***Official*** eVGA Classified SR-2 Thread

[Zaxis01]

I'm thinking about picking up an SR-2, But not sure if my cpu will work with other xeons.

I have a xeon es chip equivalent to a xeon e5645 six core 2.4ghz.

Does anyone know if it'll work with a retail chip?

[Sam_oslo]

I'm thinking about picking up an SR-2, But not sure if my cpu will work with other xeons.

I have a xeon es chip equivalent to a xeon e5645 six core 2.4ghz.

Does anyone know if it'll work with a retail chip?

Do you have more info on that? it's hard to tell without the S-Specs, but early X5650 A0's come in many different clock speeds, and if it's a variant of these, it should work.

[Zaxis01]

[Sam_oslo]

This is not a E5620, for sure, because that's 4C/8T. CPU-z gets confused on these early ES's.

It's not easy to say for sure without knowing the S-Specs, but I think it will work. I'm running a couple of these on SR-2, and these looks pretty much the same to me, except the different multi.



EDIT:
That chip got a pretty low multi, and be aware that you won't get a killer OC out of it on SR-2. You can't hit that kind of high BCLK on SR-2 with current BIOS.

[Zaxis01]

So is the SR-2 BCLK limited?

What's the highest you've reached??

[Sam_oslo]

Yep, you bet. I'm running with BCLK200 for 24/7, but I can bench up to BCLK211 with an unstable chipset.

This may get better with upcoming BIOSes, and this may vary depending on the CPU-batch# too, but you shouldn't count on BCLK225, as it is.

[Zaxis01]

TAB Sam for all you info. It definitely helped me out with my mobo decision.

I think i'll go with R3E until the SR-2 can reach a higher bclk

[overdoze]

Can someone shed light to a newbie:

Which bios version to run sr2 at such high bclk 216

my vtt=1.4V my IOH=1.4V and can only run stable at bclk=194 using A46 bios on a pair of x5650 lock uncore multiplier at 20x

How can I get a B1 and is the B1 truely has unlock uncore

[Movieman]

Can someone shed light to a newbie:

Which bios version to run sr2 at such high bclk 216

my vtt=1.4V my IOH=1.4V and can only run stable at bclk=194 using A46 bios on a pair of x5650 lock uncore multiplier at 20x

How can I get a B1 and is the B1 truely has unlock uncore

The first thing you have to understand is that Sam bought this crappy pair of chips for chump change that he's transformed into monsters just thru his tralents and persistance.
Yea Sam,Kudo's to you!

Next, yes, uncore is unlocked with B1 chips but not with B0's..

[resident1509]

You're going to need to get the chipset cooler for high bclk on the sr-2. I'm afraid a new BIOS won't help you much.

[tistou77]

Next, yes, uncore is unlocked with B1 chips but not with B0's..

but if uncor is unlocked with B1, on a R3E, not to exceed the 1600Mhz with memory

[overdoze]

any help is appreciated as you can see in the picture. I have the B1 stepping and the uncore is still lock at 20x no matter what I do in the bios ..

FYI, I'm fighting cancer and this is close to number 100th PC that I built over 10 years

[Sam_oslo]

The first thing you have to understand is that Sam bought this crappy pair of chips for chump change that he's transformed into monsters just thru his tralents and persistance.
Yea Sam,Kudo's to you!

Next, yes, uncore is unlocked with B1 chips but not with B0's..

This is how it goes when one learns from the best. I personally have learned a lot from your great pioneer efforts on double-socket platform. This is a new world with many existing and unexplored terrorists that you have been exploring and sharing with us .
I member you were very exited talking about some strange jumps in VVT-requirements in certain BCLKs. I've used your important findings as a fomentation for ideas about how to exploit this new chipset (and how these CPUs communicate with each other) to get the best out of these CPUs. I'm still not sure what exactly is happening, but it is both existing and promising, and I believe it can go even further with your pushes, experience and great pioneer efforts.

any help is appreciated as you can see in the picture. I have the B1 stepping and the uncore is still lock at 20x no matter what I do in the bios ..

