Sandy Bridge FPO Batch list (Excell sheet)

[CedricFP]

Why TMonitor? Is task manager in Windows or something like RealTemp (which has CPU load in percentage) not adequate?

[Leeghoofd]

I get 5 euros each time Tmonitor is downloaded Cedric

List is up to date and validation rules changed, specially for you Cedric, moff to bed now :p

[aver]

3) In case of LinX : all ram has to be selected ( preferably over 2.8Gb, so only option is a 64Bit OS ) and it has to be at least 20 runs and still running when taking the screenshot. So select eg a 25 loops; start the test. Once you pass 20 loops, grab a screenie with everything running. For prime9 an hour of the custom test (12 -12K) is already a good indication if the CPU is stable

What would the prime settings be exactly? I have never used the custom setting before.

[Loser777]

Not a Sandy Bridge user atm--but I have an interesting observation for Sandy Bridge when it comes to LinX (since there's so many LinX results in this thread). It seems that it has a negligible, if negative performance difference per clock from Core 2 Duo.

Looking at leeghoofd's example of ~56GFLOPS at ~3GB problem size and 4.5GHz, that's a 25.8% advantage over a Core 2 Quad Q6600 with 44.5GFLOPs at ~3GB problem size and 3.4GHz.

A 25.8% advantage is significant but 4.5GHz is a 34% higher than 3.4GHz.

It could be an HT On/Off Issue

[CedricFP]

I get 5 euros each time Tmonitor is downloaded Cedric

List is up to date and validation rules changed, specially for you Cedric, moff to bed now :p

Haha I see - touche Don't get me wrong, I wasn't challenging you, I just thought there was something I didn't know about task monitor / realtemp load monitoring on SB CPUs.

[Leeghoofd]

What would the prime settings be exactly? I have never used the custom setting before.

I always set this setting and then let it run for a few hours. Example below for 2500K, for 2600K it automatically sets 8 threads



As mentioned before there's no 100% bulletproof stress test. You can prime for hours/days, then run a simple game and the rig can hang. Same with the LinX vs Prime discussion. One of the two can be stable while the other one caps out.
I leave the option open to use either one of these two. Some will say 12 hours or even more is required ( for prime95 test ) Others say 25 runs is not enough either for the LinX test. I just find an hour of custom prime 95 and more than 20 loops of LinX more than enough to get an idea if it's stable or not. If you want to do more loops or continue stressing longer, be my guest.

Not a Sandy Bridge user atm--but I have an interesting observation for Sandy Bridge when it comes to LinX (since there's so many LinX results in this thread). It seems that it has a negligible, if negative performance difference per clock from Core 2 Duo.

Looking at leeghoofd's example of ~56GFLOPS at ~3GB problem size and 4.5GHz, that's a 25.8% advantage over a Core 2 Quad Q6600 with 44.5GFLOPs at ~3GB problem size and 3.4GHz.

A 25.8% advantage is significant but 4.5GHz is a 34% higher than 3.4GHz.

It could be an HT On/Off Issue

Comparing technologies based on a single benchmark is never a good idea. Look at the below results of my upcoming review for Sandy Bridge :

All CPUs at 4ghz with rams at 1600mhz CL8





Clock for clock mate (FSB of the s775 system was 400) yes you can squeeze more performance out the old setup with a higher FSB. But even then Intels latest creation is so much smoother, more responsive. I even swapped my i970 for a 2600K

Haha I see - touche Don't get me wrong, I wasn't challenging you, I just thought there was something I didn't know about task monitor / realtemp load monitoring on SB CPUs.

It was a good tip Cedric, hence why I added it to the ahum rules... just messing with ya

Enuf discussion : RESULTS PLZ !!!

[Bullu$]

@Leeghoofd

Do u have more benchmarks @ 4Ghz? Maybe a link?

