Core i7 Overclocking by Batch #

**hetsaq** · 10-06-2009, 04:21 PM

I ran 5 passes of memtest and 8 hours of prime blend. I will try the lower speed and see if that helps and run as HyperPI 32M test as well, do I want to use all 8 cores?

**Sam_oslo** · 10-06-2009, 04:31 PM

8 hours of prime blend means stable RAM (at least at the time you tried). Most probably it is stable now too, but you never know. HyperPi32M is fast and pushes both RAM and IMC, and yes with all 8 Cores.

**hetsaq** · 10-06-2009, 04:39 PM

Thank you Sam, hope I am not being too much trouble.

**Sam_oslo** · 10-06-2009, 04:52 PM

No problem. I will learn from your affords too

When you get bad benching result, it is always a good idea to check if it is stable.

I have been playing with a OCZ P 1333MHZ CL7 earlier, and it could bench 17000+(Everest read) with x6 RAM @6-7-7-22 1.65v. Your sticks are better and should do better. It is interesting to see why it is acting like this.

**hetsaq** · 10-06-2009, 04:59 PM

It went through HyperPI 32M on all 8 cores just fine. I still haven't tried the lower timings yet.

*EDIT* 6X ram

**To(V)bo Co(V)bo** · 10-06-2009, 05:20 PM

Originally Posted by Godmyster

Nice score lol i was thinking on getting the PROLIMATECH for my D0 what do you guys think? or with less money i can get something similer?

Thanks

I have a brand new never used megahalem in the box for sale, with extra fan clips im selling. If your intrested PM me. Ill do a good a deal.

**Sam_oslo** · 10-06-2009, 05:20 PM

@hetsag

That's good news, your RAM is stable at x8 and you can keep it there. Now you can focus on other stuff and then when I look at the first screen you posted,

your L3-latency is 14.4 ns, and it is too high. It is strange, but it gets more strange (to me at least) when your L3-read/right are "normal", and all L1 and L2 numbers are "normal" too. I'm not exactly sure what this means, but could indicate some problems with IMC or Uncoore?

I have to look this up, and you could try to follow this too.
Maybe somebody else knows what can make L3-latency to get so high?

**hetsaq** · 10-06-2009, 05:28 PM

I didn't know that was high, but when looking at Rad's screenshot it certainly is.

**Sam_oslo** · 10-06-2009, 05:57 PM

Based on what you have done so far, I'm under impression that your RAM is OK, and you should get higher bench results than this.
IMC and Uncore would be next to check, but it can take a while to iron out, and maybe you should post it i a new tread.

I have never seen this problem before, and couldn't find much on a fast googling either, but i will try look it up closer later. The new IMC have some "strange new stuff" that is always interesting to follow.

**hetsaq** · 10-06-2009, 06:00 PM

I posted it in the P6T thread as well, maybe someone will have an idea there.

**Sniper** · 10-07-2009, 01:29 AM

Do you have B2B in your bios ?

**hetsaq** · 10-07-2009, 06:38 AM

I do, by default it is set to 0. I check CPU Tweaker and it says disabled. I manually set it to 4 (Lowest option I have), makes no difference in my results.

**JasonSauce** · 10-09-2009, 08:30 PM

Anyone hear anything good or bad about this batch number?
#3920B535 D0

**Sam_oslo** · 10-11-2009, 06:43 PM

here is my latest on W3520 (#3848A352)

VVT=1.23/1.26v (BIOS/load)

3x2GB @1680MHZ 7-8-7-21 1.65v

w3520 @4.41GHZ(210x21) 1.32v, 1 hour OCCT, on air

**imo** · 10-12-2009, 11:39 AM

Sam_oslo good work but like to see if thats Linx 4096 mem 20 run / prime default small fft stable?

**Sam_oslo** · 10-12-2009, 12:09 PM

Originally Posted by imo

Sam_oslo good work but like to see if thats Linx 4096 mem 20 run / prime default small fft stable?

Thx, I have tested the @4.3GHZ(205x21) 1.28v with 1 hour prime small ftt and based on that, It would probably pass with a few degrees more core-temp. But I haven't had any prime run on this, yet.

Based on my earlier testes which LinX, it has always passed with a couple of notches less Vcore (I guess it got to something with my MB). It would probably pass the Linx too, but I haven't any Linx run yet.

**imo** · 10-12-2009, 12:19 PM

cool, for some reason i cant get mine linx all mem stable @ 4.4G with HT requires about 1.34v to be linx all mem stable

but 1.31 v ok for prime

**avercros** · 10-12-2009, 12:25 PM

Anyone hear anything good or bad about this batch number?
Xeon w3520 #3920B524 D0.

Thanks.

**Sam_oslo** · 10-12-2009, 01:32 PM

Originally Posted by imo

cool, for some reason i cant get mine linx all mem stable @ 4.4G with HT requires about 1.34v to be linx all mem stable

but 1.31 v ok for prime

That's cool, 1.31v prime is great.

Just tested the stability of RAM and IMC on 4.41GHZ setup for W3520 (#3848A352), and here vi go :

3x2GB @1680MHZ 7-8-7-21 1.65v
Uncore x18 (3780MHZ)
VVT=1.23/1.26v (BIOS/Load)
w3520 @4.41GHZ(210x21) 1.32v

HyperPi32M stable

**matcur** · 10-19-2009, 05:43 PM

Hello I have problems with my new Xeon to pass 205 bclk, no matter what the multiplier or the configuration.
If anyone can help me

Thanks a lot

Rampage 2 extreme
Xeon 3520 batch 3919A652
G.Skill Storm 2000mhz 8-8-8-21 1n
These are the bios settings I use to have it stable at 200x21

Temperatures do not exceed 70 degrees in LinX.
On Antec 850 Quattro
MCP655b pump cooling Swiftech Apogee GTZ in Cpu and Swiftech waterblock on the NB
Radiator 3x12

Well I hope somebody can help pass the wall of 205
Thanks