FYI, I'm fighting cancer and this is close to number 100th PC that I built over 10 years

http://img256.imageshack.us/img256/5591/mysr2.png[/IMG]

I think EVGA needs to deliver a better BIOS for unlocking the Uncore properly. They have done something and some aspect of it seams to be working for some, at least partially, but I think it is not complete yet.
I remember, in good old days, GA UD7 BIOS-team had released a BIOS and said they had applied Intel's new microcode, but it didn't work in first attempt. They tried a couple of times and finally got a it right with BIOS F6h (or later). I guess there is something tricky there that BIOS-programmers need to work on.

I was playing with the new HW-info and discovers this. According to this, Intel's new microcode is not updated, but you need it to unlock the Uncore on these CPUs. Maybe this got something to do with my A0's, but maybe you can check it too?

[overdoze]

Thanks Sam,
I just tried it out and this is what I get, seems like the microcode has been updated since A0

On the side note I think the max oc is now limited by uncore frequency at around 3800 and since your cpu has uncore mul at 18x you will have higher bclk oc capability ... I now wonder the 5677 has uncore lock at which multiplier

We definitely need a bios that can unlock these babies

[Sam_oslo]

Yeah I'm very happy with x18, but all X56xx-B1 retail-stepping have x20 Uncore. And yes, the VVT-requirements for that locked Uncore is a limiting factor for 24/7 OC.
But your CPU-batch# seams to need a high Vcore=1.4v with almost 80'C core-temp already, and you seams to be limited by vCore/heat too.
An unlocked Uncore at x16 (with less VVT) would shave a few degrees on your CPU, and give a safer and longer life to your CPU too, but you won't be able to get much higher OC out of it, due to vCore/heat-limitations on your CPU-batch#, I would think.

You are using VVT=1.4v (which is a bit too high for my liking for 24/7) for BCLK194, but you will probably end up with ~BCLK175 with a moderate VVT=1.34v. This is the real bad side-effect of locked Uncore.
I don't know much about X5677, but those are quad-cores, and I prefer the extra power of 12C/24T. A X5660 seams to a better option.

[overdoze]

That was my initial attemp max volt of what this baby can do ... Eventually, I will settle for something lower and more stable. It was not stable 24/7 yet, error out once a day (not bsod however)

Have done some PWM cooling improvement and this is now my setting for the summer. If it pass 1 week without failure I will give it a vcore decrease notch by notch till it is completely stable for Folding

One thing I notice is the Eleet report temperature much slower than Realtemp. Realtemp report max temp high but norminal temp is in line with Eleet. It would dance quickly to 76 and comes back down immediately

Since the cpu has thermal enable, it would downclock automatically when core temp reach 100C ... The design spec for the cpu is at 125C anyway, so I'm not too worry

[Sam_oslo]

That's a good OC for those CPUs, and the temp looks Ok too. You should be happy with it, if you are comfy with VVT=1.40v. I would try to keep it under 1.35v, at least until the next BIOS arrives (if it does at all, it has been a long wait ) in a hope for unlocked Uncore, but that's a personal preference and you can see how it works out for you, anyways.
It's good to know if these chips will/won't degrade with high-VVT over time, and it's good to have a good laboratory rabbit around.

[Sam_oslo]

New BIOS A49 is out.

Just flashed last night and still testing, but couldn't find any improvement on BCLK/OC yet.
A big bug (EventID 46, Machine Check Exception, fatal hardware error) is fixed, and some improvements on memory OC seams to be there too.

[resident1509]

New BIOS A49 is out.

Just flashed last night and still testing, but couldn't find any improvement on BCLK/OC yet.
A big bug (EventID 46, Machine Check Exception, fatal hardware error) is fixed, and some improvements on memory OC seams to be there too.

You won't get any improvements on blck. This board can go very high bclk but its more dependent on pushing more volts to the NB than X58 and keeping the NB cool, which is why you see the benchers putting the NB on ln2 for this board but not X58s. If you watercooled the NB, you would probably be able to push bclk more.

