air cooling. Tony should be able to squeeze more out of his elaborate WC set up.
 
hstuehmeyer2000 said:
are you just trying to start something?:confused:
Click to expand...
No, he just refuses to accept that we're not in love with an OC that makes the CPU put out more heat than any Prescott ever did :rolleyes:

This is nice for a record, but it's definitely not something I'd want to use every day...
 
alg7_munif said:
Considering that it is a "native" quad and AMD's CPUs normally have a lower clock limit, the OC is really impressive.
Click to expand...

drop this native crap, 4 cores is 4 cores and intels "non" native quad core is kicking AMD "native" around the world and back...


4 cores is 4 cores!!!!! period!
 
Not necessarily, AMD has taken a performance hit where perhaps they could've gone the route Intel did, but in the long run I think it'll pay off. Eventually Intel will have to do the same thing, and we'll see how well they do with yields and implementing it, along with the IMC.
 
A very nice overclock. Maybe more people would consider AMD to be an interesting CPU after all.
 
pxc said:
LOL, of course it's not that big of a difference when you consider:

1) that CPU @ 3.5GHz CPU hasn't been stability tested yet (ask me about my 3.3GHz Q6600 tests @ stock voltage!) and
2) you only accepting a higher overclock (1.2GHz vs 1GHz for the Phenom) at the higher edge of the voltage range (1.4-1.5v) for the Q6600 isn't reasonable.

If you're going to cherry pick one result as typical for AMD, why not accept Q6600 @ 3.6GHz, 1.4v results too? :rolleyes:

Anyways "Tony" already said 3.5GHz wasn't stable and noted this about the incredible amount of heat generated by that CPU:He has already backed down to 3.2GHz and finally managed to run some benchmarks. That's only a 700MHz semi-stability tested overclock, and there's probably nothing so special about his chip (or abilities) that other reviews already noted: other sites got 2.8-2.9GHz (possibly 3GHz) stable overclocks with B3 using air cooling. Tony should be able to squeeze more out of his elaborate WC set up.
Click to expand...
He was using the DFI board originaly to hit 3.5Ghz, switched to the DQ6 and went down to 3.2. No where in that quote does it say anything about that cpu putting out huge amounts of heat, he said he doesn't know if it needs water, but he only uses water to do his testing(that includes intel cpus btw)

Also you don't need to tell me about the Q6600, mine runs 3.6Ghz @ 1.3v and 3.2 @ 1.2v. I know how great they are, that doesn't mean this phenom isn't surprising, it's the fastest phenom we have ever seen.

XtremeSys said:
Im running at 3.2 as its EASY, I don't have time to tweak the tits out of this CPU tonight to get it stable at 3.5 for 3D. Its already 2:14am and im back in the workshop about to run cinebench 10 for ya.

I have a feeling I will get 3.5 3d stable, the issue is the NB frequency has to rise along with it to keep the performance good...this takes time.
Click to expand...

http://www.xtremesystems.org/forums/showpost.php?p=2891952&postcount=120

cinebench R10 @ 3.2

not bad a Q6600 @ 3.2 gets
Rendering (Single CPU): 3705 CB-CPU
Rendering (Multiple CPU): 12970 CB-CPU
Multiprocessor Speedup: 3.50
Shading (OpenGL Standard) : 6060 CB-GFX
Click to expand...
 
sadly with no bus scaling, the Core2 is still gonna stomp it.
 
MrGuvernment said:
drop this native crap, 4 cores is 4 cores and intels "non" native quad core is kicking AMD "native" around the world and back...


4 cores is 4 cores!!!!! period!
Click to expand...

It is easier to find two dual core CPUs that are capable of reaching a high clock speed and stick them together but it is much harder to find a "native" quad core CPU which all cores can reach the same high clock speed.
 
I'm new to whole idea of overclocking, but from my observations it seems like this is a natural progression with the manufacturing of most chips. The initial is the B2, and while they work, they're hotter, don't OC well, and are just the first generation. B3 comes and improves on heat and OC, fixes the errata that they can fix at that time. Then the true potential is arrived at with the G0 or G2 revisions where you see power being unleashed. Or maybe this is all in my head.
 
Swingline(Red) said:
I'm new to whole idea of overclocking, but from my observations it seems like this is a natural progression with the manufacturing of most chips. The initial is the B2, and while they work, they're hotter, don't OC well, and are just the first generation. B3 comes and improves on heat and OC, fixes the errata that they can fix at that time. Then the true potential is arrived at with the G0 or G2 revisions where you see power being unleashed. Or maybe this is all in my head.
Click to expand...

You're just being sensible. Every major silicon manufacturer goes through the same tick-tock process. It's just that some are very impatent. The whole hurry up and wait mentality.
 
I ordered one will post some results when it arrives. I really don't care about the power it eats up. I would rather drive a Porsche over a Yaris as well even though it eats more gas.
 
ChristO said:
I would rather drive a Porsche over a Yaris as well even though it eats more gas.
Click to expand...
So your getting a Q6600...?..:)

J/K...I'm mainly am Intel guy, but I'm looking forward to AMD making some nice chips, so I can add an AMD rig back into my line-up...I started on AMD rigs a long time ago, and I have had a yearning to build another one....I'm just waiting for the right time to build...It looks like its getting close.
 
definitely going to wait for the lower end b3's to show up before making my purchase now :]

about fucking time DAAMIT
 
Wow...I didn't think any of the Phenom's made it over 3.1 but now that I see 3.5, I might just have to build a new Phenom machine real quick.
 
I have a quick question/point to inject....

Is the voltage per core, per die, or per processor (package)?

If it's per package, 1.52 spread across 4 cores wouldn't be too bad I imagine (vs. stock).

However, I'm thinking it's probably per core.

Anyone know?
 
You must log in or register to reply here.
Back
Top