[PatRaceTin]

i will play mine for a while before submit

[Mrazaq]

I must have found the crappy batch already :p 2 2500K CPU's of the same batch. Tested on 3 boards ( 1 Dlx and 2 Pro boards ), will test the GB mobo next week... same boards power the 2600K deadeasy to 4.5 and beyond

So this is Batch L038A789 S/N 2V035198A2553 and S/N 2V035198A5600

One CPU does not post in windows with 4.3, other one stops at 4.4, 4.5 freezes up all the time... no matter the voltages etc... luckily the client only wanted 4ghz stable :p

(pic)

Keep them results coming lads

Yep, I can confirm that. Got an L038A789 last week, and I can't go past 4700MHz on 1.44v , which is pretty damn terrible. I'll probably try and RMA it to get a better one, I mean my luck can't get much worse than that.

[Jupiler]

I received the same batch# today.
I hope it clocks as good as yours. Waiting for board and memory to arrive.

How much VCore did you need for 4.5Ghz?

i7 2600K Batch L039B470.
MSI P67-GD65 Bios V1.8B5
Water, EK Supreme HF Cu

[st0ned]

I received the same batch# today.
I hope it clocks as good as yours. Waiting for board and memory to arrive.

How much VCore did you need for 4.5Ghz?

May the hand of the good luck be with you. Mine has the same batch and it's really lousy OC... 1.36v for 4.5Ghz, 1.53v for 5.0... On the other hand in appers to hold voltage just fine without degrading

[dengyong]

The results to chit chat ratio is pretty low here j/k ups just delivered my new system so I hope to have results by tomorrow. I'll be hunting the sweet spot under 1.3v, whatever it may be.

[Jupiler]

May the hand of the good luck be with you. Mine has the same batch and it's really lousy OC... 1.36v for 4.5Ghz, 1.53v for 5.0... On the other hand in appers to hold voltage just fine without degrading

That doesn't look to good, although it's always luck of the draw.
Some cpu's with the same batch numbers might clock better than others.
Knowing myself, I'll probably have a crap one.

[Bullu$]

I received the same batch# today.
I hope it clocks as good as yours. Waiting for board and memory to arrive.

How much VCore did you need for 4.5Ghz?

LinX 4.5GHz
i7 2600K Batch L039B470.
MSI P67-GD65 Bios V1.8B6
Water, EK Supreme HF Cu

[-=DVS=-]

Batch# L040B705 (crappy batch)

MicroATX Asus P8P67-M PRO
*Bios Version 0414
*Date 01/07/2011

VenomX + 1500 RPM SFlex Fan, inside Antec P180

[dengyong]

Batch# L040B705 (crappy batch)
MicroATX P8P67-M PRO
VenomX + 1500 RPM SFlex Fan, inside Antec P180

I got the same batch number and board today . I wanted 4.5ghz under 1.3v but I won't have time to test until tomorrow.

[Bullu$]

Seems like SP1 W7, gives a major boost on the Gflops

Not my pic

[miahallen]

I got the same batch number and board today . I wanted 4.5ghz under 1.3v but I won't have time to test until tomorrow.

I just got my 1st retail chip today as well, same batch. probably won't test till Thursday though.

[C-N]

Seems like SP1 W7, gives a major boost on the Gflops

Not my pic

AVX instruction set is now enabled

[Alex-Ro]

I found that it requires 0.05 more voltage than without and seems to produce far more heat too

[Jupiler]

Thanks, Dank je.

LinX 4.5GHz
i7 2600K Batch L039B470.
MSI P67-GD65 Bios V1.8B6
Water, EK Supreme HF Cu

[gurneyhck]

Seems like SP1 W7, gives a major boost on the Gflops

Not my pic

It's actually mine

[Leeghoofd]

all Results added to the file, keep them coming !!!

Sandy Bridge batches.xls

[Olivon]

Core i7 2600K Batch L042B074 1.45v set in BIOS
Asrock Extreme 6
GSkill PC12800 ECO CL7
Prolimatech Megahalems



I don't have core monitor in my SS. Ought I do it another time Leeghoofd ?

[PolRoger]

i5 2500K #L041C123
PLL: 1.70v
VccIO: 1.075v
DRAM : 1.625v
Vcore BIOS: offset +20
Cooling: Noctua NH-U12P with push/pull

Note: The chip seems stable x49 crunching WCG/Rosetta (4 threads) at offset +00... since the load demand doesn't seem to droop vcore below 1.344v.