[Movieman]

You won't get any improvements on blck. This board can go very high bclk but its more dependent on pushing more volts to the NB than X58 and keeping the NB cool, which is why you see the benchers putting the NB on ln2 for this board but not X58s. If you watercooled the NB, you would probably be able to push bclk more.

Realistic BCLK limit even with LN2 on the NB is app 215..Anything above that doesn't give any benefit..
It is what it is.
One hell of a dualie but it is a dualie not a single socket gamer and it does have limitations.
Now for pure computational power there isn't anything this side of a $30,000.00 quad Beckton build that will touch it.

[Sam_oslo]

Yeah you are right MM. This is a dualie and have it's own limitations. It is a great MB as it is, but it doesn't means we have to give up the hope for better BCLK/OC with new BIOSes.

You won't get any improvements on blck. This board can go very high bclk but its more dependent on pushing more volts to the NB than X58 and keeping the NB cool, which is why you see the benchers putting the NB on ln2 for this board but not X58s. If you watercooled the NB, you would probably be able to push bclk more.

How much IOH-voltage these guys are pushing trough those LN2-cooled NB?

My OC seams to scale really good up to 200x19 with IOH=1.35v, but it seams to need a lot more to make a real difference over that. I can't get the chipset to stabilize at 206x19 with IHO=1.42v. The CPU is stable, and passes many heavy CPU-benches easily, but the chipset seams to need excessive voltages to stabilize at BCLK201+.

I can't see any real impact on BCLK/OC with a moderate IOH-voltage increase for 24/7 use on my air-cooling. Do you know how much IOH-voltage those water-cooled NBs can handle, and what kind of real impact it has on BCLK/OC?

[PlatoonSgtElias]

Actually too much IOH will get the board unstable. Try to use 1.2 to 1.3 for 24\7 also use CPU PLL 1.6 if overclocking, lower pll might give better overclocking. Use bout 1.35 24\7 on vtt and 1.60 max when overclocking. But some fans on it.
Upping the PCI-E MHz will help u also but dont clock it for 24\7 useage. PCI-E u could raise up to 130 for the best effect.

[resident1509]

On ln2, 1.4 - 1.45v

Here are a couple threads both about ln2 and general bclk stability insight from shamino.

http://kingpincooling.com/forum/showthread.php?t=827

http://www.evga.com/forums/tm.aspx?m=268280

[Sam_oslo]

Actually too much IOH will get the board unstable. Try to use 1.2 to 1.3 for 24\7 also use CPU PLL 1.6 if overclocking, lower pll might give better overclocking. Use bout 1.35 24\7 on vtt and 1.60 max when overclocking. But some fans on it.
Upping the PCI-E MHz will help u also but dont clock it for 24\7 useage. PCI-E u could raise up to 130 for the best effect.

Generally, lowering the IOH is a good idea, but my current 24/7 setup gets display-artifacts with less than 1.3v. The SLI on my double-GPU (GTX 295) may play a role too, but you are right, an voltage-increase (at least a moderate one) doesn't seams to have any real effect on stabilizing the chipset at BCLK201+, at least not on my air-cooling. I suspect it won't have much effect with WC either, but I wouldn't know.

I remember my good old w3520 liked a low CPU-PLL=1.4v, but haven't played much with CPU-PLL on these CPUs yet. I had planned to test a low CPU-PLL round, and will try to see how it works soon hopefully.

I'm keeping the PICe=Auto and VVT=1.32v on my current 24/7 OC (200x19) and it's working fine.

On ln2, 1.4 - 1.45v

Here are a couple threads both about ln2 and general bclk stability insight from shamino.

http://kingpincooling.com/forum/showthread.php?t=827

http://www.evga.com/forums/tm.aspx?m=268280

Does this means he didn't see any real impact of higher IOH?
Because IOH = 1.4 - 1.45v is not much, and could be possible on good air-cooling too. I've tried up to 1.425v and temps seams to be OK on air inside a good ventilated case.

His use of IOH = 1.4 - 1.45v on LN2 adds to my supposition that increasing it over 1.4v+ doesn't have much effect on OC/BCLK.

[resident1509]

Yes, from what I've seen and what others have mentioned, keeping the NB cool will then help with stability past 210/215 bclk.