Intel 4.5GHz & 5GHz LinX Stable Club

[iTravis]

Just another sleepless night to get me to join this club. My chip needs lot of voltage to get it stable but well, my temps are ok so I guess that's all matter.
Batch #: 3912B168 with HT off and 24C case temp, Watercool as in sig:

[zalbard]

Very nice!
Is that with VDroop On or Off?

[iTravis]

Very nice!
Is that with VDroop On or Off?

Without VDroop

[SteveRo]

If I disable hyperthreading on 980x, linx gflops increases from 53.xx to 94.xx -

[Jor3lBR]

If I disable hyperthreading on 980x, linx gflops increases from 53.xx to 94.xx -

That's because Linpack is the only multi-threaded application that actually get worst results with more cores.

That explains also why AMD can turn more Gflops then Intel on the same speeds, but with less efficiency of course.

[zalbard]

Insane, huh?
I think you lose up to 10 GFlops or so with Nehalem... But with Gulftown it's a ton more it seems, lol.

[SteveRo]

Good evening, I couldn’t resist uping the vcore on this cpu, this is at 1.456v (bios), ht off, cooled via 2xboreas in series, please update my entry, thanks!

[Sniper]

Nice mhz and gflops SteveRo

[SteveRo]

Thanks Sniper, I wonder how much higher I cuold push vcore w/o doing damage.

[zalbard]

4.7 @ just 1.44V - very nice! Looks like a great chip!

[Sam_oslo]

Good evening, I couldn’t resist uping the vcore on this cpu, this is at 1.456v (bios), ht off, cooled via 2xboreas in series, please update my entry, thanks!

http://img682.imageshack.us/img682/9898/47linxstable.png[/IMG][/URL]

Good job it looks like the return of B-batch.

This is a really good results and you don't need to push it more.

[trn]

Then what do we make of the F-batches?

Good job it looks like the return of B-batch.

This is a really good results and you don't need to push it more.

[Sam_oslo]

Then what do we make of the F-batches?

That's really exotic batch#, and I haven't seen any of them in real 24/7 action yet. It is really interesting to see how good they are.
This B-batch seams really promising for now. Have you seen any good F-batch 24/7 results?.

[SteveRo]

4.7 @ just 1.44V - very nice! Looks like a great chip!

I just noticed you have my entry at pcie 115, for the 4.7 run I had pcie at auto.
Please update, thanks!

Also - question - how safe is it to continue to bench 32nm at 1.44v?
Is there any real data on 32nm safe bench voltages?

By "bench" I mean usage level of 1-2 hours here and there not 24/7.

[zalbard]

I just noticed you have my entry at pcie 115, for the 4.7 run I had pcie at auto.
Please update, thanks!

Done!

Also - question - how safe is it to continue to bench 32nm at 1.44v?
Is there any real data on 32nm safe bench voltages?

Well, imo... 1.45V should be totally safe, 1.5V should be safe also. I wouldn't use more than that 24/7.
For benching just watch your temps I guess... Voltage should be irrelevant.

[Sam_oslo]

I just noticed you have my entry at pcie 115, for the 4.7 run I had pcie at auto.
Please update, thanks!

Also - question - how safe is it to continue to bench 32nm at 1.44v?
Is there any real data on 32nm safe bench voltages?

By "bench" I mean usage level of 1-2 hours here and there not 24/7.

Vi don't have much data on how much 24/7-torture these can take, yet. You are our laboratory rabbit, you know.

What we know, Intel is recommending max 1.4v for these 32nm CPUs. Compare this to the 1.55v for 45nm, and you are running a ~1.6v relatively. It is interesting to see how long it can run without degradation.

[zalbard]

And they also said 1.35V for VTT which was a joke, so...

[Sam_oslo]

So, these 32nm may be as good as 45nm at taking torture for 24/7, and I really hope they do, but we don't know it yet.

Intel is increasing the max VVT from 1.35v to 1.4v, but decreasing the max vCore from 1.55v to 1.4v. So it is really interesting to see how much of a joke this new vCore is and how much these 32nm can take for 24/7.

[zalbard]

Intel is increasing the max VVT from 1.35v to 1.4v, but decreasing the max vCore from 1.55v to 1.4v. So it is really interesting to see how much of a joke this new vCore is and how much these 32nm can take for 24/7.

Pretty sure Intel is playing it safe in any case.
And the reduction of vCore is due to the new tech process, vCore keeps decreasing as transistors get smaller and smaller.

[SteveRo]

So for 45nm intel said 1.55 but 1.6 seems ok
For 32nm intel says 1.4 so 1.45 should be ok?

[SteveRo]

Good afternoon, this time with HT on, please update my entry, thanks -

[moopa]

Hello. Last time my test was stable at 4.5 GHz but now I have greater result which is stable at 4.6 GHz.

[zalbard]

Great results, everyone!
Updated!

[trn]

Your 980x is still under warranty, why don't you tell us

Thanks Sniper, I wonder how much higher I cuold push vcore w/o doing damage.

[SteveRo]

Your 980x is still under warranty, why don't you tell us

I hadn't thought of that - good idea.
I do wonder what the tigerdirect cpu return policy looks like, I should probably check that out.
I have been doing a fair amount of benching at 1.5 vcore the last couple days - no problems yet.